build: bump s3/gcsfs from 2021.10.1 to 2021.11.0

Bumps [gcsfs](https://github.com/fsspec/gcsfs) from 2021.10.1 to 2021.11.0.
- [Release notes](https://github.com/fsspec/gcsfs/releases)
Bumps [s3fs](https://github.com/fsspec/s3fs) from 2021.10.1 to 2021.11.0.
- [Release notes](https://github.com/fsspec/s3fs/releases)
- [Changelog](https://github.com/fsspec/s3fs/blob/main/release-procedure.md)

---
updated-dependencies:
- dependency-name: s3fs
  dependency-type: direct:development
  update-type: version-update:semver-minor
- dependency-name: gcsfs
  dependency-type: direct:development
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <[email protected]>

.github/workflows/benchmarks.yaml

@@ -0,0 +1,23 @@
+name: benchmarks
+on: [pull_request, workflow_dispatch]
+
+jobs:
+  bench:
+    name: run benchmarks
+    runs-on: ${{ matrix.os }}
+    strategy:
+      fail-fast: false
+      matrix:
+        os: [ubuntu-18.04]
+    steps:
+      - name: Cancel Previous Runs
+        uses: styfle/[email protected]
+        with:
+          access_token: ${{ github.token }}
+      - uses: actions/setup-python@v2
+        with:
+            python-version: 3.7
+      - uses: iterative/dvc-bench@master
+        with:
+            pytest_options: "-k 'test_init or test_help'"
+

.github/workflows/packages.yaml

@@ -5,11 +5,14 @@ on:
     - master
   release:
     types: [released, prereleased]
+  workflow_dispatch:
 jobs:
   deb:
     runs-on: ubuntu-18.04
     steps:
-    - uses: actions/checkout@v2.3.4
+    - uses: actions/checkout@v2.4.0
+      with:
+          fetch-depth: 0
     - name: Set up Python 3.8
       uses: actions/[email protected]
       with:
@@ -38,7 +41,9 @@ jobs:
   rpm:
     runs-on: ubuntu-18.04
     steps:
-    - uses: actions/checkout@v2.3.4
+    - uses: actions/checkout@v2.4.0
+      with:
+        fetch-depth: 0
     - name: Set up Python 3.8
       uses: actions/[email protected]
       with:
@@ -67,7 +72,9 @@ jobs:
   osxpkg:
     runs-on: macos-10.15
     steps:
-    - uses: actions/checkout@v2.3.4
+    - uses: actions/checkout@v2.4.0
+      with:
+        fetch-depth: 0
     - name: Set up Python 3.8
       uses: actions/[email protected]
       with:
@@ -99,7 +106,9 @@ jobs:
   exe:
     runs-on: windows-2019
     steps:
-    - uses: actions/checkout@v2.3.4
+    - uses: actions/checkout@v2.4.0
+      with:
+        fetch-depth: 0
     - name: Set up Python 3.8
       uses: actions/[email protected]
       with:
@@ -127,7 +136,9 @@ jobs:
   pip:
     runs-on: ubuntu-18.04
     steps:
-    - uses: actions/checkout@v2.3.4
+    - uses: actions/checkout@v2.4.0
+      with:
+        fetch-depth: 0
     - name: Set up Python 3.7
       uses: actions/[email protected]
       with:
@@ -135,7 +146,9 @@ jobs:
     - name: Install
       run: pip install ".[all,tests]"
     - name: Build packages
-      run: ./scripts/build_package.sh
+      run: |
+        pip install -U pip
+        ./scripts/build_package.sh
     - name: Publish packages to PyPI
       if: github.event_name == 'release'
       uses: pypa/gh-action-pypi-publish@master

.github/workflows/tests.yaml

@@ -1,12 +1,8 @@
 name: Tests
 
 on:
-  pull_request:
-    branches:
-      - master
-  push:
-    branches:
-      - master
+  pull_request: {}
+  push: {}
   schedule:
     - cron: '5 1 * * *'  # every day at 01:05
 
@@ -17,9 +13,16 @@ env:
 
 jobs:
   lint:
+    timeout-minutes: 10
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/[email protected]
+    - name: Cancel Previous Runs
+      uses: styfle/[email protected]
+      with:
+        access_token: ${{ github.token }}
+    - uses: actions/[email protected]
+      with:
+        fetch-depth: 0
     - name: Set up Python 3.8
       uses: actions/[email protected]
       with:
@@ -27,7 +30,7 @@ jobs:
     - name: Install requirements
       run: |
         pip install wheel
-        pip install ".[all,tests]" Pygments collective.checkdocs pre-commit
+        pip install ".[all,tests,terraform]" pre-commit
     - name: install temporary dependencies
       run: |
         pip install git+https://github.com/isidentical/gdrivefs@service-account
@@ -36,23 +39,21 @@ jobs:
       run: python setup.py checkdocs
     - uses: pre-commit/[email protected]
   tests:
+    timeout-minutes: 45
     runs-on: ${{ matrix.os }}
     strategy:
       fail-fast: false
       matrix:
-        os: [ubuntu-18.04, windows-2019, macos-10.15]
-        pyv: ["3.6", "3.7", "3.8", "3.9"]
-        include:
-          - os: "macos-10.15"
-            pyv: "3.10.0-rc.1"
+        os: [ubuntu-20.04, windows-latest, macos-latest]
+        pyv: ["3.7", "3.8", "3.9", "3.10"]
         exclude:
-          - os: macos-10.15
-            pyv: "3.9"
-          - os: macos-10.15
-            pyv: "3.7"
-
+          # no wheels for pygit2 yet
+          - os: windows-latest
+            pyv: "3.10"
     steps:
-    - uses: actions/checkout@v2.3.4
+    - uses: actions/checkout@v2.4.0
+      with:
+        fetch-depth: 0
     - name: Set up Python
       uses: actions/[email protected]
       with:
@@ -66,18 +67,14 @@ jobs:
       uses: actions/[email protected]
       with:
         path: ${{ steps.pip-cache-dir.outputs.dir }}
-        key: ${{ runner.os }}-pip-${{ hashFiles('setup.py') }}
+        key: ${{ runner.os }}-pip-${{ hashFiles('setup.cfg') }}
         restore-keys: |
           ${{ runner.os }}-pip-
-    - name: install libgit2 on Python3.10 in macOS
-      if: matrix.pyv == '3.10.0-rc.1' && matrix.os == 'macos-10.15'
-      run: |
-        brew install libgit2
     - name: install
       if: steps.cache.pip-cache-dir.cache-hit != 'true'
       run: |
         pip install --upgrade pip setuptools wheel
-        pip install -e ".[all,tests]"
+        pip install -e ".[all,tests,terraform]"
     - name: install temporary dependencies
       run: |
         pip install git+https://github.com/isidentical/gdrivefs@service-account
@@ -101,6 +98,7 @@ jobs:
       run: |
         echo "extra_test_args=--disable-gdrive" >> $GITHUB_ENV
     - name: run tests
+      timeout-minutes: 40
       env:
         GDRIVE_CREDENTIALS_DATA: ${{ secrets.GDRIVE_CREDENTIALS_DATA }}
         AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}
@@ -109,11 +107,11 @@ jobs:
         OSS_ACCESS_KEY_SECRET: ${{ secrets.OSS_ACCESS_KEY_SECRET}}
         OSS_ENDPOINT: ${{ secrets.OSS_ENDPOINT}}
       run: >-
-        python -m tests -n=4
+        python -m tests -n=auto
         --cov-report=xml --cov-report=term
-        --all ${{ env.extra_test_args }}
+        --enable-ssh ${{ env.extra_test_args }}
     - name: upload coverage report
-      uses: codecov/codecov-action@v2.0.2
+      uses: codecov/codecov-action@v2.1.0
       with:
         file: ./coverage.xml
         fail_ci_if_error: false

.gitignore

@@ -13,6 +13,7 @@ build/
 dist/
 
 *.egg-info/
+.eggs
 
 hooks/*.pyc
 
@@ -38,3 +39,4 @@ scripts/ci/gcp-creds.json
 .gcp-creds.json
 
 *~
+/dvc/_dvc_version.py

.pre-commit-config.yaml

@@ -1,12 +1,17 @@
 ci:
-  skip: [pylint, dvc-pre-commit]
+  skip: [mypy, pylint, dvc-pre-commit]
 
 repos:
   - hooks:
       - id: black
         language_version: python3
     repo: https://github.com/ambv/black
-    rev: 21.7b0
+    rev: 21.10b0
+  - repo: https://github.com/pre-commit/pre-commit-hooks
+    rev: v4.0.1
+    hooks:
+      - id: check-toml
+      - id: check-yaml
   - hooks:
       - id: codespell
         args: 
@@ -29,19 +34,14 @@ repos:
           - flake8-string-format
     repo: https://gitlab.com/pycqa/flake8
     rev: 3.9.2
-  - hooks:
-      -  id: mypy
-         language_version: python3
-         files: ^dvc/
-         additional_dependencies:
-           - types-requests
-           - types-paramiko
-           - types-tabulate
-           - types-toml
-    repo: https://github.com/pre-commit/mirrors-mypy
-    rev: 'v0.910'
   - repo: local
     hooks:
+    - id: mypy
+      name: mypy
+      entry: mypy
+      files: ^dvc/
+      language: system
+      types: [python]
     - id: pylint
       name: pylint
       entry: pylint
@@ -54,7 +54,7 @@ repos:
         id: beautysh
         language_version: python3
     repo: https://github.com/lovesegfault/beautysh
-    rev: v6.1.0
+    rev: v6.2.1
   - hooks:
       - args:
           - git-hook

.pylintrc

@@ -1,615 +0,0 @@
-[MASTER]
-
-# A comma-separated list of package or module names from where C extensions may
-# be loaded. Extensions are loading into the active Python interpreter and may
-# run arbitrary code.
-extension-pkg-whitelist=pygit2
-
-# Add files or directories to the blacklist. They should be base names, not
-# paths.
-ignore=CVS
-
-# Add files or directories matching the regex patterns to the blacklist. The
-# regex matches against base names, not paths.
-ignore-patterns=
-
-# Python code to execute, usually for sys.path manipulation such as
-# pygtk.require().
-init-hook="import sys;
-           from pathlib import Path;
-           from pylint.config import find_pylintrc;
-
-           tests_folder = Path(find_pylintrc()).parent / 'tests'
-           sys.path.append(str(tests_folder));
-           "
-
-
-# Use multiple processes to speed up Pylint. Specifying 0 will auto-detect the
-# number of processors available to use.
-jobs=1
-
-# Control the amount of potential inferred values when inferring a single
-# object. This can help the performance when dealing with large functions or
-# complex, nested conditions.
-limit-inference-results=100
-
-# List of plugins (as comma separated values of python modules names) to load,
-# usually to register additional checkers.
-load-plugins=pylint_pytest,pylint_plugin_disable
-
-# Pickle collected data for later comparisons.
-persistent=yes
-
-# Specify a configuration file.
-#rcfile=
-
-# When enabled, pylint would attempt to guess common misconfiguration and emit
-# user-friendly hints instead of false-positive error messages.
-suggestion-mode=yes
-
-# Allow loading of arbitrary C extensions. Extensions are imported into the
-# active Python interpreter and may run arbitrary code.
-unsafe-load-any-extension=no
-
-
-[MESSAGES CONTROL]
-
-# Only show warnings with the listed confidence levels. Leave empty to show
-# all. Valid levels: HIGH, INFERENCE, INFERENCE_FAILURE, UNDEFINED.
-confidence=
-
-# Disable the message, report, category or checker with the given id(s). You
-# can either give multiple identifiers separated by comma (,) or put this
-# option multiple times (only on the command line, not in the configuration
-# file where it should appear only once). You can also use "--disable=all" to
-# disable everything first and then reenable specific checks. For example, if
-# you want to run only the similarities checker, you can use "--disable=all
-# --enable=similarities". If you want to run only the classes checker, but have
-# no Warning level messages displayed, use "--disable=all --enable=classes
-# --disable=W".
-disable=format,
-        # ^ most of them are covered by flake8,
-        # and are conventions rather than warnings or errors
-        refactoring,  # will be noisy if it starts giving suggestions
-        spelling,
-        # design is too-few-* and too-many-* checks, too-much-* for us right now
-        design,
-        # following were generated by default
-        # most of them are redundant Py3 checks for Py2+3 codebase
-        print-statement,
-        parameter-unpacking,
-        unpacking-in-except,
-        old-raise-syntax,
-        backtick,
-        long-suffix,
-        old-ne-operator,
-        old-octal-literal,
-        import-star-module-level,
-        non-ascii-bytes-literal,
-        raw-checker-failed,
-        bad-inline-option,
-        locally-disabled,
-        file-ignored,
-        suppressed-message,
-        useless-suppression,
-        deprecated-pragma,
-        use-symbolic-message-instead,
-        apply-builtin,
-        basestring-builtin,
-        buffer-builtin,
-        cmp-builtin,
-        coerce-builtin,
-        execfile-builtin,
-        file-builtin,
-        long-builtin,
-        raw_input-builtin,
-        reduce-builtin,
-        standarderror-builtin,
-        unicode-builtin,
-        xrange-builtin,
-        coerce-method,
-        delslice-method,
-        getslice-method,
-        setslice-method,
-        no-absolute-import,
-        old-division,
-        dict-iter-method,
-        dict-view-method,
-        next-method-called,
-        metaclass-assignment,
-        indexing-exception,
-        raising-string,
-        reload-builtin,
-        oct-method,
-        hex-method,
-        nonzero-method,
-        cmp-method,
-        input-builtin,
-        round-builtin,
-        intern-builtin,
-        unichr-builtin,
-        map-builtin-not-iterating,
-        zip-builtin-not-iterating,
-        range-builtin-not-iterating,
-        filter-builtin-not-iterating,
-        using-cmp-argument,
-        eq-without-hash,
-        div-method,
-        idiv-method,
-        rdiv-method,
-        exception-message-attribute,
-        invalid-str-codec,
-        sys-max-int,
-        bad-python3-import,
-        deprecated-string-function,
-        deprecated-str-translate-call,
-        deprecated-itertools-function,
-        deprecated-types-field,
-        next-method-defined,
-        dict-items-not-iterating,
-        dict-keys-not-iterating,
-        dict-values-not-iterating,
-        deprecated-operator-function,
-        deprecated-urllib-function,
-        xreadlines-attribute,
-        deprecated-sys-function,
-        exception-escape,
-        comprehension-escape,
-        # custom disables start here
-        no-self-use,
-        invalid-name,
-        # already a check enabled for `wildcard-import`
-        unused-wildcard-import,
-        misplaced-comparison-constant,
-        # probably soon: https://github.com/iterative/dvc/issues/1843
-        logging-format-interpolation,
-        logging-fstring-interpolation,
-        # we don't have proper docstrings yet
-        # even if we do, we'll be doing it gradually
-        missing-function-docstring,
-        missing-module-docstring,
-        missing-class-docstring,
-        duplicate-code,
-        # too noisy, and detects issues even when import is inside a function
-        # see: https://github.com/PyCQA/pylint/issues/850
-        cyclic-import,
-        # isort should cover these:
-        wrong-import-order,
-        wrong-import-position,
-        ungrouped-imports,
-        multiple-imports,
-        # we use it to speedup/optimize
-        import-outside-toplevel,
-        fixme,
-        # we use cause of the exception for error message output
-        # so some of the times we don't use `raise from` to not show the cause
-        # of the error
-        raise-missing-from,
-
-# Enable the message, report, category or checker with the given id(s). You can
-# either give multiple identifier separated by comma (,) or put this option
-# multiple time (only on the command line, not in the configuration file where
-# it should appear only once). See also the "--disable" option for examples.
-enable=c-extension-no-member,
-       no-else-return,
-
-
-[REPORTS]
-
-# Python expression which should return a note less than 10 (10 is the highest
-# note). You have access to the variables errors warning, statement which
-# respectively contain the number of errors / warnings messages and the total
-# number of statements analyzed. This is used by the global evaluation report
-# (RP0004).
-evaluation=10.0 - ((float(5 * error + warning + refactor + convention) / statement) * 10)
-
-# Template used to display messages. This is a python new-style format string
-# used to format the message information. See doc for all details.
-#msg-template=
-
-# Set the output format. Available formats are text, parseable, colorized, json
-# and msvs (visual studio). You can also give a reporter class, e.g.
-# mypackage.mymodule.MyReporterClass.
-output-format=text
-
-# Tells whether to display a full report or only the messages.
-reports=no
-
-# Activate the evaluation score.
-score=yes
-
-
-[REFACTORING]
-
-# Maximum number of nested blocks for function / method body
-max-nested-blocks=5
-
-# Complete name of functions that never returns. When checking for
-# inconsistent-return-statements if a never returning function is called then
-# it will be considered as an explicit return statement and no message will be
-# printed.
-never-returning-functions=sys.exit
-
-
-[LOGGING]
-
-# Format style used to check logging format string. `old` means using %
-# formatting, while `new` is for `{}` formatting.
-logging-format-style=old
-
-# Logging modules to check that the string format arguments are in logging
-# function parameter format.
-logging-modules=logging
-
-
-[SPELLING]
-
-# Limits count of emitted suggestions for spelling mistakes.
-max-spelling-suggestions=4
-
-# Spelling dictionary name. Available dictionaries: none. To make it working
-# install python-enchant package..
-spelling-dict=
-
-# List of comma separated words that should not be checked.
-spelling-ignore-words=
-
-# A path to a file that contains private dictionary; one word per line.
-spelling-private-dict-file=
-
-# Tells whether to store unknown words to indicated private dictionary in
-# --spelling-private-dict-file option instead of raising a message.
-spelling-store-unknown-words=no
-
-
-[TYPECHECK]
-
-# List of decorators that produce context managers, such as
-# contextlib.contextmanager. Add to this list to register other decorators that
-# produce valid context managers.
-contextmanager-decorators=contextlib.contextmanager
-
-# List of members which are set dynamically and missed by pylint inference
-# system, and so shouldn't trigger E1101 when accessed. Python regular
-# expressions are accepted.
-generated-members=hdfs.*,pytest.lazy_fixture,logging.TRACE,logger.trace,sys.getwindowsversion
-
-# Tells whether missing members accessed in mixin class should be ignored. A
-# mixin class is detected if its name ends with "mixin" (case insensitive).
-ignore-mixin-members=yes
-
-# Tells whether to warn about missing members when the owner of the attribute
-# is inferred to be None.
-ignore-none=yes
-
-# This flag controls whether pylint should warn about no-member and similar
-# checks whenever an opaque object is returned when inferring. The inference
-# can return multiple potential results while evaluating a Python object, but
-# some branches might not be evaluated, which results in partial inference. In
-# that case, it might be useful to still emit no-member and other checks for
-# the rest of the inferred objects.
-ignore-on-opaque-inference=yes
-
-# List of class names for which member attributes should not be checked (useful
-# for classes with dynamically set attributes). This supports the use of
-# qualified names.
-ignored-classes=optparse.Values,thread._local,_thread._local,Dvcfile
-
-# List of module names for which member attributes should not be checked
-# (useful for modules/projects where namespaces are manipulated during runtime
-# and thus existing member attributes cannot be deduced by static analysis. It
-# supports qualified module names, as well as Unix pattern matching.
-ignored-modules=azure
-
-# Show a hint with possible names when a member name was not found. The aspect
-# of finding the hint is based on edit distance.
-missing-member-hint=yes
-
-# The minimum edit distance a name should have in order to be considered a
-# similar match for a missing member name.
-missing-member-hint-distance=1
-
-# The total number of similar names that should be taken in consideration when
-# showing a hint for a missing member.
-missing-member-max-choices=1
-
-signature-mutators=funcy.decorators.decorator,
-
-[VARIABLES]
-
-# List of additional names supposed to be defined in builtins. Remember that
-# you should avoid defining new builtins when possible.
-additional-builtins=
-
-# Tells whether unused global variables should be treated as a violation.
-allow-global-unused-variables=yes
-
-# List of strings which can identify a callback function by name. A callback
-# name must start or end with one of those strings.
-callbacks=cb_,
-          _cb
-
-# A regular expression matching the name of dummy variables (i.e. expected to
-# not be used).
-dummy-variables-rgx=_+$|(_[a-zA-Z0-9_]*[a-zA-Z0-9]+?$)|dummy|^ignored_|^unused_
-
-# Argument names that match this expression will be ignored. Default to name
-# with leading underscore.
-ignored-argument-names=_.*|^ignored_|^unused_|args|kwargs
-
-# Tells whether we should check for unused import in __init__ files.
-init-import=no
-
-# List of qualified module names which can have objects that can redefine
-# builtins.
-redefining-builtins-modules=six.moves,past.builtins,future.builtins,builtins,io
-
-
-[FORMAT]
-
-# Expected format of line ending, e.g. empty (any line ending), LF or CRLF.
-expected-line-ending-format=
-
-# Regexp for a line that is allowed to be longer than the limit.
-ignore-long-lines=^\s*(# )?<?https?://\S+>?$
-
-# Number of spaces of indent required inside a hanging or continued line.
-indent-after-paren=4
-
-# String used as indentation unit. This is usually "    " (4 spaces) or "\t" (1
-# tab).
-indent-string='    '
-
-# Maximum number of characters on a single line.
-max-line-length=79
-
-# Maximum number of lines in a module.
-max-module-lines=1000
-
-# List of optional constructs for which whitespace checking is disabled. `dict-
-# separator` is used to allow tabulation in dicts, etc.: {1  : 1,\n222: 2}.
-# `trailing-comma` allows a space between comma and closing bracket: (a, ).
-# `empty-line` allows space-only lines.
-no-space-check=trailing-comma,
-               dict-separator
-
-# Allow the body of a class to be on the same line as the declaration if body
-# contains single statement.
-single-line-class-stmt=no
-
-# Allow the body of an if to be on the same line as the test if there is no
-# else.
-single-line-if-stmt=no
-
-
-[BASIC]
-
-# Naming style matching correct argument names.
-argument-naming-style=snake_case
-
-# Regular expression matching correct argument names. Overrides argument-
-# naming-style.
-#argument-rgx=
-
-# Naming style matching correct attribute names.
-attr-naming-style=snake_case
-
-# Regular expression matching correct attribute names. Overrides attr-naming-
-# style.
-#attr-rgx=
-
-# Bad variable names which should always be refused, separated by a comma.
-bad-names=foo,
-          bar,
-          baz,
-          toto,
-          tutu,
-          tata
-
-# Naming style matching correct class attribute names.
-class-attribute-naming-style=any
-
-# Regular expression matching correct class attribute names. Overrides class-
-# attribute-naming-style.
-#class-attribute-rgx=
-
-# Naming style matching correct class names.
-class-naming-style=PascalCase
-
-# Regular expression matching correct class names. Overrides class-naming-
-# style.
-#class-rgx=
-
-# Naming style matching correct constant names.
-const-naming-style=UPPER_CASE
-
-# Regular expression matching correct constant names. Overrides const-naming-
-# style.
-#const-rgx=
-
-# Minimum line length for functions/classes that require docstrings, shorter
-# ones are exempt.
-docstring-min-length=-1
-
-# Naming style matching correct function names.
-function-naming-style=snake_case
-
-# Regular expression matching correct function names. Overrides function-
-# naming-style.
-#function-rgx=
-
-# Good variable names which should always be accepted, separated by a comma.
-good-names=i,
-           j,
-           k,
-           ex,
-           Run,
-           _,
-           s3,
-           gs,
-           x, y,
-           x0, y0,
-           x1, y1,
-           dx, dy
-
-# Include a hint for the correct naming format with invalid-name.
-include-naming-hint=no
-
-# Naming style matching correct inline iteration names.
-inlinevar-naming-style=any
-
-# Regular expression matching correct inline iteration names. Overrides
-# inlinevar-naming-style.
-#inlinevar-rgx=
-
-# Naming style matching correct method names.
-method-naming-style=snake_case
-
-# Regular expression matching correct method names. Overrides method-naming-
-# style.
-#method-rgx=
-
-# Naming style matching correct module names.
-module-naming-style=snake_case
-
-# Regular expression matching correct module names. Overrides module-naming-
-# style.
-#module-rgx=
-
-# Colon-delimited sets of names that determine each other's naming style when
-# the name regexes allow several styles.
-name-group=
-
-# Regular expression which should only match function or class names that do
-# not require a docstring.
-no-docstring-rgx=^_
-
-# List of decorators that produce properties, such as abc.abstractproperty. Add
-# to this list to register other decorators that produce valid properties.
-# These decorators are taken in consideration only for invalid-name.
-property-classes=abc.abstractproperty
-
-# Naming style matching correct variable names.
-variable-naming-style=snake_case
-
-# Regular expression matching correct variable names. Overrides variable-
-# naming-style.
-#variable-rgx=
-
-
-[SIMILARITIES]
-
-# Ignore comments when computing similarities.
-ignore-comments=yes
-
-# Ignore docstrings when computing similarities.
-ignore-docstrings=yes
-
-# Ignore imports when computing similarities.
-ignore-imports=no
-
-# Minimum lines number of a similarity.
-min-similarity-lines=4
-
-
-[MISCELLANEOUS]
-
-# List of note tags to take in consideration, separated by a comma.
-notes=FIXME,
-      XXX,
-      TODO
-
-
-[DESIGN]
-
-# Maximum number of arguments for function / method.
-max-args=5
-
-# Maximum number of attributes for a class (see R0902).
-max-attributes=7
-
-# Maximum number of boolean expressions in an if statement.
-max-bool-expr=5
-
-# Maximum number of branch for function / method body.
-max-branches=12
-
-# Maximum number of locals for function / method body.
-max-locals=15
-
-# Maximum number of parents for a class (see R0901).
-max-parents=7
-
-# Maximum number of public methods for a class (see R0904).
-max-public-methods=20
-
-# Maximum number of return / yield for function / method body.
-max-returns=6
-
-# Maximum number of statements in function / method body.
-max-statements=50
-
-# Minimum number of public methods for a class (see R0903).
-min-public-methods=2
-
-
-[CLASSES]
-
-# List of method names used to declare (i.e. assign) instance attributes.
-defining-attr-methods=__init__,
-                      __new__,
-                      setUp
-
-# List of member names, which should be excluded from the protected access
-# warning.
-exclude-protected=_asdict,
-                  _fields,
-                  _replace,
-                  _source,
-                  _make
-
-# List of valid names for the first argument in a class method.
-valid-classmethod-first-arg=cls
-
-# List of valid names for the first argument in a metaclass class method.
-valid-metaclass-classmethod-first-arg=cls
-
-
-[IMPORTS]
-
-# Allow wildcard imports from modules that define __all__.
-allow-wildcard-with-all=no
-
-# Analyse import fallback blocks. This can be used to support both Python 2 and
-# 3 compatible code, which means that the block might have code that exists
-# only in one or another interpreter, leading to false positives when analysed.
-analyse-fallback-blocks=no
-
-# Deprecated modules which should not be used, separated by a comma.
-deprecated-modules=optparse,tkinter.tix
-
-# Create a graph of external dependencies in the given file (report RP0402 must
-# not be disabled).
-ext-import-graph=
-
-# Create a graph of every (i.e. internal and external) dependencies in the
-# given file (report RP0402 must not be disabled).
-import-graph=
-
-# Create a graph of internal dependencies in the given file (report RP0402 must
-# not be disabled).
-int-import-graph=
-
-# Force import order to recognize a module as part of the standard
-# compatibility libraries.
-known-standard-library=
-
-# Force import order to recognize a module as part of a third party library.
-known-third-party=enchant
-
-
-[EXCEPTIONS]
-
-# Exceptions that will emit a warning when being caught. Defaults to
-# "Exception".
-overgeneral-exceptions=Exception

LICENSE

@@ -186,7 +186,7 @@
       same "printed page" as the copyright notice for easier
       identification within third-party archives.
 
-   Copyright {yyyy} {name of copyright owner}
+   Copyright 2017-2021 Iterative, Inc.
 
    Licensed under the Apache License, Version 2.0 (the "License");
    you may not use this file except in compliance with the License.

MANIFEST.in

@@ -1,3 +1,3 @@
-include fastentrypoints.py
 include LICENSE
-include test_requirements.txt
+recursive-include requirements *.txt
+recursive-include dvc/repo/plots/templates *.json

README.rst

@@ -8,7 +8,7 @@
 • `Tutorial <https://dvc.org/doc/get-started>`_
 • `Mailing List <https://sweedom.us10.list-manage.com/subscribe/post?u=a08bf93caae4063c4e6a351f6&id=24c0ecc49a>`_
 
-|Release| |CI| |Maintainability| |Coverage| |Donate| |DOI|
+|CI| |Maintainability| |Coverage| |Donate| |DOI|
 
 |PyPI| |Packages| |Brew| |Conda| |Choco| |Snap|
 
@@ -20,8 +20,9 @@ learning projects. Key features:
 #. Simple **command line** Git-like experience. Does not require installing and maintaining
    any databases. Does not depend on any proprietary online services.
 
-#. Management and versioning of **datasets** and **machine learning models**. Data is saved in
-   S3, Google cloud, Azure, Alibaba cloud, SSH server, HDFS, or even local HDD RAID.
+#. Management and versioning of **datasets** and **machine learning
+   models**. Data can be saved in S3, Google cloud, Azure, Alibaba 
+   cloud, SSH server, HDFS, or even local HDD RAID.
 
 #. Makes projects **reproducible** and **shareable**; helping to answer questions about how
    a model was built.
@@ -65,25 +66,25 @@ Quick start
 
 Please read `Get Started <https://dvc.org/doc/get-started>`_ guide for a full version. Common workflow commands include:
 
-+-----------------------------------+-------------------------------------------------------------------+
-| Step                              | Command                                                           |
-+===================================+===================================================================+
-| Track data                        | | ``$ git add train.py``                                          |
-|                                   | | ``$ dvc add images.zip``                                        |
-+-----------------------------------+-------------------------------------------------------------------+
-| Connect code and data by commands | | ``$ dvc run -d images.zip -o images/ unzip -q images.zip``      |
-|                                   | | ``$ dvc run -d images/ -d train.py -o model.p python train.py`` |
-+-----------------------------------+-------------------------------------------------------------------+
-| Make changes and reproduce        | | ``$ vi train.py``                                               |
-|                                   | | ``$ dvc repro model.p.dvc``                                     |
-+-----------------------------------+-------------------------------------------------------------------+
-| Share code                        | | ``$ git add .``                                                 |
-|                                   | | ``$ git commit -m 'The baseline model'``                        |
-|                                   | | ``$ git push``                                                  |
-+-----------------------------------+-------------------------------------------------------------------+
-| Share data and ML models          | | ``$ dvc remote add myremote -d s3://mybucket/image_cnn``        |
-|                                   | | ``$ dvc push``                                                  |
-+-----------------------------------+-------------------------------------------------------------------+
++-----------------------------------+----------------------------------------------------------------------------+
+| Step                              | Command                                                                    |
++===================================+============================================================================+
+| Track data                        | | ``$ git add train.py``                                                   |
+|                                   | | ``$ dvc add images.zip``                                                 |
++-----------------------------------+----------------------------------------------------------------------------+
+| Connect code and data by commands | | ``$ dvc run -n prepare -d images.zip -o images/ unzip -q images.zip``    |
+|                                   | | ``$ dvc run -n train -d images/ -d train.py -o model.p python train.py`` |
++-----------------------------------+----------------------------------------------------------------------------+
+| Make changes and reproduce        | | ``$ vi train.py``                                                        |
+|                                   | | ``$ dvc repro model.p.dvc``                                              |
++-----------------------------------+----------------------------------------------------------------------------+
+| Share code                        | | ``$ git add .``                                                          |
+|                                   | | ``$ git commit -m 'The baseline model'``                                 |
+|                                   | | ``$ git push``                                                           |
++-----------------------------------+----------------------------------------------------------------------------+
+| Share data and ML models          | | ``$ dvc remote add myremote -d s3://mybucket/image_cnn``                 |
+|                                   | | ``$ dvc push``                                                           |
++-----------------------------------+----------------------------------------------------------------------------+
 
 Installation
 ============
@@ -182,30 +183,37 @@ Fedora / CentOS (rpm)
 Comparison to related technologies
 ==================================
 
-#. *Data Engineering* tools such as `AirFlow <https://airflow.apache.org/>`_, `Luigi <https://github.com/spotify/luigi>`_ and
-   others - in DVC data, model and ML pipelines represent a single ML project and focuses on data scientist' experience while
-   data engineering pipeline orchestrates multiple data projects and focuses on efficient execution. DVC project can be used
-   from data pipeline as a single execution step. `AirFlow DVC <https://github.com/covid-genomics/airflow-dvc>`_ is an example
-   of such integration.
+#. Data Engineering tools such as `AirFlow <https://airflow.apache.org/>`,
+   `Luigi <https://github.com/spotify/luigi>`, and others - in DVC data,
+   model and ML pipelines represent a single ML project focused on data
+   scientists' experience.  Data engineering tools orchestrate multiple data
+   projects and focus on efficient execution. A DVC project can be used from
+   existing data pipelines as a single execution step.
 
 #. `Git-annex <https://git-annex.branchable.com/>`_ - DVC uses the idea of storing the content of large files (which should
    not be in a Git repository) in a local key-value store, and uses file hardlinks/symlinks instead of
    copying/duplicating files.
 
-#. `Git-LFS <https://git-lfs.github.com/>`_ - DVC is compatible with any remote storage (S3, Google Cloud, Azure, SSH,
-   etc). DVC also uses reflinks or hardlinks to avoid copy operations on checkouts; thus handling large data files
-   much more efficiently.
+#. `Git-LFS <https://git-lfs.github.com/>`_ - DVC is compatible with many
+   remote storage services (S3, Google Cloud, Azure, SSH, etc). DVC also
+   uses reflinks or hardlinks to avoid copy operations on checkouts; thus
+   handling large data files much more efficiently.
 
-#. *Makefile* (and analogues including ad-hoc scripts) - DVC tracks dependencies (in a directed acyclic graph).
+#. Makefile (and analogues including ad-hoc scripts) - DVC tracks
+   dependencies (in a directed acyclic graph).
 
 #. `Workflow Management Systems <https://en.wikipedia.org/wiki/Workflow_management_system>`_ - DVC is a workflow
    management system designed specifically to manage machine learning experiments. DVC is built on top of Git.
 
-#. `DAGsHub <https://dagshub.com/>`_ - online service of Git+DVC repositories with pipeline and metrics visualization and
-   DVC-specific cloud storage.
+#. `DAGsHub <https://dagshub.com/>`_ - online service to host DVC
+   projects.  It provides a useful UI around DVC repositories and integrates
+   other tools.
 
-#. `DVC Studio <https://studio.iterative.ai/>`_ - online service of DVC repositories visualisation from DVC team. Also,
-   integrated with `CML (CI/CD for ML) <https://cml.dev/>`_ for training models in clouds and Kubernetes.
+#. `DVC Studio <https://studio.iterative.ai/>`_ - official online
+   platform for DVC projects.  It can be used to manage data and models, run
+   and track experiments, and visualize and share results.  Also, it
+   integrates with `CML (CI/CD for ML) <https://cml.dev/>` for training
+   models in the cloud or Kubernetes.
 
 
 Contributing
@@ -214,7 +222,9 @@ Contributing
 |Maintainability| |Donate|
 
 Contributions are welcome! Please see our `Contributing Guide <https://dvc.org/doc/user-guide/contributing/core>`_ for more
-details.
+details. Thanks to all our contributors!
+
+|Contribs|
 
 Mailing List
 ============
@@ -244,10 +254,6 @@ Barrak, A., Eghan, E.E. and Adams, B. `On the Co-evolution of ML Pipelines and S
    :target: https://dvc.org
    :alt: DVC logo
 
-.. |Release| image:: https://img.shields.io/badge/release-ok-brightgreen
-   :target: https://travis-ci.com/iterative/dvc/branches
-   :alt: Release
-
 .. |CI| image:: https://github.com/iterative/dvc/workflows/Tests/badge.svg?branch=master
    :target: https://github.com/iterative/dvc/actions
    :alt: GHA Tests
@@ -295,3 +301,7 @@ Barrak, A., Eghan, E.E. and Adams, B. `On the Co-evolution of ML Pipelines and S
 .. |Flowchart| image:: https://dvc.org/img/flow.gif
    :target: https://dvc.org/img/flow.gif
    :alt: how_dvc_works
+
+.. |Contribs| image:: https://contrib.rocks/image?repo=iterative/dvc
+   :target: https://github.com/iterative/dvc/graphs/contributors
+   :alt: Contributors

dvc/_debug.py

@@ -1,10 +1,39 @@
 from contextlib import ExitStack, contextmanager
-from typing import TYPE_CHECKING
+from typing import TYPE_CHECKING, Callable, Union
 
 if TYPE_CHECKING:
     from argparse import Namespace
 
 
+@contextmanager
+def yappi_profile(
+    path: Union[Callable[[], str], str] = None, wall_clock: bool = True
+):
+    try:
+        import yappi  # pylint: disable=import-error
+    except ImportError:
+        print("Failed to run profiler, yappi is not installed")
+        yield
+        return
+
+    yappi.set_clock_type("wall" if wall_clock else "cpu")
+
+    yappi.start()
+    yield
+    yappi.stop()
+
+    # pylint:disable=no-member
+    if path:
+        stats = yappi.get_func_stats()
+        fpath = path() if callable(path) else path
+        stats.save(fpath, "callgrind")
+    else:
+        yappi.get_func_stats().print_all()
+        yappi.get_thread_stats().print_all()
+
+    yappi.clear_stats()
+
+
 @contextmanager
 def instrument(html_output=False):
     """Run a statistical profiler"""
@@ -70,6 +99,13 @@ def debugtools(args: "Namespace" = None, **kwargs):
             stack.enter_context(profile(kw.get("cprofile_dump")))
         if kw.get("instrument") or kw.get("instrument_open"):
             stack.enter_context(instrument(kw.get("instrument_open", False)))
+        if kw.get("yappi"):
+            from datetime import datetime
+
+            output = "callgrind.dvc-{0:%Y%m%d}_{0:%H%M%S}.out"
+            stack.enter_context(
+                yappi_profile(path=lambda: output.format(datetime.now()))
+            )
         yield
 
 
@@ -79,6 +115,9 @@ def add_debugging_flags(parser):
     parser.add_argument(
         "--cprofile", action="store_true", default=False, help=SUPPRESS
     )
+    parser.add_argument(
+        "--yappi", action="store_true", default=False, help=SUPPRESS
+    )
     parser.add_argument("--cprofile-dump", help=SUPPRESS)
     parser.add_argument(
         "--pdb", action="store_true", default=False, help=SUPPRESS

dvc/analytics.py

@@ -68,7 +68,7 @@ def send(path):
     url = "https://analytics.dvc.org"
     headers = {"content-type": "application/json"}
 
-    with open(path) as fobj:
+    with open(path, encoding="utf-8") as fobj:
         report = json.load(fobj)
 
     report.update(_runtime_info())
@@ -173,13 +173,13 @@ def _find_or_create_user_id():
     try:
         with Lock(lockfile):
             try:
-                with open(fname) as fobj:
+                with open(fname, encoding="utf-8") as fobj:
                     user_id = json.load(fobj)["user_id"]
 
             except (FileNotFoundError, ValueError, KeyError):
                 user_id = str(uuid.uuid4())
 
-                with open(fname, "w") as fobj:
+                with open(fname, "w", encoding="utf-8") as fobj:
                     json.dump({"user_id": user_id}, fobj)
 
             return user_id

dvc/api.py

@@ -113,7 +113,7 @@ def make_checkpoint():
         root_dir, Repo.DVC_DIR, "tmp", CheckpointTask.SIGNAL_FILE
     )
 
-    with builtins.open(signal_file, "w") as fobj:
+    with builtins.open(signal_file, "w", encoding="utf-8") as fobj:
         # NOTE: force flushing/writing empty file to disk, otherwise when
         # run in certain contexts (pytest) file may not actually be written
         fobj.write("")

dvc/cli.py

@@ -30,6 +30,7 @@ from .command import (
     install,
     live,
     ls,
+    machine,
     metrics,
     move,
     params,
@@ -87,6 +88,7 @@ COMMANDS = [
     experiments,
     check_ignore,
     live,
+    machine,
 ]
 
 

dvc/command/config.py

@@ -66,7 +66,8 @@ class CmdConfig(CmdBaseNoRepo):
             )
             return 1
 
-        levels = [self.args.level] if self.args.level else self.config.LEVELS
+        levels = self._get_appropriate_levels(self.args.level)
+
         for level in levels:
             conf = self.config.read(level)
             prefix = self._config_file_prefix(
@@ -81,9 +82,7 @@ class CmdConfig(CmdBaseNoRepo):
     def _get(self, remote, section, opt):
         from dvc.config import ConfigError
 
-        levels = (
-            [self.args.level] if self.args.level else self.config.LEVELS[::-1]
-        )
+        levels = self._get_appropriate_levels(self.args.level)[::-1]
 
         for level in levels:
             conf = self.config.read(level)
@@ -136,6 +135,20 @@ class CmdConfig(CmdBaseNoRepo):
                 f"option '{opt}' doesn't exist in {name} '{section}'"
             )
 
+    def _get_appropriate_levels(self, levels):
+        if levels:
+            self._validate_level_for_non_repo_operation(levels)
+            return [levels]
+        if self.config.dvc_dir is None:
+            return self.config.SYSTEM_LEVELS
+        return self.config.LEVELS
+
+    def _validate_level_for_non_repo_operation(self, level):
+        from dvc.config import ConfigError
+
+        if self.config.dvc_dir is None and level in self.config.REPO_LEVELS:
+            raise ConfigError("Not inside a DVC repo")
+
     @staticmethod
     def _format_config(config, prefix=""):
         from dvc.utils.flatten import flatten

dvc/command/dag.py

@@ -103,9 +103,8 @@ class CmdDAG(CmdBase):
         if self.args.dot:
             ui.write(_show_dot(G))
         else:
-            from dvc.utils.pager import pager
-
-            pager(_show_ascii(G))
+            with ui.pager():
+                ui.write(_show_ascii(G))
 
         return 0
 

dvc/command/data_sync.py

@@ -393,6 +393,7 @@ def add_parser(subparsers, _parent_parser):
         help="Show status of all stages in the specified directory.",
     )
     status_parser.add_argument(
+        "--json",
         "--show-json",
         action="store_true",
         default=False,

dvc/command/diff.py

@@ -1,5 +1,4 @@
 import argparse
-import json
 import logging
 import os
 
@@ -16,7 +15,7 @@ def _digest(checksum):
     return "{}..{}".format(checksum["old"][0:8], checksum["new"][0:8])
 
 
-def _show_md(diff, show_hash=False, hide_missing=False):
+def _show_markdown(diff, show_hash=False, hide_missing=False):
     headers = ["Status", "Hash", "Path"] if show_hash else ["Status", "Path"]
     rows = []
     statuses = ["added", "deleted", "renamed", "modified"]
@@ -142,10 +141,10 @@ class CmdDiff(CmdBase):
                         del entry["hash"]
                 diff[key] = entries
 
-            if self.args.show_json:
-                ui.write(json.dumps(diff))
-            elif self.args.show_md:
-                _show_md(diff, show_hash, hide_missing)
+            if self.args.json:
+                ui.write_json(diff)
+            elif self.args.markdown:
+                _show_markdown(diff, show_hash, hide_missing)
             elif diff:
                 self._show_diff(diff, hide_missing)
 
@@ -188,6 +187,7 @@ def add_parser(subparsers, parent_parser):
         nargs="?",
     )
     diff_parser.add_argument(
+        "--json",
         "--show-json",
         help="Format the output into a JSON",
         action="store_true",
@@ -200,9 +200,11 @@ def add_parser(subparsers, parent_parser):
         default=False,
     )
     diff_parser.add_argument(
+        "--md",
         "--show-md",
         help="Show tabulated output in the Markdown format (GFM).",
         action="store_true",
+        dest="markdown",
         default=False,
     )
     diff_parser.add_argument(

dvc/command/experiments.py

@@ -5,9 +5,8 @@ from datetime import date, datetime
 from fnmatch import fnmatch
 from typing import TYPE_CHECKING, Dict, Iterable, Optional
 
-from funcy import lmap
+from funcy import compact, lmap
 
-from dvc import prompt
 from dvc.command import completion
 from dvc.command.base import CmdBase, append_doc_link, fix_subparsers
 from dvc.command.metrics import DEFAULT_PRECISION
@@ -19,9 +18,8 @@ from dvc.utils.flatten import flatten
 from dvc.utils.serialize import encode_exception
 
 if TYPE_CHECKING:
-    from rich.text import Text
-
     from dvc.compare import TabularData
+    from dvc.ui import RichText
 
 
 logger = logging.getLogger(__name__)
@@ -141,6 +139,8 @@ def _collect_rows(
     precision=DEFAULT_PRECISION,
     sort_by=None,
     sort_order=None,
+    fill_value=FILL_VALUE,
+    iso=False,
 ):
     from dvc.scm.git import Git
 
