build(deps-dev): Bump types-requests from 2.27.9 to 2.27.10

Bumps [types-requests](https://github.com/python/typeshed) from 2.27.9 to 2.27.10.
- [Release notes](https://github.com/python/typeshed/releases)
- [Commits](https://github.com/python/typeshed/commits)

---
updated-dependencies:
- dependency-name: types-requests
  dependency-type: direct:development
  update-type: version-update:semver-patch
...

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

.codecov.yml

@@ -3,7 +3,7 @@ coverage:
   status:
     project:
       default:
-        target: 85
+        target: 90
         threshold: 10%
     patch:
       default:

.git_archival.txt

@@ -0,0 +1 @@
+ref-names: $Format:%D$

.gitattributes

@@ -0,0 +1 @@
+.git_archival.txt  export-subst

.github/mergify.yml

@@ -1,7 +1,7 @@
 pull_request_rules:
   - name: backport patches to 1.11
     conditions:
-      - base=master
+      - base=main
       - label=backport-1.11
     actions:
       backport:

.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@main
+        with:
+            pytest_options: "-k 'test_init or test_help'"
+

.github/workflows/packages.yaml

@@ -2,18 +2,21 @@ name: Build packages
 on:
   push:
     branches:
-    - master
+    - main
   release:
     types: [released, prereleased]
+  workflow_dispatch:
 jobs:
   deb:
     runs-on: ubuntu-18.04
     steps:
-    - uses: actions/checkout@v2.3.4
-    - name: Set up Python 3.8
+    - uses: actions/checkout@v2.4.0
+      with:
+          fetch-depth: 0
+    - name: Set up Python 3.9
       uses: actions/[email protected]
       with:
-        python-version: 3.8
+        python-version: 3.9
     - name: Set up Ruby 2.4
       uses: actions/[email protected]
       with:
@@ -38,11 +41,13 @@ jobs:
   rpm:
     runs-on: ubuntu-18.04
     steps:
-    - uses: actions/checkout@v2.3.4
-    - name: Set up Python 3.8
+    - uses: actions/checkout@v2.4.0
+      with:
+        fetch-depth: 0
+    - name: Set up Python 3.9
       uses: actions/[email protected]
       with:
-        python-version: 3.8
+        python-version: 3.9
     - name: Set up Ruby 2.4
       uses: actions/[email protected]
       with:
@@ -67,11 +72,13 @@ jobs:
   osxpkg:
     runs-on: macos-10.15
     steps:
-    - uses: actions/checkout@v2.3.4
-    - name: Set up Python 3.8
+    - uses: actions/checkout@v2.4.0
+      with:
+        fetch-depth: 0
+    - name: Set up Python 3.9
       uses: actions/[email protected]
       with:
-        python-version: 3.8
+        python-version: 3.9
     - name: Set up Ruby 2.4
       uses: actions/[email protected]
       with:
@@ -99,11 +106,13 @@ jobs:
   exe:
     runs-on: windows-2019
     steps:
-    - uses: actions/checkout@v2.3.4
-    - name: Set up Python 3.8
+    - uses: actions/checkout@v2.4.0
+      with:
+        fetch-depth: 0
+    - name: Set up Python 3.9
       uses: actions/[email protected]
       with:
-        python-version: 3.8
+        python-version: 3.9
     - name: Install deps
       run: |
         pip install .[all]
@@ -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
 
@@ -17,41 +13,39 @@ 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:
         python-version: 3.8
     - name: Install requirements
       run: |
-        pip install wheel
-        pip install ".[all,tests]" Pygments collective.checkdocs pre-commit
-    - name: install temporary dependencies
-      run: |
-        pip install git+https://github.com/isidentical/gdrivefs@service-account
-        pip install git+https://github.com/isidentical/pydata-google-auth@service-account
+        pip install --upgrade pip setuptools wheel
+        pip install ".[dev]" pre-commit
     - name: Check README
       run: python setup.py checkdocs
     - uses: pre-commit/[email protected]
   tests:
+    timeout-minutes: 45
     runs-on: ${{ matrix.os }}
     strategy:
       fail-fast: false
       matrix:
-        os: [ubuntu-18.04, windows-2019, macos-10.15]
-        pyv: ["3.6", "3.7", "3.8", "3.9"]
-# macos 11.0 workers take too long to start up
-#        include:
-#          - os: "macos-11.0"
-#            pyv: "3.9"
-        exclude:
-          - os: macos-10.15
-            pyv: "3.9"
-
+        os: [ubuntu-20.04, windows-latest, macos-latest]
+        pyv: ["3.7", "3.8", "3.9", "3.10"]
     steps:
-    - uses: actions/checkout@v2.3.4
+    - uses: actions/checkout@v2.4.0
+      with:
+        fetch-depth: 0
     - name: Set up Python
       uses: actions/[email protected]
       with:
@@ -62,55 +56,30 @@ jobs:
         echo "::set-output name=dir::$(pip cache dir)"
     - name: set pip cache
       id: pip-cache
-      uses: actions/[email protected].6
+      uses: actions/[email protected].7
       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
       if: steps.cache.pip-cache-dir.cache-hit != 'true'
       run: |
-        pip install wheel
-        pip install -e ".[all,tests]"
-    - name: install temporary dependencies
-      run: |
-        pip install git+https://github.com/isidentical/gdrivefs@service-account
-        pip install git+https://github.com/isidentical/pydata-google-auth@service-account
-    - name: setup git
-      run: |
-        git config --global user.email "[email protected]"
-        git config --global user.name "DVC Tester"
-    - name