@@ -161,8 +161,8 @@ def _collect_rows(
         elif exp.get("queued"):
             state = "Queued"
         else:
-            state = FILL_VALUE
-        executor = exp.get("executor", FILL_VALUE)
+            state = fill_value
+        executor = exp.get("executor", fill_value)
         is_baseline = rev == "baseline"
 
         if is_baseline:
@@ -202,12 +202,12 @@ def _collect_rows(
             exp_name,
             name_rev,
             typ,
-            _format_time(exp.get("timestamp")),
+            _format_time(exp.get("timestamp"), fill_value, iso),
             parent,
             state,
             executor,
         ]
-        fill_value = FILL_VALUE_ERRORED if results.get("error") else FILL_VALUE
+        fill_value = FILL_VALUE_ERRORED if results.get("error") else fill_value
         _extend_row(
             row,
             metric_names,
@@ -258,11 +258,12 @@ def _sort_column(sort_by, metric_names, param_names):
 def _sort_exp(experiments, sort_path, sort_name, typ, reverse):
     def _sort(item):
         rev, exp = item
-        tip = exp.get("checkpoint_tip")
+        exp_data = exp.get("data", {})
+        tip = exp_data.get("checkpoint_tip")
         if tip and tip != rev:
             # Sort checkpoint experiments by tip commit
             return _sort((tip, experiments[tip]))
-        data = exp.get(typ, {}).get(sort_path, {})
+        data = exp_data.get(typ, {}).get(sort_path, {}).get("data", {})
         val = flatten(data).get(sort_name)
         return val is None, val
 
@@ -274,19 +275,21 @@ def _sort_exp(experiments, sort_path, sort_name, typ, reverse):
     return ret
 
 
-def _format_time(timestamp):
-    if timestamp is None:
-        return FILL_VALUE
-    if timestamp.date() == date.today():
+def _format_time(datetime_obj, fill_value=FILL_VALUE, iso=False):
+    if datetime_obj is None:
+        return fill_value
+
+    if iso:
+        return datetime_obj.isoformat()
+
+    if datetime_obj.date() == date.today():
         fmt = "%I:%M %p"
     else:
         fmt = "%b %d, %Y"
-    return timestamp.strftime(fmt)
+    return datetime_obj.strftime(fmt)
 
 
 def _extend_row(row, names, items, precision, fill_value=FILL_VALUE):
-    from rich.text import Text
-
     from dvc.compare import _format_field, with_value
 
     if not items:
@@ -301,10 +304,10 @@ def _extend_row(row, names, items, precision, fill_value=FILL_VALUE):
                 item.get(name),
                 FILL_VALUE_ERRORED if data.get("error", None) else fill_value,
             )
-            # wrap field data in rich.Text, otherwise rich may
+            # wrap field data in ui.rich_text, otherwise rich may
             # interpret unescaped braces from list/dict types as rich
             # markup tags
-            row.append(Text(str(_format_field(value, precision))))
+            row.append(ui.rich_text(str(_format_field(value, precision))))
 
 
 def _parse_filter_list(param_list):
@@ -320,6 +323,7 @@ def _parse_filter_list(param_list):
 
 def experiments_table(
     all_experiments,
+    headers,
     metric_headers,
     metric_names,
     param_headers,
@@ -327,22 +331,15 @@ def experiments_table(
     sort_by=None,
     sort_order=None,
     precision=DEFAULT_PRECISION,
+    fill_value=FILL_VALUE,
+    iso=False,
 ) -> "TabularData":
     from funcy import lconcat
 
     from dvc.compare import TabularData
 
-    headers = [
-        "Experiment",
-        "rev",
-        "typ",
-        "Created",
-        "parent",
-        "State",
-        "Executor",
-    ]
     td = TabularData(
-        lconcat(headers, metric_headers, param_headers), fill_value=FILL_VALUE
+        lconcat(headers, metric_headers, param_headers), fill_value=fill_value
     )
     for base_rev, experiments in all_experiments.items():
         rows = _collect_rows(
@@ -353,23 +350,23 @@ def experiments_table(
             sort_by=sort_by,
             sort_order=sort_order,
             precision=precision,
+            fill_value=fill_value,
+            iso=iso,
         )
         td.extend(rows)
 
     return td
 
 
-def prepare_exp_id(kwargs) -> "Text":
-    from rich.text import Text
-
+def prepare_exp_id(kwargs) -> "RichText":
     exp_name = kwargs["Experiment"]
     rev = kwargs["rev"]
     typ = kwargs.get("typ", "baseline")
 
     if typ == "baseline" or not exp_name:
-        text = Text(exp_name or rev)
+        text = ui.rich_text(exp_name or rev)
     else:
-        text = Text.assemble(rev, " [", (exp_name, "bold"), "]")
+        text = ui.rich_text.assemble(rev, " [", (exp_name, "bold"), "]")
 
     parent = kwargs.get("parent")
     suff = f" ({parent})" if parent else ""
@@ -377,7 +374,7 @@ def prepare_exp_id(kwargs) -> "Text":
 
     tree = experiment_types[typ]
     pref = f"{tree} " if tree else ""
-    return Text(pref) + text
+    return ui.rich_text(pref) + text
 
 
 def baseline_styler(typ):
@@ -385,8 +382,15 @@ def baseline_styler(typ):
 
 
 def show_experiments(
-    all_experiments, pager=True, no_timestamp=False, **kwargs
+    all_experiments,
+    pager=True,
+    no_timestamp=False,
+    csv=False,
+    markdown=False,
+    **kwargs,
 ):
+    from funcy.seqs import flatten as flatten_list
+
     include_metrics = _parse_filter_list(kwargs.pop("include_metrics", []))
     exclude_metrics = _parse_filter_list(kwargs.pop("exclude_metrics", []))
     include_params = _parse_filter_list(kwargs.pop("include_params", []))
@@ -399,11 +403,26 @@ def show_experiments(
         include_params=include_params,
         exclude_params=exclude_params,
     )
-    metric_headers = _normalize_headers(metric_names)
-    param_headers = _normalize_headers(param_names)
+
+    headers = [
+        "Experiment",
+        "rev",
+        "typ",
+        "Created",
+        "parent",
+        "State",
+        "Executor",
+    ]
+
+    names = {**metric_names, **param_names}
+    counter = Counter(flatten_list([list(a.keys()) for a in names.values()]))
+    counter.update(headers)
+    metric_headers = _normalize_headers(metric_names, counter)
+    param_headers = _normalize_headers(param_names, counter)
 
     td = experiments_table(
         all_experiments,
+        headers,
         metric_headers,
         metric_names,
         param_headers,
@@ -411,6 +430,8 @@ def show_experiments(
         kwargs.get("sort_by"),
         kwargs.get("sort_order"),
         kwargs.get("precision"),
+        kwargs.get("fill_value"),
+        kwargs.get("iso"),
     )
 
     if no_timestamp:
@@ -422,9 +443,12 @@ def show_experiments(
 
     row_styles = lmap(baseline_styler, td.column("typ"))
 
-    merge_headers = ["Experiment", "rev", "typ", "parent"]
-    td.column("Experiment")[:] = map(prepare_exp_id, td.as_dict(merge_headers))
-    td.drop(*merge_headers[1:])
+    if not csv:
+        merge_headers = ["Experiment", "rev", "typ", "parent"]
+        td.column("Experiment")[:] = map(
+            prepare_exp_id, td.as_dict(merge_headers)
+        )
+        td.drop(*merge_headers[1:])
 
     headers = {"metrics": metric_headers, "params": param_headers}
     styles = {
@@ -453,13 +477,12 @@ def show_experiments(
         rich_table=True,
         header_styles=styles,
         row_styles=row_styles,
+        csv=csv,
+        markdown=markdown,
     )
 
 
-def _normalize_headers(names):
-    count = Counter(
-        name for path in names for name in names[path] for path in names
-    )
+def _normalize_headers(names, count):
     return [
         name if count[name] == 1 else f"{path}:{name}"
         for path in names
@@ -488,11 +511,17 @@ class CmdExperimentsShow(CmdBase):
             logger.exception("failed to show experiments")
             return 1
 
-        if self.args.show_json:
-            import json
-
-            ui.write(json.dumps(all_experiments, default=_format_json))
+        if self.args.json:
+            ui.write_json(all_experiments, default=_format_json)
         else:
+            precision = (
+                self.args.precision or None
+                if self.args.csv
+                else DEFAULT_PRECISION
+            )
+            fill_value = "" if self.args.csv else FILL_VALUE
+            iso = True if self.args.csv else False
+
             show_experiments(
                 all_experiments,
                 include_metrics=self.args.include_metrics,
@@ -502,8 +531,12 @@ class CmdExperimentsShow(CmdBase):
                 no_timestamp=self.args.no_timestamp,
                 sort_by=self.args.sort_by,
                 sort_order=self.args.sort_order,
-                precision=self.args.precision or DEFAULT_PRECISION,
+                precision=precision,
+                fill_value=fill_value,
+                iso=iso,
                 pager=not self.args.no_pager,
+                csv=self.args.csv,
+                markdown=self.args.markdown,
             )
         return 0
 
@@ -532,10 +565,8 @@ class CmdExperimentsDiff(CmdBase):
             logger.exception("failed to show experiments diff")
             return 1
 
-        if self.args.show_json:
-            import json
-
-            ui.write(json.dumps(diff))
+        if self.args.json:
+            ui.write_json(diff)
         else:
             from dvc.compare import show_diff
 
@@ -550,11 +581,13 @@ class CmdExperimentsDiff(CmdBase):
                 show_diff(
                     diff[key],
                     title=title,
-                    markdown=self.args.show_md,
+                    markdown=self.args.markdown,
                     no_path=self.args.no_path,
                     old=self.args.old,
                     on_empty_diff="diff not supported",
                     precision=precision if key == "metrics" else None,
+                    a_rev=self.args.a_rev,
+                    b_rev=self.args.b_rev,
                 )
 
         return 0
@@ -635,7 +668,7 @@ class CmdExperimentsGC(CmdRepro):
         logger.warning(msg)
 
         msg = "Are you sure you want to proceed?"
-        if not self.args.force and not prompt.confirm(msg):
+        if not self.args.force and not ui.confirm(msg):
             return 1
 
         removed = self.repo.experiments.gc(
@@ -750,11 +783,101 @@ class CmdExperimentsRemove(CmdBase):
             exp_names=self.args.experiment,
             queue=self.args.queue,
             clear_all=self.args.all,
+            remote=self.args.git_remote,
         )
 
         return 0
 
 
+class CmdExperimentsInit(CmdBase):
+    CODE = "src"
+    DATA = "data"
+    MODELS = "models"
+    DEFAULT_METRICS = "metrics.json"
+    DEFAULT_PARAMS = "params.yaml"
+    PLOTS = "plots"
+    DVCLIVE = "dvclive"
+    DEFAULTS = {
+        "code": CODE,
+        "data": DATA,
+        "models": MODELS,
+        "metrics": DEFAULT_METRICS,
+        "params": DEFAULT_PARAMS,
+        "plots": PLOTS,
+        "live": DVCLIVE,
+    }
+    EXP_LINK = (
+        "https://dvc.org/doc"
+        "/user-guide/experiment-management/running-experiments"
+    )
+
+    def run(self):
+        from dvc.command.stage import parse_cmd
+
+        cmd = parse_cmd(self.args.cmd)
+        if not self.args.interactive and not cmd:
+            raise InvalidArgumentError("command is not specified")
+
+        from dvc.repo.experiments.init import init
+
+        defaults = {}
+        if not self.args.explicit:
+            config = self.repo.config["exp"]
+            defaults.update({**self.DEFAULTS, **config})
+
+        cli_args = compact(
+            {
+                "cmd": cmd,
+                "code": self.args.code,
+                "data": self.args.data,
+                "models": self.args.models,
+                "metrics": self.args.metrics,
+                "params": self.args.params,
+                "plots": self.args.plots,
+                "live": self.args.live,
+            }
+        )
+
+        initialized_stage = init(
+            self.repo,
+            name=self.args.name,
+            type=self.args.type,
+            defaults=defaults,
+            overrides=cli_args,
+            interactive=self.args.interactive,
+            force=self.args.force,
+        )
+
+        name = self.args.name or self.args.type
+
+        text = ui.rich_text.assemble(
+            "\n" if self.args.interactive else "",
+            "Created ",
+            (name, "bright_blue"),
+            " stage in ",
+            ("dvc.yaml", "green"),
+            ".",
+        )
+        if not self.args.run:
+            text.append_text(
+                ui.rich_text.assemble(
+                    " To run, use ",
+                    ('"dvc exp run"', "green"),
+                    ".\nSee ",
+                    (self.EXP_LINK, "repr.url"),
+                    ".",
+                )
+            )
+
+        ui.write(text, styled=True)
+        if self.args.run:
+            return self.repo.experiments.run(
+                targets=[initialized_stage.addressing]
+            )
+
+        return 0
+
+
 def add_parser(subparsers, parent_parser):
     EXPERIMENTS_HELP = "Commands to run and compare experiments."
 
@@ -860,7 +983,10 @@ def add_parser(subparsers, parent_parser):
     )
     experiments_show_parser.add_argument(
         "--sort-order",
-        help="Sort order to use with --sort-by.",
+        help=(
+            "Sort order to use with --sort-by."
+            " Defaults to ascending ('asc')."
+        ),
         choices=("asc", "desc"),
         default="asc",
     )
@@ -877,11 +1003,27 @@ def add_parser(subparsers, parent_parser):
         help="Always show git commit SHAs instead of branch/tag names.",
     )
     experiments_show_parser.add_argument(
+        "--json",
         "--show-json",
         action="store_true",
         default=False,
         help="Print output in JSON format instead of a human-readable table.",
     )
+    experiments_show_parser.add_argument(
+        "--csv",
+        "--show-csv",
+        action="store_true",
+        default=False,
+        help="Print output in csv format instead of a human-readable table.",
+    )
+    experiments_show_parser.add_argument(
+        "--md",
+        "--show-md",
+        action="store_true",
+        default=False,
+        dest="markdown",
+        help="Show tabulated output in the Markdown format (GFM).",
+    )
     experiments_show_parser.add_argument(
         "--precision",
         type=int,
@@ -945,22 +1087,25 @@ def add_parser(subparsers, parent_parser):
         help="Show only params that are stage dependencies.",
     )
     experiments_diff_parser.add_argument(
+        "--json",
         "--show-json",
         action="store_true",
         default=False,
         help="Show output in JSON format.",
     )
     experiments_diff_parser.add_argument(
+        "--md",
         "--show-md",
         action="store_true",
         default=False,
+        dest="markdown",
         help="Show tabulated output in the Markdown format (GFM).",
     )
     experiments_diff_parser.add_argument(
         "--old",
         action="store_true",
         default=False,
-        help="Show old metric/param value.",
+        help="Show old ('a_rev') metric/param value.",
     )
     experiments_diff_parser.add_argument(
         "--no-path",
@@ -1111,9 +1256,9 @@ def add_parser(subparsers, parent_parser):
         nargs="?",
         default=None,
         help=(
-            "Optional Git remote name or Git URL. If provided, experiments "
-            "from the specified Git repository will be listed instead of "
-            "local experiments."
+            "Optional Git remote name or Git URL. "
+            "If provided, experiments from the specified Git repository "
+            " will be listed instead of local ones."
         ),
         metavar="[<git_remote>]",
     )
@@ -1231,7 +1376,7 @@ def add_parser(subparsers, parent_parser):
     )
     experiments_pull_parser.set_defaults(func=CmdExperimentsPull)
 
-    EXPERIMENTS_REMOVE_HELP = "Remove local experiments."
+    EXPERIMENTS_REMOVE_HELP = "Remove experiments."
     experiments_remove_parser = experiments_subparsers.add_parser(
         "remove",
         parents=[parent_parser],
@@ -1249,6 +1394,12 @@ def add_parser(subparsers, parent_parser):
         action="store_true",
         help="Remove all committed experiments.",
     )
+    remove_group.add_argument(
+        "-g",
+        "--git-remote",
+        metavar="<git_remote>",
+        help="Name or URL of the Git remote to remove the experiment from",
+    )
     experiments_remove_parser.add_argument(
         "experiment",
         nargs="*",
@@ -1257,6 +1408,91 @@ def add_parser(subparsers, parent_parser):
     )
     experiments_remove_parser.set_defaults(func=CmdExperimentsRemove)
 
+    EXPERIMENTS_INIT_HELP = "Initialize experiments."
+    experiments_init_parser = experiments_subparsers.add_parser(
+        "init",
+        parents=[parent_parser],
+        description=append_doc_link(EXPERIMENTS_INIT_HELP, "exp/init"),
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    experiments_init_parser.add_argument(
+        "cmd",
+        nargs=argparse.REMAINDER,
+        help="Command to execute.",
+        metavar="command",
+    )
+    experiments_init_parser.add_argument(
+        "--run",
+        action="store_true",
+        help="Run the experiment after initializing it",
+    )
+    experiments_init_parser.add_argument(
+        "--interactive",
+        "-i",
+        action="store_true",
+        help="Prompt for values that are not provided",
+    )
+    experiments_init_parser.add_argument(
+        "-f",
+        "--force",
+        action="store_true",
+        default=False,
+        help="Overwrite existing stage",
+    )
+    experiments_init_parser.add_argument(
+        "--explicit",
+        action="store_true",
+        default=False,
+        help="Only use the path values explicitly provided",
+    )
+    experiments_init_parser.add_argument(
+        "--name", "-n", help="Name of the stage to create"
+    )
+    experiments_init_parser.add_argument(
+        "--code",
+        help="Path to the source file or directory "
+        "which your experiments depend"
+        f" (default: {CmdExperimentsInit.CODE})",
+    )
+    experiments_init_parser.add_argument(
+        "--data",
+        help="Path to the data file or directory "
+        "which your experiments depend"
+        f" (default: {CmdExperimentsInit.DATA})",
+    )
+    experiments_init_parser.add_argument(
+        "--models",
+        help="Path to the model file or directory for your experiments"
+        f" (default: {CmdExperimentsInit.MODELS})",
+    )
+    experiments_init_parser.add_argument(
+        "--params",
+        help="Path to the parameters file for your experiments"
+        f" (default: {CmdExperimentsInit.DEFAULT_PARAMS})",
+    )
+    experiments_init_parser.add_argument(
+        "--metrics",
+        help="Path to the metrics file for your experiments"
+        f" (default: {CmdExperimentsInit.DEFAULT_METRICS})",
+    )
+    experiments_init_parser.add_argument(
+        "--plots",
+        help="Path to the plots file or directory for your experiments"
+        f" (default: {CmdExperimentsInit.PLOTS})",
+    )
+    experiments_init_parser.add_argument(
+        "--live",
+        help="Path to log dvclive outputs for your experiments"
+        f" (default: {CmdExperimentsInit.DVCLIVE})",
+    )
+    experiments_init_parser.add_argument(
+        "--type",
+        choices=["default", "live"],
+        default="default",
+        help="Select type of stage to create (default: %(default)s)",
+    )
+    experiments_init_parser.set_defaults(func=CmdExperimentsInit)
+
 
 def _add_run_common(parser):
     """Add common args for 'exp run' and 'exp resume'."""

dvc/command/live.py

@@ -1,5 +1,5 @@
 import argparse
-import os
+from pathlib import Path
 
 from dvc.command import completion
 from dvc.command.base import CmdBase, fix_subparsers
@@ -13,10 +13,11 @@ class CmdLive(CmdBase):
         metrics, plots = self.repo.live.show(target=target, revs=revs)
 
         if plots:
-            html_path = self.args.target + ".html"
-            self.repo.plots.write_html(html_path, plots, metrics)
+            html_path = Path.cwd() / (self.args.target + "_html")
+            from dvc.render.utils import render
 
-            ui.write("\nfile://", os.path.abspath(html_path), sep="")
+            index_path = render(self.repo, plots, metrics, html_path)
+            ui.write(index_path.as_uri())
             return 0
         return 1
 

dvc/command/ls/__init__.py

@@ -34,10 +34,8 @@ class CmdList(CmdBaseNoRepo):
                 recursive=self.args.recursive,
                 dvc_only=self.args.dvc_only,
             )
-            if self.args.show_json:
-                import json
-
-                ui.write(json.dumps(entries))
+            if self.args.json:
+                ui.write_json(entries)
             elif entries:
                 entries = _prettify(entries, with_color=True)
                 ui.write("\n".join(entries))
@@ -54,6 +52,7 @@ def add_parser(subparsers, parent_parser):
     )
     list_parser = subparsers.add_parser(
         "list",
+        aliases=["ls"],
         parents=[parent_parser],
         description=append_doc_link(LIST_HELP, "list"),
         help=LIST_HELP,
@@ -70,7 +69,10 @@ def add_parser(subparsers, parent_parser):
         "--dvc-only", action="store_true", help="Show only DVC outputs."
     )
     list_parser.add_argument(
-        "--show-json", action="store_true", help="Show output in JSON format."
+        "--json",
+        "--show-json",
+        action="store_true",
+        help="Show output in JSON format.",
     )
     list_parser.add_argument(
         "--rev",

dvc/command/machine.py

@@ -0,0 +1,430 @@
+import argparse
+
+from dvc.command.base import CmdBase, append_doc_link, fix_subparsers
+from dvc.command.config import CmdConfig
+from dvc.config import ConfigError
+from dvc.exceptions import DvcException
+from dvc.ui import ui
+from dvc.utils import format_link
+
+
+class MachineDisabledError(ConfigError):
+    def __init__(self):
+        super().__init__("Machine feature is disabled")
+
+
+class CmdMachineConfig(CmdConfig):
+    def __init__(self, args):
+
+        super().__init__(args)
+        if not self.config["feature"].get("machine", False):
+            raise MachineDisabledError
+
+        if getattr(self.args, "name", None):
+            self.args.name = self.args.name.lower()
+
+    def _check_exists(self, conf):
+        if self.args.name not in conf["machine"]:
+            raise ConfigError(f"machine '{self.args.name}' doesn't exist.")
+
+
+class CmdMachineAdd(CmdMachineConfig):
+    def run(self):
+        from dvc.machine import validate_name
+
+        validate_name(self.args.name)
+
+        if self.args.default:
+            ui.write(f"Setting '{self.args.name}' as a default machine.")
+
+        with self.config.edit(self.args.level) as conf:
+            if self.args.name in conf["machine"] and not self.args.force:
+                raise ConfigError(
+                    "machine '{}' already exists. Use `-f|--force` to "
+                    "overwrite it.".format(self.args.name)
+                )
+
+            conf["machine"][self.args.name] = {"cloud": self.args.cloud}
+            if self.args.default:
+                conf["core"]["machine"] = self.args.name
+
+        return 0
+
+
+class CmdMachineRemove(CmdMachineConfig):
+    def run(self):
+        with self.config.edit(self.args.level) as conf:
+            self._check_exists(conf)
+            del conf["machine"][self.args.name]
+
+        up_to_level = self.args.level or "repo"
+        # Remove core.machine refs to this machine in any shadowing configs
+        for level in reversed(self.config.LEVELS):
+            with self.config.edit(level) as conf:
+                if conf["core"].get("machine") == self.args.name:
+                    del conf["core"]["machine"]
+
+            if level == up_to_level:
+                break
+
+        return 0
+
+
+class CmdMachineList(CmdMachineConfig):
+    def run(self):
+        levels = [self.args.level] if self.args.level else self.config.LEVELS
+        for level in levels:
+            conf = self.config.read(level)["machine"]
+            if self.args.name:
+                conf = conf.get(self.args.name, {})
+            prefix = self._config_file_prefix(
+                self.args.show_origin, self.config, level
+            )
+            configs = list(self._format_config(conf, prefix))
+            if configs:
+                ui.write("\n".join(configs))
+        return 0
+
+
+class CmdMachineModify(CmdMachineConfig):
+    def run(self):
+        from dvc.config import merge
+
+        with self.config.edit(self.args.level) as conf:
+            merged = self.config.load_config_to_level(self.args.level)
+            merge(merged, conf)
+            self._check_exists(merged)
+
+            if self.args.name not in conf["machine"]:
+                conf["machine"][self.args.name] = {}
+            section = conf["machine"][self.args.name]
+            if self.args.unset:
+                section.pop(self.args.option, None)
+            else:
+                section[self.args.option] = self.args.value
+        return 0
+
+
+class CmdMachineRename(CmdBase):
+    def _check_exists(self, conf):
+        if self.args.name not in conf["machine"]:
+            raise ConfigError(f"machine '{self.args.name}' doesn't exist.")
+
+    def _rename_default(self, conf):
+        if conf["core"].get("machine") == self.args.name:
+            conf["core"]["machine"] = self.args.new
+
+    def _check_before_rename(self):
+        from dvc.machine import validate_name
+
+        validate_name(self.args.new)
+
+        all_config = self.config.load_config_to_level(None)
+        if self.args.new in all_config.get("machine", {}):
+            raise ConfigError(
+                "Rename failed. Machine '{}' already exists.".format(
+                    self.args.new
+                )
+            )
+        ui.write(f"Rename machine '{self.args.name}' to '{self.args.new}'.")
+
+    def run(self):
+
+        self._check_before_rename()
+
+        with self.config.edit(self.args.level) as conf:
+            self._check_exists(conf)
+            conf["machine"][self.args.new] = conf["machine"][self.args.name]
+            try:
+                self.repo.machine.rename(self.args.name, self.args.new)
+            except DvcException as error:
+                del conf["machine"][self.args.new]
+                raise ConfigError("terraform rename failed") from error
+            del conf["machine"][self.args.name]
+            self._rename_default(conf)
+
+        up_to_level = self.args.level or "repo"
+        for level in reversed(self.config.LEVELS):
+            if level == up_to_level:
+                break
+            with self.config.edit(level) as level_conf:
+                self._rename_default(level_conf)
+
+        return 0
+
+
+class CmdMachineDefault(CmdMachineConfig):
+    def run(self):
+        if self.args.name is None and not self.args.unset:
+            conf = self.config.read(self.args.level)
+            try:
+                print(conf["core"]["machine"])
+            except KeyError:
+                ui.write("No default machine set")
+                return 1
+        else:
+            with self.config.edit(self.args.level) as conf:
+                if self.args.unset:
+                    conf["core"].pop("machine", None)
+                else:
+                    merged_conf = self.config.load_config_to_level(
+                        self.args.level
+                    )
+                    if (
+                        self.args.name in conf["machine"]
+                        or self.args.name in merged_conf["machine"]
+                    ):
+                        conf["core"]["machine"] = self.args.name
+                    else:
+                        raise ConfigError(
+                            "default machine must be present in machine "
+                            "list."
+                        )
+        return 0
+
+
+class CmdMachineCreate(CmdBase):
+    def run(self):
+        if self.repo.machine is None:
+            raise MachineDisabledError
+
+        self.repo.machine.create(self.args.name)
+        return 0
+
+
+class CmdMachineStatus(CmdBase):
+    SHOWN_FIELD = [
+        "name",
+        "cloud",
+        "instance_ip",
+        "instance_type",
+        "instance_hdd_size",
+        "instance_gpu",
+    ]
+
+    def run(self):
+        if self.repo.machine is None:
+            raise MachineDisabledError
+
+        all_status = self.repo.machine.status(self.args.name)
+        for i, status in enumerate(all_status, start=1):
+            ui.write(f"instance_num_{i}:")
+            for field in self.SHOWN_FIELD:
+                value = status.get(field, None)
+                ui.write(f"\t{field:20}: {value}")
+        return 0
+
+
+class CmdMachineDestroy(CmdBase):
+    def run(self):
+        if self.repo.machine is None:
+            raise MachineDisabledError
+
+        self.repo.machine.destroy(self.args.name)
+        return 0
+
+
+class CmdMachineSsh(CmdBase):
+    def run(self):
+        if self.repo.machine is None:
+            raise MachineDisabledError
+
+        self.repo.machine.run_shell(self.args.name)
+        return 0
+
+
+def add_parser(subparsers, parent_parser):
+    from dvc.command.config import parent_config_parser
+
+    machine_HELP = "Set up and manage cloud machines."
+    machine_parser = subparsers.add_parser(
+        "machine",
+        parents=[parent_parser],
+        description=append_doc_link(machine_HELP, "machine"),
+        # NOTE: suppress help during development to hide command
+        # help=machine_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+
+    machine_subparsers = machine_parser.add_subparsers(
+        dest="cmd",
+        help="Use `dvc machine CMD --help` for " "command-specific help.",
+    )
+
+    fix_subparsers(machine_subparsers)
+
+    machine_ADD_HELP = "Add a new data machine."
+    machine_add_parser = machine_subparsers.add_parser(
+        "add",
+        parents=[parent_config_parser, parent_parser],
+        description=append_doc_link(machine_ADD_HELP, "machine/add"),
+        help=machine_ADD_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    machine_add_parser.add_argument("name", help="Name of the machine")
+    machine_add_parser.add_argument(
+        "cloud",
+        help="Machine cloud. See full list of supported clouds at {}".format(
+            format_link(
+                "https://github.com/iterative/"
+                "terraform-provider-iterative#machine"
+            )
+        ),
+    )
+    machine_add_parser.add_argument(
+        "-d",
+        "--default",
+        action="store_true",
+        default=False,
+        help="Set as default machine.",
+    )
+    machine_add_parser.add_argument(
+        "-f",
+        "--force",
+        action="store_true",
+        default=False,
+        help="Force overwriting existing configs",
+    )
+    machine_add_parser.set_defaults(func=CmdMachineAdd)
+
+    machine_DEFAULT_HELP = "Set/unset the default machine."
+    machine_default_parser = machine_subparsers.add_parser(
+        "default",
+        parents=[parent_config_parser, parent_parser],
+        description=append_doc_link(machine_DEFAULT_HELP, "machine/default"),
+        help=machine_DEFAULT_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    machine_default_parser.add_argument(
+        "name", nargs="?", help="Name of the machine"
+    )
+    machine_default_parser.add_argument(
+        "-u",
+        "--unset",
+        action="store_true",
+        default=False,
+        help="Unset default machine.",
+    )
+    machine_default_parser.set_defaults(func=CmdMachineDefault)
+
+    machine_LIST_HELP = "List the configuration of one/all machines."
+    machine_list_parser = machine_subparsers.add_parser(
+        "list",
+        parents=[parent_config_parser, parent_parser],
+        description=append_doc_link(machine_LIST_HELP, "machine/list"),
+        help=machine_LIST_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    machine_list_parser.add_argument(
+        "--show-origin",
+        default=False,
+        action="store_true",
+        help="Show the source file containing each config value.",
+    )
+    machine_list_parser.add_argument(
+        "name",
+        nargs="?",
+        type=str,
+        help="name of machine to specify",
+    )
+    machine_list_parser.set_defaults(func=CmdMachineList)
+    machine_MODIFY_HELP = "Modify the configuration of an machine."
+    machine_modify_parser = machine_subparsers.add_parser(
+        "modify",
+        parents=[parent_config_parser, parent_parser],
+        description=append_doc_link(machine_MODIFY_HELP, "machine/modify"),
+        help=machine_MODIFY_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    machine_modify_parser.add_argument("name", help="Name of the machine")
+    machine_modify_parser.add_argument(
+        "option", help="Name of the option to modify."
+    )
+    machine_modify_parser.add_argument(
+        "value", nargs="?", help="(optional) Value of the option."
+    )
+    machine_modify_parser.add_argument(
+        "-u",
+        "--unset",
+        default=False,
+        action="store_true",
+        help="Unset option.",
+    )
+    machine_modify_parser.set_defaults(func=CmdMachineModify)
+
+    machine_RENAME_HELP = "Rename a machine "
+    machine_rename_parser = machine_subparsers.add_parser(
+        "rename",
+        parents=[parent_config_parser, parent_parser],
+        description=append_doc_link(machine_RENAME_HELP, "remote/rename"),
+        help=machine_RENAME_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    machine_rename_parser.add_argument("name", help="Machine to be renamed")
+    machine_rename_parser.add_argument("new", help="New name of the machine")
+    machine_rename_parser.set_defaults(func=CmdMachineRename)
+
+    machine_REMOVE_HELP = "Remove an machine."
+    machine_remove_parser = machine_subparsers.add_parser(
+        "remove",
+        parents=[parent_config_parser, parent_parser],
+        description=append_doc_link(machine_REMOVE_HELP, "machine/remove"),
+        help=machine_REMOVE_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    machine_remove_parser.add_argument(
+        "name", help="Name of the machine to remove."
+    )
+    machine_remove_parser.set_defaults(func=CmdMachineRemove)
+
+    machine_CREATE_HELP = "Create and start a machine instance."
+    machine_create_parser = machine_subparsers.add_parser(
+        "create",
+        parents=[parent_config_parser, parent_parser],
+        description=append_doc_link(machine_CREATE_HELP, "machine/create"),
+        help=machine_CREATE_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    machine_create_parser.add_argument(
+        "name", help="Name of the machine to create."
+    )
+    machine_create_parser.set_defaults(func=CmdMachineCreate)
+
+    machine_STATUS_HELP = "List the status of a running machine."
+    machine_status_parser = machine_subparsers.add_parser(
+        "status",
+        parents=[parent_config_parser, parent_parser],
+        description=append_doc_link(machine_STATUS_HELP, "machine/status"),
+        help=machine_STATUS_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    machine_status_parser.add_argument(
+        "name", help="Name of the running machine."
+    )
+    machine_status_parser.set_defaults(func=CmdMachineStatus)
+
+    machine_DESTROY_HELP = "Destroy an machine instance."
+    machine_destroy_parser = machine_subparsers.add_parser(
+        "destroy",
+        parents=[parent_config_parser, parent_parser],
+        description=append_doc_link(machine_DESTROY_HELP, "machine/destroy"),
+        help=machine_DESTROY_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    machine_destroy_parser.add_argument(
+        "name", help="Name of the machine instance to destroy."
+    )
+    machine_destroy_parser.set_defaults(func=CmdMachineDestroy)
+
+    machine_SSH_HELP = "Connect to a machine via SSH."
+    machine_ssh_parser = machine_subparsers.add_parser(
+        "ssh",
+        parents=[parent_config_parser, parent_parser],
+        description=append_doc_link(machine_SSH_HELP, "machine/ssh"),
+        help=machine_SSH_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    machine_ssh_parser.add_argument(
+        "name", help="Name of the machine instance to connect to."
+    )
+    machine_ssh_parser.set_defaults(func=CmdMachineSsh)

dvc/command/metrics.py

@@ -31,16 +31,14 @@ class CmdMetricsShow(CmdMetricsBase):
             logger.exception("")
             return 1
 
-        if self.args.show_json:
-            import json
-
-            ui.write(json.dumps(metrics, default=encode_exception))
+        if self.args.json:
+            ui.write_json(metrics, default=encode_exception)
         else:
             from dvc.compare import show_metrics
 
             show_metrics(
                 metrics,
-                markdown=self.args.show_md,
+                markdown=self.args.markdown,
                 all_branches=self.args.all_branches,
                 all_tags=self.args.all_tags,
                 all_commits=self.args.all_commits,
@@ -65,20 +63,20 @@ class CmdMetricsDiff(CmdMetricsBase):
             logger.exception("failed to show metrics diff")
             return 1
 
-        if self.args.show_json:
-            import json
-
-            ui.write(json.dumps(diff))
+        if self.args.json:
+            ui.write_json(diff)
         else:
             from dvc.compare import show_diff
 
             show_diff(
                 diff,
                 title="Metric",
-                markdown=self.args.show_md,
+                markdown=self.args.markdown,
                 no_path=self.args.no_path,
                 precision=self.args.precision or DEFAULT_PRECISION,
                 round_digits=True,
+                a_rev=self.args.a_rev,
+                b_rev=self.args.b_rev,
             )
 
         return 0
@@ -140,15 +138,18 @@ def add_parser(subparsers, parent_parser):
         help="Show metrics for all commits.",
     )
     metrics_show_parser.add_argument(
+        "--json",
         "--show-json",
         action="store_true",
         default=False,
         help="Show output in JSON format.",
     )
     metrics_show_parser.add_argument(
+        "--md",
         "--show-md",
         action="store_true",
         default=False,
+        dest="markdown",
         help="Show tabulated output in the Markdown format (GFM).",
     )
     metrics_show_parser.add_argument(
@@ -220,15 +221,18 @@ def add_parser(subparsers, parent_parser):
         help="Show unchanged metrics as well.",
     )
     metrics_diff_parser.add_argument(
+        "--json",
         "--show-json",
         action="store_true",
         default=False,
         help="Show output in JSON format.",
     )
     metrics_diff_parser.add_argument(
+        "--md",
         "--show-md",
         action="store_true",
         default=False,
+        dest="markdown",
         help="Show tabulated output in the Markdown format (GFM).",
     )
     metrics_diff_parser.add_argument(

dvc/command/params.py

@@ -25,19 +25,19 @@ class CmdParamsDiff(CmdBase):
             logger.exception("failed to show params diff")
             return 1
 
-        if self.args.show_json:
-            import json
-
-            ui.write(json.dumps(diff))
+        if self.args.json:
+            ui.write_json(diff)
         else:
             from dvc.compare import show_diff
 
             show_diff(
                 diff,
                 title="Param",
-                markdown=self.args.show_md,
+                markdown=self.args.markdown,
                 no_path=self.args.no_path,
                 show_changes=False,
+                a_rev=self.args.a_rev,
+                b_rev=self.args.b_rev,
             )
 
         return 0
@@ -103,15 +103,18 @@ def add_parser(subparsers, parent_parser):
         help="Show only params that are stage dependencies.",
     )
     params_diff_parser.add_argument(
+        "--json",
         "--show-json",
         action="store_true",
         default=False,
         help="Show output in JSON format.",
     )
     params_diff_parser.add_argument(
+        "--md",
         "--show-md",
         action="store_true",
         default=False,
+        dest="markdown",
         help="Show tabulated output in the Markdown format (GFM).",
     )
     params_diff_parser.add_argument(

dvc/command/plots.py

@@ -4,6 +4,7 @@ import logging
 from dvc.command import completion
 from dvc.command.base import CmdBase, append_doc_link, fix_subparsers
 from dvc.exceptions import DvcException
+from dvc.render.utils import find_vega, render
 from dvc.ui import ui
 from dvc.utils import format_link
 
@@ -35,43 +36,54 @@ class CmdPlots(CmdBase):
                 return 1
 
         try:
-            plots = self._func(targets=self.args.targets, props=self._props())
+            plots_data = self._func(
+                targets=self.args.targets, props=self._props()
+            )
+
+            if not plots_data:
+                ui.error_write(
+                    "No plots were loaded, "
+                    "visualization file will not be created."
+                )
 
             if self.args.show_vega:
                 target = self.args.targets[0]
-                ui.write(plots[target])
+                plot_json = find_vega(self.repo, plots_data, target)
+                if plot_json:
+                    ui.write(plot_json)
                 return 0
 
-        except DvcException:
-            logger.exception("")
-            return 1
-
-        if plots:
-            rel: str = self.args.out or "plots.html"
+            rel: str = self.args.out or "dvc_plots"
             path: Path = (Path.cwd() / rel).resolve()
-            self.repo.plots.write_html(
-                path, plots=plots, html_template_path=self.args.html_template
+            index_path = render(
+                self.repo,
+                plots_data,
+                path=path,
+                html_template_path=self.args.html_template,
             )
 
-            assert (
-                path.is_absolute()
-            )  # as_uri throws ValueError if not absolute
-            url = path.as_uri()
+            assert index_path.is_absolute()
+            url = index_path.as_uri()
             ui.write(url)
+
             if self.args.open:
                 import webbrowser
+                from platform import uname
 
-                opened = webbrowser.open(rel)
+                if "Microsoft" in uname().release:
+                    url = Path(rel) / "index.html"
+
+                opened = webbrowser.open(url)
                 if not opened:
                     ui.error_write(
                         "Failed to open. Please try opening it manually."
                     )
                     return 1
-        else:
-            ui.error_write(
-                "No plots were loaded, visualization file will not be created."
-            )
-        return 0
+            return 0
+
+        except DvcException:
+            logger.exception("")
+            return 1
 
 
 class CmdPlotsShow(CmdPlots):
@@ -175,7 +187,10 @@ def add_parser(subparsers, parent_parser):
     _add_output_arguments(plots_diff_parser)
     plots_diff_parser.set_defaults(func=CmdPlotsDiff)
 
-    PLOTS_MODIFY_HELP = "Modify display properties of plot metrics files."
+    PLOTS_MODIFY_HELP = (
+        "Modify display properties of data-series plots "
+        "(has no effect on image-type plots)."
+    )
     plots_modify_parser = plots_subparsers.add_parser(
         "modify",
         parents=[parent_parser],

dvc/command/remote.py

@@ -86,7 +86,7 @@ class CmdRemoteDefault(CmdRemote):
         if self.args.name is None and not self.args.unset:
             conf = self.config.read(self.args.level)
             try:
-                print(conf["core"]["remote"])
+                ui.write(conf["core"]["remote"])
             except KeyError:
                 ui.write("No default remote set")
                 return 1

dvc/command/status.py

@@ -69,10 +69,8 @@ class CmdDataStatus(CmdDataBase):
             if self.args.quiet:
                 return bool(st)
 
-            if self.args.show_json:
-                import json
-
-                ui.write(json.dumps(st))
+            if self.args.json:
+                ui.write_json(st)
                 return 0
 
             if st:

dvc/compare.py

@@ -12,6 +12,7 @@ from typing import (
     Mapping,
     MutableSequence,
     Sequence,
+    Set,
     Tuple,
     Union,
     overload,
@@ -168,7 +169,10 @@ class TabularData(MutableSequence[Sequence["CellT"]]):
     def render(self, **kwargs: Any):
         from dvc.ui import ui
 
-        ui.table(self, headers=self.keys(), **kwargs)
+        if kwargs.pop("csv", False):
+            ui.write(self.to_csv(), end="")
+        else:
+            ui.table(self, headers=self.keys(), **kwargs)
 
     def as_dict(
         self, cols: Iterable[str] = None
@@ -178,6 +182,35 @@ class TabularData(MutableSequence[Sequence["CellT"]]):
             {k: self._columns[k][i] for k in keys} for i in range(len(self))
         ]
 
+    def dropna(self, axis: str = "rows"):
+        if axis not in ["rows", "cols"]:
+            raise ValueError(
+                f"Invalid 'axis' value {axis}."
+                "Choose one of ['rows', 'cols']"
+            )
+
+        to_drop: Set = set()
+        for n_row, row in enumerate(self):
+            for n_col, col in enumerate(row):
+                if col == self._fill_value:
+                    if axis == "rows":
+                        to_drop.add(n_row)
+                        break
+                    else:
+                        to_drop.add(self.keys()[n_col])
+
+        if axis == "rows":
+            for name in self.keys():
+                self._columns[name] = Column(
+                    [
+                        x
+                        for n, x in enumerate(self._columns[name])
+                        if n not in to_drop
+                    ]
+                )
+        else:
+            self.drop(*to_drop)
+
 
 def _normalize_float(val: float, precision: int):
     return f"{val:.{precision}g}"
@@ -208,8 +241,12 @@ def diff_table(
     precision: int = None,
     round_digits: bool = False,
     on_empty_diff: str = None,
+    a_rev: str = None,
+    b_rev: str = None,
 ) -> TabularData:
-    headers: List[str] = ["Path", title, "Old", "New", "Change"]
+    a_rev = a_rev or "HEAD"
+    b_rev = b_rev or "workspace"
+    headers: List[str] = ["Path", title, a_rev, b_rev, "Change"]
     fill_value = "-"
     td = TabularData(headers, fill_value=fill_value)
 
@@ -237,8 +274,8 @@ def diff_table(
         td.drop("Change")
 
     if not old:
-        td.drop("Old")
-        td.rename("New", "Value")
+        td.drop(a_rev)
+        td.rename(b_rev, "Value")
 
     return td
 
@@ -253,6 +290,8 @@ def show_diff(
     round_digits: bool = False,
     on_empty_diff: str = None,
     markdown: bool = False,
+    a_rev: str = None,
+    b_rev: str = None,
 ) -> None:
     td = diff_table(
         diff,
@@ -263,6 +302,8 @@ def show_diff(
         precision=precision,
         round_digits=round_digits,
         on_empty_diff=on_empty_diff,
+        a_rev=a_rev,
+        b_rev=b_rev,
     )
     td.render(markdown=markdown)
 

dvc/config.py

@@ -32,6 +32,18 @@ class RemoteNotFoundError(RemoteConfigError):
     pass
 
 
+class MachineConfigError(ConfigError):
+    pass
+
+
+class NoMachineError(MachineConfigError):
+    pass
+
+
+class MachineNotFoundError(MachineConfigError):
+    pass
+
+
 @memoize
 def get_compiled_schema():
     from voluptuous import Schema
@@ -64,8 +76,10 @@ class Config(dict):
     APPNAME = "dvc"
     APPAUTHOR = "iterative"
 
+    SYSTEM_LEVELS = ("system", "global")
+    REPO_LEVELS = ("repo", "local")
     # In the order they shadow each other
-    LEVELS = ("system", "global", "repo", "local")
+    LEVELS = SYSTEM_LEVELS + REPO_LEVELS
 
     CONFIG = "config"
     CONFIG_LOCAL = "config.local"
@@ -127,7 +141,7 @@ class Config(dict):
             dvc.config.Config: config object.
         """
         config_file = os.path.join(dvc_dir, Config.CONFIG)
-        open(config_file, "w+").close()
+        open(config_file, "w+", encoding="utf-8").close()
         return Config(dvc_dir)
 
     def load(self, validate=True, config=None):
@@ -167,7 +181,7 @@ class Config(dict):
                 conf_obj = ConfigObj(fobj)
         else:
             conf_obj = ConfigObj()
-        return _parse_remotes(_lower_keys(conf_obj.dict()))
+        return _parse_named(_lower_keys(conf_obj.dict()))
 
     def _save_config(self, level, conf_dict):
         from configobj import ConfigObj
@@ -179,7 +193,7 @@ class Config(dict):
 
         fs.makedirs(os.path.dirname(filename))
 
-        config = ConfigObj(_pack_remotes(conf_dict))
+        config = ConfigObj(_pack_named(conf_dict))
         with fs.open(filename, "wb") as fobj:
             config.write(fobj)
         config.filename = filename
@@ -251,6 +265,7 @@ class Config(dict):
                     "key_path": func,
                 }
             },
+            "machine": {str: {"startup_script": func}},
         }
         return Schema(dirs_schema, extra=ALLOW_EXTRA)(conf)
 
@@ -273,7 +288,7 @@ class Config(dict):
     def edit(self, level=None, validate=True):
         # NOTE: we write to repo config by default, same as git config
         level = level or "repo"
-        if level in {"repo", "local"} and self.dvc_dir is None:
+        if self.dvc_dir is None and level in self.REPO_LEVELS:
             raise ConfigError("Not inside a DVC repo")
 
         conf = self.load_one(level)
@@ -300,27 +315,29 @@ class Config(dict):
             raise ConfigError(str(exc)) from None
 
 
-def _parse_remotes(conf):
-    result = {"remote": {}}
+def _parse_named(conf):
+    result = {"remote": {}, "machine": {}}
 
     for section, val in conf.items():
-        name = re_find(r'^\s*remote\s*"(.*)"\s*$', section)
-        if name:
-            result["remote"][name] = val
+        match = re_find(r'^\s*(remote|machine)\s*"(.*)"\s*$', section)
+        if match:
+            key, name = match
+            result[key][name] = val
         else:
             result[section] = val
 
     return result
 
 
-def _pack_remotes(conf):
+def _pack_named(conf):
     # Drop empty sections
     result = compact(conf)
 
     # Transform remote.name -> 'remote "name"'
-    for name, val in conf["remote"].items():
-        result[f'remote "{name}"'] = val
-    result.pop("remote", None)
+    for key in ("remote", "machine"):
+        for name, val in conf[key].items():
+            result[f'{key} "{name}"'] = val
+        result.pop(key, None)
 
     return result
 

dvc/config_schema.py

@@ -118,6 +118,7 @@ SCHEMA = {
         Optional("autostage", default=False): Bool,
         Optional("experiments"): Bool,  # obsoleted
         Optional("check_update", default=True): Bool,
+        "machine": Lower,
     },
     "cache": {
         "local": str,
@@ -154,6 +155,9 @@ SCHEMA = {
                     "grant_read_acp": str,
                     "grant_write_acp": str,
                     "grant_full_control": str,
+                    "cache_regions": bool,
+                    "read_timeout": Coerce(int),
+                    "connect_timeout": Coerce(int),
                     **REMOTE_COMMON,
                 },
                 "gs": {
@@ -175,10 +179,12 @@ SCHEMA = {
                 },
                 "hdfs": {"user": str, "kerb_ticket": str, **REMOTE_COMMON},
                 "webhdfs": {
-                    "hdfscli_config": str,
-                    "webhdfs_token": str,
-                    "user": str,
-                    "webhdfs_alias": str,
+                    "kerberos": Bool,
+                    "kerberos_principal": str,
+                    "proxy_to": str,
+                    "ssl_verify": Any(Bool, str),
+                    "token": str,
+                    "use_https": Bool,
                     **REMOTE_COMMON,
                 },
                 "azure": {
@@ -221,13 +227,44 @@ SCHEMA = {
         )
     },
     "state": {
+        "dir": str,
         "row_limit": All(Coerce(int), Range(1)),  # obsoleted
         "row_cleanup_quota": All(Coerce(int), Range(0, 100)),  # obsoleted
     },
+    "index": {
+        "dir": str,
+    },
+    "machine": {
+        str: {
+            "cloud": All(Lower, Choices("aws", "azure")),
+            "region": All(
+                Lower, Choices("us-west", "us-east", "eu-west", "eu-north")
+            ),
+            "image": str,
+            "name": str,
+            "spot": Bool,
+            "spot_price": Coerce(float),
+            "instance_hdd_size": Coerce(int),
+            "instance_type": Lower,
+            "instance_gpu": Lower,
+            "ssh_private": str,
+            "startup_script": str,
+        },
+    },
     # section for experimental features
     "feature": {
+        Optional("machine", default=False): Bool,
         # enabled by default. It's of no use, kept for backward compatibility.
-        Optional("parametrization", default=True): Bool
+        Optional("parametrization", default=True): Bool,
     },
     "plots": {"html_template": str},
+    "exp": {
+        "code": str,
+        "data": str,
+        "models": str,
+        "metrics": str,
+        "params": str,
+        "plots": str,
+        "live": str,
+    },
 }

dvc/daemon.py

@@ -12,29 +12,33 @@ from dvc.utils import fix_env, is_binary
 
 logger = logging.getLogger(__name__)
 
-CREATE_NEW_PROCESS_GROUP = 0x00000200
-DETACHED_PROCESS = 0x00000008
-
 
 def _popen(cmd, **kwargs):
     prefix = [sys.executable]
     if not is_binary():
-        prefix += [sys.argv[0]]
-
+        main_entrypoint = os.path.join(
+            os.path.abspath(os.path.dirname(__file__)), "__main__.py"
+        )
+        prefix += [main_entrypoint]
     return Popen(prefix + cmd, close_fds=True, shell=False, **kwargs)
 
 
 def _spawn_windows(cmd, env):
-    from subprocess import STARTF_USESHOWWINDOW, STARTUPINFO
+    from subprocess import (
+        CREATE_NEW_PROCESS_GROUP,
+        CREATE_NO_WINDOW,
+        STARTF_USESHOWWINDOW,
+        STARTUPINFO,
+    )
 
-    creationflags = CREATE_NEW_PROCESS_GROUP | DETACHED_PROCESS
+    # https://stackoverflow.com/a/7006424
+    # https://bugs.python.org/issue41619
+    creationflags = CREATE_NEW_PROCESS_GROUP | CREATE_NO_WINDOW
 
     startupinfo = STARTUPINFO()
     startupinfo.dwFlags |= STARTF_USESHOWWINDOW
 
-    _popen(
-        cmd, env=env, creationflags=creationflags, startupinfo=startupinfo
-    ).communicate()
+    _popen(cmd, env=env, creationflags=creationflags, startupinfo=startupinfo)
 
 
 def _spawn_posix(cmd, env):
@@ -44,17 +48,17 @@ def _spawn_posix(cmd, env):
     # with PyInstaller has trouble with SystemExit exception and throws
     # errors such as "[26338] Failed to execute script __main__"
     try:
-        pid = os.fork()
+        pid = os.fork()  # pylint: disable=no-member
         if pid > 0:
             return
     except OSError:
         logger.exception("failed at first fork")
         os._exit(1)  # pylint: disable=protected-access
 
-    os.setsid()
+    os.setsid()  # pylint: disable=no-member
 
     try:
-        pid = os.fork()
+        pid = os.fork()  # pylint: disable=no-member
         if pid > 0:
             os._exit(0)  # pylint: disable=protected-access
     except OSError:

dvc/data_cloud.py

@@ -59,7 +59,7 @@ class DataCloud:
         from dvc.objects.db import get_odb
 
         cls, config, path_info = get_cloud_fs(self.repo, name=name)
-        config["tmp_dir"] = self.repo.tmp_dir
+        config["tmp_dir"] = self.repo.index_db_dir
         return get_odb(cls(**config), path_info, **config)
 
     def push(

dvc/dependency/param.py

@@ -1,3 +1,4 @@
+import logging
 import os
 from collections import defaultdict
 
@@ -10,6 +11,8 @@ from dvc.utils.serialize import LOADERS, ParseError
 
 from .base import Dependency
 
+logger = logging.getLogger(__name__)
+
 
 class MissingParamsError(DvcException):
     pass
@@ -133,3 +136,20 @@ class ParamsDependency(Dependency):
             )
 
         return HashInfo(self.PARAM_PARAMS, info)
+
+    def save(self):
+        if not self.exists:
+            raise self.DoesNotExistError(self)
+
+        if not self.isfile and not self.isdir:
+            raise self.IsNotFileOrDirError(self)
+
+        if self.is_empty:
+            logger.warning(f"'{self}' is empty.")
+
+        self.ignore()
+
+        if self.metric or self.plot:
+            self.verify_metric()
+
+        self.hash_info = self.get_hash()

dvc/dependency/repo.py

@@ -63,7 +63,7 @@ class RepoDependency(Dependency):
         return {self.PARAM_PATH: self.def_path, self.PARAM_REPO: self.def_repo}
 
     def download(self, to, jobs=None):
-        from dvc.checkout import checkout
+        from dvc.objects.checkout import checkout
 
         for odb, objs in self.get_used_objs().items():
             self.repo.cloud.pull(objs, jobs=jobs, odb=odb)
@@ -132,7 +132,7 @@ class RepoDependency(Dependency):
                     pass
 
             try:
-                staging, staged_obj = stage(
+                staging, _, staged_obj = stage(
                     local_odb,
                     path_info,
                     repo.repo_fs,
@@ -148,9 +148,7 @@ class RepoDependency(Dependency):
             self._staged_objs[rev] = staged_obj
             used_obj_ids[staging].add(staged_obj.hash_info)
             if isinstance(staged_obj, Tree):
-                used_obj_ids[staging].update(
-                    entry.hash_info for _, entry in staged_obj
-                )
+                used_obj_ids[staging].update(oid for _, _, oid in staged_obj)
             return used_obj_ids, staged_obj
 
     def _check_circular_import(self, odb, obj_ids):
@@ -166,7 +164,7 @@ class RepoDependency(Dependency):
             for hash_info in obj_ids:
                 if hash_info.isdir:
                     tree = Tree.load(odb, hash_info)
-                    yield from (odb.get(entry.hash_info) for _, entry in tree)
+                    yield from (odb.get(oid) for _, _, oid in tree)
                 else:
                     yield odb.get(hash_info)
 

dvc/dvcfile.py

@@ -1,10 +1,7 @@
-import collections
 import contextlib
 import logging
 import os
-from typing import TYPE_CHECKING, Any, Union
-
-from voluptuous import MultipleInvalid
+from typing import TYPE_CHECKING, Any, Callable, Tuple, TypeVar, Union
 
 from dvc.exceptions import DvcException
 from dvc.parsing.versions import LOCKFILE_VERSION, SCHEMA_KWD
@@ -12,23 +9,18 @@ from dvc.stage import serialize
 from dvc.stage.exceptions import (
     StageFileBadNameError,
     StageFileDoesNotExistError,
-    StageFileFormatError,
     StageFileIsNotDvcFileError,
 )
 from dvc.types import AnyPath
 from dvc.utils import relpath
 from dvc.utils.collections import apply_diff
-from dvc.utils.serialize import (
-    dump_yaml,
-    modify_yaml,
-    parse_yaml,
-    parse_yaml_for_update,
-)
+from dvc.utils.serialize import dump_yaml, modify_yaml
 
 if TYPE_CHECKING:
     from dvc.repo import Repo
 
 logger = logging.getLogger(__name__)
+_T = TypeVar("_T")
 
 DVC_FILE = "Dvcfile"
 DVC_FILE_SUFFIX = ".dvc"
@@ -45,10 +37,6 @@ class FileIsGitIgnored(DvcException):
         )
 
 
-class LockfileCorruptedError(DvcException):
-    pass
-
-
 class ParametrizedDumpError(DvcException):
     pass
 
@@ -96,7 +84,7 @@ def check_dvcfile_path(repo, path):
 
 
 class FileMixin:
-    SCHEMA = None
+    SCHEMA: Callable[[_T], _T]
 
     def __init__(self, repo, path, verify=True, **kwargs):
         self.repo = repo
@@ -138,7 +126,11 @@ class FileMixin:
         if self._is_git_ignored():
             raise FileIsGitIgnored(self.path)
 
-    def _load(self):
+    def load(self, **kwargs: Any) -> Any:
+        d, _ = self._load(**kwargs)
+        return d
+
+    def _load(self, **kwargs: Any) -> Tuple[Any, str]:
         # it raises the proper exceptions by priority:
         # 1. when the file doesn't exists
         # 2. filename is not a DVC file
@@ -155,18 +147,23 @@ class FileMixin:
             raise StageFileIsNotDvcFileError(self.path)
 
         self._check_gitignored()
-        with self.repo.fs.open(self.path, encoding="utf-8") as fd:
-            stage_text = fd.read()
-        d = parse_yaml(stage_text, self.path)
-        return self.validate(d, self.relpath), stage_text
+        return self._load_yaml(**kwargs)
 
     @classmethod
-    def validate(cls, d, fname=None):
-        assert isinstance(cls.SCHEMA, collections.abc.Callable)
-        try:
-            return cls.SCHEMA(d)  # pylint: disable=not-callable
-        except MultipleInvalid as exc:
-            raise StageFileFormatError(f"'{fname}' format error: {exc}")
+    def validate(cls, d: _T, fname: str = None) -> _T:
+        from dvc.utils.strictyaml import validate
+
+        return validate(d, cls.SCHEMA, path=fname)  # type: ignore[arg-type]
+
+    def _load_yaml(self, **kwargs: Any) -> Tuple[Any, str]:
+        from dvc.utils import strictyaml
+
+        return strictyaml.load(
+            self.path,
+            self.SCHEMA,  # type: ignore[arg-type]
+            self.repo.fs,
+            **kwargs,
+        )
 
     def remove(self, force=False):  # pylint: disable=unused-argument
         with contextlib.suppress(FileNotFoundError):
@@ -298,10 +295,7 @@ class PipelineFile(FileMixin):
         if not self.exists():
             return
 
-        with open(self.path, encoding="utf-8") as f:
-            d = parse_yaml_for_update(f.read(), self.path)
-
-        self.validate(d, self.path)
+        d, _ = self._load_yaml(round_trip=True)
         if stage.name not in d.get("stages", {}):
             return
 
@@ -343,34 +337,25 @@ def migrate_lock_v1_to_v2(d, version_info):
     d["stages"] = stages
 
 
+def lockfile_schema(data: _T) -> _T:
+    schema = get_lockfile_schema(data)
+    return schema(data)
+
+
 class Lockfile(FileMixin):
-    @classmethod
-    def validate(cls, d, fname=None):
-        schema = get_lockfile_schema(d)
-        try:
-            return schema(d)
-        except MultipleInvalid as exc:
-            raise StageFileFormatError(f"'{fname}' format error: {exc}")
+    SCHEMA = staticmethod(lockfile_schema)  # type: ignore[assignment]
 
     def _verify_filename(self):
         pass  # lockfile path is hardcoded, so no need to verify here
 
-    def _load(self):
+    def _load(self, **kwargs: Any):
         try:
-            return super()._load()
+            return super()._load(**kwargs)
         except StageFileDoesNotExistError:
             # we still need to account for git-ignored dvc.lock file
             # even though it may not exist or have been .dvcignored
             self._check_gitignored()
             return {}, ""
-        except StageFileFormatError as exc:
-            raise LockfileCorruptedError(
-                f"Lockfile '{self.relpath}' is corrupted."
-            ) from exc
-
-    def load(self):
-        d, _ = self._load()
-        return d
 
     @property
     def latest_version_info(self):
@@ -409,10 +394,7 @@ class Lockfile(FileMixin):
         if not self.exists():
             return
 
-        with open(self.path, encoding="utf-8") as f:
-            d = parse_yaml_for_update(f.read(), self.path)
-        self.validate(d, self.path)
-
+        d, _ = self._load_yaml(round_trip=True)
         version = LOCKFILE_VERSION.from_dict(d)
         data = d if version == LOCKFILE_VERSION.V1 else d.get("stages", {})
         if stage.name not in data:

dvc/exceptions.py

@@ -364,3 +364,8 @@ class CircularImportError(DvcException):
             f"'{dep}' contains invalid circular import. "
             f"DVC repo '{a}' already imports from '{b}'."
         )
+
+
+class PrettyDvcException(DvcException):
+    def __pretty_exc__(self, **kwargs):
+        """Print prettier exception message."""

dvc/fs/__init__.py

@@ -2,6 +2,7 @@ import posixpath
 from urllib.parse import urlparse
 
 from ..scheme import Schemes
+from . import utils  # noqa: F401
 from .azure import AzureFileSystem
 from .gdrive import GDriveFileSystem
 from .gs import GSFileSystem

dvc/fs/azure.py

@@ -11,7 +11,7 @@ from dvc.path_info import CloudURLInfo
 from dvc.scheme import Schemes
 from dvc.utils import format_link
 
-from .fsspec_wrapper import ObjectFSWrapper
+from .fsspec_wrapper import CallbackMixin, ObjectFSWrapper
 
 logger = logging.getLogger(__name__)
 _DEFAULT_CREDS_STEPS = (
@@ -39,7 +39,7 @@ def _az_config():
 
 
 # pylint:disable=abstract-method
-class AzureFileSystem(ObjectFSWrapper):
+class AzureFileSystem(CallbackMixin, ObjectFSWrapper):
     scheme = Schemes.AZURE
     PATH_CLS = CloudURLInfo
     PARAM_CHECKSUM = "etag"

dvc/fs/base.py

@@ -1,15 +1,19 @@
+import contextlib
 import logging
+import os
 from concurrent.futures import ThreadPoolExecutor, as_completed
-from functools import partial, partialmethod
+from functools import partialmethod
 from multiprocessing import cpu_count
 from typing import Any, ClassVar, Dict, FrozenSet, Optional
 
+from tqdm.utils import CallbackIOWrapper
+
 from dvc.exceptions import DvcException
 from dvc.path_info import URLInfo
-from dvc.progress import Tqdm
+from dvc.progress import DEFAULT_CALLBACK, FsspecCallback
+from dvc.ui import ui
 from dvc.utils import tmp_fname
 from dvc.utils.fs import makedirs, move
-from dvc.utils.http import open_url
 
 logger = logging.getLogger(__name__)
 
@@ -125,13 +129,10 @@ class BaseFileSystem:
             f"dependencies: {missing}. {hint}"
         )
 
-    def open(self, path_info, mode: str = "r", encoding: str = None, **kwargs):
-        if hasattr(self, "_generate_download_url"):
-            # pylint:disable=no-member
-            func = self._generate_download_url  # type: ignore[attr-defined]
-            get_url = partial(func, path_info)
-            return open_url(get_url, mode=mode, encoding=encoding, **kwargs)
+    def checksum(self, path_info) -> str:
+        raise NotImplementedError
 
+    def open(self, path_info, mode: str = "r", encoding: str = None, **kwargs):
         raise RemoteActionNotImplemented("open", self.scheme)
 
     def exists(self, path_info) -> bool:
@@ -174,12 +175,6 @@ class BaseFileSystem:
         """
         raise NotImplementedError
 
-    def ls(self, path_info, detail=False):
-        raise RemoteActionNotImplemented("ls", self.scheme)
-
-    def find(self, path_info, detail=False, prefix=None):
-        raise RemoteActionNotImplemented("find", self.scheme)
-
     def is_empty(self, path_info):
         return False
 
@@ -221,43 +216,58 @@ class BaseFileSystem:
             return False
         return hash_.endswith(cls.CHECKSUM_DIR_SUFFIX)
 
-    def upload(self, from_info, to_info, name=None, no_progress_bar=False):
-        if not hasattr(self, "_upload"):
-            raise RemoteActionNotImplemented("upload", self.scheme)
+    def upload(
+        self,
+        from_info,
+        to_info,
+        total=None,
+        desc=None,
+        callback=None,
+        no_progress_bar=False,
+        **pbar_args,
+    ):
+        is_file_obj = hasattr(from_info, "read")
+        method = "upload_fobj" if is_file_obj else "put_file"
+        if not hasattr(self, method):
+            raise RemoteActionNotImplemented(method, self.scheme)
 
         if to_info.scheme != self.scheme:
             raise NotImplementedError
 
-        if from_info.scheme != "local":
-            raise NotImplementedError
+        if not is_file_obj:
+            desc = desc or from_info.name
 
-        logger.debug("Uploading '%s' to '%s'", from_info, to_info)
+        stack = contextlib.ExitStack()
+        if not callback:
+            pbar = ui.progress(
+                desc=desc,
+                disable=no_progress_bar,
+                bytes=True,
+                total=total or -1,
+                **pbar_args,
+            )
+            stack.enter_context(pbar)
+            callback = pbar.as_callback()
+            if total:
+                callback.set_size(total)
 
-        name = name or from_info.name
+        with stack:
+            if is_file_obj:
+                wrapped = CallbackIOWrapper(
+                    callback.relative_update, from_info
+                )
+                # `size` is used to provide hints to the WebdavFileSystem
+                # for legacy servers.
+                # pylint: disable=no-member
+                return self.upload_fobj(wrapped, to_info, size=total)
 
-        self._upload(  # noqa, pylint: disable=no-member
-            from_info.fspath,
-            to_info,
-            name=name,
-            no_progress_bar=no_progress_bar,
-        )
+            if from_info.scheme != "local":
+                raise NotImplementedError
 
-    def upload_fobj(
-        self, fobj, to_info, no_progress_bar=False, size=None, **pbar_args
-    ):
-        if not hasattr(self, "_upload_fobj"):
-            raise RemoteActionNotImplemented("upload_fobj", self.scheme)
-
-        with Tqdm.wrapattr(
-            fobj,
-            "read",
-            disable=no_progress_bar,
-            bytes=True,
-            total=size,
-            **pbar_args,
-        ) as wrapped:
-            self._upload_fobj(  # pylint: disable=no-member
-                wrapped, to_info, size=size
+            logger.debug("Uploading '%s' to '%s'", from_info, to_info)
+            # pylint: disable=no-member
+            return self.put_file(
+                os.fspath(from_info), to_info, callback=callback
             )
 
     def download(
@@ -265,84 +275,105 @@ class BaseFileSystem:
         from_info,
         to_info,
         name=None,
+        callback=None,
         no_progress_bar=False,
         jobs=None,
         _only_file=False,
         **kwargs,
     ):
-        if not hasattr(self, "_download"):
-            raise RemoteActionNotImplemented("download", self.scheme)
+        if not hasattr(self, "get_file"):
+            raise RemoteActionNotImplemented("get_file", self.scheme)
 
         if from_info.scheme != self.scheme:
             raise NotImplementedError
 
-        if to_info.scheme == self.scheme != "local":
-            self.copy(from_info, to_info)
-            return 0
-
         if to_info.scheme != "local":
             raise NotImplementedError
 
-        if not _only_file and self.isdir(from_info):
-            return self._download_dir(
-                from_info, to_info, name, no_progress_bar, jobs, **kwargs
+        download_dir = not _only_file and self.isdir(from_info)
+
+        desc = name or to_info.name
+        stack = contextlib.ExitStack()
+        if not callback:
+            pbar_kwargs = {"unit": "Files"} if download_dir else {}
+            pbar = ui.progress(
+                total=-1,
+                desc="Downloading directory" if download_dir else desc,
+                bytes=not download_dir,
+                disable=no_progress_bar,
+                **pbar_kwargs,
             )
-        return self._download_file(from_info, to_info, name, no_progress_bar)
+            stack.enter_context(pbar)
+            callback = pbar.as_callback()
+
+        with stack:
+            if download_dir:
+                return self._download_dir(
+                    from_info, to_info, callback=callback, jobs=jobs, **kwargs
+                )
+            return self._download_file(from_info, to_info, callback=callback)
 
     download_file = partialmethod(download, _only_file=True)
 
     def _download_dir(
-        self, from_info, to_info, name, no_progress_bar, jobs, **kwargs
+        self,
+        from_info,
+        to_info,
+        callback=DEFAULT_CALLBACK,
+        jobs=None,
+        **kwargs,
     ):
         from_infos = list(self.walk_files(from_info, **kwargs))
         if not from_infos:
             makedirs(to_info, exist_ok=True)
             return None
+
         to_infos = (
             to_info / info.relative_to(from_info) for info in from_infos
         )
+        callback.set_size(len(from_infos))
 
-        with Tqdm(
-            total=len(from_infos),
-            desc="Downloading directory",
-            unit="Files",
-            disable=no_progress_bar,
-        ) as pbar:
-            download_files = pbar.wrap_fn(
-                partial(self._download_file, name=name, no_progress_bar=True)
-            )
-            max_workers = jobs or self.jobs
-            with ThreadPoolExecutor(max_workers=max_workers) as executor:
-                futures = [
-                    executor.submit(download_files, from_info, to_info)
-                    for from_info, to_info in zip(from_infos, to_infos)
-                ]
+        download_files = FsspecCallback.wrap_fn(callback, self._download_file)
+        max_workers = jobs or self.jobs
+        with ThreadPoolExecutor(max_workers=max_workers) as executor:
+            futures = [
+                executor.submit(download_files, from_info, to_info)
+                for from_info, to_info in zip(from_infos, to_infos)
+            ]
 
-                # NOTE: unlike pulling/fetching cache, where we need to
-                # download everything we can, not raising an error here might
-                # turn very ugly, as the user might think that he has
-                # downloaded a complete directory, while having a partial one,
-                # which might cause unexpected results in his pipeline.
-                for future in as_completed(futures):
-                    # NOTE: executor won't let us raise until all futures that
-                    # it has are finished, so we need to cancel them ourselves
-                    # before re-raising.
-                    exc = future.exception()
-                    if exc:
-                        for entry in futures:
-                            entry.cancel()
-                        raise exc
+            # NOTE: unlike pulling/fetching cache, where we need to
+            # download everything we can, not raising an error here might
+            # turn very ugly, as the user might think that he has
+            # downloaded a complete directory, while having a partial one,
+            # which might cause unexpected results in his pipeline.
+            for future in as_completed(futures):
+                # NOTE: executor won't let us raise until all futures that
+                # it has are finished, so we need to cancel them ourselves
+                # before re-raising.
+                exc = future.exception()
+                if exc:
+                    for entry in futures:
+                        entry.cancel()
+                    raise exc
 
-    def _download_file(self, from_info, to_info, name, no_progress_bar):
+    def _download_file(
+        self,
+        from_info,
+        to_info,
+        callback=DEFAULT_CALLBACK,
+    ):
         makedirs(to_info.parent, exist_ok=True)
-
-        logger.debug("Downloading '%s' to '%s'", from_info, to_info)
-        name = name or to_info.name
-
         tmp_file = tmp_fname(to_info)
 
-        self._download(  # noqa, pylint: disable=no-member
-            from_info, tmp_file, name=name, no_progress_bar=no_progress_bar
-        )
+        logger.debug("Downloading '%s' to '%s'", from_info, to_info)
+        try:
+            # noqa, pylint: disable=no-member
+            self.get_file(from_info, tmp_file, callback=callback)
+        except Exception:  # pylint: disable=broad-except
+            # do we need to rollback makedirs for previously not-existing
+            # directories?
+            with contextlib.suppress(FileNotFoundError):
+                os.unlink(tmp_file)
+            raise
 
         move(tmp_file, to_info)

dvc/fs/dvc.py

@@ -6,6 +6,7 @@ from dvc.exceptions import OutputNotFoundError
 from dvc.path_info import PathInfo
 from dvc.utils import relpath
 
+from ..progress import DEFAULT_CALLBACK
 from ._metadata import Metadata
 from .base import BaseFileSystem
 
@@ -57,9 +58,9 @@ class DvcFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
         out.get_dir_cache(remote=remote)
         if out.obj is None:
             raise FileNotFoundError
-        obj = out.obj.trie.get(key)
-        if obj:
-            return obj.hash_info
+        (_, oid) = out.obj.trie.get(key) or (None, None)
+        if oid:
+            return oid
         raise FileNotFoundError
 
     def _get_fs_path(self, path: PathInfo, remote=None):
@@ -153,7 +154,7 @@ class DvcFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
         self._fetch_dir(out, **kwargs)
 
         base = out.path_info.parts
-        for key, _ in out.obj:  # noqa: B301
+        for key, _, _ in out.obj:  # noqa: B301
             trie[base + key] = None
 
     def _walk(self, root, trie, topdown=True, **kwargs):
@@ -165,16 +166,19 @@ class DvcFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
             self._add_dir(trie, out, **kwargs)
 
         root_len = len(root.parts)
-        for key, out in trie.iteritems(prefix=root.parts):  # noqa: B301
-            if key == root.parts:
-                continue
+        try:
+            for key, out in trie.iteritems(prefix=root.parts):  # noqa: B301
+                if key == root.parts:
+                    continue
 
-            name = key[root_len]
-            if len(key) > root_len + 1 or (out and out.is_dir_checksum):
-                dirs.add(name)
-                continue
+                name = key[root_len]
+                if len(key) > root_len + 1 or (out and out.is_dir_checksum):
+                    dirs.add(name)
+                    continue
 
-            files.append(name)
+                files.append(name)
+        except KeyError:
+            pass
 
         assert topdown
         dirs = list(dirs)
@@ -243,21 +247,30 @@ class DvcFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
         meta = self.metadata(path_info)
         ret = {"type": "directory" if meta.isdir else "file"}
         if meta.is_output and len(meta.outs) == 1 and meta.outs[0].hash_info:
-            hash_info = meta.outs[0].hash_info
-            ret["size"] = hash_info.size
-            ret[hash_info.name] = hash_info.value
+            out = meta.outs[0]
+            ret["size"] = out.meta.size
+            ret[out.hash_info.name] = out.hash_info.value
         elif meta.part_of_output:
             (out,) = meta.outs
             key = path_info.relative_to(out.path_info).parts
-            obj = out.obj.trie.get(key)
-            if obj:
-                ret["size"] = obj.size
-                ret[obj.hash_info.name] = obj.hash_info.value
+            (obj_meta, oid) = out.obj.trie.get(key) or (None, None)
+            if oid:
+                ret["size"] = obj_meta.size if obj_meta else 0
+                ret[oid.name] = oid.value
 
         return ret
 
-    def _download(self, from_info, to_file, **kwargs):
+    def get_file(
+        self, from_info, to_file, callback=DEFAULT_CALLBACK, **kwargs
+    ):
         fs, path = self._get_fs_path(from_info)
-        fs._download(  # pylint: disable=protected-access
-            path, to_file, **kwargs
+        fs.get_file(  # pylint: disable=protected-access
+            path, to_file, callback=callback, **kwargs
         )
+
+    def checksum(self, path_info):
+        info = self.info(path_info)
+        md5 = info.get("md5")
+        if md5:
+            return md5
+        raise NotImplementedError

dvc/fs/fsspec_wrapper.py

@@ -1,13 +1,37 @@
 import os
 import shutil
 from functools import lru_cache
+from typing import (
+    IO,
+    TYPE_CHECKING,
+    Any,
+    Dict,
+    Iterator,
+    Optional,
+    Union,
+    overload,
+)
 
 from funcy import cached_property
+from tqdm.utils import CallbackIOWrapper
 
-from dvc.progress import Tqdm
+from dvc.path_info import URLInfo
+from dvc.progress import DEFAULT_CALLBACK
 
 from .base import BaseFileSystem
-from .local import LocalFileSystem
+
+FSPath = URLInfo
+AnyFSPath = Union[FSPath, str]
+
+if TYPE_CHECKING:
+    from typing import Iterable
+
+    from typing_extensions import Literal
+
+
+# An info() entry, might evolve to a TypedDict
+# in the future (e.g for properly type 'size' etc).
+Entry = Dict[str, Any]
 
 
 # pylint: disable=no-member
@@ -19,18 +43,8 @@ class FSSpecWrapper(BaseFileSystem):
         self.fs_args = {"skip_instance_cache": True}
         self.fs_args.update(self._prepare_credentials(**kwargs))
 
-    @cached_property
-    def fs(self):
-        raise NotImplementedError
-
-    def _with_bucket(self, path):
-        return str(path)
-
-    def _strip_bucket(self, entry):
-        return entry
-
     @staticmethod
-    def _get_kwargs_from_urls(urlpath):
+    def _get_kwargs_from_urls(urlpath: str) -> "Dict[str, Any]":
         from fsspec.utils import infer_storage_options
 
         options = infer_storage_options(urlpath)
@@ -38,6 +52,28 @@ class FSSpecWrapper(BaseFileSystem):
         options.pop("protocol", None)
         return options
 
+    @cached_property
+    def fs(self):
+        raise NotImplementedError
+
+    def _with_bucket(self, path: AnyFSPath) -> str:
+        return str(path)
+
+    def _strip_bucket(self, entry: str) -> str:
+        return entry
+
+    @overload
+    def _strip_buckets(
+        self, entries: "Iterable[Entry]", detail: "Literal[True]"
+    ) -> "Iterator[Entry]":
+        ...
+
+    @overload
+    def _strip_buckets(
+        self, entries: "Iterable[str]", detail: "Literal[False]" = False
+    ) -> "Iterator[str]":
+        ...
+
     def _strip_buckets(self, entries, detail=False):
         for entry in entries:
             if detail:
@@ -47,56 +83,91 @@ class FSSpecWrapper(BaseFileSystem):
                 entry = self._strip_bucket(entry)
             yield entry
 
-    def _entry_hook(self, entry):
+    def _entry_hook(self, entry: "Entry") -> "Entry":
         """Simple hook method to be overridden when wanted to process
         entries within info() and ls(detail=True) calls"""
         return entry
 
     def _prepare_credentials(
-        self, **config
-    ):  # pylint: disable=unused-argument
+        self, **config: Dict[str, Any]  # pylint: disable=unused-argument
+    ) -> Dict[str, Any]:
         """Prepare the arguments for authentication to the
         host filesystem"""
         return {}
 
-    def _isdir(self, path_info):
+    def _isdir(self, path_info: FSPath) -> bool:
         return self.fs.isdir(self._with_bucket(path_info))
 
-    def isdir(self, path_info):
+    def isdir(self, path_info: FSPath) -> bool:
         try:
             return self._isdir(path_info)
         except FileNotFoundError:
             return False
 
-    def isfile(self, path_info):
+    def isfile(self, path_info: FSPath) -> bool:
         try:
             return not self._isdir(path_info)
         except FileNotFoundError:
             return False
 
-    def is_empty(self, path_info):
+    def is_empty(self, path_info: FSPath) -> bool:
         entry = self.info(path_info)
         if entry["type"] == "directory":
             return not self.fs.ls(self._with_bucket(path_info))
         return entry["size"] == 0
 
     def open(
-        self, path_info, mode="r", **kwargs
-    ):  # pylint: disable=arguments-differ
+        self,
+        path_info: FSPath,
+        mode: str = "r",
+        encoding: Optional[str] = None,
+        **kwargs,
+    ) -> "IO":  # pylint: disable=arguments-differ
         return self.fs.open(self._with_bucket(path_info), mode=mode)
 
-    def copy(self, from_info, to_info):
+    def checksum(self, path_info: FSPath) -> str:
+        return self.fs.checksum(self._with_bucket(path_info))
+
+    def copy(self, from_info: FSPath, to_info: FSPath) -> None:
         self.makedirs(to_info.parent)
         self.fs.copy(self._with_bucket(from_info), self._with_bucket(to_info))
 
-    def exists(self, path_info) -> bool:
+    def exists(self, path_info: FSPath) -> bool:
         return self.fs.exists(self._with_bucket(path_info))
 
+    @overload
+    def ls(
+        self, path_info: FSPath, detail: "Literal[True]"
+    ) -> "Iterator[Entry]":
+        ...
+
+    @overload
+    def ls(self, path_info: FSPath, detail: "Literal[False]") -> Iterator[str]:
+        ...
+
     def ls(self, path_info, detail=False):
         path = self._with_bucket(path_info)
         files = self.fs.ls(path, detail=detail)
         yield from self._strip_buckets(files, detail=detail)
 
+    @overload
+    def find(
+        self,
+        path_info: FSPath,
+        detail: "Literal[True]",
+        prefix: Optional[bool],
+    ) -> "Iterator[Entry]":
+        ...
+
+    @overload
+    def find(
+        self,
+        path_info: FSPath,
+        detail: "Literal[False]",
+        prefix: Optional[bool],
+    ) -> Iterator[str]:
+        ...
+
     # pylint: disable=unused-argument
     def find(self, path_info, detail=False, prefix=None):
         path = self._with_bucket(path_info)
@@ -106,63 +177,56 @@ class FSSpecWrapper(BaseFileSystem):
 
         yield from self._strip_buckets(files, detail=detail)
 
-    def walk_files(self, path_info, **kwargs):
+    def walk_files(self, path_info: FSPath, **kwargs) -> Iterator[FSPath]:
         for file in self.find(path_info, **kwargs):
             yield path_info.replace(path=file)
 
-    def move(self, from_info, to_info):
+    def move(self, from_info: FSPath, to_info: FSPath) -> None:
         self.fs.move(self._with_bucket(from_info), self._with_bucket(to_info))
 
-    def remove(self, path_info):
+    def remove(self, path_info: FSPath) -> None:
         self.fs.rm_file(self._with_bucket(path_info))
 
-    def info(self, path_info):
+    def info(self, path_info: FSPath) -> "Entry":
         info = self.fs.info(self._with_bucket(path_info)).copy()
         info = self._entry_hook(info)
         info["name"] = self._strip_bucket(info["name"])
         return info
 
-    def makedirs(self, path_info, **kwargs):
+    def makedirs(self, path_info: FSPath, **kwargs) -> None:
         self.fs.makedirs(
             self._with_bucket(path_info), exist_ok=kwargs.pop("exist_ok", True)
         )
 
-    def _upload_fobj(self, fobj, to_info, size=None):
-        self.makedirs(to_info.parent)
-        with self.open(to_info, "wb") as fdest:
-            shutil.copyfileobj(fobj, fdest, length=fdest.blocksize)
-
-    def _upload(
-        self, from_file, to_info, name=None, no_progress_bar=False, **kwargs
-    ):
-        self.makedirs(to_info.parent)
-        size = os.path.getsize(from_file)
-        with open(from_file, "rb") as fobj:
-            self.upload_fobj(
-                fobj,
-                to_info,
-                size=size,
-                desc=name,
-                no_progress_bar=no_progress_bar,
-            )
+    def put_file(
+        self,
+        from_file: FSPath,
+        to_info: FSPath,
+        callback: Any = DEFAULT_CALLBACK,
+        **kwargs,
+    ) -> None:
+        self.fs.put_file(
+            from_file, self._with_bucket(to_info), callback=callback, **kwargs
+        )
         self.fs.invalidate_cache(self._with_bucket(to_info.parent))
 
-    def _download(
-        self, from_info, to_file, name=None, no_progress_bar=False, **pbar_args
-    ):
-        total = self.getsize(self._with_bucket(from_info))
-        with self.open(from_info, "rb") as fobj:
-            with Tqdm.wrapattr(
-                fobj,
-                "read",
-                desc=name,
-                disable=no_progress_bar,
-                bytes=True,
-                total=total,
-                **pbar_args,
-            ) as wrapped:
-                with open(to_file, "wb") as fdest:
-                    shutil.copyfileobj(wrapped, fdest, length=fobj.blocksize)
+    def get_file(
+        self,
+        from_info: FSPath,
+        to_info: FSPath,
+        callback: Any = DEFAULT_CALLBACK,
+        **kwargs,
+    ) -> None:
+        self.fs.get_file(
+            self._with_bucket(from_info), to_info, callback=callback, **kwargs
+        )
+
+    def upload_fobj(self, fobj: IO, to_info: FSPath, **kwargs) -> None:
+        self.makedirs(to_info.parent)
+        with self.open(to_info, "wb") as fdest:
+            shutil.copyfileobj(
+                fobj, fdest, length=getattr(fdest, "blocksize", None)
+            )
 
 
 # pylint: disable=abstract-method
@@ -170,21 +234,21 @@ class ObjectFSWrapper(FSSpecWrapper):
     TRAVERSE_PREFIX_LEN = 3
 
     @lru_cache(512)
-    def _with_bucket(self, path):
+    def _with_bucket(self, path: AnyFSPath) -> str:
         if isinstance(path, self.PATH_CLS):
             return f"{path.bucket}/{path.path}"
-        return path
+        return str(path)
 
-    def _strip_bucket(self, entry):
+    def _strip_bucket(self, entry: str) -> str:
         try:
             bucket, path = entry.split("/", 1)
         except ValueError:
             # If there is no path attached, only returns
             # the bucket (top-level).
-            bucket, path = entry, None
+            bucket, path = entry, ""
         return path or bucket
 
-    def makedirs(self, path_info, **kwargs):
+    def makedirs(self, path_info: FSPath, **kwargs) -> None:
         # For object storages make this method a no-op. The original
         # fs.makedirs() method will only check if the bucket exists
         # and create if it doesn't though we don't want to support
@@ -192,7 +256,7 @@ class ObjectFSWrapper(FSSpecWrapper):
         # simply ignore all mkdir()/makedirs() calls.
         return None
 
-    def _isdir(self, path_info):
+    def _isdir(self, path_info: FSPath) -> bool:
         # Directory in object storages are interpreted differently
         # among different fsspec providers, so this logic is a temporary
         # measure for us to adapt as of now. It checks whether it is a
@@ -227,45 +291,64 @@ class ObjectFSWrapper(FSSpecWrapper):
         if len(files) == 1 and files[0] == path and self.isdir(path_info):
             return None
 
-        yield from self._strip_buckets(files, detail=detail)
+        yield from self._strip_buckets(  # pylint: disable=not-an-iterable
+            files, detail=detail
+        )
 
 
-_LOCAL_FS = LocalFileSystem()
+# pylint: disable=arguments-differ
+class NoDirectoriesMixin:
+    def isdir(self, *args, **kwargs):
+        return False
+
+    def isfile(self, *args, **kwargs):
+        return True
+
+    def find(self, *args, **kwargs):
+        raise NotImplementedError
+
+    def walk(self, *args, **kwargs):
+        raise NotImplementedError
+
+    def walk_files(self, *args, **kwargs):
+        raise NotImplementedError
+
+    def ls(self, *args, **kwargs):
+        raise NotImplementedError
 
 
 class CallbackMixin:
-    """Use the native ``get_file()``/``put_file()`` APIs
-    if the target filesystem supports callbacks."""
+    """Provides callback support for the filesystem that don't support yet."""
 
-    def _upload(
-        self, from_file, to_info, name=None, no_progress_bar=False, **pbar_args
+    def put_file(
+        self,
+        from_file,
+        to_info,
+        callback=DEFAULT_CALLBACK,
+        **kwargs,
     ):
-        with Tqdm(
-            desc=name,
-            disable=no_progress_bar,
-            bytes=True,
-            total=-1,
-            **pbar_args,
-        ) as pbar:
-            self.fs.put_file(
-                os.fspath(from_file),
-                self._with_bucket(to_info),
-                callback=pbar.as_callback(_LOCAL_FS, from_file),
-            )
-        self.fs.invalidate_cache(self._with_bucket(to_info.parent))
+        """Add compatibility support for Callback."""
+        # pylint: disable=protected-access
+        self.makedirs(to_info.parent)
+        size = os.path.getsize(from_file)
+        with open(from_file, "rb") as fobj:
+            callback.set_size(size)
+            wrapped = CallbackIOWrapper(callback.relative_update, fobj)
+            self.upload_fobj(wrapped, to_info)
+            self.fs.invalidate_cache(self._with_bucket(to_info.parent))
 
-    def _download(
-        self, from_info, to_file, name=None, no_progress_bar=False, **pbar_args
+    def get_file(
+        self,
+        from_info,
+        to_info,
+        callback=DEFAULT_CALLBACK,
+        **kwargs,
     ):
-        with Tqdm(
-            desc=name,
-            disable=no_progress_bar,
-            bytes=True,
-            total=-1,
-            **pbar_args,
-        ) as pbar:
-            self.fs.get_file(
-                self._with_bucket(from_info),
-                os.fspath(to_file),
-                callback=pbar.as_callback(self, from_info),
-            )
+        # pylint: disable=protected-access
+        total: int = self.getsize(from_info)
+        if total:
+            callback.set_size(total)
+
+        with self.open(from_info, "rb") as fobj, open(to_info, "wb") as fdest:
+            wrapped = CallbackIOWrapper(callback.relative_update, fobj)
+            shutil.copyfileobj(wrapped, fdest, length=fobj.blocksize)

dvc/fs/gdrive.py

@@ -5,14 +5,14 @@ import re
 import threading
 from urllib.parse import urlparse
 
-from funcy import cached_property, retry, wrap_prop
+from funcy import cached_property, wrap_prop
 
 from dvc.exceptions import DvcException
 from dvc.path_info import CloudURLInfo
 from dvc.scheme import Schemes
 from dvc.utils import format_link, tmp_fname
 
-from .fsspec_wrapper import CallbackMixin, FSSpecWrapper
+from .fsspec_wrapper import FSSpecWrapper
 
 logger = logging.getLogger(__name__)
 FOLDER_MIME_TYPE = "application/vnd.google-apps.folder"
@@ -34,36 +34,6 @@ class GDriveAuthError(DvcException):
         super().__init__(message)
 
 
-def _gdrive_retry(func):
-    def should_retry(exc):
-        from pydrive2.files import ApiRequestError
-
-        if not isinstance(exc, ApiRequestError):
-            return False
-
-        error_code = exc.error.get("code", 0)
-        result = False
-        if 500 <= error_code < 600:
-            result = True
-
-        if error_code == 403:
-            result = exc.GetField("reason") in [
-                "userRateLimitExceeded",
-                "rateLimitExceeded",
-            ]
-        if result:
-            logger.debug(f"Retrying GDrive API call, error: {exc}.")
-
-        return result
-
-    # 16 tries, start at 0.5s, multiply by golden ratio, cap at 20s
-    return retry(
-        16,
-        timeout=lambda a: min(0.5 * 1.618 ** a, 20),
-        filter_errors=should_retry,
-    )(func)
-
-
 class GDriveURLInfo(CloudURLInfo):
     def __init__(self, url):
         super().__init__(url)
@@ -79,9 +49,7 @@ class GDriveURLInfo(CloudURLInfo):
         self._spath = re.sub("/{2,}", "/", self._spath.rstrip("/"))
 
 
-class GDriveFileSystem(
-    CallbackMixin, FSSpecWrapper
-):  # pylint:disable=abstract-method
+class GDriveFileSystem(FSSpecWrapper):  # pylint:disable=abstract-method
     scheme = Schemes.GDRIVE
     PATH_CLS = GDriveURLInfo
     PARAM_CHECKSUM = "checksum"
@@ -221,7 +189,7 @@ class GDriveFileSystem(
             temporary_save_path = self._gdrive_service_credentials_path
 
         if is_credentials_temp:
-            with open(temporary_save_path, "w") as cred_file:
+            with open(temporary_save_path, "w", encoding="utf-8") as cred_file:
                 cred_file.write(
                     os.getenv(GDriveFileSystem.GDRIVE_CREDENTIALS_DATA)
                 )
@@ -298,7 +266,16 @@ class GDriveFileSystem(
 
         return super()._with_bucket(path)
 
-    def _upload_fobj(self, fobj, to_info, size: int = None):
+    def _strip_bucket(self, entry):
+        try:
+            bucket, path = entry.split("/", 1)
+        except ValueError:
+            # If there is no path attached, only returns
+            # the bucket (top-level).
+            bucket, path = entry, None
+        return path or bucket
+
+    def upload_fobj(self, fobj, to_info, **kwargs):
         rpath = self._with_bucket(to_info)
         self.makedirs(os.path.dirname(rpath))
-        return self.fs.upload_fobj(fobj, rpath, size=size)
+        return self.fs.upload_fobj(fobj, rpath, **kwargs)

dvc/fs/git.py

@@ -1,9 +1,11 @@
 import errno
 import os
-import stat
+
+from tqdm.utils import CallbackIOWrapper
 
 from dvc.utils import is_exec, relpath
 
+from ..progress import DEFAULT_CALLBACK
 from .base import BaseFileSystem
 
 
@@ -96,30 +98,20 @@ class GitFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
         if not self.exists(path_info):
             return False
 
-        mode = self.stat(path_info).st_mode
+        mode = self.info(path_info)["mode"]
         return is_exec(mode)
 
     def info(self, path_info):
         key = self._get_key(path_info)
         try:
-            st = self.trie.stat(key)
+            return self.trie.info(key)
         except KeyError:
             raise FileNotFoundError(
                 errno.ENOENT, os.strerror(errno.ENOENT), path_info
             )
-        return {
-            "size": st.st_size,
-            "type": "directory" if stat.S_ISDIR(st.st_mode) else "file",
-        }
 
-    def stat(self, path):
-        key = self._get_key(path)
-        try:
-            return self.trie.stat(key)
-        except KeyError:
-            raise FileNotFoundError(
-                errno.ENOENT, os.strerror(errno.ENOENT), path
-            )
+    def checksum(self, path_info):
+        return self.info(path_info)["sha"]
 
     def walk_files(self, path_info, **kwargs):
         for root, _, files in self.walk(path_info, **kwargs):
@@ -127,16 +119,18 @@ class GitFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
                 # NOTE: os.path.join is ~5.5 times slower
                 yield f"{root}{os.sep}{file}"
 
-    def _download(
-        self, from_info, to_file, name=None, no_progress_bar=False, **kwargs
+    def get_file(
+        self, from_info, to_file, callback=DEFAULT_CALLBACK, **kwargs
     ):
         import shutil
 
-        from dvc.progress import Tqdm
+        total = self.getsize(from_info)
+        if total:
+            callback.set_size(total)
 
-        with open(to_file, "wb+") as to_fobj:
-            with Tqdm.wrapattr(
-                to_fobj, "write", desc=name, disable=no_progress_bar
-            ) as wrapped:
-                with self.open(from_info, "rb", **kwargs) as from_fobj:
-                    shutil.copyfileobj(from_fobj, wrapped)
+        with self.open(from_info, "rb", **kwargs) as from_fobj:
+            with open(to_file, "wb+") as to_fobj:
+                wrapped = CallbackIOWrapper(
+                    callback.relative_update, from_fobj
+                )
+                shutil.copyfileobj(wrapped, to_fobj)

dvc/fs/gs.py

@@ -6,17 +6,16 @@ from funcy import cached_property, wrap_prop
 from dvc.path_info import CloudURLInfo
 from dvc.scheme import Schemes
 
-from .fsspec_wrapper import ObjectFSWrapper
-
-
 # pylint:disable=abstract-method
-class GSFileSystem(ObjectFSWrapper):
+from .fsspec_wrapper import CallbackMixin, ObjectFSWrapper
+
+
+class GSFileSystem(CallbackMixin, ObjectFSWrapper):
     scheme = Schemes.GS
     PATH_CLS = CloudURLInfo
     REQUIRES = {"gcsfs": "gcsfs"}
     PARAM_CHECKSUM = "etag"
     DETAIL_FIELDS = frozenset(("etag", "size"))
-    TRAVERSE_PREFIX_LEN = 2
 
     def _prepare_credentials(self, **config):
         login_info = {"consistency": None}

dvc/fs/hdfs.py

@@ -1,283 +1,90 @@
-import errno
-import io
-import logging
 import os
 import re
-import shutil
 import subprocess
-from collections import deque
-from contextlib import closing, contextmanager
+import threading
+
+from funcy import cached_property, wrap_prop
 
-from dvc.hash_info import HashInfo
-from dvc.progress import Tqdm
 from dvc.scheme import Schemes
-from dvc.utils import fix_env, tmp_fname
+from dvc.utils import fix_env
 
-from .base import BaseFileSystem, RemoteCmdError
-from .pool import get_connection
+from .fsspec_wrapper import CallbackMixin, FSSpecWrapper
 
-logger = logging.getLogger(__name__)
+CHECKSUM_REGEX = re.compile(r".*\t.*\t(?P<checksum>.*)")
 
 
-def _hadoop_fs(cmd, user=None):
-    cmd = "hadoop fs -" + cmd
-    if user:
-        cmd = f"HADOOP_USER_NAME={user} " + cmd
-
-    # NOTE: close_fds doesn't work with redirected stdin/stdout/stderr.
-    # See https://github.com/iterative/dvc/issues/1197.
-    close_fds = os.name != "nt"
-
-    executable = os.getenv("SHELL") if os.name != "nt" else None
-    p = subprocess.Popen(
-        cmd,
-        shell=True,
-        close_fds=close_fds,
-        executable=executable,
-        env=fix_env(os.environ),
-        stdin=subprocess.PIPE,
-        stdout=subprocess.PIPE,
-        stderr=subprocess.PIPE,
-    )
-    out, err = p.communicate()
-    if p.returncode != 0:
-        raise RemoteCmdError("hdfs", cmd, p.returncode, err)
-    return out.decode("utf-8")
-
-
-def _group(regex, s, gname):
-    match = re.match(regex, s)
-    assert match is not None
-    return match.group(gname)
-
-
-def _hadoop_fs_checksum(path_info):
-    # NOTE: pyarrow doesn't support checksum, so we need to use hadoop
-    regex = r".*\t.*\t(?P<checksum>.*)"
-    stdout = _hadoop_fs(f"checksum {path_info.url}", user=path_info.user)
-    return _group(regex, stdout, "checksum")
-
-
-class HDFSFileSystem(BaseFileSystem):
+# pylint: disable=abstract-method
+class HDFSFileSystem(CallbackMixin, FSSpecWrapper):
     scheme = Schemes.HDFS
-    REQUIRES = {"pyarrow": "pyarrow"}
-    REGEX = r"^hdfs://((?P<user>.*)@)?.*$"
+    REQUIRES = {"fsspec": "fsspec", "pyarrow": "pyarrow"}
     PARAM_CHECKSUM = "checksum"
-    TRAVERSE_PREFIX_LEN = 2
 
-    def __init__(self, **config):
-        super().__init__(**config)
-
-        self.host = config["host"]
-        self.user = config.get("user")
-        self.port = config.get("port")
-
-        self.kerb_ticket = config.get("kerb_ticket")
+    def _with_bucket(self, path):
+        if isinstance(path, self.PATH_CLS):
+            return path.path
+        return super()._with_bucket(path)
 
     @staticmethod
     def _get_kwargs_from_urls(urlpath):
-        from fsspec.implementations.hdfs import PyArrowHDFS
+        from fsspec.implementations.arrow import HadoopFileSystem
 
         # pylint:disable=protected-access
-        return PyArrowHDFS._get_kwargs_from_urls(urlpath)
+        return HadoopFileSystem._get_kwargs_from_urls(urlpath)
 
-    def hdfs(self, path_info):
-        import pyarrow.fs
+    def _prepare_credentials(self, **config):
+        return config
 
-        # NOTE: HadoopFileSystem is not meant to be closed by us and doesn't
-        # have close() method or alternative, so we add a noop one to satisfy
-        # our connection pool.
-        class HDFSConnection(pyarrow.fs.HadoopFileSystem):
-            def close(self):
-                pass
+    @wrap_prop(threading.Lock())
+    @cached_property
+    def fs(self):
+        from fsspec.implementations.arrow import HadoopFileSystem
 
-        return get_connection(
-            HDFSConnection,
-            path_info.host,
-            path_info.port,
-            user=path_info.user,
-            kerb_ticket=self.kerb_ticket,
-        )
-
-    @contextmanager
-    def open(self, path_info, mode="r", encoding=None, **kwargs):
-        assert mode in {"r", "rt", "rb"}
-
-        try:
-            with self.hdfs(path_info) as hdfs, closing(
-                hdfs.open_input_stream(path_info.path)
-            ) as fd:
-                if mode == "rb":
-                    yield fd
-                else:
-                    yield io.TextIOWrapper(fd, encoding=encoding)
-        except OSError as e:
-            # Empty .errno and not specific enough error class in pyarrow,
-            # see https://issues.apache.org/jira/browse/ARROW-6248
-            if "file does not exist" in str(e):
-                raise FileNotFoundError(*e.args)
-            raise
-
-    def exists(self, path_info) -> bool:
-        assert not isinstance(path_info, list)
-        assert path_info.scheme == "hdfs"
-        with self.hdfs(path_info) as hdfs:
-            import pyarrow.fs
-
-            file_info = hdfs.get_file_info(path_info.path)
-            return file_info.type != pyarrow.fs.FileType.NotFound
-
-    def _walk(self, hdfs, root, topdown=True):
-        import posixpath
-
-        from pyarrow.fs import FileSelector, FileType
-
-        dirs = deque()
-        nondirs = deque()
-
-        selector = FileSelector(root)
-        for entry in hdfs.get_file_info(selector):
-            if entry.type == FileType.Directory:
-                dirs.append(entry.base_name)
-            else:
-                nondirs.append(entry.base_name)
-
-        if topdown:
-            yield root, dirs, nondirs
-
-        for dname in dirs:
-            # NOTE: posixpath.join() is slower
-            yield from self._walk(
-                hdfs, f"{root}{posixpath.sep}{dname}", topdown=topdown
-            )
-
-        if not topdown:
-            yield root, dirs, nondirs
-
-    def walk(self, top, topdown=True, onerror=None, **kwargs):
-        if not self.isdir(top):
-            return
-
-        with self.hdfs(top) as hdfs:
-            for root, dnames, fnames in self._walk(
-                hdfs, top.path, topdown=topdown
-            ):
-                yield top.replace(path=root), dnames, fnames
-
-    def walk_files(self, path_info, **kwargs):
-        for root, _, fnames in self.walk(path_info):
-            yield from (root / fname for fname in fnames)
-
-    def remove(self, path_info):
-        if path_info.scheme != "hdfs":
-            raise NotImplementedError
-
-        if self.exists(path_info):
-            logger.debug(f"Removing {path_info.path}")
-            with self.hdfs(path_info) as hdfs:
-                import pyarrow.fs
-
-                file_info = hdfs.get_file_info(path_info.path)
-                if file_info.type == pyarrow.fs.FileType.Directory:
-                    hdfs.delete_dir(path_info.path)
-                else:
-                    hdfs.delete_file(path_info.path)
-
-    def makedirs(self, path_info, **kwargs):
-        with self.hdfs(path_info) as hdfs:
-            # NOTE: fs.create_dir creates parents by default
-            hdfs.create_dir(path_info.path)
-
-    def copy(self, from_info, to_info, **_kwargs):
-        # NOTE: hdfs.copy_file is not supported yet in pyarrow
-        with self.hdfs(to_info) as hdfs:
-            # NOTE: this is how `hadoop fs -cp` works too: it copies through
-            # your local machine.
-            with closing(hdfs.open_input_stream(from_info.path)) as from_fobj:
-                tmp_info = to_info.parent / tmp_fname(to_info.name)
-                try:
-                    with closing(
-                        hdfs.open_output_stream(tmp_info.path)
-                    ) as tmp_fobj:
-                        shutil.copyfileobj(from_fobj, tmp_fobj)
-                    hdfs.move(tmp_info.path, to_info.path)
-                except Exception:
-                    self.remove(tmp_info)
-                    raise
-
-    def isfile(self, path_info):
-        with self.hdfs(path_info) as hdfs:
-            import pyarrow.fs
-
-            file_info = hdfs.get_file_info(path_info.path)
-            return file_info.type == pyarrow.fs.FileType.File
-
-    def isdir(self, path_info):
-        with self.hdfs(path_info) as hdfs:
-            import pyarrow.fs
-
-            file_info = hdfs.get_file_info(path_info.path)
-            return file_info.type == pyarrow.fs.FileType.Directory
-
-    def info(self, path_info):
-        with self.hdfs(path_info) as hdfs:
-            from pyarrow.fs import FileType
-
-            file_info = hdfs.get_file_info(path_info.path)
-            if file_info.type is FileType.Directory:
-                kind = "directory"
-            elif file_info.type is FileType.File:
-                kind = "file"
-            else:
-                raise FileNotFoundError(
-                    errno.ENOENT, os.strerror(errno.ENOENT), path_info
-                )
-
-            return {"size": file_info.size, "type": kind}
+        return HadoopFileSystem(**self.fs_args)
 
     def checksum(self, path_info):
-        return HashInfo(
-            "checksum",
-            _hadoop_fs_checksum(path_info),
-            size=self.getsize(path_info),
+        return self._checksum(path_info)
+
+    def _checksum(self, path_info, **kwargs):
+        # PyArrow doesn't natively support retrieving the
+        # checksum, so we have to use hadoop fs
+
+        result = self._run_command(
+            f"checksum {path_info.url}",
+            env=fix_env(os.environ),
+            user=path_info.user,
         )
+        if result is None:
+            return None
 
-    def _upload_fobj(self, fobj, to_info, **kwargs):
-        with self.hdfs(to_info) as hdfs:
-            with hdfs.open_output_stream(to_info.path) as fdest:
-                shutil.copyfileobj(fobj, fdest)
+        match = CHECKSUM_REGEX.match(result)
+        if match is None:
+            return None
 
-    def _upload(
-        self, from_file, to_info, name=None, no_progress_bar=False, **_kwargs
-    ):
-        with self.hdfs(to_info) as hdfs:
-            tmp_file = tmp_fname(to_info.path)
-            total = os.path.getsize(from_file)
-            with open(from_file, "rb") as fobj:
-                with Tqdm.wrapattr(
-                    fobj,
-                    "read",
-                    desc=name,
-                    total=total,
-                    disable=no_progress_bar,
-                ) as wrapped:
-                    with hdfs.open_output_stream(tmp_file) as sobj:
-                        sobj.write(wrapped.read())
-            hdfs.move(tmp_file, to_info.path)
+        return match.group("checksum")
 
-    def _download(
-        self, from_info, to_file, name=None, no_progress_bar=False, **_kwargs
-    ):
-        with self.hdfs(from_info) as hdfs:
-            file_info = hdfs.get_file_info(from_info.path)
-            total = file_info.size
-            with open(to_file, "wb+") as fobj:
-                with Tqdm.wrapattr(
-                    fobj,
-                    "write",
-                    desc=name,
-                    total=total,
-                    disable=no_progress_bar,
-                ) as wrapped:
-                    with hdfs.open_input_stream(from_info.path) as sobj:
-                        wrapped.write(sobj.read())
+    def _run_command(self, cmd, env=None, user=None):
+        cmd = "hadoop fs -" + cmd
+        if user:
+            cmd = f"HADOOP_USER_NAME={user} " + cmd
+
+        # NOTE: close_fds doesn't work with redirected stdin/stdout/stderr.
+        # See https://github.com/iterative/dvc/issues/1197.
+        close_fds = os.name != "nt"
+
+        executable = os.getenv("SHELL") if os.name != "nt" else None
+        p = subprocess.Popen(
+            cmd,
+            shell=True,
+            close_fds=close_fds,
+            executable=executable,
+            env=env or os.environ,
+            stdin=subprocess.PIPE,
+            stdout=subprocess.PIPE,
+            stderr=subprocess.PIPE,
+        )
+        out, err = p.communicate()
+
+        if p.returncode != 0:
+            raise subprocess.CalledProcessError(p.returncode, cmd, out, err)
+        else:
+            return out.decode("utf-8")

dvc/fs/http.py

@@ -1,20 +1,12 @@
-import logging
-import os.path
 import threading
-from typing import Optional
-from urllib.parse import urlparse
 
-from funcy import cached_property, memoize, wrap_prop, wrap_with
+from funcy import cached_property, memoize, wrap_with
 
 from dvc import prompt
-from dvc.exceptions import DvcException, HTTPError
 from dvc.path_info import HTTPURLInfo
-from dvc.progress import Tqdm
 from dvc.scheme import Schemes
 
-from .base import BaseFileSystem
-
-logger = logging.getLogger(__name__)
+from .fsspec_wrapper import FSSpecWrapper, NoDirectoriesMixin
 
 
 @wrap_with(threading.Lock())
@@ -26,171 +18,115 @@ def ask_password(host, user):
     )
 
 
-class HTTPFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
+def make_context(ssl_verify):
+    if isinstance(ssl_verify, bool) or ssl_verify is None:
+        return ssl_verify
+
+    # If this is a path, then we will create an
+    # SSL context for it, and load the given certificate.
+    import ssl
+
+    context = ssl.create_default_context()
+    context.load_verify_locations(ssl_verify)
+    return context
+
+
+# pylint: disable=abstract-method
+class HTTPFileSystem(NoDirectoriesMixin, FSSpecWrapper):
     scheme = Schemes.HTTP
     PATH_CLS = HTTPURLInfo
-    PARAM_CHECKSUM = "etag"
+    PARAM_CHECKSUM = "checksum"
+    REQUIRES = {"aiohttp": "aiohttp", "aiohttp-retry": "aiohttp_retry"}
     CAN_TRAVERSE = False
-    REQUIRES = {"requests": "requests"}
 
     SESSION_RETRIES = 5
     SESSION_BACKOFF_FACTOR = 0.1
     REQUEST_TIMEOUT = 60
-    CHUNK_SIZE = 2 ** 16
 
-    def __init__(self, **config):
-        super().__init__(**config)
+    def _prepare_credentials(self, **config):
+        import aiohttp
+        from fsspec.asyn import fsspec_loop
 
-        self.user = config.get("user", None)
+        from dvc.config import ConfigError
 
-        self.auth = config.get("auth", None)
-        self.custom_auth_header = config.get("custom_auth_header", None)
-        self.password = config.get("password", None)
-        self.ask_password = config.get("ask_password", False)
-        self.headers = {}
-        self.ssl_verify = config.get("ssl_verify", True)
-        self.method = config.get("method", "POST")
+        credentials = {}
+        client_kwargs = credentials.setdefault("client_kwargs", {})
 
-    def _auth_method(self, url):
-        from requests.auth import HTTPBasicAuth, HTTPDigestAuth
+        if config.get("auth"):
+            user = config.get("user")
+            password = config.get("password")
+            custom_auth_header = config.get("custom_auth_header")
 
-        if self.auth:
-            if self.ask_password and self.password is None:
-                self.password = ask_password(urlparse(url).hostname, self.user)
-            if self.auth == "basic":
-                return HTTPBasicAuth(self.user, self.password)
-            if self.auth == "digest":
-                return HTTPDigestAuth(self.user, self.password)
-            if self.auth == "custom" and self.custom_auth_header:
-                self.headers.update({self.custom_auth_header: self.password})
-        return None
+            if password is None and config.get("ask_password"):
+                password = ask_password(config.get("url"), user or "custom")
 
-    def _generate_download_url(self, path_info):
-        return path_info.url
+            auth_method = config["auth"]
+            if auth_method == "basic":
+                if user is None or password is None:
+                    raise ConfigError(
+                        "HTTP 'basic' authentication require both "
+                        "'user' and 'password'"
+                    )
 
-    @wrap_prop(threading.Lock())
-    @cached_property
-    def _session(self):
-        import requests
-        from requests.adapters import HTTPAdapter
-        from urllib3.util.retry import Retry
+                client_kwargs["auth"] = aiohttp.BasicAuth(user, password)
+            elif auth_method == "custom":
+                if custom_auth_header is None or password is None:
+                    raise ConfigError(
+                        "HTTP 'custom' authentication require both "
+                        "'custom_auth_header' and 'password'"
+                    )
+                credentials["headers"] = {custom_auth_header: password}
+            else:
+                raise NotImplementedError(
+                    f"Auth method {auth_method!r} is not supported."
+                )
 
-        session = requests.Session()
+        if "ssl_verify" in config:
+            with fsspec_loop():
+                client_kwargs["connector"] = aiohttp.TCPConnector(
+                    ssl=make_context(config["ssl_verify"])
+                )
 
-        session.verify = self.ssl_verify
+        # Allow reading proxy configurations from the environment.
+        client_kwargs["trust_env"] = True
 
-        retries = Retry(
-            total=self.SESSION_RETRIES,
-            backoff_factor=self.SESSION_BACKOFF_FACTOR,
+        credentials["get_client"] = self.get_client
+        self.upload_method = config.get("method", "POST")
+        return credentials
+
+    async def get_client(self, **kwargs):
+        import aiohttp
+        from aiohttp_retry import ExponentialRetry, RetryClient
+
+        kwargs["retry_options"] = ExponentialRetry(
+            attempts=self.SESSION_RETRIES,
+            factor=self.SESSION_BACKOFF_FACTOR,
+            max_timeout=self.REQUEST_TIMEOUT,
+            exceptions=[aiohttp.ClientError],
         )
 
-        session.mount("http://", HTTPAdapter(max_retries=retries))
-        session.mount("https://", HTTPAdapter(max_retries=retries))
+        # The default timeout for the aiohttp is 300 seconds
+        # which is too low for DVC's interactions (especially
+        # on the read) when dealing with large data blobs. We
+        # unlimit the total time to read, and only limit the
+        # time that is spent when connecting to the remote server.
+        kwargs["timeout"] = aiohttp.ClientTimeout(
+            total=None,
+            connect=self.REQUEST_TIMEOUT,
+            sock_connect=self.REQUEST_TIMEOUT,
+            sock_read=None,
+        )
 
-        return session
+        return RetryClient(**kwargs)
 
-    def request(self, method, url, **kwargs):
-        import requests
+    @cached_property
+    def fs(self):
+        from fsspec.implementations.http import (
+            HTTPFileSystem as _HTTPFileSystem,
+        )
 
-        kwargs.setdefault("allow_redirects", True)
-        kwargs.setdefault("timeout", self.REQUEST_TIMEOUT)
+        return _HTTPFileSystem(**self.fs_args)
 
-        try:
-            res = self._session.request(
-                method,
-                url,
-                auth=self._auth_method(url),
-                headers=self.headers,
-                **kwargs,
-            )
-
-            redirect_no_location = (
-                kwargs["allow_redirects"]
-                and res.status_code in (301, 302)
-                and "location" not in res.headers
-            )
-
-            if redirect_no_location:
-                # AWS s3 doesn't like to add a location header to its redirects
-                # from https://s3.amazonaws.com/<bucket name>/* type URLs.
-                # This should be treated as an error
-                raise requests.exceptions.RequestException
-
-            return res
-
-        except requests.exceptions.RequestException:
-            raise DvcException(f"could not perform a {method} request")
-
-    def _head(self, url):
-        response = self.request("HEAD", url)
-        if response.ok:
-            return response
-
-        # Sometimes servers are configured to forbid HEAD requests
-        # Context: https://github.com/iterative/dvc/issues/4131
-        with self.request("GET", url, stream=True) as r:
-            if r.ok:
-                return r
-
-        return response
-
-    def exists(self, path_info) -> bool:
-        res = self._head(path_info.url)
-        if res.status_code == 404:
-            return False
-        if bool(res):
-            return True
-        raise HTTPError(res.status_code, res.reason)
-
-    def info(self, path_info):
-        resp = self._head(path_info.url)
-        etag = resp.headers.get("ETag") or resp.headers.get("Content-MD5")
-        size = self._content_length(resp)
-        return {"etag": etag, "size": size, "type": "file"}
-
-    def _upload_fobj(self, fobj, to_info, **kwargs):
-        def chunks(fobj):
-            while True:
-                chunk = fobj.read(self.CHUNK_SIZE)
-                if not chunk:
-                    break
-                yield chunk
-
-        response = self.request(self.method, to_info.url, data=chunks(fobj))
-        if response.status_code not in (200, 201):
-            raise HTTPError(response.status_code, response.reason)
-
-    def _download(self, from_info, to_file, name=None, no_progress_bar=False):
-        response = self.request("GET", from_info.url, stream=True)
-        if response.status_code != 200:
-            raise HTTPError(response.status_code, response.reason)
-        with open(to_file, "wb") as fd:
-            with Tqdm.wrapattr(
-                fd,
-                "write",
-                total=None
-                if no_progress_bar
-                else self._content_length(response),
-                leave=False,
-                desc=from_info.url if name is None else name,
-                disable=no_progress_bar,
-            ) as fd_wrapped:
-                for chunk in response.iter_content(chunk_size=self.CHUNK_SIZE):
-                    fd_wrapped.write(chunk)
-
-    def _upload(
-        self, from_file, to_info, name=None, no_progress_bar=False, **_kwargs
-    ):
-        with open(from_file, "rb") as fobj:
-            self.upload_fobj(
-                fobj,
-                to_info,
-                size=None if no_progress_bar else os.path.getsize(from_file),
-                no_progress_bar=no_progress_bar,
-                desc=name or to_info.url,
-            )
-
-    @staticmethod
-    def _content_length(response) -> Optional[int]:
-        res = response.headers.get("Content-Length")
-        return int(res) if res else None
+    def _entry_hook(self, entry):
+        entry["checksum"] = entry.get("ETag") or entry.get("Content-MD5")
+        return entry

dvc/fs/local.py

@@ -1,6 +1,5 @@
 import logging
 import os
-import stat
 
 from dvc.path_info import PathInfo
 from dvc.scheme import Schemes
@@ -8,6 +7,7 @@ from dvc.system import System
 from dvc.utils import is_exec, tmp_fname
 from dvc.utils.fs import copy_fobj_to_file, copyfile, makedirs, move, remove
 
+from ..progress import DEFAULT_CALLBACK
 from .base import BaseFileSystem
 
 logger = logging.getLogger(__name__)
@@ -34,7 +34,16 @@ class LocalFileSystem(BaseFileSystem):
 
     def exists(self, path_info) -> bool:
         assert isinstance(path_info, str) or path_info.scheme == "local"
-        return self.fs.exists(path_info)
+        # TODO: replace this with os.path.exists once the problem is fixed on
+        # the fsspec https://github.com/intake/filesystem_spec/issues/742
+        return os.path.lexists(path_info)
+
+    def checksum(self, path_info) -> str:
+        from fsspec.utils import tokenize
+
+        st = os.stat(path_info)
+
+        return str(int(tokenize([st.st_ino, st.st_mtime, st.st_size]), 16))
 
     def isfile(self, path_info) -> bool:
         return os.path.isfile(path_info)
@@ -83,12 +92,9 @@ class LocalFileSystem(BaseFileSystem):
         makedirs(path_info, exist_ok=kwargs.pop("exist_ok", True))
 
     def isexec(self, path_info):
-        mode = self.stat(path_info).st_mode
+        mode = self.info(path_info)["mode"]
         return is_exec(mode)
 
-    def stat(self, path):
-        return os.stat(path)
-
     def move(self, from_info, to_info):
         if (
             isinstance(from_info, PathInfo) and from_info.scheme != "local"
@@ -107,7 +113,7 @@ class LocalFileSystem(BaseFileSystem):
             self.remove(tmp_info)
             raise
 
-    def _upload_fobj(self, fobj, to_info, **kwargs):
+    def upload_fobj(self, fobj, to_info, **kwargs):
         self.makedirs(to_info.parent)
         tmp_info = to_info.parent / tmp_fname("")
         try:
@@ -154,29 +160,18 @@ class LocalFileSystem(BaseFileSystem):
     def reflink(self, from_info, to_info):
         System.reflink(from_info, to_info)
 
-    @staticmethod
-    def info(path_info):
-        st = os.stat(path_info)
-        return {
-            "size": st.st_size,
-            "type": "directory" if stat.S_ISDIR(st.st_mode) else "file",
-        }
+    def info(self, path_info):
+        return self.fs.info(path_info)
 
-    def _upload(
-        self, from_file, to_info, name=None, no_progress_bar=False, **_kwargs
+    def put_file(
+        self, from_file, to_info, callback=DEFAULT_CALLBACK, **kwargs
     ):
         makedirs(to_info.parent, exist_ok=True)
-
         tmp_file = tmp_fname(to_info)
-        copyfile(
-            from_file, tmp_file, name=name, no_progress_bar=no_progress_bar
-        )
+        copyfile(from_file, tmp_file, callback=callback)
         os.replace(tmp_file, to_info)
 
-    @staticmethod
-    def _download(
-        from_info, to_file, name=None, no_progress_bar=False, **_kwargs
+    def get_file(
+        self, from_info, to_file, callback=DEFAULT_CALLBACK, **kwargs
     ):
-        copyfile(
-            from_info, to_file, no_progress_bar=no_progress_bar, name=name
-        )
+        copyfile(from_info, to_file, callback=callback)

dvc/fs/oss.py

@@ -7,13 +7,13 @@ from funcy import cached_property, wrap_prop
 from dvc.path_info import CloudURLInfo
 from dvc.scheme import Schemes
 
-from .fsspec_wrapper import CallbackMixin, ObjectFSWrapper
+from .fsspec_wrapper import ObjectFSWrapper
 
 logger = logging.getLogger(__name__)
 
 
 # pylint:disable=abstract-method
-class OSSFileSystem(CallbackMixin, ObjectFSWrapper):
+class OSSFileSystem(ObjectFSWrapper):
     scheme = Schemes.OSS
     PATH_CLS = CloudURLInfo
     REQUIRES = {"ossfs": "ossfs"}

dvc/fs/repo.py

@@ -1,6 +1,5 @@
 import logging
 import os
-import stat
 import threading
 from contextlib import suppress
 from itertools import takewhile
@@ -10,6 +9,7 @@ from funcy import lfilter, wrap_with
 
 from dvc.path_info import PathInfo
 
+from ..progress import DEFAULT_CALLBACK
 from .base import BaseFileSystem
 from .dvc import DvcFileSystem
 
@@ -256,8 +256,8 @@ class RepoFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
             return False
 
         try:
-            st = fs.stat(path)
-            return stat.S_ISDIR(st.st_mode)
+            info = fs.info(path)
+            return info["type"] == "directory"
         except (OSError, ValueError):
             # from CPython's os.path.isdir()
             pass
@@ -287,8 +287,8 @@ class RepoFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
             return False
 
         try:
-            st = fs.stat(path)
-            return stat.S_ISREG(st.st_mode)
+            info = fs.info(path)
+            return info["type"] == "file"
         except (OSError, ValueError):
             # from CPython's os.path.isfile()
             pass
@@ -313,10 +313,6 @@ class RepoFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
             return dvc_fs.isexec(path_info)
         return fs.isexec(path_info)
 
-    def stat(self, path):
-        fs, _ = self._get_fs_pair(path)
-        return fs.stat(path)
-
     def _dvc_walk(self, walk):
         try:
             root, dirs, files = next(walk)
@@ -453,19 +449,21 @@ class RepoFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
             for fname in files:
                 yield PathInfo(root) / fname
 
-    def _download(self, from_info, to_file, **kwargs):
+    def get_file(
+        self, from_info, to_file, callback=DEFAULT_CALLBACK, **kwargs
+    ):
         fs, dvc_fs = self._get_fs_pair(from_info)
         try:
-            fs._download(  # pylint: disable=protected-access
-                from_info, to_file, **kwargs
+            fs.get_file(  # pylint: disable=protected-access
+                from_info, to_file, callback=callback, **kwargs
             )
             return
         except FileNotFoundError:
             if not dvc_fs:
                 raise
 
-        dvc_fs._download(  # pylint: disable=protected-access
-            from_info, to_file, **kwargs
+        dvc_fs.get_file(  # pylint: disable=protected-access
+            from_info, to_file, callback=callback, **kwargs
         )
 
     def metadata(self, path):
@@ -478,11 +476,11 @@ class RepoFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
             with suppress(FileNotFoundError):
                 dvc_meta = dvc_fs.metadata(path_info)
 
-        stat_result = None
+        info_result = None
         with suppress(FileNotFoundError):
-            stat_result = fs.stat(path_info)
+            info_result = fs.info(path_info)
 
-        if not stat_result and not dvc_meta:
+        if not info_result and not dvc_meta:
             raise FileNotFoundError
 
         from ._metadata import Metadata
@@ -492,13 +490,13 @@ class RepoFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
             repo=self._get_repo(abspath) or self._main_repo,
         )
 
-        isdir = bool(stat_result) and stat.S_ISDIR(stat_result.st_mode)
+        isdir = bool(info_result) and info_result["type"] == "directory"
         meta.isdir = meta.isdir or isdir
 
         if not dvc_meta:
             from dvc.utils import is_exec
 
-            meta.is_exec = bool(stat_result) and is_exec(stat_result.st_mode)
+            meta.is_exec = bool(info_result) and is_exec(info_result["mode"])
         return meta
 
     def info(self, path_info):
@@ -508,3 +506,11 @@ class RepoFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
             return fs.info(path_info)
         except FileNotFoundError:
             return dvc_fs.info(path_info)
+
+    def checksum(self, path_info):
+        fs, dvc_fs = self._get_fs_pair(path_info)
+
+        try:
+            return fs.checksum(path_info)
+        except FileNotFoundError:
+            return dvc_fs.checksum(path_info)

dvc/fs/s3.py

@@ -6,7 +6,7 @@ from collections import defaultdict
 from funcy import cached_property, wrap_prop
 
 from dvc.path_info import CloudURLInfo
-from dvc.progress import Tqdm
+from dvc.progress import DEFAULT_CALLBACK
 from dvc.scheme import Schemes
 
 from .fsspec_wrapper import ObjectFSWrapper
@@ -68,12 +68,12 @@ class BaseS3FileSystem(ObjectFSWrapper):
         self._transfer_config = None
         config_path = os.environ.get("AWS_CONFIG_FILE", _AWS_CONFIG_PATH)
         if not os.path.exists(config_path):
-            return None
+            return {}
 
         config = load_config(config_path)
         profile_config = config["profiles"].get(profile or "default")
         if not profile_config:
-            return None
+            return {}
 
         s3_config = profile_config.get("s3", {})
         return self._split_s3_config(s3_config)
@@ -99,6 +99,11 @@ class BaseS3FileSystem(ObjectFSWrapper):
         client["endpoint_url"] = config.get("endpointurl")
         client["verify"] = config.get("ssl_verify")
 
+        # timeout configuration
+        config_kwargs = login_info["config_kwargs"]
+        config_kwargs["read_timeout"] = config.get("read_timeout")
+        config_kwargs["connect_timeout"] = config.get("connect_timeout")
+
         # encryptions
         additional = login_info["s3_additional_kwargs"]
         additional["ServerSideEncryption"] = config.get("sse")
@@ -123,6 +128,14 @@ class BaseS3FileSystem(ObjectFSWrapper):
         if shared_creds:
             os.environ.setdefault("AWS_SHARED_CREDENTIALS_FILE", shared_creds)
 
+        if (
+            client["region_name"] is None
+            and session_config["s3"].get("region_name") is None
+            and os.getenv("AWS_REGION") is None
+        ):
+            # Enable bucket region caching
+            login_info["cache_regions"] = config.get("cache_regions", True)
+
         config_path = config.get("configpath")
         if config_path:
             os.environ.setdefault("AWS_CONFIG_FILE", config_path)
@@ -196,36 +209,23 @@ class S3FileSystem(BaseS3FileSystem):  # pylint:disable=abstract-method
         return bucket.Object(path_info.path)
 
     @_translate_exceptions
-    def _upload(
-        self, from_file, to_info, name=None, no_progress_bar=False, **pbar_args
+    def put_file(
+        self, from_file, to_info, callback=DEFAULT_CALLBACK, **kwargs
     ):
-        total = os.path.getsize(from_file)
-        with Tqdm(
-            disable=no_progress_bar,
-            total=total,
-            bytes=True,
-            desc=name,
-            **pbar_args,
-        ) as pbar:
-            obj = self._get_obj(to_info)
-            obj.upload_file(
-                from_file,
-                Callback=pbar.update,
-                ExtraArgs=self.fs_args.get("s3_additional_kwargs"),
-                Config=self._transfer_config,
-            )
+        callback.set_size(os.path.getsize(from_file))
+        obj = self._get_obj(to_info)
+        obj.upload_file(
+            from_file,
+            Callback=callback.relative_update,
+            ExtraArgs=self.fs_args.get("s3_additional_kwargs"),
+            Config=self._transfer_config,
+        )
         self.fs.invalidate_cache(self._with_bucket(to_info.parent))
 
     @_translate_exceptions
-    def _download(
-        self, from_info, to_file, name=None, no_progress_bar=False, **pbar_args
+    def get_file(
+        self, from_info, to_info, callback=DEFAULT_CALLBACK, **kwargs
     ):
         obj = self._get_obj(from_info)
-        with Tqdm(
-            disable=no_progress_bar,
-            total=obj.content_length,
-            bytes=True,
-            desc=name,
-            **pbar_args,
-        ) as pbar:
-            obj.download_file(to_file, Callback=pbar.update)
+        callback.set_size(obj.content_length)
+        obj.download_file(to_info, Callback=callback.relative_update)

dvc/fs/ssh.py

@@ -2,13 +2,14 @@ import getpass
 import os.path
 import threading
 
-from funcy import cached_property, first, memoize, silent, wrap_prop, wrap_with
+from funcy import cached_property, memoize, silent, wrap_prop, wrap_with
 
 from dvc import prompt
 from dvc.scheme import Schemes
 from dvc.utils.fs import as_atomic
 
-from .fsspec_wrapper import CallbackMixin, FSSpecWrapper
+from ..progress import DEFAULT_CALLBACK
+from .fsspec_wrapper import FSSpecWrapper
 
 _SSH_TIMEOUT = 60 * 30
 _SSH_CONFIG_FILE = os.path.expanduser(os.path.join("~", ".ssh", "config"))
@@ -26,7 +27,7 @@ def ask_password(host, user, port):
 
 
 # pylint:disable=abstract-method
-class SSHFileSystem(CallbackMixin, FSSpecWrapper):
+class SSHFileSystem(FSSpecWrapper):
     scheme = Schemes.SSH
     REQUIRES = {"sshfs": "sshfs"}
 
@@ -35,10 +36,8 @@ class SSHFileSystem(CallbackMixin, FSSpecWrapper):
 
     def _with_bucket(self, path):
         if isinstance(path, self.PATH_CLS):
-            path_str = path.path
-        else:
-            path_str = super()._with_bucket(path)
-        return os.path.normpath(path_str)
+            return path.path
+        return super()._with_bucket(path)
 
     def _prepare_credentials(self, **config):
         self.CAN_TRAVERSE = True
@@ -66,16 +65,25 @@ class SSHFileSystem(CallbackMixin, FSSpecWrapper):
             or self.DEFAULT_PORT
         )
 
-        login_info["password"] = config.get("password")
-
-        if user_ssh_config.get("IdentityFile"):
-            config.setdefault(
-                "keyfile", first(user_ssh_config.get("IdentityFile"))
+        if config.get("ask_password") and config.get("password") is None:
+            config["password"] = ask_password(
+                login_info["host"], login_info["username"], login_info["port"]
             )
 
-        keyfile = config.get("keyfile")
-        if keyfile:
-            login_info["client_keys"] = [os.path.expanduser(keyfile)]
+        login_info["password"] = config.get("password")
+        login_info["passphrase"] = config.get("password")
+
+        raw_keys = []
+        if config.get("keyfile"):
+            raw_keys.append(config.get("keyfile"))
+        elif user_ssh_config.get("IdentityFile"):
+            raw_keys.extend(user_ssh_config.get("IdentityFile"))
+
+        if raw_keys:
+            login_info["client_keys"] = [
+                os.path.expanduser(key) for key in raw_keys
+            ]
+
         login_info["timeout"] = config.get("timeout", _SSH_TIMEOUT)
 
         # These two settings fine tune the asyncssh to use the
@@ -96,11 +104,6 @@ class SSHFileSystem(CallbackMixin, FSSpecWrapper):
         login_info["agent_forwarding"] = config.get("agent_forwarding", True)
         login_info["proxy_command"] = user_ssh_config.get("ProxyCommand")
 
-        if config.get("ask_password") and login_info["password"] is None:
-            login_info["password"] = ask_password(
-                login_info["host"], login_info["username"], login_info["port"]
-            )
-
         # We are going to automatically add stuff to known_hosts
         # something like paramiko's AutoAddPolicy()
         login_info["known_hosts"] = None
@@ -116,10 +119,12 @@ class SSHFileSystem(CallbackMixin, FSSpecWrapper):
     # Ensure that if an interrupt happens during the transfer, we don't
     # pollute the cache.
 
-    def _upload_fobj(self, fobj, to_info, *args, **kwargs):
+    def upload_fobj(self, fobj, to_info, **kwargs):
         with as_atomic(self, to_info) as tmp_file:
-            super()._upload_fobj(fobj, tmp_file, *args, **kwargs)
+            super().upload_fobj(fobj, tmp_file, **kwargs)
 
-    def _upload(self, from_file, to_info, *args, **kwargs):
+    def put_file(
+        self, from_file, to_info, callback=DEFAULT_CALLBACK, **kwargs
+    ):
         with as_atomic(self, to_info) as tmp_file:
-            super()._upload(from_file, tmp_file, *args, **kwargs)
+            super().put_file(from_file, tmp_file, callback=callback, **kwargs)

dvc/fs/utils.py

@@ -0,0 +1,87 @@
+import errno
+import logging
+import os
+from typing import TYPE_CHECKING
+
+from .base import RemoteActionNotImplemented
+from .local import LocalFileSystem
+
+if TYPE_CHECKING:
+    from dvc.types import AnyPath, DvcPath
+
+    from .base import BaseFileSystem
+
+logger = logging.getLogger(__name__)
+
+
+def _link(
+    from_fs: "BaseFileSystem",
+    from_info: "AnyPath",
+    to_fs: "BaseFileSystem",
+    to_info: "DvcPath",
+    hardlink: bool = False,
+) -> None:
+    if not isinstance(from_fs, type(to_fs)):
+        raise OSError(errno.EXDEV, "can't link across filesystems")
+
+    links = ["reflink"] + ["hardlink"] if hardlink else []
+    while links:
+        link = links.pop(0)
+        try:
+            func = getattr(to_fs, link)
+        except AttributeError:
+            continue
+
+        try:
+            return func(from_info, to_info)
+        except RemoteActionNotImplemented:
+            continue
+        except OSError as exc:
+            if exc.errno not in [errno.EXDEV, errno.ENOTSUP]:
+                raise
+
+    raise OSError(errno.ENOTSUP, "reflink and hardlink are not supported")
+
+
+def transfer(
+    from_fs: "BaseFileSystem",
+    from_info: "AnyPath",
+    to_fs: "BaseFileSystem",
+    to_info: "DvcPath",
+    hardlink: bool = False,
+) -> None:
+    try:
+        try:
+            return _link(from_fs, from_info, to_fs, to_info, hardlink=hardlink)
+        except OSError as exc:
+            if exc.errno not in [errno.EXDEV, errno.ENOTSUP]:
+                raise
+
+        if isinstance(from_fs, LocalFileSystem):
+            if not isinstance(from_info, from_fs.PATH_CLS):
+                from_info = from_fs.PATH_CLS(from_info)
+            return to_fs.upload(from_info, to_info)
+
+        if isinstance(to_fs, LocalFileSystem):
+            return from_fs.download_file(from_info, to_info)
+
+        with from_fs.open(from_info, mode="rb") as fobj:
+            size = from_fs.getsize(from_info)
+            return to_fs.upload(fobj, to_info, total=size)
+    except OSError as exc:
+        # If the target file already exists, we are going to simply
+        # ignore the exception (#4992).
+        #
+        # On Windows, it is not always guaranteed that you'll get
+        # FileExistsError (it might be PermissionError or a bare OSError)
+        # but all of those exceptions raised from the original
+        # FileExistsError so we have a separate check for that.
+        if isinstance(exc, FileExistsError) or (
+            os.name == "nt"
+            and exc.__context__
+            and isinstance(exc.__context__, FileExistsError)
+        ):
+            logger.debug("'%s' file already exists, skipping", to_info)
+            return None
+
+        raise

dvc/fs/webdav.py

@@ -1,5 +1,4 @@
 import logging
-import os
 import threading
 from functools import lru_cache
 
@@ -71,15 +70,13 @@ class WebDAVFileSystem(FSSpecWrapper):  # pylint:disable=abstract-method
 
         return WebdavFileSystem(**self.fs_args)
 
-    def _upload_fobj(self, fobj, to_info, size: int = None):
+    def upload_fobj(self, fobj, to_info, **kwargs):
         rpath = self.translate_path_info(to_info)
-        self.makedirs(os.path.dirname(rpath))
+        size = kwargs.get("size")
         # using upload_fileobj to directly upload fileobj rather than buffering
         # and using overwrite=True to avoid check for an extra exists call,
         # as caller should ensure that the file does not exist beforehand.
-        return self.fs.client.upload_fileobj(
-            fobj, rpath, overwrite=True, size=size
-        )
+        return self.fs.upload_fileobj(fobj, rpath, overwrite=True, size=size)
 
     @lru_cache(512)
     def translate_path_info(self, path):
@@ -87,7 +84,7 @@ class WebDAVFileSystem(FSSpecWrapper):  # pylint:disable=abstract-method
             return path.path[len(self.prefix) :].lstrip("/")
         return path
 
-    _with_bucket = translate_path_info
+    _with_bucket = translate_path_info  # type: ignore
 
 
 class WebDAVSFileSystem(WebDAVFileSystem):  # pylint:disable=abstract-method

dvc/fs/webhdfs.py

@@ -1,51 +1,24 @@
-import logging
-import os
-import posixpath
-import shutil
 import threading
-from contextlib import contextmanager
 
 from funcy import cached_property, wrap_prop
 
-from dvc.hash_info import HashInfo
 from dvc.path_info import CloudURLInfo
-from dvc.progress import Tqdm
 from dvc.scheme import Schemes
 
-from .base import BaseFileSystem
-
-logger = logging.getLogger(__name__)
+# pylint:disable=abstract-method
+from .fsspec_wrapper import CallbackMixin, FSSpecWrapper
 
 
-def update_pbar(pbar, total):
-    """Update pbar to accept the two arguments passed by hdfs"""
-
-    def update(_, bytes_transfered):
-        if bytes_transfered == -1:
-            pbar.update_to(total)
-            return
-        pbar.update_to(bytes_transfered)
-
-    return update
-
-
-class WebHDFSFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
+class WebHDFSFileSystem(CallbackMixin, FSSpecWrapper):
     scheme = Schemes.WEBHDFS
     PATH_CLS = CloudURLInfo
-    REQUIRES = {"hdfs": "hdfs"}
+    REQUIRES = {"fsspec": "fsspec"}
     PARAM_CHECKSUM = "checksum"
-    TRAVERSE_PREFIX_LEN = 2
 
-    def __init__(self, **config):
-        super().__init__(**config)
-
-        self.host = config["host"]
-        self.user = config.get("user")
-        self.port = config.get("port")
-
-        self.hdfscli_config = config.get("hdfscli_config")
-        self.token = config.get("webhdfs_token")
-        self.alias = config.get("webhdfs_alias")
+    def _with_bucket(self, path):
+        if isinstance(path, self.PATH_CLS):
+            return f"/{path.path.rstrip('/')}"
+        return path
 
     @staticmethod
     def _get_kwargs_from_urls(urlpath):
@@ -57,115 +30,23 @@ class WebHDFSFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
             )
         )
 
+    def _prepare_credentials(self, **config):
+        self._ssl_verify = config.pop("ssl_verify", True)
+        principal = config.pop("kerberos_principal", None)
+        if principal:
+            config["kerb_kwargs"] = {"principal": principal}
+        return config
+
     @wrap_prop(threading.Lock())
     @cached_property
-    def hdfs_client(self):
-        import hdfs
+    def fs(self):
+        from fsspec.implementations.webhdfs import WebHDFS
 
-        logger.debug("HDFSConfig: %s", self.hdfscli_config)
-
-        try:
-            return hdfs.config.Config(self.hdfscli_config).get_client(
-                self.alias
-            )
-        except hdfs.util.HdfsError as exc:
-            exc_msg = str(exc)
-            errors = (
-                "No alias specified",
-                "Invalid configuration file",
-                f"Alias {self.alias} not found",
-            )
-            if not any(err in exc_msg for err in errors):
-                raise
-
-            http_url = f"http://{self.host}:{self.port}"
-            logger.debug("URL: %s", http_url)
-
-            if self.token is not None:
-                client = hdfs.TokenClient(http_url, token=self.token, root="/")
-            else:
-                client = hdfs.InsecureClient(
-                    http_url, user=self.user, root="/"
-                )
-
-        return client
-
-    @contextmanager
-    def open(self, path_info, mode="r", encoding=None, **kwargs):
-        assert mode in {"r", "rt", "rb"}
-
-        with self.hdfs_client.read(
-            path_info.path, encoding=encoding
-        ) as reader:
-            yield reader
-
-    def walk_files(self, path_info, **kwargs):
-        if not self.exists(path_info):
-            return
-
-        root = path_info.path
-        for path, _, fnames in self.hdfs_client.walk(root):
-            for fname in fnames:
-                yield path_info.replace(path=posixpath.join(path, fname))
-
-    def remove(self, path_info):
-        if path_info.scheme != self.scheme:
-            raise NotImplementedError
-
-        self.hdfs_client.delete(path_info.path)
-
-    def exists(self, path_info) -> bool:
-        assert not isinstance(path_info, list)
-        assert path_info.scheme == "webhdfs"
-
-        status = self.hdfs_client.status(path_info.path, strict=False)
-        return status is not None
-
-    def info(self, path_info):
-        st = self.hdfs_client.status(path_info.path)
-        return {"size": st["length"], "type": "file"}
+        fs = WebHDFS(**self.fs_args)
+        fs.session.verify = self._ssl_verify
+        return fs
 
     def checksum(self, path_info):
-        return HashInfo(
-            "checksum",
-            self.hdfs_client.checksum(path_info.path)["bytes"],
-            size=self.hdfs_client.status(path_info.path)["length"],
-        )
-
-    def copy(self, from_info, to_info, **_kwargs):
-        with self.hdfs_client.read(from_info.path) as reader:
-            with self.hdfs_client.write(to_info.path) as writer:
-                shutil.copyfileobj(reader, writer)
-
-    def move(self, from_info, to_info):
-        self.hdfs_client.makedirs(to_info.parent.path)
-        self.hdfs_client.rename(from_info.path, to_info.path)
-
-    def _upload_fobj(self, fobj, to_info, **kwargs):
-        with self.hdfs_client.write(to_info.path) as fdest:
-            shutil.copyfileobj(fobj, fdest)
-
-    def _upload(
-        self, from_file, to_info, name=None, no_progress_bar=False, **_kwargs
-    ):
-        total = os.path.getsize(from_file)
-        with Tqdm(
-            desc=name, total=total, disable=no_progress_bar, bytes=True
-        ) as pbar:
-            self.hdfs_client.upload(
-                to_info.path,
-                from_file,
-                overwrite=True,
-                progress=update_pbar(pbar, total),
-            )
-
-    def _download(
-        self, from_info, to_file, name=None, no_progress_bar=False, **_kwargs
-    ):
-        total = self.getsize(from_info)
-        with Tqdm(
-            desc=name, total=total, disable=no_progress_bar, bytes=True
-        ) as pbar:
-            self.hdfs_client.download(
-                from_info.path, to_file, progress=update_pbar(pbar, total)
-            )
+        path = self._with_bucket(path_info)
+        ukey = self.fs.ukey(path)
+        return ukey["bytes"]

dvc/hash_info.py

@@ -7,13 +7,8 @@ HASH_DIR_SUFFIX = ".dir"
 
 @dataclass
 class HashInfo:
-    PARAM_SIZE = "size"
-    PARAM_NFILES = "nfiles"
-
     name: Optional[str]
     value: Optional[str]
-    size: Optional[int] = field(default=None, compare=False)
-    nfiles: Optional[int] = field(default=None, compare=False)
     obj_name: Optional[str] = field(default=None, compare=False)
 
     def __bool__(self):
@@ -27,15 +22,11 @@ class HashInfo:
 
     @classmethod
     def from_dict(cls, d):
-        _d = d.copy() if d else {}
-        size = _d.pop(cls.PARAM_SIZE, None)
-        nfiles = _d.pop(cls.PARAM_NFILES, None)
-
-        if not _d:
+        if not d:
             return cls(None, None)
 
-        ((name, value),) = _d.items()
-        return cls(name, value, size=size, nfiles=nfiles)
+        ((name, value),) = d.items()
+        return cls(name, value)
 
     def to_dict(self):
         ret = OrderedDict()
@@ -43,10 +34,6 @@ class HashInfo:
             return ret
 
         ret[self.name] = self.value
-        if self.size is not None:
-            ret[self.PARAM_SIZE] = self.size
-        if self.nfiles is not None:
-            ret[self.PARAM_NFILES] = self.nfiles
         return ret
 
     @property

dvc/ignore.py

@@ -11,7 +11,6 @@ from dvc.fs.base import BaseFileSystem
 from dvc.path_info import PathInfo
 from dvc.pathspec_math import PatternInfo, merge_patterns
 from dvc.scheme import Schemes
-from dvc.system import System
 from dvc.types import AnyPath, List, Optional
 from dvc.utils import relpath
 from dvc.utils.collections import PathStringTrie
@@ -84,7 +83,7 @@ class DvcIgnorePatterns(DvcIgnore):
         else:
             return False
 
-        if not System.is_unix():
+        if os.name == "nt":
             path = normalize_file(path)
         return path
 
@@ -394,7 +393,7 @@ def init(path):
     if os.path.exists(dvcignore):
         return dvcignore
 
-    with open(dvcignore, "w") as fobj:
+    with open(dvcignore, "w", encoding="utf-8") as fobj:
         fobj.write(
             "# Add patterns of files dvc should ignore, which could improve\n"
             "# the performance. Learn more at\n"

dvc/info.py

@@ -1,3 +1,4 @@
+import errno
 import itertools
 import os
 import pathlib
@@ -6,19 +7,14 @@ import uuid
 
 import psutil
 
-from dvc.exceptions import DvcException, NotDvcRepoError
+from dvc import __version__
+from dvc.exceptions import NotDvcRepoError
 from dvc.fs import FS_MAP, get_fs_cls, get_fs_config
 from dvc.repo import Repo
 from dvc.scm.base import SCMError
 from dvc.system import System
 from dvc.utils import error_link
 from dvc.utils.pkg import PKG
-from dvc.version import __version__
-
-if PKG is None:
-    package = ""
-else:
-    package = f"({PKG})"
 
 try:
     import importlib.metadata as importlib_metadata
@@ -26,6 +22,9 @@ except ImportError:  # < 3.8
     import importlib_metadata  # type: ignore[no-redef]
 
 
+package = "" if PKG is None else f"({PKG})"
+
+
 def get_dvc_info():
     info = [
         f"DVC version: {__version__} {package}",
@@ -93,7 +92,7 @@ def _get_linktype_support_info(repo):
 
     fname = "." + str(uuid.uuid4())
     src = os.path.join(repo.odb.local.cache_dir, fname)
-    open(src, "w").close()
+    open(src, "w", encoding="utf-8").close()
     dst = os.path.join(repo.root_dir, fname)
 
     cache = []
@@ -105,7 +104,9 @@ def _get_linktype_support_info(repo):
             if is_link and not is_link(dst):
                 status = "broken"
             os.unlink(dst)
-        except DvcException:
+        except OSError as exc:
+            if exc.errno != errno.ENOTSUP:
+                raise
             status = "not supported"
 
         if status == "supported":
@@ -136,11 +137,14 @@ def _get_supported_remotes():
 
 
 def get_fs_type(path):
-
-    partition = {
-        pathlib.Path(part.mountpoint): (part.fstype + " on " + part.device)
-        for part in psutil.disk_partitions(all=True)
-    }
+    partition = {}
+    for part in psutil.disk_partitions(all=True):
+        if part.fstype != "":
+            try:
+                mountpoint = pathlib.Path(part.mountpoint).resolve()
+                partition[mountpoint] = part.fstype + " on " + part.device
+            except PermissionError:
+                pass
 
     # need to follow the symlink: https://github.com/iterative/dvc/issues/5065
     path = pathlib.Path(path).resolve()

dvc/lock.py

@@ -176,25 +176,14 @@ class HardlinkLock(flufl.lock.Lock, LockBase):
         except flufl.lock.TimeOutError:
             raise LockError(FAILED_TO_LOCK_MESSAGE)
 
-    def _set_claimfile(self, pid=None):
-        super()._set_claimfile(pid)
+    def _set_claimfile(self):
+        super()._set_claimfile()
 
         if self._tmp_dir is not None:
             # Under Windows file path length is limited so we hash it
             filename = hashlib.md5(self._claimfile.encode()).hexdigest()
             self._claimfile = os.path.join(self._tmp_dir, filename + ".lock")
 
-    # Fix for __del__ bug in flufl.lock [1] which is causing errors on
-    # Python shutdown [2].
-    # [1] https://gitlab.com/warsaw/flufl.lock/issues/7
-    # [2] https://github.com/iterative/dvc/issues/2573
-    def __del__(self):
-        try:
-            if self._owned:
-                self.finalize()
-        except ImportError:
-            pass
-
 
 def make_lock(lockfile, tmp_dir=None, friendly=False, hardlink_lock=False):
     cls = HardlinkLock if hardlink_lock else Lock

dvc/logger.py

@@ -142,9 +142,22 @@ class LoggerHandler(logging.StreamHandler):
         super().handleError(record)
         raise LoggingException(record)
 
+    def emit_pretty_exception(self, exc, verbose: bool = False):
+        return exc.__pretty_exc__(verbose=verbose)
+
     def emit(self, record):
         """Write to Tqdm's stream so as to not break progress-bars"""
         try:
+            if record.exc_info:
+                _, exc, *_ = record.exc_info
+                if hasattr(exc, "__pretty_exc__"):
+                    try:
+                        self.emit_pretty_exception(exc, verbose=_is_verbose())
+                        if not _is_verbose():
+                            return
+                    except Exception:  # noqa, pylint: disable=broad-except
+                        pass
+
             msg = self.format(record)
             Tqdm.write(
                 msg, file=self.stream, end=getattr(self, "terminator", "\n")

dvc/machine/__init__.py

@@ -0,0 +1,217 @@
+import logging
+import os
+from typing import (
+    TYPE_CHECKING,
+    Any,
+    Dict,
+    Iterable,
+    Iterator,
+    Mapping,
+    Optional,
+    Tuple,
+    Type,
+)
+
+from dvc.exceptions import DvcException
+from dvc.types import StrPath
+
+if TYPE_CHECKING:
+    from dvc.repo import Repo
+
+    from .backend.base import BaseMachineBackend
+
+    BackendCls = Type[BaseMachineBackend]
+
+logger = logging.getLogger(__name__)
+
+
+RESERVED_NAMES = {"local", "localhost"}
+
+DEFAULT_STARTUP_SCRIPT = """#!/bin/bash
+sudo apt-get update
+sudo apt-get install --yes python3 python3-pip
+python3 -m pip install --upgrade pip
+
+pushd /etc/apt/sources.list.d
+sudo wget https://dvc.org/deb/dvc.list
+sudo apt-get update
+sudo apt-get install --yes dvc
+popd
+"""
+
+
+def validate_name(name: str):
+    from dvc.exceptions import InvalidArgumentError
+
+    name = name.lower()
+    if name in RESERVED_NAMES:
+        raise InvalidArgumentError(
+            f"Machine name '{name}' is reserved for internal DVC use."
+        )
+
+
+class MachineBackends(Mapping):
+    try:
+        from .backend.terraform import TerraformBackend
+    except ImportError:
+        TerraformBackend = None  # type: ignore[assignment, misc]
+
+    DEFAULT: Dict[str, Optional["BackendCls"]] = {
+        "terraform": TerraformBackend,
+    }
+
+    def __getitem__(self, key: str) -> "BaseMachineBackend":
+        """Lazily initialize backends and cache it afterwards"""
+        initialized = self.initialized.get(key)
+        if not initialized:
+            backend = self.backends[key]
+            initialized = backend(
+                os.path.join(self.tmp_dir, key), **self.kwargs
+            )
+            self.initialized[key] = initialized
+        return initialized
+
+    def __init__(
+        self,
+        selected: Optional[Iterable[str]],
+        tmp_dir: StrPath,
+        **kwargs,
+    ) -> None:
+        selected = selected or list(self.DEFAULT)
+        self.backends: Dict[str, "BackendCls"] = {}
+        for key in selected:
+            cls = self.DEFAULT.get(key)
+            if cls is None:
+                raise DvcException(
+                    f"'dvc machine' backend '{key}' is missing required "
+                    "dependencies. Install them with:\n"
+                    f"\tpip install dvc[{key}]"
+                )
+            self.backends[key] = cls
+
+        self.initialized: Dict[str, "BaseMachineBackend"] = {}
+
+        self.tmp_dir = tmp_dir
+        self.kwargs = kwargs
+
+    def __iter__(self):
+        return iter(self.backends)
+
+    def __len__(self) -> int:
+        return len(self.backends)
+
+    def close_initialized(self) -> None:
+        for backend in self.initialized.values():
+            backend.close()
+
+
+class MachineManager:
+    """Class that manages dvc cloud machines.
+
+    Args:
+        repo (dvc.repo.Repo): repo instance that belongs to the repo that
+            we are working on.
+
+    Raises:
+        config.ConfigError: thrown when config has invalid format.
+    """
+
+    CLOUD_BACKENDS = {
+        "aws": "terraform",
+        "azure": "terraform",
+    }
+
+    def __init__(
+        self, repo: "Repo", backends: Optional[Iterable[str]] = None, **kwargs
+    ):
+        self.repo = repo
+        tmp_dir = os.path.join(self.repo.tmp_dir, "machine")
+        self.backends = MachineBackends(backends, tmp_dir=tmp_dir, **kwargs)
+
+    def get_config_and_backend(
+        self,
+        name: Optional[str] = None,
+    ) -> Tuple[dict, "BaseMachineBackend"]:
+        from dvc.config import NoMachineError
+
+        if not name:
+            name = self.repo.config["core"].get("machine")
+
+        if name:
+            config = self._get_config(name=name)
+            backend = self._get_backend(config["cloud"])
+            return config, backend
+
+        if bool(self.repo.config["machine"]):
+            error_msg = (
+                "no machine specified. Setup default machine with\n"
+                "    dvc machine default <name>\n"
+            )
+        else:
+            error_msg = (
+                "no machine specified. Create a default machine with\n"
+                "    dvc machine add -d <name> <cloud>"
+            )
+
+        raise NoMachineError(error_msg)
+
+    def _get_config(self, **kwargs):
+        config = self.repo.config
+        name = kwargs.get("name")
+        if name:
+            try:
+                conf = config["machine"][name.lower()]
+                conf["name"] = name
+            except KeyError:
+                from dvc.config import MachineNotFoundError
+
+                raise MachineNotFoundError(f"machine '{name}' doesn't exist")
+        else:
+            conf = kwargs
+        return conf
+
+    def _get_backend(self, cloud: str) -> "BaseMachineBackend":
+        from dvc.config import NoMachineError
+
+        try:
+            backend = self.CLOUD_BACKENDS[cloud]
+            return self.backends[backend]
+        except KeyError:
+            raise NoMachineError(f"Machine platform '{cloud}' unsupported")
+
+    def create(self, name: Optional[str]):
+        """Create and start the specified machine instance."""
+        config, backend = self.get_config_and_backend(name)
+        if "startup_script" in config:
+            with open(config["startup_script"], encoding="utf-8") as fobj:
+                startup_script = fobj.read()
+        else:
+            startup_script = DEFAULT_STARTUP_SCRIPT
+        config["startup_script"] = startup_script
+        return backend.create(**config)
+
+    def destroy(self, name: Optional[str]):
+        """Destroy the specified machine instance."""
+        config, backend = self.get_config_and_backend(name)
+        return backend.destroy(**config)
+
+    def get_sshfs(self, name: Optional[str]):
+        config, backend = self.get_config_and_backend(name)
+        return backend.get_sshfs(**config)
+
+    def run_shell(self, name: Optional[str]):
+        config, backend = self.get_config_and_backend(name)
+        return backend.run_shell(**config)
+
+    def status(self, name: str) -> Iterator[Dict[Any, Any]]:
+        config, backend = self.get_config_and_backend(name)
+        return backend.instances(**config)
+
+    def rename(self, name: str, new: str):
+        """move configuration to a new location if the machine is running."""
+        config, backend = self.get_config_and_backend(name)
+        return backend.rename(new=new, **config)
+
+    def get_executor_kwargs(self, name: Optional[str]):
+        config, backend = self.get_config_and_backend(name)
+        return backend.get_executor_kwargs(**config)

dvc/machine/backend/base.py

@@ -0,0 +1,51 @@
+from abc import ABC, abstractmethod
+from contextlib import contextmanager
+from typing import TYPE_CHECKING, Iterator, Optional
+
+if TYPE_CHECKING:
+    from dvc.fs.ssh import SSHFileSystem
+    from dvc.types import StrPath
+
+
+class BaseMachineBackend(ABC):
+    def __init__(self, tmp_dir: "StrPath", **kwargs):
+        self.tmp_dir = tmp_dir
+
+    @abstractmethod
+    def create(self, name: Optional[str] = None, **config):
+        """Create and start an instance of the specified machine."""
+
+    @abstractmethod
+    def destroy(self, name: Optional[str] = None, **config):
+        """Stop and destroy all instances of the specified machine."""
+
+    @abstractmethod
+    def instances(
+        self, name: Optional[str] = None, **config
+    ) -> Iterator[dict]:
+        """Iterate over status of all instances of the specified machine."""
+
+    def close(self):
+        pass
+
+    @abstractmethod
+    def run_shell(self, name: Optional[str] = None, **config):
+        """Spawn an interactive SSH shell for the specified machine."""
+
+    @abstractmethod
+    def get_executor_kwargs(self, name: str, **config) -> dict:
+        """Return SSHExecutor kwargs which can be used for DVC
+        experiment/pipeline execution on the specified machine.
+        """
+
+    @abstractmethod
+    @contextmanager
+    def get_sshfs(
+        self, name: Optional[str] = None, **config
+    ) -> Iterator["SSHFileSystem"]:
+        """Return an sshfs instance for the default directory on the
+        specified machine."""
+
+    @abstractmethod
+    def rename(self, name: str, new: str, **config):
+        """Rename a machine instance."""

dvc/machine/backend/terraform.py

@@ -0,0 +1,103 @@
+import os
+from contextlib import contextmanager
+from functools import partial, partialmethod
+from typing import TYPE_CHECKING, Iterator, Optional
+
+from dvc.exceptions import DvcException
+from dvc.fs.ssh import SSHFileSystem
+from dvc.utils.fs import makedirs
+
+from .base import BaseMachineBackend
+
+if TYPE_CHECKING:
+    from dvc.types import StrPath
+
+
+@contextmanager
+def _sshfs(resource: dict):
+    from tpi import TerraformProviderIterative
+
+    with TerraformProviderIterative.pemfile(resource) as pem:
+        fs = SSHFileSystem(
+            host=resource["instance_ip"],
+            user="ubuntu",
+            keyfile=pem,
+        )
+        yield fs
+
+
+class TerraformBackend(BaseMachineBackend):
+    def __init__(self, tmp_dir: "StrPath", **kwargs):
+        super().__init__(tmp_dir, **kwargs)
+        makedirs(tmp_dir, exist_ok=True)
+
+    @contextmanager
+    def make_tpi(self, name: str):
+        from tpi import TPIError
+        from tpi.terraform import TerraformBackend as TPIBackend
+
+        try:
+            working_dir = os.path.join(self.tmp_dir, name)
+            makedirs(working_dir, exist_ok=True)
+            yield TPIBackend(working_dir=working_dir)
+        except TPIError as exc:
+            raise DvcException("TPI operation failed") from exc
+
+    def _tpi_func(self, fname, name: Optional[str] = None, **config):
+        from tpi import TPIError
+
+        assert name
+        with self.make_tpi(name) as tpi:
+            func = getattr(tpi, fname)
+            try:
+                return func(name=name, **config)
+            except TPIError as exc:
+                raise DvcException(f"TPI {fname} failed") from exc
+
+    create = partialmethod(_tpi_func, "create")  # type: ignore[assignment]
+    destroy = partialmethod(_tpi_func, "destroy")  # type: ignore[assignment]
+    instances = partialmethod(
+        _tpi_func, "instances"
+    )  # type: ignore[assignment]
+    run_shell = partialmethod(
+        _tpi_func, "run_shell"
+    )  # type: ignore[assignment]
+
+    def get_executor_kwargs(self, name: str, **config) -> dict:
+        with self.make_tpi(name) as tpi:
+            resource = tpi.default_resource(name)
+        return {
+            "host": resource["instance_ip"],
+            "port": SSHFileSystem.DEFAULT_PORT,
+            "username": "ubuntu",
+            "fs_factory": partial(_sshfs, dict(resource)),
+        }
+
+    @contextmanager
+    def get_sshfs(  # pylint: disable=unused-argument
+        self, name: Optional[str] = None, **config
+    ) -> Iterator["SSHFileSystem"]:
+        assert name
+        with self.make_tpi(name) as tpi:
+            resource = tpi.default_resource(name)
+        with _sshfs(resource) as fs:
+            yield fs
+
+    def rename(self, name: str, new: str, **config):
+        """rename a dvc machine instance to a new name"""
+        import shutil
+
+        mtype = "iterative_machine"
+
+        new_dir = os.path.join(self.tmp_dir, new)
+        old_dir = os.path.join(self.tmp_dir, name)
+        if os.path.exists(new_dir):
+            raise DvcException(f"rename failed: path {new_dir} already exists")
+
+        if not os.path.exists(old_dir):
+            return
+
+        with self.make_tpi(name) as tpi:
+            tpi.state_mv(f"{mtype}.{name}", f"{mtype}.{new}", **config)
+
+        shutil.move(old_dir, new_dir)

dvc/objects/__init__.py

@@ -14,8 +14,8 @@ logger = logging.getLogger(__name__)
 
 def check(odb: "ObjectDB", obj: "HashFile", **kwargs):
     if isinstance(obj, Tree):
-        for _, entry in obj:
-            odb.check(entry.hash_info, **kwargs)
+        for _, _, oid in obj:
+            odb.check(oid, **kwargs)
 
     odb.check(obj.hash_info, **kwargs)
 

dvc/objects/checkout.py

@@ -1,4 +1,6 @@
+import errno
 import logging
+from itertools import chain
 
 from shortuuid import uuid
 
@@ -10,52 +12,18 @@ from dvc.exceptions import (
     DvcException,
 )
 from dvc.ignore import DvcIgnoreFilter
-from dvc.objects import check, load
 from dvc.objects.db.slow_link_detection import (  # type: ignore[attr-defined]
     slow_link_guard,
 )
-from dvc.objects.errors import ObjectFormatError
+from dvc.objects.diff import ROOT
+from dvc.objects.diff import diff as odiff
 from dvc.objects.stage import stage
 from dvc.types import Optional
 
 logger = logging.getLogger(__name__)
 
 
-def _changed(path_info, fs, obj, cache, state=None):
-    logger.trace("checking if '%s'('%s') has changed.", path_info, obj)
-
-    try:
-        check(cache, obj)
-    except (FileNotFoundError, ObjectFormatError):
-        logger.debug(
-            "cache for '%s'('%s') has changed.", path_info, obj.hash_info
-        )
-        return True
-
-    actual = state.get(path_info, fs) if state else None
-    if actual is None:
-        try:
-            actual = stage(
-                cache, path_info, fs, obj.hash_info.name, dry_run=True
-            )[1].hash_info
-        except FileNotFoundError:
-            logger.debug("'%s' doesn't exist.", path_info)
-            return True
-
-    if obj.hash_info != actual:
-        logger.debug(
-            "hash value '%s' for '%s' has changed (actual '%s').",
-            obj.hash_info,
-            actual,
-            path_info,
-        )
-        return True
-
-    logger.trace("'%s' hasn't changed.", path_info)
-    return False
-
-
-def _remove(path_info, fs, cache, force=False):
+def _remove(path_info, fs, in_cache, force=False):
     if not fs.exists(path_info):
         return
 
@@ -63,13 +31,7 @@ def _remove(path_info, fs, cache, force=False):
         fs.remove(path_info)
         return
 
-    current = stage(cache, path_info, fs, fs.PARAM_CHECKSUM, dry_run=True)[
-        1
-    ].hash_info
-    try:
-        obj = load(cache, current)
-        check(cache, obj)
-    except (FileNotFoundError, ObjectFormatError):
+    if not in_cache:
         msg = (
             f"file/directory '{path_info}' is going to be removed. "
             "Are you sure you want to proceed?"
@@ -98,7 +60,7 @@ def _verify_link(cache, path_info, link_type):
 
 def _do_link(cache, from_info, to_info, link_method):
     if cache.fs.exists(to_info):
-        raise DvcException(f"Link '{to_info}' already exists!")
+        cache.fs.remove(to_info)  # broken symlink
 
     link_method(from_info, to_info)
 
@@ -116,7 +78,9 @@ def _try_links(cache, from_info, to_info, link_types):
             _verify_link(cache, to_info, link_types[0])
             return
 
-        except DvcException as exc:
+        except OSError as exc:
+            if exc.errno not in [errno.EXDEV, errno.ENOTSUP]:
+                raise
             logger.debug(
                 "Cache type '%s' is not supported: %s", link_types[0], exc
             )
@@ -130,16 +94,16 @@ def _link(cache, from_info, to_info):
     try:
         _try_links(cache, from_info, to_info, cache.cache_types)
     except FileNotFoundError as exc:
-        raise CheckoutError([to_info]) from exc
+        raise CheckoutError([str(to_info)]) from exc
 
 
-def _cache_is_copy(cache, path_info):
+def _confirm_cache_type(cache, path_info):
     """Checks whether cache uses copies."""
     if cache.cache_type_confirmed:
-        return cache.cache_types[0] == "copy"
+        return
 
     if set(cache.cache_types) <= {"copy"}:
-        return True
+        return
 
     workspace_file = path_info.with_name("." + uuid())
     test_cache_file = cache.path_info / ".cache_type_test_file"
@@ -154,13 +118,22 @@ def _cache_is_copy(cache, path_info):
         cache.fs.remove(test_cache_file)
 
     cache.cache_type_confirmed = True
-    return cache.cache_types[0] == "copy"
+
+
+def _relink(cache, cache_info, fs, path_info, in_cache, force):
+    _remove(path_info, fs, in_cache, force=force)
+    _link(cache, cache_info, path_info)
+    # NOTE: Depending on a file system (e.g. on NTFS), `_remove` might reset
+    # read-only permissions in order to delete a hardlink to protected object,
+    # which will also reset it for the object itself, making it unprotected,
+    # so we need to protect it back.
+    cache.protect(cache_info)
 
 
 def _checkout_file(
     path_info,
     fs,
-    obj,
+    change,
     cache,
     force,
     progress_callback=None,
@@ -169,24 +142,39 @@ def _checkout_file(
 ):
     """The file is changed we need to checkout a new copy"""
     modified = False
-    cache_info = cache.hash_to_path_info(obj.hash_info.value)
-    if fs.exists(path_info):
-        if not relink and _changed(path_info, fs, obj, cache, state=state):
-            modified = True
-            _remove(path_info, fs, cache, force=force)
-            _link(cache, cache_info, path_info)
-        else:
-            if fs.iscopy(path_info) and _cache_is_copy(cache, path_info):
+
+    _confirm_cache_type(cache, path_info)
+
+    cache_info = cache.hash_to_path_info(change.new.oid.value)
+    if change.old.oid:
+        if relink:
+            if fs.iscopy(path_info) and cache.cache_types[0] == "copy":
                 cache.unprotect(path_info)
             else:
-                _remove(path_info, fs, cache, force=force)
-                _link(cache, cache_info, path_info)
+                _relink(
+                    cache,
+                    cache_info,
+                    fs,
+                    path_info,
+                    change.old.in_cache,
+                    force=force,
+                )
+        else:
+            modified = True
+            _relink(
+                cache,
+                cache_info,
+                fs,
+                path_info,
+                change.old.in_cache,
+                force=force,
+            )
     else:
         _link(cache, cache_info, path_info)
         modified = True
 
     if state:
-        state.save(path_info, fs, obj.hash_info)
+        state.save(path_info, fs, change.new.oid)
 
     if progress_callback:
         progress_callback(str(path_info))
@@ -194,109 +182,83 @@ def _checkout_file(
     return modified
 
 
-def _remove_redundant_files(
-    path_info, fs, obj, cache, force, dvcignore: Optional[DvcIgnoreFilter]
-):
-    if dvcignore:
-        existing_files = set(dvcignore.walk_files(fs, path_info))
-    else:
-        existing_files = set(fs.walk_files(path_info))
-
-    needed_files = {path_info.joinpath(*key) for key, _ in obj}
-    redundant_files = existing_files - needed_files
-    for path in redundant_files:
-        _remove(path, fs, cache, force)
-
-    return bool(redundant_files)
-
-
-def _checkout_dir(
+def _diff(
     path_info,
     fs,
     obj,
     cache,
-    force,
-    progress_callback=None,
     relink=False,
     dvcignore: Optional[DvcIgnoreFilter] = None,
-    state=None,
 ):
-    modified = False
-    # Create dir separately so that dir is created
-    # even if there are no files in it
-    if not fs.exists(path_info):
-        modified = True
-        fs.makedirs(path_info)
-
-    logger.debug("Linking directory '%s'.", path_info)
-
-    failed = []
-    for entry_key, entry_obj in obj:
-        try:
-            entry_modified = _checkout_file(
-                path_info.joinpath(*entry_key),
-                fs,
-                entry_obj,
-                cache,
-                force,
-                progress_callback,
-                relink,
-                state=None,
-            )
-            if entry_modified:
-                modified = True
-        except CheckoutError as exc:
-            failed.extend(exc.target_infos)
-
-    modified = (
-        _remove_redundant_files(
-            path_info, fs, obj, cache, force, dvcignore=dvcignore
+    old = None
+    try:
+        _, _, old = stage(
+            cache,
+            path_info,
+            fs,
+            obj.hash_info.name if obj else cache.fs.PARAM_CHECKSUM,
+            dry_run=True,
+            dvcignore=dvcignore,
         )
-        or modified
-    )
+    except FileNotFoundError:
+        pass
 
-    if failed:
-        raise CheckoutError(failed)
+    diff = odiff(old, obj, cache)
 
-    if state:
-        state.save(path_info, fs, obj.hash_info)
+    if relink:
+        diff.modified.extend(diff.unchanged)
 
-    # relink is not modified, assume it as nochange
-    return modified and not relink
+    return diff
 
 
 def _checkout(
+    diff,
     path_info,
     fs,
-    obj,
     cache,
     force=False,
     progress_callback=None,
     relink=False,
-    dvcignore: Optional[DvcIgnoreFilter] = None,
     state=None,
 ):
-    if not obj.hash_info.isdir:
-        ret = _checkout_file(
-            path_info, fs, obj, cache, force, progress_callback, relink, state
-        )
-    else:
-        ret = _checkout_dir(
-            path_info,
-            fs,
-            obj,
-            cache,
-            force,
-            progress_callback,
-            relink,
-            dvcignore=dvcignore,
-            state=state,
-        )
+    if not diff:
+        return
 
-    if state:
-        state.save_link(path_info, fs)
+    for change in diff.deleted:
+        entry_path = (
+            path_info.joinpath(*change.old.key)
+            if change.old.key != ROOT
+            else path_info
+        )
+        _remove(entry_path, fs, change.old.in_cache, force=force)
 
-    return ret
+    failed = []
+    for change in chain(diff.added, diff.modified):
+        entry_path = (
+            path_info.joinpath(*change.new.key)
+            if change.new.key != ROOT
+            else path_info
+        )
+        if change.new.oid.isdir:
+            fs.makedirs(entry_path)
+            continue
+
+        try:
+            _checkout_file(
+                entry_path,
+                fs,
+                change,
+                cache,
+                force,
+                progress_callback,
+                relink,
+                state=state,
+            )
+        except CheckoutError as exc:
+            failed.extend(exc.target_infos)
+
+    if failed:
+        raise CheckoutError(failed)
 
 
 def checkout(
@@ -311,41 +273,54 @@ def checkout(
     dvcignore: Optional[DvcIgnoreFilter] = None,
     state=None,
 ):
+
     if path_info.scheme not in ["local", cache.fs.scheme]:
         raise NotImplementedError
 
-    failed = None
-    skip = False
+    diff = _diff(
+        path_info,
+        fs,
+        obj,
+        cache,
+        relink=relink,
+        dvcignore=dvcignore,
+    )
+
+    failed = []
     if not obj:
         if not quiet:
             logger.warning(
                 "No file hash info found for '%s'. It won't be created.",
                 path_info,
             )
-        _remove(path_info, fs, cache, force=force)
-        failed = path_info
-
-    elif not relink and not _changed(path_info, fs, obj, cache, state=state):
+        failed.append(str(path_info))
+    elif not diff:
         logger.trace("Data '%s' didn't change.", path_info)  # type: ignore
-        skip = True
 
-    if failed or skip:
+    try:
+        _checkout(
+            diff,
+            path_info,
+            fs,
+            cache,
+            force=force,
+            progress_callback=progress_callback,
+            relink=relink,
+            state=state,
+        )
+    except CheckoutError as exc:
+        failed.extend(exc.target_infos)
+
+    if diff and state:
+        state.save_link(path_info, fs)
+        if not failed:
+            state.save(path_info, fs, obj.hash_info)
+
+    if failed or not diff:
         if progress_callback and obj:
             progress_callback(str(path_info), len(obj))
         if failed:
-            raise CheckoutError([failed])
+            raise CheckoutError(failed)
         return
 
-    logger.debug("Checking out '%s' with cache '%s'.", path_info, obj)
-
-    return _checkout(
-        path_info,
-        fs,
-        obj,
-        cache,
-        force,
-        progress_callback,
-        relink,
-        dvcignore=dvcignore,
-        state=state,
-    )
+    return bool(diff) and not relink

dvc/objects/db/base.py

@@ -1,12 +1,10 @@
 import itertools
 import logging
-import os
 from concurrent.futures import ThreadPoolExecutor
 from contextlib import suppress
 from copy import copy
 from typing import TYPE_CHECKING, Optional
 
-from dvc.fs.local import LocalFileSystem
 from dvc.objects.errors import ObjectDBPermissionError, ObjectFormatError
 from dvc.objects.file import HashFile
 from dvc.progress import Tqdm
@@ -83,49 +81,26 @@ class ObjectDB:
         from_info: "AnyPath",
         to_info: "DvcPath",
         _hash_info: "HashInfo",
-        move: bool = False,
+        hardlink: bool = False,
     ):
+        from dvc import fs
+
         self.makedirs(to_info.parent)
-        use_move = isinstance(from_fs, type(self.fs)) and move
-        try:
-            if use_move:
-                self.fs.move(from_info, to_info)
-            elif isinstance(from_fs, LocalFileSystem):
-                if not isinstance(from_info, from_fs.PATH_CLS):
-                    from_info = from_fs.PATH_CLS(from_info)
-                self.fs.upload(from_info, to_info)
-            elif isinstance(self.fs, LocalFileSystem):
-                from_fs.download_file(from_info, to_info)
-            else:
-                with from_fs.open(from_info, mode="rb") as fobj:
-                    self.fs.upload_fobj(fobj, to_info)
-        except OSError as exc:
-            # If the target file already exists, we are going to simply
-            # ignore the exception (#4992).
-            #
-            # On Windows, it is not always guaranteed that you'll get
-            # FileExistsError (it might be PermissionError or a bare OSError)
-            # but all of those exceptions raised from the original
-            # FileExistsError so we have a separate check for that.
-            if isinstance(exc, FileExistsError) or (
-                os.name == "nt"
-                and exc.__context__
-                and isinstance(exc.__context__, FileExistsError)
-            ):
-                logger.debug("'%s' file already exists, skipping", to_info)
-                if use_move:
-                    from_fs.remove(from_info)
-            else:
-                raise
+        return fs.utils.transfer(
+            from_fs,
+            from_info,
+            self.fs,
+            to_info,
+            hardlink=hardlink,
+        )
 
     def add(
         self,
         path_info: "AnyPath",
         fs: "BaseFileSystem",
         hash_info: "HashInfo",
-        move: bool = True,
+        hardlink: bool = False,
         verify: Optional[bool] = None,
-        **kwargs,
     ):
         if self.read_only:
             raise ObjectDBPermissionError("Cannot add to read-only ODB")
@@ -139,7 +114,7 @@ class ObjectDB:
             pass
 
         cache_info = self.hash_to_path_info(hash_info.value)
-        self._add_file(fs, path_info, cache_info, hash_info, move)
+        self._add_file(fs, path_info, cache_info, hash_info, hardlink=hardlink)
 
         try:
             if verify:
@@ -149,10 +124,6 @@ class ObjectDB:
         except (ObjectFormatError, FileNotFoundError):
             pass
 
-        callback = kwargs.get("download_callback")
-        if callback:
-            callback(1)
-
     def hash_to_path_info(self, hash_) -> "DvcPath":
         return self.path_info / hash_[0:2] / hash_[2:]
 

dvc/objects/db/reference.py

@@ -58,19 +58,23 @@ class ReferenceObjectDB(ObjectDB):
         from_info: "AnyPath",
         to_info: "DvcPath",
         hash_info: "HashInfo",
-        move: bool = False,
+        hardlink: bool = False,
     ):
         self.makedirs(to_info.parent)
         if hash_info.isdir:
             return super()._add_file(
-                from_fs, from_info, to_info, hash_info, move
+                from_fs,
+                from_info,
+                to_info,
+                hash_info,
+                hardlink=hardlink,
             )
         ref_file = ReferenceHashFile(from_info, from_fs, hash_info)
         self._obj_cache[hash_info] = ref_file
         ref_fobj = io.BytesIO(ref_file.to_bytes())
         ref_fobj.seek(0)
         try:
-            self.fs.upload_fobj(ref_fobj, to_info)
+            self.fs.upload(ref_fobj, to_info)
         except OSError as exc:
             if isinstance(exc, FileExistsError) or (
                 os.name == "nt"

dvc/objects/diff.py

@@ -0,0 +1,134 @@
+from dataclasses import dataclass, field
+from typing import TYPE_CHECKING, List, Optional, Tuple
+
+if TYPE_CHECKING:
+    from dvc.hash_info import HashInfo
+
+    from .file import HashFile
+
+ADD = "add"
+MODIFY = "modify"
+DELETE = "delete"
+UNCHANGED = "unchanged"
+
+
+@dataclass
+class TreeEntry:
+    in_cache: bool
+    key: Tuple[str]
+    oid: Optional["HashInfo"] = field(default=None)
+
+    def __bool__(self):
+        return bool(self.oid)
+
+    def __eq__(self, other):
+        if not isinstance(other, TreeEntry):
+            return False
+
+        if self.key != other.key:
+            return False
+
+        return self.oid == other.oid
+
+
+@dataclass
+class Change:
+    old: TreeEntry
+    new: TreeEntry
+
+    @property
+    def typ(self):
+        if not self.old and not self.new:
+            return UNCHANGED
+
+        if self.old and not self.new:
+            return DELETE
+
+        if not self.old and self.new:
+            return ADD
+
+        if self.old != self.new:
+            return MODIFY
+
+        return UNCHANGED
+
+    def __bool__(self):
+        return self.typ != UNCHANGED
+
+
+@dataclass
+class DiffResult:
+    added: List[Change] = field(default_factory=list, compare=True)
+    modified: List[Change] = field(default_factory=list, compare=True)
+    deleted: List[Change] = field(default_factory=list, compare=True)
+    unchanged: List[Change] = field(default_factory=list, compare=True)
+
+    def __bool__(self):
+        return bool(self.added or self.modified or self.deleted)
+
+
+ROOT = ("",)
+
+
+def diff(
+    old: Optional["HashFile"], new: Optional["HashFile"], cache
+) -> DiffResult:
+    from .tree import Tree
+
+    if old is None and new is None:
+        return DiffResult()
+
+    def _get_keys(obj):
+        if not obj:
+            return []
+        return [ROOT] + (
+            [key for key, _, _ in obj] if isinstance(obj, Tree) else []
+        )
+
+    old_keys = set(_get_keys(old))
+    new_keys = set(_get_keys(new))
+
+    def _get_oid(obj, key):
+        if not obj or key == ROOT:
+            return obj.hash_info if obj else None
+
+        entry_obj = obj.get(cache, key)
+        return entry_obj.hash_info if entry_obj else None
+
+    def _in_cache(oid, cache):
+        from .errors import ObjectFormatError
+
+        if not oid:
+            return False
+
+        try:
+            cache.check(oid)
+            return True
+        except (FileNotFoundError, ObjectFormatError):
+            return False
+
+    ret = DiffResult()
+    for key in old_keys | new_keys:
+        old_oid = _get_oid(old, key)
+        new_oid = _get_oid(new, key)
+
+        change = Change(
+            old=TreeEntry(_in_cache(old_oid, cache), key, old_oid),
+            new=TreeEntry(_in_cache(new_oid, cache), key, new_oid),
+        )
+
+        if change.typ == ADD:
+            ret.added.append(change)
+        elif change.typ == MODIFY:
+            ret.modified.append(change)
+        elif change.typ == DELETE:
+            ret.deleted.append(change)
+        else:
+            assert change.typ == UNCHANGED
+            if not change.new.in_cache and not (
+                change.new.oid and change.new.oid.isdir
+            ):
+                ret.modified.append(change)
+            else:
+                ret.unchanged.append(change)
+    return ret

dvc/objects/file.py

@@ -28,10 +28,6 @@ class HashFile:
         self.hash_info = hash_info
         self.name = name
 
-    @property
-    def size(self):
-        return self.hash_info.size
-
     def __len__(self):
         return 1
 
@@ -74,7 +70,7 @@ class HashFile:
     def _check_hash(self, odb):
         from .stage import get_file_hash
 
-        actual = get_file_hash(
+        _, actual = get_file_hash(
             self.path_info, self.fs, self.hash_info.name, odb.state
         )
 

dvc/objects/meta.py

@@ -0,0 +1,32 @@
+from collections import OrderedDict
+from dataclasses import dataclass, field
+from typing import Optional
+
+
+@dataclass
+class Meta:
+    PARAM_SIZE = "size"
+    PARAM_NFILES = "nfiles"
+
+    size: Optional[int] = field(default=None)
+    nfiles: Optional[int] = field(default=None)
+
+    @classmethod
+    def from_dict(cls, d):
+        if not d:
+            return cls()
+
+        size = d.pop(cls.PARAM_SIZE, None)
+        nfiles = d.pop(cls.PARAM_NFILES, None)
+        return cls(size=size, nfiles=nfiles)
+
+    def to_dict(self):
+        ret = OrderedDict()
+
+        if self.size is not None:
+            ret[self.PARAM_SIZE] = self.size
+
+        if self.nfiles is not None:
+            ret[self.PARAM_NFILES] = self.nfiles
+
+        return ret

dvc/objects/reference.py

@@ -2,7 +2,7 @@ import errno
 import logging
 import os
 import pickle
-from typing import TYPE_CHECKING, Optional, Tuple
+from typing import TYPE_CHECKING, Optional
 
 from .errors import ObjectFormatError
 from .file import HashFile
@@ -20,8 +20,7 @@ logger = logging.getLogger(__name__)
 class ReferenceHashFile(HashFile):
     PARAM_PATH = "path"
     PARAM_HASH = "hash"
-    PARAM_MTIME = "mtime"
-    PARAM_SIZE = "size"
+    PARAM_CHECKSUM = "checksum"
     PARAM_FS_CONFIG = "fs_config"
 
     def __init__(
@@ -29,14 +28,11 @@ class ReferenceHashFile(HashFile):
         path_info: "AnyPath",
         fs: "BaseFileSystem",
         hash_info: "HashInfo",
-        mtime: Optional[str] = None,
-        size: Optional[int] = None,
+        checksum: Optional[str] = None,
         **kwargs,
     ):
         super().__init__(path_info, fs, hash_info, **kwargs)
-        cur_mtime, cur_size = self._get_mtime_and_size()
-        self.mtime = cur_mtime if mtime is None else mtime
-        self.hash_info.size = cur_size if size is None else size
+        self.checksum = checksum or fs.checksum(path_info)
 
     def __str__(self):
         return f"ref object {self.hash_info} -> {self.path_info}"
@@ -47,21 +43,14 @@ class ReferenceHashFile(HashFile):
             raise FileNotFoundError(
                 errno.ENOENT, os.strerror(errno.ENOENT), self.path_info
             )
-        if not (self.mtime, self.size) == self._get_mtime_and_size():
+        if self.checksum != self._get_checksum():
             raise ObjectFormatError(f"{self} is changed")
         if check_hash:
             self._check_hash(odb)
 
-    def _get_mtime_and_size(self) -> Tuple[Optional[str], int]:
-        from dvc.utils.fs import get_mtime_and_size
-
+    def _get_checksum(self) -> str:
         assert self.fs
-        if hasattr(self.fs, "stat"):
-            try:
-                return get_mtime_and_size(self.path_info, self.fs)
-            except FileNotFoundError:
-                pass
-        return None, self.fs.getsize(self.path_info)
+        return self.fs.checksum(self.path_info)
 
     def to_bytes(self):
         from dvc.fs.repo import RepoFileSystem
@@ -79,8 +68,7 @@ class ReferenceHashFile(HashFile):
         dict_ = {
             self.PARAM_PATH: path_info,
             self.PARAM_HASH: self.hash_info,
-            self.PARAM_MTIME: self.mtime,
-            self.PARAM_SIZE: self.size,
+            self.PARAM_CHECKSUM: self.checksum,
             self.PARAM_FS_CONFIG: self.config_tuple(self.fs),
         }
         try:
@@ -118,8 +106,7 @@ class ReferenceHashFile(HashFile):
             path_info,
             fs,
             hash_info,
-            mtime=dict_.get(cls.PARAM_MTIME),
-            size=dict_.get(cls.PARAM_SIZE),
+            checksum=dict_.get(cls.PARAM_CHECKSUM),
         )
 
     @staticmethod

dvc/objects/stage.py

@@ -14,6 +14,7 @@ from dvc.utils import file_md5
 
 from .db.reference import ReferenceObjectDB
 from .file import HashFile
+from .meta import Meta
 
 if TYPE_CHECKING:
     from dvc.fs.base import BaseFileSystem
@@ -34,54 +35,52 @@ def _upload_file(path_info, fs, odb, upload_odb):
     tmp_info = upload_odb.path_info / tmp_fname()
     with fs.open(path_info, mode="rb", chunk_size=fs.CHUNK_SIZE) as stream:
         stream = HashedStreamReader(stream)
-        upload_odb.fs.upload_fobj(
-            stream, tmp_info, desc=path_info.name, size=fs.getsize(path_info)
-        )
+        size = fs.getsize(path_info)
+        upload_odb.fs.upload(stream, tmp_info, desc=path_info.name, total=size)
 
     odb.add(tmp_info, upload_odb.fs, stream.hash_info)
-    return path_info, odb.get(stream.hash_info)
+    meta = Meta(size=size)
+    return path_info, meta, odb.get(stream.hash_info)
 
 
 def _get_file_hash(path_info, fs, name):
     info = fs.info(path_info)
     if name in info:
         assert not info[name].endswith(".dir")
-        return HashInfo(name, info[name], size=info["size"])
+        hash_value = info[name]
+    elif hasattr(fs, name):
+        func = getattr(fs, name)
+        hash_value = func(path_info)
+    elif name == "md5":
+        hash_value = file_md5(path_info, fs)
+    else:
+        raise NotImplementedError
 
-    func = getattr(fs, name, None)
-    if func:
-        return func(path_info)
-
-    if name == "md5":
-        return HashInfo(
-            name, file_md5(path_info, fs), size=fs.getsize(path_info)
-        )
-
-    raise NotImplementedError
+    meta = Meta(size=info["size"])
+    hash_info = HashInfo(name, hash_value)
+    return meta, hash_info
 
 
 def get_file_hash(path_info, fs, name, state=None):
     if state:
-        hash_info = state.get(  # pylint: disable=assignment-from-none
+        meta, hash_info = state.get(  # pylint: disable=assignment-from-none
             path_info, fs
         )
         if hash_info:
-            return hash_info
+            return meta, hash_info
 
-    hash_info = _get_file_hash(path_info, fs, name)
+    meta, hash_info = _get_file_hash(path_info, fs, name)
 
     if state:
         assert ".dir" not in hash_info.value
         state.save(path_info, fs, hash_info)
 
-    return hash_info
+    return meta, hash_info
 
 
-def _get_file_obj(
-    path_info, fs, name, odb=None, upload_odb=None, dry_run=False
-):
+def _stage_file(path_info, fs, name, odb=None, upload_odb=None, dry_run=False):
     state = odb.state if odb else None
-    hash_info = get_file_hash(path_info, fs, name, state=state)
+    meta, hash_info = get_file_hash(path_info, fs, name, state=state)
     if upload_odb and not dry_run:
         assert odb and name == "md5"
         return _upload_file(path_info, fs, odb, upload_odb)
@@ -89,9 +88,10 @@ def _get_file_obj(
     if dry_run:
         obj = HashFile(path_info, fs, hash_info)
     else:
-        odb.add(path_info, fs, hash_info, move=False)
+        odb.add(path_info, fs, hash_info, hardlink=False)
         obj = odb.get(hash_info)
-    return path_info, obj
+
+    return path_info, meta, obj
 
 
 def _build_objects(
@@ -114,7 +114,7 @@ def _build_objects(
     ) as pbar:
         worker = pbar.wrap_fn(
             partial(
-                _get_file_obj,
+                _stage_file,
                 fs=fs,
                 name=name,
                 **kwargs,
@@ -133,8 +133,9 @@ def _iter_objects(path_info, fs, name, **kwargs):
 def _build_tree(path_info, fs, name, **kwargs):
     from .tree import Tree
 
+    tree_meta = Meta(size=0, nfiles=0)
     tree = Tree(None, None, None)
-    for file_info, obj in _iter_objects(path_info, fs, name, **kwargs):
+    for file_info, meta, obj in _iter_objects(path_info, fs, name, **kwargs):
         if DvcIgnore.DVCIGNORE_FILE == file_info.name:
             raise DvcIgnoreInCollectedDirError(file_info.parent)
 
@@ -146,11 +147,15 @@ def _build_tree(path_info, fs, name, **kwargs):
         #
         # Yes, this is a BUG, as long as we permit "/" in
         # filenames on Windows and "\" on Unix
-        tree.add(file_info.relative_to(path_info).parts, obj)
-    return tree
+        tree.add(file_info.relative_to(path_info).parts, meta, obj.hash_info)
+
+        tree_meta.size += meta.size
+        tree_meta.nfiles += 1
+
+    return tree_meta, tree
 
 
-def _get_tree_obj(path_info, fs, fs_info, name, odb=None, **kwargs):
+def _stage_tree(path_info, fs, fs_info, name, odb=None, **kwargs):
     from .tree import Tree
 
     value = fs_info.get(name)
@@ -158,25 +163,19 @@ def _get_tree_obj(path_info, fs, fs_info, name, odb=None, **kwargs):
         hash_info = HashInfo(name, value)
         try:
             tree = Tree.load(odb, hash_info)
-            # NOTE: loaded entries are naive objects with hash_infos but no
-            # path_info. For staging trees, obj.path_info should be relative
-            # to the staging src `path_info` and src fs
-            for key, entry in tree:
-                entry.fs = fs
-                entry.path_info = path_info.joinpath(*key)
-            return tree
+            return Meta(nfiles=len(tree)), tree
         except FileNotFoundError:
             pass
 
-    tree = _build_tree(path_info, fs, name, odb=odb, **kwargs)
+    meta, tree = _build_tree(path_info, fs, name, odb=odb, **kwargs)
     state = odb.state if odb and odb.state else None
     hash_info = None
     if state:
-        hash_info = state.get(  # pylint: disable=assignment-from-none
+        _, hash_info = state.get(  # pylint: disable=assignment-from-none
             path_info, fs
         )
     tree.digest(hash_info=hash_info)
-    odb.add(tree.path_info, tree.fs, tree.hash_info, move=True)
+    odb.add(tree.path_info, tree.fs, tree.hash_info, hardlink=False)
     raw = odb.get(tree.hash_info)
     # cleanup unneeded memfs tmpfile and return tree based on the
     # ODB fs/path
@@ -184,7 +183,7 @@ def _get_tree_obj(path_info, fs, fs_info, name, odb=None, **kwargs):
         tree.fs.remove(tree.path_info)
     tree.fs = raw.fs
     tree.path_info = raw.path_info
-    return tree
+    return meta, tree
 
 
 _url_cache: Dict[str, str] = {}
@@ -226,7 +225,7 @@ def _load_from_state(odb, staging, path_info, fs, name):
     from .tree import Tree
 
     state = odb.state
-    hash_info = state.get(path_info, fs)
+    meta, hash_info = state.get(path_info, fs)
     if hash_info:
         for odb_ in (odb, staging):
             if odb_.exists(hash_info):
@@ -234,16 +233,9 @@ def _load_from_state(odb, staging, path_info, fs, name):
                     obj = load(odb_, hash_info)
                     check(odb_, obj, check_hash=False)
                     if isinstance(obj, Tree):
-                        obj.hash_info.nfiles = len(obj)
-                        for key, entry in obj:
-                            entry.fs = fs
-                            entry.path_info = path_info.joinpath(*key)
-                    else:
-                        obj.fs = fs
-                        obj.path_info = path_info
+                        meta.nfiles = len(obj)
                     assert obj.hash_info.name == name
-                    obj.hash_info.size = hash_info.size
-                    return odb_, obj
+                    return odb_, meta, obj
                 except (ObjectFormatError, FileNotFoundError):
                     pass
     raise FileNotFoundError
@@ -258,7 +250,7 @@ def _stage_external_tree_info(odb, tree, name):
 
     odb.add(tree.path_info, tree.fs, tree.hash_info)
     raw = odb.get(tree.hash_info)
-    hash_info = get_file_hash(raw.path_info, raw.fs, name, state=odb.state)
+    _, hash_info = get_file_hash(raw.path_info, raw.fs, name, state=odb.state)
     tree.path_info = raw.path_info
     tree.fs = raw.fs
     tree.hash_info.name = hash_info.name
@@ -276,7 +268,7 @@ def stage(
     upload: bool = False,
     dry_run: bool = False,
     **kwargs,
-) -> Tuple["ObjectDB", "HashFile"]:
+) -> Tuple["ObjectDB", "Meta", "HashFile"]:
     """Stage (prepare) objects from the given path for addition to an ODB.
 
     Returns at tuple of (staging_odb, object) where addition to the ODB can
@@ -301,7 +293,7 @@ def stage(
             pass
 
     if details["type"] == "directory":
-        obj = _get_tree_obj(
+        meta, obj = _stage_tree(
             path_info,
             fs,
             details,
@@ -315,7 +307,7 @@ def stage(
         if name != "md5":
             obj = _stage_external_tree_info(odb, obj, name)
     else:
-        _, obj = _get_file_obj(
+        _, meta, obj = _stage_file(
             path_info,
             fs,
             name,
@@ -327,4 +319,4 @@ def stage(
     if odb and odb.state and obj.hash_info:
         odb.state.save(path_info, fs, obj.hash_info)
 
-    return staging, obj
+    return staging, meta, obj

dvc/objects/status.py

@@ -56,7 +56,7 @@ def _indexed_dir_hashes(odb, index, dir_objs, name, cache_odb):
                 tree = Tree.load(cache_odb, HashInfo(name, dir_hash))
             except FileNotFoundError:
                 continue
-        file_hashes = [entry.hash_info.value for _, entry in tree]
+        file_hashes = [oid.value for _, _, oid in tree]
         if dir_hash not in index:
             logger.debug(
                 "Indexing new .dir '%s' with '%s' nested files",
@@ -102,9 +102,9 @@ def status(
                 tree = None
             else:
                 tree = Tree.load(cache_odb, hash_info)
-                for _, entry in tree:
-                    assert entry.hash_info and entry.hash_info.value
-                    hash_infos[entry.hash_info.value] = entry.hash_info
+                for _, _, oid in tree:
+                    assert oid and oid.value
+                    hash_infos[oid.value] = oid
             if index:
                 dir_objs[hash_info.value] = tree
         hash_infos[hash_info.value] = hash_info

dvc/objects/transfer.py

@@ -1,20 +1,19 @@
 import errno
-import itertools
 import logging
-from concurrent import futures
-from concurrent.futures import ThreadPoolExecutor
 from functools import partial, wraps
-from typing import TYPE_CHECKING, Callable, Iterable, Optional
+from typing import TYPE_CHECKING, Any, Callable, Iterable, Optional
 
 from funcy import split
 
 from dvc.progress import Tqdm
+from dvc.utils.threadpool import ThreadPoolExecutor
 
 if TYPE_CHECKING:
     from dvc.hash_info import HashInfo
 
     from .db.base import ObjectDB
     from .db.index import ObjectDBIndexBase
+    from .tree import Tree
 
 logger = logging.getLogger(__name__)
 
@@ -38,109 +37,48 @@ def _log_exceptions(func):
     return wrapper
 
 
-def _transfer(
-    src, dest, dir_ids, file_ids, missing_ids, jobs, verify, move, **kwargs
-):
-    func = _log_exceptions(dest.add)
-    total = len(dir_ids) + len(file_ids)
-    if total == 0:
-        return 0
-    with Tqdm(total=total, unit="file", desc="Transferring") as pbar:
-        func = pbar.wrap_fn(func)
-        with ThreadPoolExecutor(max_workers=jobs) as executor:
-            processor = partial(
-                _create_tasks,
-                executor,
-                jobs,
-                func,
-                src,
-                verify,
-                move,
-            )
-            processor.add_func = func
-            _do_transfer(
-                src,
-                dest,
-                dir_ids,
-                file_ids,
-                missing_ids,
-                processor,
-                verify=verify,
-                move=move,
-                **kwargs,
-            )
-    return total
+def find_tree_by_obj_id(
+    odbs: Iterable[Optional["ObjectDB"]], obj_id: "HashInfo"
+) -> Optional["Tree"]:
+    from .errors import ObjectFormatError
+    from .tree import Tree
 
-
-def _create_tasks(executor, jobs, func, src, verify, move, obj_ids):
-    fails = 0
-    hash_iter = iter(obj_ids)
-
-    def submit(hash_info):
-        obj = src.get(hash_info)
-        return executor.submit(
-            func,
-            obj.path_info,
-            obj.fs,
-            obj.hash_info,
-            verify=verify,
-            move=move,
-        )
-
-    def create_taskset(amount):
-        return {
-            submit(hash_info)
-            for hash_info in itertools.islice(hash_iter, amount)
-        }
-
-    tasks = create_taskset(jobs * 5)
-    while tasks:
-        done, tasks = futures.wait(tasks, return_when=futures.FIRST_COMPLETED)
-        fails += sum(task.result() for task in done)
-        tasks.update(create_taskset(len(done)))
-    return fails
+    for odb in odbs:
+        if odb is not None:
+            try:
+                return Tree.load(odb, obj_id)
+            except (FileNotFoundError, ObjectFormatError):
+                pass
+    return None
 
 
 def _do_transfer(
     src: "ObjectDB",
     dest: "ObjectDB",
-    dir_ids: Iterable["HashInfo"],
-    file_ids: Iterable["HashInfo"],
+    obj_ids: Iterable["HashInfo"],
     missing_ids: Iterable["HashInfo"],
     processor: Callable,
     src_index: Optional["ObjectDBIndexBase"] = None,
     dest_index: Optional["ObjectDBIndexBase"] = None,
     cache_odb: Optional["ObjectDB"] = None,
-    **kwargs,
+    **kwargs: Any,
 ):
     from dvc.exceptions import FileTransferError
-    from dvc.objects.errors import ObjectFormatError
 
+    dir_ids, file_ids = split(lambda hash_info: hash_info.isdir, obj_ids)
     total_fails = 0
     succeeded_dir_objs = []
     all_file_ids = set(file_ids)
 
     for dir_hash in dir_ids:
-        from .tree import Tree
-
-        bound_file_ids = set()
-        dir_obj: Optional["Tree"] = None
-        for odb in (cache_odb, src):
-            if odb is not None:
-                try:
-                    dir_obj = Tree.load(odb, dir_hash)
-                    break
-                except (FileNotFoundError, ObjectFormatError):
-                    pass
+        dir_obj = find_tree_by_obj_id([cache_odb, src], dir_hash)
         assert dir_obj
-        entry_ids = {entry.hash_info for _, entry in dir_obj}
 
-        for file_hash in all_file_ids.copy():
-            if file_hash in entry_ids:
-                bound_file_ids.add(file_hash)
-                all_file_ids.remove(file_hash)
+        entry_ids = {oid for _, _, oid in dir_obj}
+        bound_file_ids = all_file_ids & entry_ids
+        all_file_ids -= entry_ids
 
-        dir_fails = processor(bound_file_ids)
+        dir_fails = sum(processor(bound_file_ids))
         if dir_fails:
             logger.debug(
                 "failed to upload full contents of '%s', "
@@ -164,19 +102,13 @@ def _do_transfer(
                 dir_obj.name,
             )
         else:
-            raw_obj = src.get(dir_obj.hash_info)
-            is_dir_failed = processor.add_func(  # type: ignore[attr-defined]
-                raw_obj.path_info,
-                raw_obj.fs,
-                raw_obj.hash_info,
-                **kwargs,
-            )
+            is_dir_failed = sum(processor([dir_obj.hash_info]))
             total_fails += is_dir_failed
             if not is_dir_failed:
                 succeeded_dir_objs.append(dir_obj)
 
     # insert the rest
-    total_fails += processor(all_file_ids)
+    total_fails += sum(processor(all_file_ids))
     if total_fails:
         if src_index:
             src_index.clear()
@@ -185,7 +117,7 @@ def _do_transfer(
     # index successfully pushed dirs
     if dest_index:
         for dir_obj in succeeded_dir_objs:
-            file_hashes = {entry.hash_info.value for _, entry in dir_obj}
+            file_hashes = {oid.value for _, _, oid in dir_obj}
             logger.debug(
                 "Indexing pushed dir '%s' with '%s' nested files",
                 dir_obj.hash_info,
@@ -201,7 +133,7 @@ def transfer(
     obj_ids: Iterable["HashInfo"],
     jobs: Optional[int] = None,
     verify: bool = False,
-    move: bool = False,
+    hardlink: bool = False,
     **kwargs,
 ) -> int:
     """Transfer (copy) the specified objects from one ODB to another.
@@ -222,18 +154,23 @@ def transfer(
     if not status.new:
         return 0
 
-    dir_ids, file_ids = split(lambda hash_info: hash_info.isdir, status.new)
-    if jobs is None:
-        jobs = dest.fs.jobs
+    def func(hash_info: "HashInfo") -> None:
+        obj = src.get(hash_info)
+        return dest.add(
+            obj.path_info,
+            obj.fs,
+            obj.hash_info,
+            verify=verify,
+            hardlink=hardlink,
+        )
 
-    return _transfer(
-        src,
-        dest,
-        set(dir_ids),
-        set(file_ids),
-        status.missing,
-        jobs,
-        verify,
-        move,
-        **kwargs,
-    )
+    total = len(status.new)
+    jobs = jobs or dest.fs.jobs
+    with Tqdm(total=total, unit="file", desc="Transferring") as pbar:
+        with ThreadPoolExecutor(max_workers=jobs) as executor:
+            wrapped_func = pbar.wrap_fn(_log_exceptions(func))
+            processor = partial(executor.imap_unordered, wrapped_func)
+            _do_transfer(
+                src, dest, status.new, status.missing, processor, **kwargs
+            )
+    return total

dvc/objects/tree.py

@@ -12,6 +12,8 @@ from .stage import get_file_hash
 if TYPE_CHECKING:
     from dvc.hash_info import HashInfo
 
+    from .meta import Meta
+
 logger = logging.getLogger(__name__)
 
 
@@ -20,7 +22,7 @@ class Tree(HashFile):
 
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
-        self._dict: Dict[Tuple[str], "HashFile"] = {}
+        self._dict: Dict[Tuple[str], Tuple["Meta", "HashInfo"]] = {}
 
     @cached_property
     def trie(self):
@@ -28,9 +30,9 @@ class Tree(HashFile):
 
         return Trie(self._dict)
 
-    def add(self, key, obj):
+    def add(self, key: Tuple[str], meta: "Meta", oid: "HashInfo"):
         self.__dict__.pop("trie", None)
-        self._dict[key] = obj
+        self._dict[key] = (meta, oid)
 
     def digest(self, hash_info: Optional["HashInfo"] = None):
         from dvc.fs.memory import MemoryFileSystem
@@ -46,20 +48,17 @@ class Tree(HashFile):
         if hash_info:
             self.hash_info = hash_info
         else:
-            self.hash_info = get_file_hash(path_info, memfs, "md5")
+            _, self.hash_info = get_file_hash(path_info, memfs, "md5")
             assert self.hash_info.value
             self.hash_info.value += ".dir"
-        try:
-            self.hash_info.size = sum(obj.size for _, obj in self)
-        except TypeError:
-            self.hash_info.size = None
-        self.hash_info.nfiles = len(self)
 
     def __len__(self):
         return len(self._dict)
 
     def __iter__(self):
-        yield from self._dict.items()
+        yield from (
+            (key, value[0], value[1]) for key, value in self._dict.items()
+        )
 
     def as_dict(self):
         return self._dict.copy()
@@ -73,10 +72,10 @@ class Tree(HashFile):
                 {
                     # NOTE: not using hash_info.to_dict() because we don't want
                     # size/nfiles fields at this point.
-                    obj.hash_info.name: obj.hash_info.value,
+                    oid.name: oid.value,
                     self.PARAM_RELPATH: posixpath.sep.join(parts),
                 }
-                for parts, obj in self._dict.items()  # noqa: B301
+                for parts, _, oid in self  # noqa: B301
             ),
             key=itemgetter(self.PARAM_RELPATH),
         )
@@ -94,8 +93,7 @@ class Tree(HashFile):
             relpath = entry.pop(cls.PARAM_RELPATH)
             parts = tuple(relpath.split(posixpath.sep))
             hash_info = HashInfo.from_dict(entry)
-            obj = HashFile(None, None, hash_info)
-            tree.add(parts, obj)
+            tree.add(parts, None, hash_info)
         return tree
 
     @classmethod
@@ -118,8 +116,6 @@ class Tree(HashFile):
         tree = cls.from_list(raw)
         tree.path_info = obj.path_info
         tree.fs = obj.fs
-        for _, entry_obj in tree:
-            entry_obj.fs = obj.fs
         tree.hash_info = hash_info
 
         return tree
@@ -135,37 +131,37 @@ class Tree(HashFile):
         """
         tree = Tree(self.path_info, self.fs, self.hash_info)
         try:
-            for key, obj in self.trie.items(prefix):
-                tree.add(key, obj)
+            for key, (meta, oid) in self.trie.items(prefix):
+                tree.add(key, meta, oid)
         except KeyError:
             pass
         return tree
 
-    def get(self, prefix: Tuple[str]) -> Optional[HashFile]:
+    def get(self, odb, prefix: Tuple[str]) -> Optional[HashFile]:
         """Return object at the specified prefix in this tree.
 
         Returns None if no object exists at the specified prefix.
         """
-        obj = self._dict.get(prefix)
-        if obj:
-            return obj
+        _, oid = self._dict.get(prefix) or (None, None)
+        if oid:
+            return odb.get(oid)
 
         tree = Tree(None, None, None)
         depth = len(prefix)
         try:
-            for key, obj in self.trie.items(prefix):
-                tree.add(key[depth:], obj)
+            for key, (meta, entry_oid) in self.trie.items(prefix):
+                tree.add(key[depth:], meta, entry_oid)
         except KeyError:
             return None
         tree.digest()
         return tree
 
 
-def _get_dir_size(odb, tree):
+def du(odb, tree):
     try:
         return sum(
-            odb.fs.getsize(odb.hash_to_path_info(obj.hash_info.value))
-            for _, obj in tree
+            odb.fs.getsize(odb.hash_to_path_info(oid.value))
+            for _, _, oid in tree
         )
     except FileNotFoundError:
         return None
@@ -226,11 +222,8 @@ def merge(odb, ancestor_info, our_info, their_info):
     merged_dict = _merge(ancestor.as_dict(), our.as_dict(), their.as_dict())
 
     merged = Tree(None, None, None)
-    for key, hi in merged_dict.items():
-        merged.add(key, hi)
+    for key, (meta, oid) in merged_dict.items():
+        merged.add(key, meta, oid)
     merged.digest()
 
-    odb.add(merged.path_info, merged.fs, merged.hash_info)
-    hash_info = merged.hash_info
-    hash_info.size = _get_dir_size(odb, merged)
-    return hash_info
+    return merged

dvc/output.py

@@ -1,7 +1,6 @@
 import logging
 import os
 from collections import defaultdict
-from copy import copy
 from typing import TYPE_CHECKING, Dict, Optional, Set, Type
 from urllib.parse import urlparse
 
@@ -9,7 +8,6 @@ from funcy import collecting, project
 from voluptuous import And, Any, Coerce, Length, Lower, Required, SetTo
 
 from dvc import objects, prompt
-from dvc.checkout import checkout
 from dvc.exceptions import (
     CheckoutError,
     CollectCacheError,
@@ -17,6 +15,7 @@ from dvc.exceptions import (
     MergeError,
     RemoteCacheRequiredError,
 )
+from dvc.objects.checkout import checkout
 
 from .fs import get_cloud_fs
 from .fs.hdfs import HDFSFileSystem
@@ -26,6 +25,7 @@ from .hash_info import HashInfo
 from .istextfile import istextfile
 from .objects import Tree
 from .objects.errors import ObjectFormatError
+from .objects.meta import Meta
 from .objects.stage import stage as ostage
 from .objects.transfer import transfer as otransfer
 from .scheme import Schemes
@@ -81,6 +81,7 @@ def loadd_from(stage, d_list):
         desc = d.pop(Output.PARAM_DESC, False)
         isexec = d.pop(Output.PARAM_ISEXEC, False)
         live = d.pop(Output.PARAM_LIVE, False)
+        remote = d.pop(Output.PARAM_REMOTE, None)
         ret.append(
             _get(
                 stage,
@@ -94,6 +95,7 @@ def loadd_from(stage, d_list):
                 desc=desc,
                 isexec=isexec,
                 live=live,
+                remote=remote,
             )
         )
     return ret
@@ -109,6 +111,7 @@ def loads_from(
     checkpoint=False,
     isexec=False,
     live=False,
+    remote=None,
 ):
     return [
         _get(
@@ -122,6 +125,7 @@ def loads_from(
             checkpoint=checkpoint,
             isexec=isexec,
             live=live,
+            remote=remote,
         )
         for s in s_list
     ]
@@ -162,7 +166,6 @@ def load_from_pipeline(stage, data, typ="outs"):
     metric = typ == stage.PARAM_METRICS
     plot = typ == stage.PARAM_PLOTS
     live = typ == stage.PARAM_LIVE
-
     if live:
         # `live` is single object
         data = [data]
@@ -185,6 +188,7 @@ def load_from_pipeline(stage, data, typ="outs"):
                 Output.PARAM_CACHE,
                 Output.PARAM_PERSIST,
                 Output.PARAM_CHECKPOINT,
+                Output.PARAM_REMOTE,
             ],
         )
 
@@ -255,6 +259,7 @@ class Output:
     PARAM_LIVE = "live"
     PARAM_LIVE_SUMMARY = "summary"
     PARAM_LIVE_HTML = "html"
+    PARAM_REMOTE = "remote"
 
     METRIC_SCHEMA = Any(
         None,
@@ -283,6 +288,7 @@ class Output:
         live=False,
         desc=None,
         isexec=False,
+        remote=None,
     ):
         self.repo = stage.repo if stage else None
 
@@ -310,6 +316,7 @@ class Output:
         # By resolved path, which contains actual location,
         # should be absolute and don't contain remote:// refs.
         self.stage = stage
+        self.meta = Meta.from_dict(info)
         self.hash_info = HashInfo.from_dict(info)
         self.use_cache = False if self.IS_DEPENDENCY else cache
         self.metric = False if self.IS_DEPENDENCY else metric
@@ -326,7 +333,7 @@ class Output:
         self.obj = None
         self.isexec = False if self.IS_DEPENDENCY else isexec
 
-        self.def_remote = None
+        self.remote = remote
 
     def _parse_path(self, fs, path_info):
         if fs.scheme != "local":
@@ -409,7 +416,7 @@ class Output:
         else:
             odb = self.repo.odb.local
             name = self.fs.PARAM_CHECKSUM
-        _, obj = ostage(
+        _, _, obj = ostage(
             odb,
             self.path_info,
             self.fs,
@@ -531,7 +538,15 @@ class Output:
             self.verify_metric()
 
         if not self.use_cache:
-            self.hash_info = self.get_hash()
+            _, self.meta, obj = ostage(
+                self.repo.odb.local,
+                self.path_info,
+                self.fs,
+                self.fs.PARAM_CHECKSUM,
+                dvcignore=self.dvcignore,
+                dry_run=True,
+            )
+            self.hash_info = obj.hash_info
             if not self.IS_DEPENDENCY:
                 logger.debug(
                     "Output '%s' doesn't use cache. Skipping saving.", self
@@ -540,11 +555,7 @@ class Output:
 
         assert not self.IS_DEPENDENCY
 
-        if not self.changed():
-            logger.debug("Output '%s' didn't change. Skipping saving.", self)
-            return
-
-        _, self.obj = ostage(
+        _, self.meta, self.obj = ostage(
             self.odb,
             self.path_info,
             self.fs,
@@ -573,7 +584,7 @@ class Output:
             if granular:
                 obj = self._commit_granular_dir(filter_info)
             else:
-                staging, obj = ostage(
+                staging, _, obj = ostage(
                     self.odb,
                     filter_info or self.path_info,
                     self.fs,
@@ -585,7 +596,7 @@ class Output:
                     self.odb,
                     {obj.hash_info},
                     shallow=False,
-                    move=True,
+                    hardlink=True,
                 )
             checkout(
                 filter_info or self.path_info,
@@ -600,7 +611,7 @@ class Output:
 
     def _commit_granular_dir(self, filter_info):
         prefix = filter_info.relative_to(self.path_info).parts
-        staging, save_obj = ostage(
+        staging, _, save_obj = ostage(
             self.odb,
             self.path_info,
             self.fs,
@@ -608,18 +619,18 @@ class Output:
             dvcignore=self.dvcignore,
         )
         save_obj = save_obj.filter(prefix)
-        checkout_obj = save_obj.get(prefix)
+        checkout_obj = save_obj.get(self.odb, prefix)
         otransfer(
             staging,
             self.odb,
-            {save_obj.hash_info} | {entry.hash_info for _, entry in save_obj},
+            {save_obj.hash_info} | {oid for _, _, oid in save_obj},
             shallow=True,
-            move=True,
+            hardlink=True,
         )
         return checkout_obj
 
     def dumpd(self):
-        ret = copy(self.hash_info.to_dict())
+        ret = {**self.hash_info.to_dict(), **self.meta.to_dict()}
 
         if self.is_in_repo:
             path = self.path_info.relpath(self.stage.wdir).as_posix()
@@ -703,7 +714,7 @@ class Output:
 
         if filter_info and filter_info != self.path_info:
             prefix = filter_info.relative_to(self.path_info).parts
-            obj = obj.get(prefix)
+            obj = obj.get(self.odb, prefix)
 
         return obj
 
@@ -797,7 +808,7 @@ class Output:
 
         upload = not (update and from_fs.isdir(from_info))
         jobs = jobs or min((from_fs.jobs, odb.fs.jobs))
-        staging, obj = ostage(
+        staging, self.meta, obj = ostage(
             odb,
             from_info,
             from_fs,
@@ -811,7 +822,7 @@ class Output:
             odb,
             {obj.hash_info},
             jobs=jobs,
-            move=upload,
+            hardlink=False,
             shallow=False,
         )
 
@@ -826,7 +837,7 @@ class Output:
             return 1
 
         if not filter_info or filter_info == self.path_info:
-            return self.hash_info.nfiles or 0
+            return self.meta.nfiles or 0
 
         obj = self.get_obj(filter_info=filter_info)
         return len(obj) if obj else 0
@@ -843,6 +854,8 @@ class Output:
         try:
             objects.check(self.odb, obj)
         except FileNotFoundError:
+            if self.remote:
+                kwargs["remote"] = self.remote
             self.repo.cloud.pull([obj.hash_info], **kwargs)
 
         if self.obj:
@@ -855,9 +868,9 @@ class Output:
 
         return self.obj
 
-    def collect_used_dir_cache(
+    def _collect_used_dir_cache(
         self, remote=None, force=False, jobs=None, filter_info=None
-    ) -> Dict[Optional["ObjectDB"], Set["HashInfo"]]:
+    ) -> Optional["Tree"]:
         """Fetch dir cache and return used object IDs for this out."""
 
         try:
@@ -878,13 +891,13 @@ class Output:
                     "unable to fully collect used cache"
                     " without cache for directory '{}'".format(self)
                 )
-            return {}
+            return None
 
         obj = self.get_obj()
         if filter_info and filter_info != self.path_info:
             prefix = filter_info.relative_to(self.path_info).parts
             obj = obj.filter(prefix)
-        return {None: set(self._named_obj_ids(obj))}
+        return obj
 
     def get_used_objs(
         self, **kwargs
@@ -917,22 +930,31 @@ class Output:
             return {}
 
         if self.is_dir_checksum:
-            return self.collect_used_dir_cache(**kwargs)
+            obj = self._collect_used_dir_cache(**kwargs)
+        else:
+            obj = self.get_obj(filter_info=kwargs.get("filter_info"))
+            if not obj:
+                obj = self.odb.get(self.hash_info)
 
-        obj = self.get_obj(filter_info=kwargs.get("filter_info"))
         if not obj:
-            obj = self.odb.get(self.hash_info)
+            return {}
 
-        return {None: set(self._named_obj_ids(obj))}
+        if self.remote:
+            remote = self.repo.cloud.get_remote_odb(name=self.remote)
+        else:
+            remote = None
+
+        return {remote: self._named_obj_ids(obj)}
 
     def _named_obj_ids(self, obj):
         name = str(self)
         obj.hash_info.obj_name = name
-        yield obj.hash_info
+        oids = {obj.hash_info}
         if isinstance(obj, Tree):
-            for key, entry_obj in obj:
-                entry_obj.hash_info.obj_name = self.fs.sep.join([name, *key])
-                yield entry_obj.hash_info
+            for key, _, oid in obj:
+                oid.obj_name = self.fs.sep.join([name, *key])
+                oids.add(oid)
+        return oids
 
     def get_used_external(
         self, **kwargs
@@ -964,8 +986,8 @@ class Output:
 
         ignored = [
             self.fs.PARAM_CHECKSUM,
-            HashInfo.PARAM_SIZE,
-            HashInfo.PARAM_NFILES,
+            Meta.PARAM_SIZE,
+            Meta.PARAM_NFILES,
         ]
 
         for opt in ignored:
@@ -983,7 +1005,7 @@ class Output:
             )
 
     def merge(self, ancestor, other):
-        from dvc.objects.tree import merge
+        from dvc.objects.tree import du, merge
 
         assert other
 
@@ -996,9 +1018,16 @@ class Output:
         self._check_can_merge(self)
         self._check_can_merge(other)
 
-        self.hash_info = merge(
+        merged = merge(
             self.odb, ancestor_info, self.hash_info, other.hash_info
         )
+        self.odb.add(merged.path_info, merged.fs, merged.hash_info)
+
+        self.hash_info = merged.hash_info
+        self.meta = Meta(
+            size=du(self.odb, merged),
+            nfiles=len(merged),
+        )
 
     @property
     def fspath(self):
@@ -1023,8 +1052,8 @@ ARTIFACT_SCHEMA = {
     Output.PARAM_PLOT: bool,
     Output.PARAM_PERSIST: bool,
     Output.PARAM_CHECKPOINT: bool,
-    HashInfo.PARAM_SIZE: int,
-    HashInfo.PARAM_NFILES: int,
+    Meta.PARAM_SIZE: int,
+    Meta.PARAM_NFILES: int,
     Output.PARAM_ISEXEC: bool,
 }
 
@@ -1033,4 +1062,5 @@ SCHEMA = {
     Output.PARAM_CACHE: bool,
     Output.PARAM_METRIC: Output.METRIC_SCHEMA,
     Output.PARAM_DESC: str,
+    Output.PARAM_REMOTE: str,
 }

dvc/progress.py

@@ -1,5 +1,4 @@
 """Manages progress bars for DVC repo."""
-
 import logging
 import sys
 from threading import RLock
@@ -137,8 +136,8 @@ class Tqdm(tqdm):
 
         return wrapped
 
-    def as_callback(self, fs, path_info):
-        return FsspecCallback(fs, path_info, self)
+    def as_callback(self):
+        return FsspecCallback(self)
 
     def close(self):
         self.postfix["info"] = ""
@@ -168,23 +167,46 @@ class Tqdm(tqdm):
 
 
 class FsspecCallback(fsspec.Callback):
-    def __init__(self, fs, path_info, progress_bar):
-        self.fs = fs
-        self.path_info = path_info
+    def __init__(self, progress_bar):
         self.progress_bar = progress_bar
         super().__init__()
 
     def set_size(self, size):
-        """``set_size()`` is an API that might be called with ``None`` if the
-        information is not already present on the caller. In that case, we'll
-        retrieve the size from an ``info()`` call."""
-        if size is None:
-            size = self.fs.info(self.path_info)["size"]
-        self.progress_bar.total = size
-        self.progress_bar.refresh()
+        if size is not None:
+            self.progress_bar.total = size
+            self.progress_bar.refresh()
+            super().set_size(size)
 
     def relative_update(self, inc=1):
         self.progress_bar.update(inc)
+        super().relative_update(inc)
 
     def absolute_update(self, value):
         self.progress_bar.update_to(value)
+        super().absolute_update(value)
+
+    @staticmethod
+    def wrap_fn(cb, fn):
+        def wrapped(*args, **kwargs):
+            res = fn(*args, **kwargs)
+            cb.relative_update()
+            return res
+
+        return wrapped
+
+
+def tdqm_or_callback_wrapped(
+    fobj, method, total, callback=None, **pbar_kwargs
+):
+    if callback:
+        from funcy import nullcontext
+        from tqdm.utils import CallbackIOWrapper
+
+        callback.set_size(total)
+        wrapper = CallbackIOWrapper(callback.relative_update, fobj, method)
+        return nullcontext(wrapper)
+
+    return Tqdm.wrapattr(fobj, method, total=total, bytes=True, **pbar_kwargs)
+
+
+DEFAULT_CALLBACK = fsspec.callbacks.NoOpCallback()

dvc/render/__init__.py

@@ -0,0 +1,33 @@
+import logging
+from typing import TYPE_CHECKING, Dict
+
+if TYPE_CHECKING:
+    from dvc.types import StrPath
+
+logger = logging.getLogger(__name__)
+
+
+class Renderer:
+    def __init__(self, data: Dict):
+        self.data = data
+
+        from dvc.render.utils import get_files
+
+        files = get_files(self.data)
+
+        # we assume comparison of same file between revisions for now
+        assert len(files) == 1
+        self.filename = files.pop()
+
+    def _convert(self, path: "StrPath"):
+        raise NotImplementedError
+
+    @property
+    def DIV(self):
+        raise NotImplementedError
+
+    def generate_html(self, path: "StrPath"):
+        """this method might edit content of path"""
+        partial = self._convert(path)
+        div_id = f"plot_{self.filename.replace('.', '_').replace('/', '_')}"
+        return self.DIV.format(id=div_id, partial=partial)

dvc/render/html.py

@@ -1,11 +1,17 @@
-from typing import Dict, List, Optional
+import os
+from pathlib import Path
+from typing import TYPE_CHECKING, Dict, List, Optional
 
 from dvc.exceptions import DvcException
-from dvc.types import StrPath
+
+if TYPE_CHECKING:
+    from dvc.render import Renderer
+    from dvc.types import StrPath
 
 PAGE_HTML = """<!DOCTYPE html>
 <html>
 <head>
+    {refresh_tag}
     <title>DVC Plot</title>
     <script src="https://cdn.jsdelivr.net/npm/[email protected]"></script>
     <script src="https://cdn.jsdelivr.net/npm/[email protected]"></script>
@@ -16,12 +22,6 @@ PAGE_HTML = """<!DOCTYPE html>
 </body>
 </html>"""
 
-VEGA_DIV_HTML = """<div id = "{id}"></div>
-<script type = "text/javascript">
-    var spec = {vega_json};
-    vegaEmbed('#{id}', spec);
-</script>"""
-
 
 class MissingPlaceholderError(DvcException):
     def __init__(self, placeholder):
@@ -31,14 +31,23 @@ class MissingPlaceholderError(DvcException):
 class HTML:
     PLACEHOLDER = "plot_divs"
     PLACEHOLDER_FORMAT_STR = f"{{{PLACEHOLDER}}}"
+    REFRESH_PLACEHOLDER = "refresh_tag"
+    REFRESH_TAG = '<meta http-equiv="refresh" content="{}">'
 
-    def __init__(self, template: str = None):
+    def __init__(
+        self,
+        template: Optional[str] = None,
+        refresh_seconds: Optional[int] = None,
+    ):
         template = template or PAGE_HTML
         if self.PLACEHOLDER_FORMAT_STR not in template:
             raise MissingPlaceholderError(self.PLACEHOLDER_FORMAT_STR)
 
         self.template = template
         self.elements: List[str] = []
+        self.refresh_tag = ""
+        if refresh_seconds is not None:
+            self.refresh_tag = self.REFRESH_TAG.format(refresh_seconds)
 
     def with_metrics(self, metrics: Dict[str, Dict]) -> "HTML":
         import tabulate
@@ -56,41 +65,43 @@ class HTML:
         self.elements.append(tabulate.tabulate(rows, header, tablefmt="html"))
         return self
 
-    def with_plots(self, plots: Dict[str, Dict]) -> "HTML":
-        self.elements.extend(
-            [
-                VEGA_DIV_HTML.format(id=f"plot{i}", vega_json=plot)
-                for i, plot in enumerate(plots.values())
-            ]
-        )
-        return self
-
     def with_element(self, html: str) -> "HTML":
         self.elements.append(html)
         return self
 
     def embed(self) -> str:
-        kwargs = {self.PLACEHOLDER: "\n".join(self.elements)}
+        kwargs = {
+            self.PLACEHOLDER: "\n".join(self.elements),
+            self.REFRESH_PLACEHOLDER: self.refresh_tag,
+        }
         return self.template.format(**kwargs)
 
 
 def write(
-    path: StrPath,
-    plots: Dict[str, Dict],
+    path: "StrPath",
+    renderers: List["Renderer"],
     metrics: Optional[Dict[str, Dict]] = None,
-    template_path: Optional[StrPath] = None,
+    template_path: Optional["StrPath"] = None,
+    refresh_seconds: Optional[int] = None,
 ):
+
+    os.makedirs(path, exist_ok=True)
+
     page_html = None
     if template_path:
-        with open(template_path) as fobj:
+        with open(template_path, encoding="utf-8") as fobj:
             page_html = fobj.read()
 
-    document = HTML(page_html)
+    document = HTML(page_html, refresh_seconds=refresh_seconds)
     if metrics:
         document.with_metrics(metrics)
         document.with_element("<br>")
 
-    document.with_plots(plots)
+    for renderer in renderers:
+        document.with_element(renderer.generate_html(path))
 
-    with open(path, "w") as fd:
+    index = Path(os.path.join(path, "index.html"))
+
+    with open(index, "w", encoding="utf-8") as fd:
         fd.write(document.embed())
+    return index

dvc/render/image.py

@@ -0,0 +1,63 @@
+import os
+from typing import TYPE_CHECKING
+
+from dvc.render import Renderer
+from dvc.render.utils import get_files
+from dvc.utils import relpath
+
+if TYPE_CHECKING:
+    from dvc.types import StrPath
+
+
+class ImageRenderer(Renderer):
+    DIV = """
+        <div
+            id="{id}"
+            style="border: 1px solid;">
+            {partial}
+        </div>"""
+
+    def _write_image(
+        self,
+        page_dir_path: "StrPath",
+        revision: str,
+        filename: str,
+        image_data: bytes,
+    ):
+        static = os.path.join(page_dir_path, "static")
+        os.makedirs(static, exist_ok=True)
+
+        img_path = os.path.join(
+            static, f"{revision}_{filename.replace(os.sep, '_')}"
+        )
+        with open(img_path, "wb") as fd:
+            fd.write(image_data)
+        return """
+        <div>
+            <p>{title}</p>
+            <img src="{src}">
+        </div>""".format(
+            title=revision, src=(relpath(img_path, page_dir_path))
+        )
+
+    def _convert(self, path: "StrPath"):
+        div_content = []
+        for rev, rev_data in self.data.items():
+            if "data" in rev_data:
+                for file, file_data in rev_data.get("data", {}).items():
+                    if "data" in file_data:
+                        div_content.append(
+                            self._write_image(
+                                path, rev, file, file_data["data"]
+                            )
+                        )
+        if div_content:
+            div_content.insert(0, f"<p>{self.filename}</p>")
+            return "\n".join(div_content)
+        return ""
+
+    @staticmethod
+    def matches(data):
+        files = get_files(data)
+        extensions = set(map(lambda f: os.path.splitext(f)[1], files))
+        return extensions.issubset({".jpg", ".jpeg", ".gif", ".png"})

dvc/render/utils.py

@@ -0,0 +1,71 @@
+import os.path
+from typing import Dict, List
+
+import dpath.util
+
+
+def get_files(data: Dict) -> List:
+    files = set()
+    for rev in data.keys():
+        for file in data[rev].get("data", {}).keys():
+            files.add(file)
+    sorted_files = sorted(files)
+    return sorted_files
+
+
+def group_by_filename(plots_data: Dict) -> List[Dict]:
+    files = get_files(plots_data)
+    grouped = []
+    for file in files:
+        grouped.append(dpath.util.search(plots_data, ["*", "*", file]))
+    return grouped
+
+
+def find_vega(repo, plots_data, target):
+    found = dpath.util.search(plots_data, ["*", "*", target])
+    from dvc.render.vega import VegaRenderer
+
+    if found and VegaRenderer.matches(found):
+        return VegaRenderer(found, repo.plots.templates).get_vega()
+    return ""
+
+
+def match_renderers(plots_data, templates):
+    from dvc.render.image import ImageRenderer
+    from dvc.render.vega import VegaRenderer
+
+    renderers = []
+    for g in group_by_filename(plots_data):
+        if VegaRenderer.matches(g):
+            renderers.append(VegaRenderer(g, templates))
+        if ImageRenderer.matches(g):
+            renderers.append(ImageRenderer(g))
+    return renderers
+
+
+def render(
+    repo,
+    plots_data,
+    metrics=None,
+    path=None,
+    html_template_path=None,
+    refresh_seconds=None,
+):
+    # TODO we could probably remove repo usages (here and in VegaRenderer)
+    renderers = match_renderers(plots_data, repo.plots.templates)
+    if not html_template_path:
+        html_template_path = repo.config.get("plots", {}).get(
+            "html_template", None
+        )
+        if html_template_path and not os.path.isabs(html_template_path):
+            html_template_path = os.path.join(repo.dvc_dir, html_template_path)
+
+    from dvc.render.html import write
+
+    return write(
+        path,
+        renderers,
+        metrics=metrics,
+        template_path=html_template_path,
+        refresh_seconds=refresh_seconds,
+    )

dvc/render/vega.py

@@ -0,0 +1,175 @@
+import os
+from copy import copy, deepcopy
+from typing import TYPE_CHECKING, Dict, List, Optional, Union
+
+from funcy import first
+
+from dvc.exceptions import DvcException
+from dvc.render import Renderer
+from dvc.render.utils import get_files
+
+if TYPE_CHECKING:
+    from dvc.repo.plots.template import PlotTemplates
+    from dvc.types import StrPath
+
+REVISION_FIELD = "rev"
+INDEX_FIELD = "step"
+
+
+class PlotMetricTypeError(DvcException):
+    def __init__(self, file):
+        super().__init__(
+            "'{}' - file type error\n"
+            "Only JSON, YAML, CSV and TSV formats are supported.".format(file)
+        )
+
+
+class PlotDataStructureError(DvcException):
+    def __init__(self):
+        super().__init__(
+            "Plot data extraction failed. Please see "
+            "https://man.dvc.org/plots for supported data formats."
+        )
+
+
+def _filter_fields(
+    datapoints: List[Dict], filename, revision, fields=None
+) -> List[Dict]:
+    if not fields:
+        return datapoints
+    assert isinstance(fields, set)
+
+    new_data = []
+    for data_point in datapoints:
+        new_dp = copy(data_point)
+
+        keys = set(data_point.keys())
+        if keys & fields != fields:
+            raise DvcException(
+                "Could not find fields: '{}' for '{}' at '{}'.".format(
+                    ", ".join(fields), filename, revision
+                )
+            )
+
+        to_del = keys - fields
+        for key in to_del:
+            del new_dp[key]
+        new_data.append(new_dp)
+    return new_data
+
+
+def _lists(dictionary):
+    for _, value in dictionary.items():
+        if isinstance(value, dict):
+            yield from _lists(value)
+        elif isinstance(value, list):
+            yield value
+
+
+def _find_data(data: Union[Dict, List], fields=None) -> List[Dict]:
+    if not isinstance(data, dict):
+        return data
+
+    if not fields:
+        # just look for first list of dicts
+        fields = set()
+
+    for lst in _lists(data):
+        if (
+            all(isinstance(dp, dict) for dp in lst)
+            and set(first(lst).keys()) & fields == fields
+        ):
+            return lst
+    raise PlotDataStructureError()
+
+
+def _append_index(datapoints: List[Dict], append_index=False) -> List[Dict]:
+    if not append_index or INDEX_FIELD in first(datapoints).keys():
+        return datapoints
+
+    for index, data_point in enumerate(datapoints):
+        data_point[INDEX_FIELD] = index
+    return datapoints
+
+
+def _append_revision(datapoints: List[Dict], revision) -> List[Dict]:
+    for data_point in datapoints:
+        data_point[REVISION_FIELD] = revision
+    return datapoints
+
+
+class VegaRenderer(Renderer):
+    DIV = """
+    <div id = "{id}">
+        <script type = "text/javascript">
+            var spec = {partial};
+            vegaEmbed('#{id}', spec);
+        </script>
+    </div>
+    """
+
+    def __init__(self, data: Dict, templates: "PlotTemplates"):
+        super().__init__(data)
+        self.templates = templates
+
+    def _squash_props(self) -> Dict:
+        resolved: Dict[str, str] = {}
+        for rev_data in self.data.values():
+            for file_data in rev_data.get("data", {}).values():
+                props = file_data.get("props", {})
+                resolved = {**resolved, **props}
+        return resolved
+
+    def _datapoints(self, props: Dict):
+        fields = props.get("fields", set())
+        if fields:
+            fields = {*fields, props.get("x"), props.get("y")} - {None}
+
+        datapoints = []
+        for revision, rev_data in self.data.items():
+            for filename, file_data in rev_data.get("data", {}).items():
+                if "data" in file_data:
+                    tmp = deepcopy(file_data.get("data"))
+                    tmp = _find_data(tmp, fields=fields - {INDEX_FIELD})
+                    tmp = _append_index(
+                        tmp, append_index=props.get("append_index", False)
+                    )
+                    tmp = _filter_fields(
+                        tmp,
+                        filename=filename,
+                        revision=revision,
+                        fields=fields,
+                    )
+                    tmp = _append_revision(tmp, revision=revision)
+                    datapoints.extend(tmp)
+        return datapoints
+
+    def get_vega(self) -> Optional[str]:
+        props = self._squash_props()
+
+        template = self.templates.load(props.get("template", None))
+
+        if not props.get("x") and template.has_anchor("x"):
+            props["append_index"] = True
+            props["x"] = INDEX_FIELD
+
+        datapoints = self._datapoints(props)
+
+        if datapoints:
+            if not props.get("y") and template.has_anchor("y"):
+                fields = list(first(datapoints))
+                skip = (REVISION_FIELD, props.get("x"))
+                props["y"] = first(
+                    f for f in reversed(fields) if f not in skip
+                )
+            return template.render(datapoints, props=props)
+        return None
+
+    def _convert(self, path: "StrPath"):
+        return self.get_vega()
+
+    @staticmethod
+    def matches(data):
+        files = get_files(data)
+        extensions = set(map(lambda f: os.path.splitext(f)[1], files))
+        return extensions.issubset({".yml", ".yaml", ".json", ".csv", ".tsv"})

dvc/repo/__init__.py

@@ -12,6 +12,7 @@ from dvc.exceptions import IsADirectoryError as DvcIsADirectoryError
 from dvc.exceptions import NotDvcRepoError, OutputNotFoundError
 from dvc.ignore import DvcIgnoreFilter
 from dvc.path_info import PathInfo
+from dvc.utils import env2bool
 from dvc.utils.fs import path_isin
 
 if TYPE_CHECKING:
@@ -119,6 +120,34 @@ class Repo:
 
         return root_dir, dvc_dir, tmp_dir
 
+    def _get_database_dir(self, db_name):
+        # NOTE: by default, store SQLite-based remote indexes and state's
+        # `links` and `md5s` caches in the repository itself to avoid any
+        # possible state corruption in 'shared cache dir' scenario, but allow
+        # user to override this through config when, say, the repository is
+        # located on a mounted volume — see
+        # https://github.com/iterative/dvc/issues/4420
+        base_db_dir = self.config.get(db_name, {}).get("dir", None)
+        if not base_db_dir:
+            return self.tmp_dir
+
+        import hashlib
+
+        from dvc.utils.fs import makedirs
+
+        root_dir_hash = hashlib.sha224(
+            self.root_dir.encode("utf-8")
+        ).hexdigest()
+
+        db_dir = os.path.join(
+            base_db_dir,
+            self.DVC_DIR,
+            f"{os.path.basename(self.root_dir)}-{root_dir_hash[0:7]}",
+        )
+
+        makedirs(db_dir, exist_ok=True)
+        return db_dir
+
     def __init__(
         self,
         root_dir=None,
@@ -181,10 +210,8 @@ class Repo:
                 friendly=True,
             )
 
-            # NOTE: storing state and link_state in the repository itself to
-            # avoid any possible state corruption in 'shared cache dir'
-            # scenario.
-            self.state = State(self.root_dir, self.tmp_dir, self.dvcignore)
+            state_db_dir = self._get_database_dir("state")
+            self.state = State(self.root_dir, state_db_dir, self.dvcignore)
             self.odb = ODBManager(self)
 
             self.stage_cache = StageCache(self)
@@ -262,6 +289,17 @@ class Repo:
 
         return Experiments(self)
 
+    @cached_property
+    def machine(self):
+        from dvc.machine import MachineManager
+
+        if self.tmp_dir and (
+            self.config["feature"].get("machine", False)
+            or env2bool("DVC_TEST")
+        ):
+            return MachineManager(self)
+        return None
+
     @property
     def fs(self) -> "BaseFileSystem":
         return self._fs
@@ -442,6 +480,10 @@ class Repo:
 
         return RepoFileSystem(self, subrepos=self.subrepos, **self._fs_conf)
 
+    @cached_property
+    def index_db_dir(self):
+        return self._get_database_dir("index")
+
     @contextmanager
     def open_by_relpath(self, path, remote=None, mode="r", encoding=None):
         """Opens a specified resource as a file descriptor"""

dvc/repo/collect.py

@@ -49,7 +49,6 @@ def _collect_paths(
                 )
             else:
                 logger.warning("'%s' was not found at: '%s'.", path_info, rev)
-            continue
         target_infos.append(path_info)
     return target_infos
 

dvc/repo/diff.py

@@ -4,7 +4,6 @@ from collections import defaultdict
 from typing import Dict, List
 
 from dvc.exceptions import PathMissingError
-from dvc.objects.stage import get_file_hash
 from dvc.repo import locked
 from dvc.repo.experiments.utils import fix_exp_head
 
@@ -45,7 +44,7 @@ def diff(self, a_rev="HEAD", b_rev=None, targets=None):
                 repo_fs, targets
             )
 
-        results[rev] = _paths_checksums(self, repo_fs, targets_path_infos)
+        results[rev] = _paths_checksums(self, targets_path_infos)
 
     if targets is not None:
         # check for overlapping missing targets between a_rev and b_rev
@@ -98,7 +97,7 @@ def diff(self, a_rev="HEAD", b_rev=None, targets=None):
     return ret if any(ret.values()) else {}
 
 
-def _paths_checksums(repo, repo_fs, targets):
+def _paths_checksums(repo, targets):
     """
     A dictionary of checksums addressed by relpaths collected from
     the current fs outputs.
@@ -110,10 +109,10 @@ def _paths_checksums(repo, repo_fs, targets):
         file:      "data"
     """
 
-    return dict(_output_paths(repo, repo_fs, targets))
+    return dict(_output_paths(repo, targets))
 
 
-def _output_paths(repo, repo_fs, targets):
+def _output_paths(repo, targets):
     from dvc.fs.local import LocalFileSystem
     from dvc.objects.stage import stage as ostage
 
@@ -131,45 +130,47 @@ def _output_paths(repo, repo_fs, targets):
             else os.path.join(str(output), "")
         )
 
-    def _to_checksum(output):
-        if on_working_fs:
-            return ostage(
-                repo.odb.local,
-                output.path_info,
-                repo.odb.local.fs,
-                "md5",
-                dry_run=True,
-            )[1].hash_info.value
-        return output.hash_info.value
-
     for output in repo.index.outs:
         if _exists(output):
             yield_output = targets is None or any(
                 output.path_info.isin_or_eq(target) for target in targets
             )
 
+            if on_working_fs:
+                _, _, obj = ostage(
+                    repo.odb.local,
+                    output.path_info,
+                    repo.odb.local.fs,
+                    "md5",
+                    dry_run=True,
+                    dvcignore=output.dvcignore,
+                )
+                hash_info = obj.hash_info
+            else:
+                hash_info = output.hash_info
+                obj = output.get_obj()
+
             if yield_output:
-                yield _to_path(output), _to_checksum(output)
+                yield _to_path(output), hash_info.value
+
+            if not obj:
+                continue
 
             if output.is_dir_checksum and (
                 yield_output
                 or any(target.isin(output.path_info) for target in targets)
             ):
-                yield from _dir_output_paths(repo_fs, output, targets)
+                yield from _dir_output_paths(output.path_info, obj, targets)
 
 
-def _dir_output_paths(repo_fs, output, targets=None):
-    from dvc.config import NoRemoteError
-
-    try:
-        for fname in repo_fs.walk_files(output.path_info):
-            if targets is None or any(
-                fname.isin_or_eq(target) for target in targets
-            ):
-                # pylint: disable=no-member
-                yield str(fname), get_file_hash(fname, repo_fs, "md5").value
-    except NoRemoteError:
-        logger.warning("dir cache entry for '%s' is missing", output)
+def _dir_output_paths(path_info, obj, targets=None):
+    for key, _, oid in obj:
+        fname = path_info.joinpath(*key)
+        if targets is None or any(
+            fname.isin_or_eq(target) for target in targets
+        ):
+            # pylint: disable=no-member
+            yield str(fname), oid.value
 
 
 def _filter_missing(repo_fs, paths):

dvc/repo/experiments/__init__.py

@@ -10,6 +10,7 @@ from typing import Dict, Iterable, Mapping, Optional
 
 from funcy import cached_property, first
 
+from dvc.dependency.param import MissingParamsError
 from dvc.env import DVCLIVE_RESUME
 from dvc.exceptions import DvcException
 from dvc.path_info import PathInfo
@@ -350,9 +351,19 @@ class Experiments:
         )
         self.scm.add(self.args_file)
 
+    def _format_new_params_msg(self, new_params, config_path):
+        """Format an error message for when new parameters are identified"""
+        new_param_count = len(new_params)
+        pluralise = "s are" if new_param_count > 1 else " is"
+        param_list = ", ".join(new_params)
+        return (
+            f"{new_param_count} parameter{pluralise} missing "
+            f"from '{config_path}': {param_list}"
+        )
+
     def _update_params(self, params: dict):
         """Update experiment params files with the specified values."""
-        from dvc.utils.collections import merge_params
+        from dvc.utils.collections import NewParamsFound, merge_params
         from dvc.utils.serialize import MODIFIERS
 
         logger.debug("Using experiment params '%s'", params)
@@ -362,7 +373,11 @@ class Experiments:
             suffix = path.suffix.lower()
             modify_data = MODIFIERS[suffix]
             with modify_data(path, fs=self.repo.fs) as data:
-                merge_params(data, params[params_fname])
+                try:
+                    merge_params(data, params[params_fname], allow_new=False)
+                except NewParamsFound as e:
+                    msg = self._format_new_params_msg(e.new_params, path)
+                    raise MissingParamsError(msg)
 
         # Force params file changes to be staged in git
         # Otherwise in certain situations the changes to params file may be
@@ -458,6 +473,17 @@ class Experiments:
             "\tdvc exp branch <exp> <branch>\n"
         )
 
+    def _validate_new_ref(self, exp_ref: ExpRefInfo):
+        from .utils import check_ref_format
+
+        if not exp_ref.name:
+            return
+
+        check_ref_format(self.scm, exp_ref)
+
+        if self.scm.get_ref(str(exp_ref)):
+            raise ExperimentExistsError(exp_ref.name)
+
     @scm_locked
     def new(self, *args, checkpoint_resume: Optional[str] = None, **kwargs):
         """Create a new experiment.
@@ -470,6 +496,16 @@ class Experiments:
                 *args, resume_rev=checkpoint_resume, **kwargs
             )
 
+        name = kwargs.get("name", None)
+        baseline_sha = kwargs.get("baseline_rev") or self.repo.scm.get_rev()
+        exp_ref = ExpRefInfo(baseline_sha=baseline_sha, name=name)
+
+        try:
+            self._validate_new_ref(exp_ref)
+        except ExperimentExistsError as err:
+            if not (kwargs.get("force", False) or kwargs.get("reset", False)):
+                raise err
+
         return self._stash_exp(*args, **kwargs)
 
     def _resume_checkpoint(
@@ -638,8 +674,8 @@ class Experiments:
                 name=item.name,
                 branch=item.branch,
                 tmp_dir=base_tmp_dir,
-                cache_dir=self.repo.odb.local.cache_dir,
             )
+            executor.init_cache(self.repo, stash_rev)
             executors[stash_rev] = executor
 
         for ref in (EXEC_HEAD, EXEC_MERGE, EXEC_BASELINE):

dvc/repo/experiments/base.py

@@ -36,7 +36,7 @@ class BaselineMismatchError(DvcException):
 class ExperimentExistsError(DvcException):
     def __init__(self, name: str):
         msg = (
-            "Reproduced experiment conflicts with existing experiment "
+            "Experiment conflicts with existing experiment "
             f"'{name}'. To overwrite the existing experiment run:\n\n"
             "\tdvc exp run -f ...\n\n"
             "To run this experiment with a different name run:\n\n"

dvc/repo/experiments/executor/base.py

@@ -14,8 +14,6 @@ from typing import (
     Union,
 )
 
-from funcy import cached_property
-
 from dvc.env import DVC_EXP_AUTO_PUSH, DVC_EXP_GIT_REMOTE
 from dvc.exceptions import DvcException
 from dvc.path_info import PathInfo
@@ -24,8 +22,6 @@ from dvc.repo.experiments.base import (
     EXEC_BASELINE,
     EXEC_BRANCH,
     EXEC_CHECKPOINT,
-    EXEC_HEAD,
-    EXEC_MERGE,
     EXEC_NAMESPACE,
     EXPS_NAMESPACE,
     EXPS_STASH,
@@ -34,7 +30,6 @@ from dvc.repo.experiments.base import (
     ExpRefInfo,
     UnchangedExperimentError,
 )
-from dvc.scm import SCM
 from dvc.stage import PipelineStage
 from dvc.stage.monitor import CheckpointKilledError
 from dvc.stage.serialize import to_lockfile
@@ -117,44 +112,21 @@ class BaseExecutor(ABC):
         self._init_git(src, branch)
         self.name = name
 
-    def _init_git(self, scm: "Git", branch: Optional[str] = None):
-        """Init git repo and collect executor refs from the specified SCM."""
-        from dulwich.repo import Repo as DulwichRepo
-
-        DulwichRepo.init(os.fspath(self.root_dir))
-
-        cwd = os.getcwd()
-        os.chdir(self.root_dir)
-        try:
-            refspec = f"{EXEC_NAMESPACE}/"
-            scm.push_refspec(self.git_url, refspec, refspec)
-            if branch:
-                scm.push_refspec(self.git_url, branch, branch)
-                self.scm.set_ref(EXEC_BRANCH, branch, symbolic=True)
-            elif self.scm.get_ref(EXEC_BRANCH):
-                self.scm.remove_ref(EXEC_BRANCH)
-
-            if self.scm.get_ref(EXEC_CHECKPOINT):
-                self.scm.remove_ref(EXEC_CHECKPOINT)
-
-            # checkout EXEC_HEAD and apply EXEC_MERGE on top of it without
-            # committing
-            head = EXEC_BRANCH if branch else EXEC_HEAD
-            self.scm.checkout(head, detach=True)
-            merge_rev = self.scm.get_ref(EXEC_MERGE)
-            self.scm.merge(merge_rev, squash=True, commit=False)
-        finally:
-            os.chdir(cwd)
-
-    @cached_property
-    def scm(self):
-        return SCM(self.root_dir)
+    @abstractmethod
+    def _init_git(self, scm: "Git", branch: Optional[str] = None, **kwargs):
+        """Init git repo and populate it using exp refs from the specified
+        SCM instance.
+        """
 
     @property
     @abstractmethod
     def git_url(self) -> str:
         pass
 
+    @abstractmethod
+    def init_cache(self, dvc: "Repo", rev: str, run_cache: bool = True):
+        """Initialize DVC (cache)."""
+
     @property
     def dvc_dir(self) -> str:
         return os.path.join(self.root_dir, self._dvc_dir)
@@ -168,8 +140,7 @@ class BaseExecutor(ABC):
         return dict_sha256(exp_data)
 
     def cleanup(self):
-        self.scm.close()
-        del self.scm
+        pass
 
     # TODO: come up with better way to stash repro arguments
     @staticmethod
@@ -490,7 +461,7 @@ class BaseExecutor(ABC):
                 push_cache=push_cache,
                 run_cache=run_cache,
             )
-        except BaseException as exc:
+        except BaseException as exc:  # pylint: disable=broad-except
             logger.warning(
                 "Something went wrong while auto pushing experiment "
                 f"to the remote '{git_remote}': {exc}"

dvc/repo/experiments/executor/local.py

@@ -1,10 +1,25 @@
 import logging
 import os
 from tempfile import TemporaryDirectory
-from typing import Optional
+from typing import TYPE_CHECKING, Optional
+
+from funcy import cached_property
+
+from dvc.repo.experiments.base import (
+    EXEC_BRANCH,
+    EXEC_CHECKPOINT,
+    EXEC_HEAD,
+    EXEC_MERGE,
+    EXEC_NAMESPACE,
+)
+from dvc.scm import SCM
 
 from .base import BaseExecutor
 
+if TYPE_CHECKING:
+    from dvc.repo import Repo
+    from dvc.scm.git import Git
+
 logger = logging.getLogger(__name__)
 
 
@@ -36,6 +51,15 @@ class BaseLocalExecutor(BaseExecutor):
             root_dir = root_dir.replace(os.sep, "/")
         return f"file://{root_dir}"
 
+    @cached_property
+    def scm(self):
+        return SCM(self.root_dir)
+
+    def cleanup(self):
+        super().cleanup()
+        self.scm.close()
+        del self.scm
+
 
 class TempDirExecutor(BaseLocalExecutor):
     """Temp directory experiment executor."""
@@ -51,22 +75,46 @@ class TempDirExecutor(BaseLocalExecutor):
         self,
         *args,
         tmp_dir: Optional[str] = None,
-        cache_dir: Optional[str] = None,
         **kwargs,
     ):
         self._tmp_dir = ExpTemporaryDirectory(dir=tmp_dir)
         kwargs["root_dir"] = self._tmp_dir.name
         super().__init__(*args, **kwargs)
-        if cache_dir:
-            self._config(cache_dir)
         logger.debug("Init temp dir executor in dir '%s'", self._tmp_dir)
 
+    def _init_git(self, scm: "Git", branch: Optional[str] = None, **kwargs):
+        from dulwich.repo import Repo as DulwichRepo
+
+        DulwichRepo.init(os.fspath(self.root_dir))
+
+        refspec = f"{EXEC_NAMESPACE}/"
+        scm.push_refspec(self.git_url, refspec, refspec, **kwargs)
+        if branch:
+            scm.push_refspec(self.git_url, branch, branch, **kwargs)
+            self.scm.set_ref(EXEC_BRANCH, branch, symbolic=True)
+        elif self.scm.get_ref(EXEC_BRANCH):
+            self.scm.remove_ref(EXEC_BRANCH)
+
+        if self.scm.get_ref(EXEC_CHECKPOINT):
+            self.scm.remove_ref(EXEC_CHECKPOINT)
+
+        # checkout EXEC_HEAD and apply EXEC_MERGE on top of it without
+        # committing
+        head = EXEC_BRANCH if branch else EXEC_HEAD
+        self.scm.checkout(head, detach=True)
+        merge_rev = self.scm.get_ref(EXEC_MERGE)
+        self.scm.merge(merge_rev, squash=True, commit=False)
+
     def _config(self, cache_dir):
         local_config = os.path.join(self.dvc_dir, "config.local")
         logger.debug("Writing experiments local config '%s'", local_config)
-        with open(local_config, "w") as fobj:
+        with open(local_config, "w", encoding="utf-8") as fobj:
             fobj.write(f"[cache]\n    dir = {cache_dir}")
 
+    def init_cache(self, dvc: "Repo", rev: str, run_cache: bool = True):
+        """Initialize DVC (cache)."""
+        self._config(dvc.odb.local.cache_dir)
+
     def cleanup(self):
         super().cleanup()
         logger.debug("Removing tmpdir '%s'", self._tmp_dir)

dvc/repo/experiments/executor/ssh.py

@@ -0,0 +1,159 @@
+import logging
+import posixpath
+from contextlib import contextmanager
+from typing import TYPE_CHECKING, Callable, Optional
+
+from funcy import first
+
+from dvc.fs.ssh import SSHFileSystem
+from dvc.repo.experiments.base import (
+    EXEC_BRANCH,
+    EXEC_CHECKPOINT,
+    EXEC_HEAD,
+    EXEC_MERGE,
+    EXEC_NAMESPACE,
+)
+
+from .base import BaseExecutor
+
+if TYPE_CHECKING:
+    from dvc.machine import MachineManager
+    from dvc.repo import Repo
+    from dvc.scm.git import Git
+
+logger = logging.getLogger(__name__)
+
+
+@contextmanager
+def _sshfs(fs_factory, **kwargs):
+    if fs_factory:
+        with fs_factory() as fs:
+            yield fs
+        return
+    yield SSHFileSystem(**kwargs)
+
+
+class SSHExecutor(BaseExecutor):
+    """SSH experiment executor."""
+
+    WARN_UNTRACKED = True
+    QUIET = True
+
+    def __init__(
+        self,
+        *args,
+        host: Optional[str] = None,
+        port: Optional[int] = None,
+        username: Optional[str] = None,
+        fs_factory: Optional[Callable] = None,
+        **kwargs,
+    ):
+        assert host
+
+        self.host: str = host
+        self.port = port
+        self.username = username
+        self._fs_factory = fs_factory
+
+        super().__init__(*args, **kwargs)
+        logger.debug("Init SSH executor for host '%s'", self.host)
+
+    @classmethod
+    def gen_dirname(cls, name: Optional[str] = None):
+        from shortuuid import uuid
+
+        return "-".join([name or "dvc-exp", "executor", uuid()])
+
+    @classmethod
+    def from_machine(
+        cls,
+        manager: "MachineManager",
+        machine_name: Optional[str],
+        *args,
+        **kwargs,
+    ):
+        kwargs["root_dir"] = cls.gen_dirname(kwargs.get("name"))
+        kwargs.update(manager.get_executor_kwargs(machine_name))
+        return cls(*args, **kwargs)
+
+    def sshfs(self):
+        return _sshfs(self._fs_factory, host=self.host, port=self.port)
+
+    @property
+    def git_url(self) -> str:
+        user = f"{self.username}@" if self.username else ""
+        port = f":{self.port}" if self.port is not None else ""
+        path = f"{self.root_dir}" if self.root_dir else ""
+        if path and not posixpath.isabs(path):
+            path = f"/~/{path}"
+        return f"ssh://{user}{self.host}{port}{path}"
+
+    @property
+    def abs_url(self) -> str:
+        assert self._repo_abspath
+        user = f"{self.username}@" if self.username else ""
+        port = f":{self.port}" if self.port is not None else ""
+        return f"ssh://{user}{self.host}{port}{self._repo_abspath}"
+
+    @staticmethod
+    def _git_client_args(fs):
+        kwargs = {
+            "password": fs.fs_args.get("password"),
+            "key_filename": first(fs.fs_args.get("client_keys", [])),
+        }
+        return kwargs
+
+    def _init_git(self, scm: "Git", branch: Optional[str] = None, **kwargs):
+        with self.sshfs() as fs:
+            fs.makedirs(self.root_dir)
+            self._ssh_cmd(fs, "git init .")
+            self._ssh_cmd(fs, "git config user.name dvc-exp")
+            self._ssh_cmd(
+                fs, "git config user.email [email protected]"
+            )
+
+            result = self._ssh_cmd(fs, "pwd")
+            path = result.stdout.strip()
+            self._repo_abspath = path
+
+            # TODO: support multiple client key retries in git backends
+            # (see https://github.com/iterative/dvc/issues/6508)
+            kwargs.update(self._git_client_args(fs))
+            refspec = f"{EXEC_NAMESPACE}/"
+            scm.push_refspec(self.git_url, refspec, refspec, **kwargs)
+            if branch:
+                scm.push_refspec(self.git_url, branch, branch, **kwargs)
+                self._ssh_cmd(fs, f"git symbolic-ref {EXEC_BRANCH} {branch}")
+            else:
+                self._ssh_cmd(
+                    fs, f"git symbolic-ref -d {EXEC_BRANCH}", check=False
+                )
+            self._ssh_cmd(
+                fs, f"git update-ref -d {EXEC_CHECKPOINT}", check=False
+            )
+
+            # checkout EXEC_HEAD and apply EXEC_MERGE on top of it without
+            # committing
+            head = EXEC_BRANCH if branch else EXEC_HEAD
+            self._ssh_cmd(fs, f"git checkout {head}")
+            merge_rev = scm.get_ref(EXEC_MERGE)
+            self._ssh_cmd(fs, f"git merge --squash --no-commit {merge_rev}")
+
+    def _ssh_cmd(self, sshfs, cmd, chdir=None, **kwargs):
+        working_dir = chdir or self.root_dir
+        return sshfs.fs.execute(f"cd {working_dir};{cmd}", **kwargs)
+
+    def init_cache(self, dvc: "Repo", rev: str, run_cache: bool = True):
+        from dvc.objects.db import ODBManager, get_odb
+        from dvc.repo import Repo
+        from dvc.repo.push import push
+
+        cache_url = posixpath.join(
+            self.abs_url,
+            self._dvc_dir,
+            Repo.DVC_DIR,
+            ODBManager.CACHE_DIR,
+        )
+        with self.sshfs() as fs:
+            odb = get_odb(fs, fs.PATH_CLS(cache_url), **fs.config)
+            push(dvc, revs=[rev], run_cache=run_cache, odb=odb)

dvc/repo/experiments/init.py

@@ -0,0 +1,293 @@
+import logging
+import os
+from contextlib import contextmanager
+from typing import TYPE_CHECKING, Dict, Iterable, cast
+
+from funcy import compact, lremove, post_processing
+from rich.prompt import Confirm, Prompt
+from rich.rule import Rule
+from rich.syntax import Syntax
+
+from dvc.exceptions import DvcException
+from dvc.stage import PipelineStage
+from dvc.stage.serialize import to_pipeline_file
+from dvc.types import OptStr
+from dvc.utils.serialize import dumps_yaml
+
+if TYPE_CHECKING:
+    from dvc.repo import Repo
+    from dvc.dvcfile import DVCFile
+    from dvc.stage import Stage
+
+from dvc.ui import ui
+
+PROMPTS = {
+    "cmd": "[b]Command[/b] to execute",
+    "code": "Path to a [b]code[/b] file/directory",
+    "data": "Path to a [b]data[/b] file/directory",
+    "models": "Path to a [b]model[/b] file/directory",
+    "params": "Path to a [b]parameters[/b] file",
+    "metrics": "Path to a [b]metrics[/b] file",
+    "plots": "Path to a [b]plots[/b] file/directory",
+    "live": "Path to log [b]dvclive[/b] outputs",
+}
+
+
+class RichInputMixin:
+    """Prevents exc message from printing in the same line on Ctrl + D/C."""
+
+    @classmethod
+    def get_input(cls, *args, **kwargs) -> str:
+        try:
+            return super().get_input(*args, **kwargs)  # type: ignore[misc]
+        except (KeyboardInterrupt, EOFError):
+            ui.error_write()
+            raise
+
+
+class ConfirmationPrompt(RichInputMixin, Confirm):
+    def make_prompt(self, default):
+        prompt = self.prompt.copy()
+        prompt.end = ""
+
+        prompt.append(" [")
+        yes, no = self.choices
+        for idx, (val, choice) in enumerate(((True, yes), (False, no))):
+            if idx:
+                prompt.append("/")
+            if val == default:
+                prompt.append(choice.upper(), "green")
+            else:
+                prompt.append(choice)
+        prompt.append("] ")
+        return prompt
+
+
+class RequiredPrompt(RichInputMixin, Prompt):
+    def process_response(self, value: str):
+        from rich.prompt import InvalidResponse
+
+        ret = super().process_response(value)
+        if not ret:
+            raise InvalidResponse(
+                "[prompt.invalid]Response required. Please try again."
+            )
+        return ret
+
+    def render_default(self, default):
+        from rich.text import Text
+
+        return Text(f"{default!s}", "green")
+
+
+class SkippablePrompt(RequiredPrompt):
+    skip_value: str = "n"
+
+    def process_response(self, value: str):
+        ret = super().process_response(value)
+        return None if ret == self.skip_value else ret
+
+    def make_prompt(self, default):
+        prompt = self.prompt.copy()
+        prompt.end = ""
+
+        prompt.append(" [")
+        if (
+            default is not ...
+            and self.show_default
+            and isinstance(default, (str, self.response_type))
+        ):
+            _default = self.render_default(default)
+            prompt.append(_default)
+            prompt.append(", ")
+
+        prompt.append(f"{self.skip_value} to omit", style="italic")
+        prompt.append("]")
+        prompt.append(self.prompt_suffix)
+        return prompt
+
+
+@post_processing(dict)
+def _prompts(keys: Iterable[str], defaults: Dict[str, OptStr]):
+    for key in keys:
+        if key == "cmd":
+            prompt_cls = RequiredPrompt
+        else:
+            prompt_cls = SkippablePrompt
+        kwargs = {"default": defaults[key]} if key in defaults else {}
+        prompt = PROMPTS[key]
+        value = prompt_cls.ask(  # type: ignore[call-overload]
+            prompt, console=ui.error_console, **kwargs
+        )
+        yield key, value
+
+
+@contextmanager
+def _disable_logging(highest_level=logging.CRITICAL):
+    previous_level = logging.root.manager.disable
+
+    logging.disable(highest_level)
+
+    try:
+        yield
+    finally:
+        logging.disable(previous_level)
+
+
+PIPELINE_FILE_LINK = (
+    "https://dvc.org/doc/user-guide/project-structure/pipelines-files"
+)
+
+
+def init_interactive(
+    name: str,
+    defaults: Dict[str, str],
+    provided: Dict[str, str],
+    show_tree: bool = False,
+    live: bool = False,
+) -> Dict[str, str]:
+    command = provided.pop("cmd", None)
+    primary = lremove(provided.keys(), ["code", "data", "models", "params"])
+    secondary = lremove(
+        provided.keys(), ["live"] if live else ["metrics", "plots"]
+    )
+
+    ret: Dict[str, str] = {}
+    if not (command or primary or secondary):
+        return ret
+
+    message = ui.rich_text.assemble(
+        "This command will guide you to set up a ",
+        (name, "bright_blue"),
+        " stage in ",
+        ("dvc.yaml", "green"),
+        ".",
+    )
+    doc_link = ui.rich_text.assemble(
+        "See ", (PIPELINE_FILE_LINK, "repr.url"), "."
+    )
+    ui.error_write(message, doc_link, "", sep="\n", styled=True)
+
+    if not command:
+        ret.update(_prompts(["cmd"], defaults))
+        ui.write()
+    else:
+        ret.update({"cmd": command})
+
+    ui.write("Enter the paths for dependencies and outputs of the command.")
+    workspace = {**defaults, **provided}
+    if show_tree and workspace:
+        from rich.tree import Tree
+
+        tree = Tree(
+            "DVC assumes the following workspace structure:",
+            highlight=True,
+        )
+        if not live and "live" not in provided:
+            workspace.pop("live", None)
+        for value in sorted(workspace.values()):
+            tree.add(f"[green]{value}[/green]")
+        ui.error_write(tree, styled=True)
+
+    ui.error_write()
+    ret.update(_prompts(primary, defaults))
+    ret.update(_prompts(secondary, defaults))
+    return compact(ret)
+
+
+def _check_stage_exists(
+    dvcfile: "DVCFile", name: str, force: bool = False
+) -> None:
+    if not force and dvcfile.exists() and name in dvcfile.stages:
+        from dvc.stage.exceptions import DuplicateStageName
+
+        hint = "Use '--force' to overwrite."
+        raise DuplicateStageName(
+            f"Stage '{name}' already exists in 'dvc.yaml'. {hint}"
+        )
+
+
+def init(
+    repo: "Repo",
+    name: str = None,
+    type: str = "default",  # pylint: disable=redefined-builtin
+    defaults: Dict[str, str] = None,
+    overrides: Dict[str, str] = None,
+    interactive: bool = False,
+    force: bool = False,
+) -> "Stage":
+    from dvc.dvcfile import make_dvcfile
+
+    dvcfile = make_dvcfile(repo, "dvc.yaml")
+    name = name or type
+
+    _check_stage_exists(dvcfile, name, force=force)
+
+    defaults = defaults or {}
+    overrides = overrides or {}
+
+    with_live = type == "live"
+    if interactive:
+        defaults = init_interactive(
+            name,
+            defaults=defaults,
+            live=with_live,
+            provided=overrides,
+            show_tree=True,
+        )
+    else:
+        if with_live:
+            # suppress `metrics`/`params` if live is selected, unless
+            # it is also provided via overrides/cli.
+            # This makes output to be a checkpoint as well.
+            defaults.pop("metrics", None)
+            defaults.pop("params", None)
+        else:
+            defaults.pop("live", None)  # suppress live otherwise
+
+    context: Dict[str, str] = {**defaults, **overrides}
+    assert "cmd" in context
+
+    params_kv = []
+    params = context.get("params")
+    if params:
+        from dvc.utils.serialize import LOADERS
+
+        assert isinstance(params, str)
+        _, ext = os.path.splitext(params)
+        params_kv = [{params: list(LOADERS[ext](params))}]
+
+    checkpoint_out = bool(context.get("live"))
+    models = context.get("models")
+    stage = repo.stage.create(
+        name=name,
+        cmd=context["cmd"],
+        deps=compact([context.get("code"), context.get("data")]),
+        params=params_kv,
+        metrics_no_cache=compact([context.get("metrics")]),
+        plots_no_cache=compact([context.get("plots")]),
+        live=context.get("live"),
+        force=force,
+        **{"checkpoints" if checkpoint_out else "outs": compact([models])},
+    )
+
+    if interactive:
+        ui.write(Rule(style="green"), styled=True)
+        _yaml = dumps_yaml(to_pipeline_file(cast(PipelineStage, stage)))
+        syn = Syntax(_yaml, "yaml", theme="ansi_dark")
+        ui.error_write(syn, styled=True)
+
+    if not interactive or ConfirmationPrompt.ask(
+        "Do you want to add the above contents to dvc.yaml?",
+        console=ui.error_console,
+        default=True,
+    ):
+        scm = repo.scm
+        with _disable_logging(), scm.track_file_changes(autostage=True):
+            stage.dump(update_lock=False)
+            stage.ignore_outs()
+            if params:
+                scm.track_file(params)
+    else:
+        raise DvcException("Aborting ...")
+    return stage

dvc/repo/experiments/pull.py

@@ -4,7 +4,7 @@ from dvc.exceptions import DvcException, InvalidArgumentError
 from dvc.repo import locked
 from dvc.repo.scm_context import scm_context
 
-from .utils import exp_commits, remote_exp_refs_by_name
+from .utils import exp_commits, resolve_exp_ref
 
 logger = logging.getLogger(__name__)
 
@@ -14,7 +14,11 @@ logger = logging.getLogger(__name__)
 def pull(
     repo, git_remote, exp_name, *args, force=False, pull_cache=False, **kwargs
 ):
-    exp_ref = _get_exp_ref(repo, git_remote, exp_name)
+    exp_ref = resolve_exp_ref(repo.scm, exp_name, git_remote)
+    if not exp_ref:
+        raise InvalidArgumentError(
+            f"Experiment '{exp_name}' does not exist in '{git_remote}'"
+        )
 
     def on_diverged(refname: str, rev: str) -> bool:
         if repo.scm.get_ref(refname) == rev:
@@ -35,33 +39,6 @@ def pull(
         _pull_cache(repo, exp_ref, **kwargs)
 
 
-def _get_exp_ref(repo, git_remote, exp_name):
-    if exp_name.startswith("refs/"):
-        return exp_name
-
-    exp_refs = list(remote_exp_refs_by_name(repo.scm, git_remote, exp_name))
-    if not exp_refs:
-        raise InvalidArgumentError(
-            f"Experiment '{exp_name}' does not exist in '{git_remote}'"
-        )
-    if len(exp_refs) > 1:
-        cur_rev = repo.scm.get_rev()
-        for info in exp_refs:
-            if info.baseline_sha == cur_rev:
-                return info
-        msg = [
-            (
-                f"Ambiguous name '{exp_name}' refers to multiple "
-                "experiments in '{git_remote}'. Use full refname to pull one "
-                "of the following:"
-            ),
-            "",
-        ]
-        msg.extend([f"\t{info}" for info in exp_refs])
-        raise InvalidArgumentError("\n".join(msg))
-    return exp_refs[0]
-
-
 def _pull_cache(repo, exp_ref, dvc_remote=None, jobs=None, run_cache=False):
     revs = list(exp_commits(repo.scm, [exp_ref]))
     logger.debug("dvc fetch experiment '%s'", exp_ref)

dvc/repo/experiments/push.py

@@ -2,10 +2,9 @@ import logging
 
 from dvc.exceptions import DvcException, InvalidArgumentError
 from dvc.repo import locked
-from dvc.repo.experiments.base import ExpRefInfo
 from dvc.repo.scm_context import scm_context
 
-from .utils import exp_commits, exp_refs_by_name
+from .utils import exp_commits, resolve_exp_ref
 
 logger = logging.getLogger(__name__)
 
@@ -21,7 +20,11 @@ def push(
     push_cache=False,
     **kwargs,
 ):
-    exp_ref = _get_exp_ref(repo, exp_name)
+    exp_ref = resolve_exp_ref(repo.scm, exp_name)
+    if not exp_ref:
+        raise InvalidArgumentError(
+            f"'{exp_name}' is not a valid experiment name"
+        )
 
     def on_diverged(refname: str, rev: str) -> bool:
         if repo.scm.get_ref(refname) == rev:
@@ -42,33 +45,6 @@ def push(
         _push_cache(repo, exp_ref, **kwargs)
 
 
-def _get_exp_ref(repo, exp_name: str) -> ExpRefInfo:
-    if exp_name.startswith("refs/"):
-        return ExpRefInfo.from_ref(exp_name)
-
-    exp_refs = list(exp_refs_by_name(repo.scm, exp_name))
-    if not exp_refs:
-        raise InvalidArgumentError(
-            f"'{exp_name}' is not a valid experiment name"
-        )
-    if len(exp_refs) > 1:
-        cur_rev = repo.scm.get_rev()
-        for info in exp_refs:
-            if info.baseline_sha == cur_rev:
-                return info
-        msg = [
-            (
-                f"Ambiguous name '{exp_name}' refers to multiple "
-                "experiments. Use full refname to push one of the "
-                "following:"
-            ),
-            "",
-        ]
-        msg.extend([f"\t{info}" for info in exp_refs])
-        raise InvalidArgumentError("\n".join(msg))
-    return exp_refs[0]
-
-
 def _push_cache(repo, exp_ref, dvc_remote=None, jobs=None, run_cache=False):
     revs = list(exp_commits(repo.scm, [exp_ref]))
     logger.debug("dvc push experiment '%s'", exp_ref)

dvc/repo/experiments/remove.py

@@ -6,8 +6,7 @@ from dvc.repo import locked
 from dvc.repo.scm_context import scm_context
 from dvc.scm.base import RevError
 
-from .base import EXPS_NAMESPACE, ExpRefInfo
-from .utils import exp_refs, exp_refs_by_name, remove_exp_refs
+from .utils import exp_refs, remove_exp_refs, resolve_exp_ref
 
 logger = logging.getLogger(__name__)
 
@@ -19,6 +18,7 @@ def remove(
     exp_names=None,
     queue=False,
     clear_all=False,
+    remote=None,
     **kwargs,
 ):
     if not any([exp_names, queue, clear_all]):
@@ -31,13 +31,7 @@ def remove(
         removed += _clear_all(repo)
 
     if exp_names:
-        remained = _remove_commited_exps(repo, exp_names)
-        remained = _remove_queued_exps(repo, remained)
-        if remained:
-            raise InvalidArgumentError(
-                "'{}' is not a valid experiment".format(";".join(remained))
-            )
-        removed += len(exp_names) - len(remained)
+        removed += _remove_exp_by_names(repo, remote, exp_names)
     return removed
 
 
@@ -67,46 +61,24 @@ def _get_exp_stash_index(repo, ref_or_rev: str) -> Optional[int]:
     return None
 
 
-def _get_exp_ref(repo, exp_name: str) -> Optional[ExpRefInfo]:
-    cur_rev = repo.scm.get_rev()
-    if exp_name.startswith(EXPS_NAMESPACE):
-        if repo.scm.get_ref(exp_name):
-            return ExpRefInfo.from_ref(exp_name)
-    else:
-        exp_ref_list = list(exp_refs_by_name(repo.scm, exp_name))
-        if exp_ref_list:
-            return _get_ref(exp_ref_list, exp_name, cur_rev)
-    return None
-
-
-def _get_ref(ref_infos, name, cur_rev) -> Optional[ExpRefInfo]:
-    if len(ref_infos) > 1:
-        for info in ref_infos:
-            if info.baseline_sha == cur_rev:
-                return info
-        msg = [
-            (
-                f"Ambiguous name '{name}' refers to multiple "
-                "experiments. Use full refname to remove one of "
-                "the following:"
-            )
-        ]
-        msg.extend([f"\t{info}" for info in ref_infos])
-        raise InvalidArgumentError("\n".join(msg))
-    return ref_infos[0]
-
-
-def _remove_commited_exps(repo, refs: List[str]) -> List[str]:
+def _remove_commited_exps(
+    repo, remote: Optional[str], exp_names: List[str]
+) -> List[str]:
     remain_list = []
     remove_list = []
-    for ref in refs:
-        ref_info = _get_exp_ref(repo, ref)
+    for exp_name in exp_names:
+        ref_info = resolve_exp_ref(repo.scm, exp_name, remote)
+
         if ref_info:
             remove_list.append(ref_info)
         else:
-            remain_list.append(ref)
+            remain_list.append(exp_name)
     if remove_list:
-        remove_exp_refs(repo.scm, remove_list)
+        if not remote:
+            remove_exp_refs(repo.scm, remove_list)
+        else:
+            for ref_info in remove_list:
+                repo.scm.push_refspec(remote, None, str(ref_info))
     return remain_list
 
 
@@ -119,3 +91,14 @@ def _remove_queued_exps(repo, refs_or_revs: List[str]) -> List[str]:
         else:
             repo.experiments.stash.drop(stash_index)
     return remain_list
+
+
+def _remove_exp_by_names(repo, remote, exp_names: List[str]) -> int:
+    remained = _remove_commited_exps(repo, remote, exp_names)
+    if not remote:
+        remained = _remove_queued_exps(repo, remained)
+    if remained:
+        raise InvalidArgumentError(
+            "'{}' is not a valid experiment".format(";".join(remained))
+        )
+    return len(exp_names) - len(remained)

dvc/repo/experiments/utils.py

@@ -1,5 +1,6 @@
 from typing import Generator, Iterable, Optional, Set
 
+from dvc.exceptions import InvalidArgumentError
 from dvc.scm.git import Git
 
 from .base import (
@@ -115,3 +116,53 @@ def fix_exp_head(scm: "Git", ref: Optional[str]) -> Optional[str]:
         if name == "HEAD" and scm.get_ref(EXEC_BASELINE):
             return "".join((EXEC_BASELINE, tail))
     return ref
+
+
+def resolve_exp_ref(
+    scm, exp_name: str, git_remote: Optional[str] = None
+) -> Optional[ExpRefInfo]:
+    if exp_name.startswith("refs/"):
+        return ExpRefInfo.from_ref(exp_name)
+
+    if git_remote:
+        exp_ref_list = list(remote_exp_refs_by_name(scm, git_remote, exp_name))
+    else:
+        exp_ref_list = list(exp_refs_by_name(scm, exp_name))
+
+    if not exp_ref_list:
+        return None
+    if len(exp_ref_list) > 1:
+        cur_rev = scm.get_rev()
+        for info in exp_ref_list:
+            if info.baseline_sha == cur_rev:
+                return info
+        if git_remote:
+            msg = [
+                (
+                    f"Ambiguous name '{exp_name}' refers to multiple "
+                    "experiments. Use full refname to push one of the "
+                    "following:"
+                ),
+                "",
+            ]
+        else:
+            msg = [
+                (
+                    f"Ambiguous name '{exp_name}' refers to multiple "
+                    f"experiments in '{git_remote}'. Use full refname to pull "
+                    "one of the following:"
+                ),
+                "",
+            ]
+        msg.extend([f"\t{info}" for info in exp_ref_list])
+        raise InvalidArgumentError("\n".join(msg))
+    return exp_ref_list[0]
+
+
+def check_ref_format(scm: "Git", ref: ExpRefInfo):
+    # "/" forbidden, only in dvc exp as we didn't support it for now.
+    if not scm.check_ref_format(str(ref)) or "/" in ref.name:
+        raise InvalidArgumentError(
+            f"Invalid exp name {ref.name}, the exp name must follow rules in "
+            "https://git-scm.com/docs/git-check-ref-format"
+        )

dvc/repo/install.py

@@ -1,2 +1,69 @@
-def install(self, use_pre_commit_tool):
-    self.scm.install(use_pre_commit_tool)
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from dvc.repo import Repo
+    from dvc.scm.git import Git
+
+
+def pre_commit_install(scm: "Git") -> None:
+    import os
+
+    from dvc.utils.serialize import modify_yaml
+
+    config_path = os.path.join(scm.root_dir, ".pre-commit-config.yaml")
+    with modify_yaml(config_path) as config:
+        entry = {
+            "repo": "https://github.com/iterative/dvc",
+            "rev": "master",
+            "hooks": [
+                {
+                    "id": "dvc-pre-commit",
+                    "language_version": "python3",
+                    "stages": ["commit"],
+                },
+                {
+                    "id": "dvc-pre-push",
+                    "language_version": "python3",
+                    "stages": ["push"],
+                },
+                {
+                    "id": "dvc-post-checkout",
+                    "language_version": "python3",
+                    "stages": ["post-checkout"],
+                    "always_run": True,
+                },
+            ],
+        }
+
+        config["repos"] = config.get("repos", [])
+        if entry not in config["repos"]:
+            config["repos"].append(entry)
+
+
+def install_hooks(scm: "Git") -> None:
+    hooks = ["post-checkout", "pre-commit", "pre-push"]
+    for hook in hooks:
+        scm.verify_hook(hook)
+    for hook in hooks:
+        scm.install_hook(hook, f"exec dvc git-hook {hook} $@")
+
+
+def install(self: "Repo", use_pre_commit_tool: bool = False) -> None:
+    """Adds dvc commands to SCM hooks for the repo.
+
+    If use_pre_commit_tool is set and pre-commit is installed it will be used
+    to install the hooks.
+    """
+    from dvc.scm.git import Git
+
+    scm = self.scm
+    if not isinstance(scm, Git):
+        return
+
+    driver = "dvc git-hook merge-driver --ancestor %O --our %A --their %B "
+    scm.install_merge_driver("dvc", "DVC merge driver", driver)
+
+    if use_pre_commit_tool:
+        return pre_commit_install(scm)
+
+    return install_hooks(scm)

dvc/repo/live.py

@@ -1,7 +1,8 @@
 import logging
-import os
 from typing import TYPE_CHECKING, List, Optional
 
+from dvc.render.utils import render
+
 logger = logging.getLogger(__name__)
 
 if TYPE_CHECKING:
@@ -15,10 +16,9 @@ def create_summary(out):
 
     metrics, plots = out.repo.live.show(str(out.path_info))
 
-    html_path = out.path_info.with_suffix(".html")
+    html_path = out.path_info.fspath + "_dvc_plots"
 
-    out.repo.plots.write_html(html_path, plots, metrics)
-    logger.info(f"\nfile://{os.path.abspath(html_path)}")
+    render(out.repo, plots, metrics=metrics, path=html_path, refresh_seconds=5)
 
 
 def summary_path_info(out: "Output") -> Optional["PathInfo"]:

dvc/repo/plots/__init__.py

@@ -3,13 +3,12 @@ import io
 import logging
 import os
 from collections import OrderedDict
-from typing import TYPE_CHECKING, Callable, Dict, List, Optional
+from typing import TYPE_CHECKING, Callable, Dict, Generator, List, Optional
 
 from funcy import cached_property, first, project
 
 from dvc.exceptions import DvcException
-from dvc.repo.plots.data import PlotMetricTypeError
-from dvc.types import StrPath
+from dvc.render.vega import PlotMetricTypeError
 from dvc.utils import (
     error_handler,
     errored_revisions,
@@ -21,6 +20,7 @@ from dvc.utils.serialize import LOADERS
 if TYPE_CHECKING:
     from dvc.output import Output
     from dvc.repo import Repo
+    from dvc.types import DvcPath
 
 logger = logging.getLogger(__name__)
 
@@ -48,10 +48,10 @@ class Plots:
         recursive: bool = False,
         onerror: Optional[Callable] = None,
         props: Optional[Dict] = None,
-    ) -> Dict[str, Dict]:
+    ) -> Generator[Dict, None, None]:
         """Collects all props and data for plots.
 
-        Returns a structure like:
+        Generator yielding a structure like:
             {rev: {plots.csv: {
                 props: {x: ..., "header": ..., ...},
                 data: "unstructured data (as stored for given extension)",
@@ -60,30 +60,20 @@ class Plots:
         from dvc.utils.collections import ensure_list
 
         targets = ensure_list(targets)
-        data: Dict[str, Dict] = {}
         for rev in self.repo.brancher(revs=revs):
             # .brancher() adds unwanted workspace
             if revs is not None and rev not in revs:
                 continue
             rev = rev or "workspace"
-            data[rev] = self._collect_from_revision(
-                revision=rev,
-                targets=targets,
-                recursive=recursive,
-                onerror=onerror,
-                props=props,
-            )
-
-        errored = errored_revisions(data)
-        if errored:
-            from dvc.ui import ui
-
-            ui.error_write(
-                "DVC failed to load some plots for following revisions: "
-                f"'{', '.join(errored)}'."
-            )
-
-        return data
+            yield {
+                rev: self._collect_from_revision(
+                    revision=rev,
+                    targets=targets,
+                    recursive=recursive,
+                    onerror=onerror,
+                    props=props,
+                )
+            }
 
     @error_handler
     def _collect_from_revision(
@@ -113,57 +103,45 @@ class Plots:
             props = props or {}
 
             for path, repo_path in plot_files:
-                res[repo_path] = {"props": rev_props}
+                joined_props = {**rev_props, **props}
+                res[repo_path] = {"props": joined_props}
                 res[repo_path].update(
                     parse(
                         fs,
                         path,
-                        rev_props=rev_props,
-                        props=props,
+                        props=joined_props,
                         onerror=onerror,
                     )
                 )
         return res
 
-    @staticmethod
-    def render(data, revs=None, props=None, templates=None):
-        """Renders plots"""
-        props = props or {}
-
-        # Merge data by plot file and apply overriding props
-        plots = _prepare_plots(data, revs, props)
-
-        result = {}
-        for datafile, desc in plots.items():
-            rendered = _render(
-                datafile,
-                desc["data"],
-                desc["props"],
-                templates,
-            )
-            if rendered:
-                result[datafile] = rendered
-
-        return result
-
     def show(
         self,
         targets: List[str] = None,
         revs=None,
         props=None,
-        templates=None,
         recursive=False,
         onerror=None,
     ):
         if onerror is None:
             onerror = onerror_collect
-        data = self.collect(
-            targets, revs, recursive, onerror=onerror, props=props
-        )
 
-        if templates is None:
-            templates = self.templates
-        return self.render(data, revs, props, templates)
+        data: Dict[str, Dict] = {}
+        for rev_data in self.collect(
+            targets, revs, recursive, onerror=onerror, props=props
+        ):
+            data.update(rev_data)
+
+        errored = errored_revisions(data)
+        if errored:
+            from dvc.ui import ui
+
+            ui.error_write(
+                "DVC failed to load some plots for following revisions: "
+                f"'{', '.join(errored)}'."
+            )
+
+        return data
 
     def diff(self, *args, **kwargs):
         from .diff import diff
@@ -187,7 +165,7 @@ class Plots:
         props = props or {}
         template = props.get("template")
         if template:
-            self.templates.get_template(template)
+            self.templates.load(template)
 
         (out,) = self.repo.find_outs_by_path(path)
         if not out.plot and unset is not None:
@@ -217,26 +195,6 @@ class Plots:
 
         return PlotTemplates(self.repo.dvc_dir)
 
-    def write_html(
-        self,
-        path: StrPath,
-        plots: Dict[str, Dict],
-        metrics: Optional[Dict[str, Dict]] = None,
-        html_template_path: Optional[StrPath] = None,
-    ):
-        if not html_template_path:
-            html_template_path = self.repo.config.get("plots", {}).get(
-                "html_template", None
-            )
-            if html_template_path and not os.path.isabs(html_template_path):
-                html_template_path = os.path.join(
-                    self.repo.dvc_dir, html_template_path
-                )
-
-        from dvc.utils.html import write
-
-        write(path, plots, metrics, html_template_path)
-
 
 def _is_plot(out: "Output") -> bool:
     return bool(out.plot) or bool(out.live)
@@ -247,7 +205,7 @@ def _collect_plots(
     targets: List[str] = None,
     rev: str = None,
     recursive: bool = False,
-) -> Dict[str, Dict]:
+) -> Dict["DvcPath", Dict]:
     from dvc.repo.collect import collect
 
     plots, path_infos = collect(
@@ -264,17 +222,19 @@ def _collect_plots(
 
 
 @error_handler
-def parse(fs, path, props=None, rev_props=None, **kwargs):
+def parse(fs, path, props=None, **kwargs):
     props = props or {}
-    rev_props = rev_props or {}
     _, extension = os.path.splitext(path)
     if extension in (".tsv", ".csv"):
-        header = {**rev_props, **props}.get("header", True)
+        header = props.get("header", True)
         if extension == ".csv":
             return _load_sv(path=path, fs=fs, delimiter=",", header=header)
         return _load_sv(path=path, fs=fs, delimiter="\t", header=header)
     if extension in LOADERS or extension in (".yml", ".yaml"):
         return LOADERS[extension](path=path, fs=fs)
+    if extension in (".jpeg", ".jpg", ".gif", ".png"):
+        with fs.open(path, "rb") as fd:
+            return fd.read()
     raise PlotMetricTypeError(path)
 
 
@@ -291,89 +251,6 @@ def _plot_props(out: "Output") -> Dict:
     return project(out.plot, PLOT_PROPS)
 
 
-def _prepare_plots(data, revs, props):
-    """Groups data by plot file.
-
-    Also resolves props conflicts between revs and applies global props.
-    """
-    # we go in order revs are supplied on props conflict first ones win.
-    revs = iter(data) if revs is None else revs
-
-    plots, props_revs = {}, {}
-    for rev in revs:
-        # Asked for revision without data
-        if rev not in data:
-            continue
-
-        for datafile, desc in data[rev].get("data", {}).items():
-            # We silently skip on an absent data file,
-            # see also try/except/pass in .collect()
-            if "data" not in desc:
-                continue
-
-            # props from command line overwrite plot props from out definition
-            full_props = {**desc["props"], **props}
-
-            if datafile in plots:
-                saved = plots[datafile]
-                if saved["props"] != full_props:
-                    logger.warning(
-                        f"Inconsistent plot props for '{datafile}' in "
-                        f"'{props_revs[datafile]}' and '{rev}'. "
-                        f"Going to use ones from '{props_revs[datafile]}'"
-                    )
-
-                saved["data"][rev] = desc["data"]
-            else:
-                plots[datafile] = {
-                    "props": full_props,
-                    "data": {rev: desc["data"]},
-                }
-                # Save rev we got props from
-                props_revs[datafile] = rev
-
-    return plots
-
-
-def _render(datafile, datas, props, templates):
-    from .data import PlotData, plot_data
-
-    # Copy it to not modify a passed value
-    props = props.copy()
-
-    # Add x and y to fields if set
-    fields = props.get("fields")
-    if fields is not None:
-        fields = {*fields, props.get("x"), props.get("y")} - {None}
-
-    template = templates.load(props.get("template") or "default")
-
-    # If x is not set add index field
-    if not props.get("x") and template.has_anchor("x"):
-        props["append_index"] = True
-        props["x"] = PlotData.INDEX_FIELD
-
-    # Parse all data, preprocess it and collect as a list of dicts
-    data = []
-    for rev, unstructured_data in datas.items():
-        rev_data = plot_data(datafile, rev, unstructured_data).to_datapoints(
-            fields=fields,
-            path=props.get("path"),
-            append_index=props.get("append_index", False),
-        )
-        data.extend(rev_data)
-
-    # If y is not set then use last field not used yet
-    if data:
-        if not props.get("y") and template.has_anchor("y"):
-            fields = list(first(data))
-            skip = (PlotData.REVISION_FIELD, props.get("x"))
-            props["y"] = first(f for f in reversed(fields) if f not in skip)
-
-        return template.render(data, props=props)
-    return None
-
-
 def _load_sv(path, fs, delimiter=",", header=True):
     with fs.open(path, "r") as fd:
         content = fd.read()

dvc/repo/plots/data.py

@@ -1,138 +0,0 @@
-import os
-from copy import copy
-
-from funcy import first
-
-from dvc.exceptions import DvcException
-from dvc.utils.serialize import ParseError
-
-
-class PlotMetricTypeError(DvcException):
-    def __init__(self, file):
-        super().__init__(
-            "'{}' - file type error\n"
-            "Only JSON, YAML, CSV and TSV formats are supported.".format(file)
-        )
-
-
-class PlotDataStructureError(DvcException):
-    def __init__(self):
-        super().__init__(
-            "Plot data extraction failed. Please see "
-            "https://man.dvc.org/plot for supported data formats."
-        )
-
-
-class PlotParsingError(ParseError):
-    def __init__(self, path, revision):
-        self.path = path
-        self.revision = revision
-
-        super().__init__(path, f"revision: '{revision}'")
-
-
-def plot_data(filename, revision, content):
-    _, extension = os.path.splitext(filename.lower())
-    if extension in (".json", ".yaml"):
-        return DictData(filename, revision, content)
-    if extension in (".csv", ".tsv"):
-        return ListData(filename, revision, content)
-    raise PlotMetricTypeError(filename)
-
-
-def _filter_fields(data_points, filename, revision, fields=None, **kwargs):
-    if not fields:
-        return data_points
-    assert isinstance(fields, set)
-
-    new_data = []
-    for data_point in data_points:
-        new_dp = copy(data_point)
-
-        keys = set(data_point.keys())
-        if keys & fields != fields:
-            raise DvcException(
-                "Could not find fields: '{}' for '{}' at '{}'.".format(
-                    ", ".join(fields), filename, revision
-                )
-            )
-
-        to_del = keys - fields
-        for key in to_del:
-            del new_dp[key]
-        new_data.append(new_dp)
-    return new_data
-
-
-def _lists(dictionary):
-    for _, value in dictionary.items():
-        if isinstance(value, dict):
-            yield from _lists(value)
-        elif isinstance(value, list):
-            yield value
-
-
-def _find_data(data, fields=None, **kwargs):
-    if not isinstance(data, dict):
-        return data
-
-    if not fields:
-        # just look for first list of dicts
-        fields = set()
-
-    for lst in _lists(data):
-        if (
-            all(isinstance(dp, dict) for dp in lst)
-            and set(first(lst).keys()) & fields == fields
-        ):
-            return lst
-    raise PlotDataStructureError()
-
-
-def _append_index(data_points, append_index=False, **kwargs):
-    if not append_index or PlotData.INDEX_FIELD in first(data_points).keys():
-        return data_points
-
-    for index, data_point in enumerate(data_points):
-        data_point[PlotData.INDEX_FIELD] = index
-    return data_points
-
-
-def _append_revision(data_points, revision, **kwargs):
-    for data_point in data_points:
-        data_point[PlotData.REVISION_FIELD] = revision
-    return data_points
-
-
-class PlotData:
-    REVISION_FIELD = "rev"
-    INDEX_FIELD = "step"
-
-    def __init__(self, filename, revision, content, **kwargs):
-        self.filename = filename
-        self.revision = revision
-        self.content = content
-
-    def _processors(self):
-        return [_filter_fields, _append_index, _append_revision]
-
-    def to_datapoints(self, **kwargs):
-        data = self.content
-
-        for data_proc in self._processors():
-            data = data_proc(
-                data, filename=self.filename, revision=self.revision, **kwargs
-            )
-        return data
-
-
-class DictData(PlotData):
-    # For files usually parsed as dicts: eg JSON, Yaml
-    def _processors(self):
-        parent_processors = super()._processors()
-        return [_find_data] + parent_processors
-
-
-class ListData(PlotData):
-    # For files parsed as list: CSV, TSV
-    pass

dvc/repo/plots/template.py

@@ -1,15 +1,23 @@
 import json
 import os
-from typing import Any, Dict, Optional
+from typing import TYPE_CHECKING, Iterable, Optional
+
+try:
+    import importlib_resources
+except ImportError:
+    import importlib.resources as importlib_resources  # type: ignore[no-redef]
 
 from funcy import cached_property
 
 from dvc.exceptions import DvcException
 
+if TYPE_CHECKING:
+    from dvc.types import StrPath
+
 
 class TemplateNotFoundError(DvcException):
-    def __init__(self, path):
-        super().__init__(f"Template '{path}' not found.")
+    def __init__(self, name):
+        super().__init__(f"Template '{name}' not found.")
 
 
 class BadTemplateError(DvcException):
@@ -29,25 +37,9 @@ class Template:
     EXTENSION = ".json"
     ANCHOR = "<DVC_METRIC_{}>"
 
-    DEFAULT_CONTENT: Optional[Dict[str, Any]] = None
-    DEFAULT_NAME: Optional[str] = None
-
-    def __init__(self, content=None, name=None):
-        if content:
-            self.content = content
-        else:
-            self.content = (
-                json.dumps(
-                    self.DEFAULT_CONTENT,
-                    indent=self.INDENT,
-                    separators=self.SEPARATORS,
-                )
-                + "\n"
-            )
-
-        self.name = name or self.DEFAULT_NAME
-        assert self.content and self.name
-        self.filename = self.name + self.EXTENSION
+    def __init__(self, content, name):
+        self.content = content
+        self.name = name
 
     def render(self, data, props=None):
         props = props or {}
@@ -55,7 +47,7 @@ class Template:
         if self._anchor_str("data") not in self.content:
             anchor = self.anchor("data")
             raise BadTemplateError(
-                f"Template '{self.filename}' is not using '{anchor}' anchor"
+                f"Template '{self.name}' is not using '{anchor}' anchor"
             )
 
         if props.get("x"):
@@ -106,466 +98,86 @@ class Template:
             raise NoFieldInDataError(field)
 
 
-class DefaultTemplate(Template):
-    DEFAULT_NAME = "default"
-
-    DEFAULT_CONTENT = {
-        "$schema": "https://vega.github.io/schema/vega-lite/v4.json",
-        "data": {"values": Template.anchor("data")},
-        "title": Template.anchor("title"),
-        "width": 300,
-        "height": 300,
-        "mark": {"type": "line"},
-        "encoding": {
-            "x": {
-                "field": Template.anchor("x"),
-                "type": "quantitative",
-                "title": Template.anchor("x_label"),
-            },
-            "y": {
-                "field": Template.anchor("y"),
-                "type": "quantitative",
-                "title": Template.anchor("y_label"),
-                "scale": {"zero": False},
-            },
-            "color": {"field": "rev", "type": "nominal"},
-        },
-    }
-
-
-class ConfusionTemplate(Template):
-    DEFAULT_NAME = "confusion"
-    DEFAULT_CONTENT = {
-        "$schema": "https://vega.github.io/schema/vega-lite/v4.json",
-        "data": {"values": Template.anchor("data")},
-        "title": Template.anchor("title"),
-        "facet": {"field": "rev", "type": "nominal"},
-        "spec": {
-            "transform": [
-                {
-                    "aggregate": [{"op": "count", "as": "xy_count"}],
-                    "groupby": [Template.anchor("y"), Template.anchor("x")],
-                },
-                {
-                    "impute": "xy_count",
-                    "groupby": ["rev", Template.anchor("y")],
-                    "key": Template.anchor("x"),
-                    "value": 0,
-                },
-                {
-                    "impute": "xy_count",
-                    "groupby": ["rev", Template.anchor("x")],
-                    "key": Template.anchor("y"),
-                    "value": 0,
-                },
-                {
-                    "joinaggregate": [
-                        {"op": "max", "field": "xy_count", "as": "max_count"}
-                    ],
-                    "groupby": [],
-                },
-                {
-                    "calculate": "datum.xy_count / datum.max_count",
-                    "as": "percent_of_max",
-                },
-            ],
-            "encoding": {
-                "x": {
-                    "field": Template.anchor("x"),
-                    "type": "nominal",
-                    "sort": "ascending",
-                    "title": Template.anchor("x_label"),
-                },
-                "y": {
-                    "field": Template.anchor("y"),
-                    "type": "nominal",
-                    "sort": "ascending",
-                    "title": Template.anchor("y_label"),
-                },
-            },
-            "layer": [
-                {
-                    "mark": "rect",
-                    "width": 300,
-                    "height": 300,
-                    "encoding": {
-                        "color": {
-                            "field": "xy_count",
-                            "type": "quantitative",
-                            "title": "",
-                            "scale": {"domainMin": 0, "nice": True},
-                        }
-                    },
-                },
-                {
-                    "mark": "text",
-                    "encoding": {
-                        "text": {"field": "xy_count", "type": "quantitative"},
-                        "color": {
-                            "condition": {
-                                "test": "datum.percent_of_max > 0.5",
-                                "value": "white",
-                            },
-                            "value": "black",
-                        },
-                    },
-                },
-            ],
-        },
-    }
-
-
-class NormalizedConfusionTemplate(Template):
-    DEFAULT_NAME = "confusion_normalized"
-    DEFAULT_CONTENT = {
-        "$schema": "https://vega.github.io/schema/vega-lite/v4.json",
-        "data": {"values": Template.anchor("data")},
-        "title": Template.anchor("title"),
-        "facet": {"field": "rev", "type": "nominal"},
-        "spec": {
-            "transform": [
-                {
-                    "aggregate": [{"op": "count", "as": "xy_count"}],
-                    "groupby": [Template.anchor("y"), Template.anchor("x")],
-                },
-                {
-                    "impute": "xy_count",
-                    "groupby": ["rev", Template.anchor("y")],
-                    "key": Template.anchor("x"),
-                    "value": 0,
-                },
-                {
-                    "impute": "xy_count",
-                    "groupby": ["rev", Template.anchor("x")],
-                    "key": Template.anchor("y"),
-                    "value": 0,
-                },
-                {
-                    "joinaggregate": [
-                        {"op": "sum", "field": "xy_count", "as": "sum_y"}
-                    ],
-                    "groupby": [Template.anchor("y")],
-                },
-                {
-                    "calculate": "datum.xy_count / datum.sum_y",
-                    "as": "percent_of_y",
-                },
-            ],
-            "encoding": {
-                "x": {
-                    "field": Template.anchor("x"),
-                    "type": "nominal",
-                    "sort": "ascending",
-                    "title": Template.anchor("x_label"),
-                },
-                "y": {
-                    "field": Template.anchor("y"),
-                    "type": "nominal",
-                    "sort": "ascending",
-                    "title": Template.anchor("y_label"),
-                },
-            },
-            "layer": [
-                {
-                    "mark": "rect",
-                    "width": 300,
-                    "height": 300,
-                    "encoding": {
-                        "color": {
-                            "field": "percent_of_y",
-                            "type": "quantitative",
-                            "title": "",
-                            "scale": {"domain": [0, 1]},
-                        }
-                    },
-                },
-                {
-                    "mark": "text",
-                    "encoding": {
-                        "text": {
-                            "field": "percent_of_y",
-                            "type": "quantitative",
-                            "format": ".2f",
-                        },
-                        "color": {
-                            "condition": {
-                                "test": "datum.percent_of_y > 0.5",
-                                "value": "white",
-                            },
-                            "value": "black",
-                        },
-                    },
-                },
-            ],
-        },
-    }
-
-
-class ScatterTemplate(Template):
-    DEFAULT_NAME = "scatter"
-
-    DEFAULT_CONTENT = {
-        "$schema": "https://vega.github.io/schema/vega-lite/v4.json",
-        "data": {"values": Template.anchor("data")},
-        "title": Template.anchor("title"),
-        "width": 300,
-        "height": 300,
-        "layer": [
-            {
-                "encoding": {
-                    "x": {
-                        "field": Template.anchor("x"),
-                        "type": "quantitative",
-                        "title": Template.anchor("x_label"),
-                    },
-                    "y": {
-                        "field": Template.anchor("y"),
-                        "type": "quantitative",
-                        "title": Template.anchor("y_label"),
-                        "scale": {"zero": False},
-                    },
-                    "color": {"field": "rev", "type": "nominal"},
-                },
-                "layer": [
-                    {"mark": "point"},
-                    {
-                        "selection": {
-                            "label": {
-                                "type": "single",
-                                "nearest": True,
-                                "on": "mouseover",
-                                "encodings": ["x"],
-                                "empty": "none",
-                                "clear": "mouseout",
-                            }
-                        },
-                        "mark": "point",
-                        "encoding": {
-                            "opacity": {
-                                "condition": {
-                                    "selection": "label",
-                                    "value": 1,
-                                },
-                                "value": 0,
-                            }
-                        },
-                    },
-                ],
-            },
-            {
-                "transform": [{"filter": {"selection": "label"}}],
-                "layer": [
-                    {
-                        "encoding": {
-                            "text": {
-                                "type": "quantitative",
-                                "field": Template.anchor("y"),
-                            },
-                            "x": {
-                                "field": Template.anchor("x"),
-                                "type": "quantitative",
-                            },
-                            "y": {
-                                "field": Template.anchor("y"),
-                                "type": "quantitative",
-                            },
-                        },
-                        "layer": [
-                            {
-                                "mark": {
-                                    "type": "text",
-                                    "align": "left",
-                                    "dx": 5,
-                                    "dy": -5,
-                                },
-                                "encoding": {
-                                    "color": {
-                                        "type": "nominal",
-                                        "field": "rev",
-                                    }
-                                },
-                            }
-                        ],
-                    }
-                ],
-            },
-        ],
-    }
-
-
-class SmoothLinearTemplate(Template):
-    DEFAULT_NAME = "smooth"
-
-    DEFAULT_CONTENT = {
-        "$schema": "https://vega.github.io/schema/vega-lite/v4.json",
-        "data": {"values": Template.anchor("data")},
-        "title": Template.anchor("title"),
-        "mark": {"type": "line"},
-        "encoding": {
-            "x": {
-                "field": Template.anchor("x"),
-                "type": "quantitative",
-                "title": Template.anchor("x_label"),
-            },
-            "y": {
-                "field": Template.anchor("y"),
-                "type": "quantitative",
-                "title": Template.anchor("y_label"),
-                "scale": {"zero": False},
-            },
-            "color": {"field": "rev", "type": "nominal"},
-        },
-        "transform": [
-            {
-                "loess": Template.anchor("y"),
-                "on": Template.anchor("x"),
-                "groupby": ["rev"],
-                "bandwidth": 0.3,
-            }
-        ],
-    }
-
-
-class LinearTemplate(Template):
-    DEFAULT_NAME = "linear"
-
-    DEFAULT_CONTENT = {
-        "$schema": "https://vega.github.io/schema/vega-lite/v4.json",
-        "data": {"values": Template.anchor("data")},
-        "title": Template.anchor("title"),
-        "width": 300,
-        "height": 300,
-        "layer": [
-            {
-                "encoding": {
-                    "x": {
-                        "field": Template.anchor("x"),
-                        "type": "quantitative",
-                        "title": Template.anchor("x_label"),
-                    },
-                    "y": {
-                        "field": Template.anchor("y"),
-                        "type": "quantitative",
-                        "title": Template.anchor("y_label"),
-                        "scale": {"zero": False},
-                    },
-                    "color": {"field": "rev", "type": "nominal"},
-                },
-                "layer": [
-                    {"mark": "line"},
-                    {
-                        "selection": {
-                            "label": {
-                                "type": "single",
-                                "nearest": True,
-                                "on": "mouseover",
-                                "encodings": ["x"],
-                                "empty": "none",
-                                "clear": "mouseout",
-                            }
-                        },
-                        "mark": "point",
-                        "encoding": {
-                            "opacity": {
-                                "condition": {
-                                    "selection": "label",
-                                    "value": 1,
-                                },
-                                "value": 0,
-                            }
-                        },
-                    },
-                ],
-            },
-            {
-                "transform": [{"filter": {"selection": "label"}}],
-                "layer": [
-                    {
-                        "mark": {"type": "rule", "color": "gray"},
-                        "encoding": {
-                            "x": {
-                                "field": Template.anchor("x"),
-                                "type": "quantitative",
-                            }
-                        },
-                    },
-                    {
-                        "encoding": {
-                            "text": {
-                                "type": "quantitative",
-                                "field": Template.anchor("y"),
-                            },
-                            "x": {
-                                "field": Template.anchor("x"),
-                                "type": "quantitative",
-                            },
-                            "y": {
-                                "field": Template.anchor("y"),
-                                "type": "quantitative",
-                            },
-                        },
-                        "layer": [
-                            {
-                                "mark": {
-                                    "type": "text",
-                                    "align": "left",
-                                    "dx": 5,
-                                    "dy": -5,
-                                },
-                                "encoding": {
-                                    "color": {
-                                        "type": "nominal",
-                                        "field": "rev",
-                                    }
-                                },
-                            }
-                        ],
-                    },
-                ],
-            },
-        ],
-    }
-
-
 class PlotTemplates:
     TEMPLATES_DIR = "plots"
-    TEMPLATES = [
-        DefaultTemplate,
-        LinearTemplate,
-        ConfusionTemplate,
-        NormalizedConfusionTemplate,
-        ScatterTemplate,
-        SmoothLinearTemplate,
-    ]
+    PKG_TEMPLATES_DIR = "templates"
 
     @cached_property
     def templates_dir(self):
         return os.path.join(self.dvc_dir, self.TEMPLATES_DIR)
 
-    def get_template(self, path):
-        if os.path.exists(path):
-            return path
+    @staticmethod
+    def _find(templates, template_name):
+        for template in templates:
+            if template.endswith(template_name) or template.endswith(
+                template_name + ".json"
+            ):
+                return template
+        return None
 
-        if self.dvc_dir and os.path.exists(self.dvc_dir):
-            t_path = os.path.join(self.templates_dir, path)
-            if os.path.exists(t_path):
-                return t_path
+    def _find_in_project(self, name: str) -> Optional["StrPath"]:
+        full_path = os.path.abspath(name)
+        if os.path.exists(full_path):
+            return full_path
 
-            all_templates = [
+        if os.path.exists(self.templates_dir):
+            templates = [
                 os.path.join(root, file)
                 for root, _, files in os.walk(self.templates_dir)
                 for file in files
             ]
-            matches = [
-                template
-                for template in all_templates
-                if os.path.splitext(template)[0] == t_path
-            ]
-            if matches:
-                assert len(matches) == 1
-                return matches[0]
+            found = self._find(templates, name)
+            if found:
+                return os.path.join(self.templates_dir, found)
+        return None
 
-        raise TemplateNotFoundError(path)
+    @staticmethod
+    def _get_templates() -> Iterable[str]:
+        if (
+            importlib_resources.files(__package__)
+            .joinpath(PlotTemplates.PKG_TEMPLATES_DIR)
+            .is_dir()
+        ):
+            entries = (
+                importlib_resources.files(__package__)
+                .joinpath(PlotTemplates.PKG_TEMPLATES_DIR)
+                .iterdir()
+            )
+            return [entry.name for entry in entries]
+        return []
+
+    @staticmethod
+    def _load_from_pkg(name):
+        templates = PlotTemplates._get_templates()
+        found = PlotTemplates._find(templates, name)
+        if found:
+            return (
+                (
+                    importlib_resources.files(__package__)
+                    / PlotTemplates.PKG_TEMPLATES_DIR
+                    / found
+                )
+                .read_bytes()
+                .decode("utf-8")
+            )
+        return None
+
+    def load(self, name: str = None) -> Template:
+
+        if name is not None:
+            template_path = self._find_in_project(name)
+            if template_path:
+                with open(template_path, "r", encoding="utf-8") as fd:
+                    content = fd.read()
+                return Template(content, name)
+        else:
+            name = "linear"
+
+        content = self._load_from_pkg(name)
+        if content:
+            return Template(content, name)
+
+        raise TemplateNotFoundError(name)
 
     def __init__(self, dvc_dir):
         self.dvc_dir = dvc_dir
@@ -574,24 +186,18 @@ class PlotTemplates:
         from dvc.utils.fs import makedirs
 
         makedirs(self.templates_dir, exist_ok=True)
-        for t in self.TEMPLATES:
-            self._dump(t())
 
-    def _dump(self, template):
-        path = os.path.join(self.templates_dir, template.filename)
-        with open(path, "w") as fd:
-            fd.write(template.content)
-
-    def load(self, name):
-        try:
-            path = self.get_template(name)
-
-            with open(path) as fd:
-                content = fd.read()
-
-            return Template(content, name=name)
-        except TemplateNotFoundError:
-            for template in self.TEMPLATES:
-                if template.DEFAULT_NAME == name:
-                    return template()
-            raise
+        templates = self._get_templates()
+        for template in templates:
+            content = (
+                importlib_resources.files(__package__)
+                .joinpath(PlotTemplates.PKG_TEMPLATES_DIR)
+                .joinpath(template)
+                .read_text()
+            )
+            with open(
+                os.path.join(self.templates_dir, template),
+                "w",
+                encoding="utf-8",
+            ) as fd:
+                fd.write(content)

dvc/repo/plots/templates/confusion.json

@@ -0,0 +1,107 @@
+{
+    "$schema": "https://vega.github.io/schema/vega-lite/v5.json",
+    "data": {
+        "values": "<DVC_METRIC_DATA>"
+    },
+    "title": "<DVC_METRIC_TITLE>",
+    "facet": {
+        "field": "rev",
+        "type": "nominal"
+    },
+    "spec": {
+        "transform": [
+            {
+                "aggregate": [
+                    {
+                        "op": "count",
+                        "as": "xy_count"
+                    }
+                ],
+                "groupby": [
+                    "<DVC_METRIC_Y>",
+                    "<DVC_METRIC_X>"
+                ]
+            },
+            {
+                "impute": "xy_count",
+                "groupby": [
+                    "rev",
+                    "<DVC_METRIC_Y>"
+                ],
+                "key": "<DVC_METRIC_X>",
+                "value": 0
+            },
+            {
+                "impute": "xy_count",
+                "groupby": [
+                    "rev",
+                    "<DVC_METRIC_X>"
+                ],
+                "key": "<DVC_METRIC_Y>",
+                "value": 0
+            },
+            {
+                "joinaggregate": [
+                    {
+                        "op": "max",
+                        "field": "xy_count",
+                        "as": "max_count"
+                    }
+                ],
+                "groupby": []
+            },
+            {
+                "calculate": "datum.xy_count / datum.max_count",
+                "as": "percent_of_max"
+            }
+        ],
+        "encoding": {
+            "x": {
+                "field": "<DVC_METRIC_X>",
+                "type": "nominal",
+                "sort": "ascending",
+                "title": "<DVC_METRIC_X_LABEL>"
+            },
+            "y": {
+                "field": "<DVC_METRIC_Y>",
+                "type": "nominal",
+                "sort": "ascending",
+                "title": "<DVC_METRIC_Y_LABEL>"
+            }
+        },
+        "layer": [
+            {
+                "mark": "rect",
+                "width": 300,
+                "height": 300,
+                "encoding": {
+                    "color": {
+                        "field": "xy_count",
+                        "type": "quantitative",
+                        "title": "",
+                        "scale": {
+                            "domainMin": 0,
+                            "nice": true
+                        }
+                    }
+                }
+            },
+            {
+                "mark": "text",
+                "encoding": {
+                    "text": {
+                        "field": "xy_count",
+                        "type": "quantitative"
+                    },
+                    "color": {
+                        "condition": {
+                            "test": "datum.percent_of_max > 0.5",
+                            "value": "white"
+                        },
+                        "value": "black"
+                    }
+                }
+            }
+        ]
+    }
+}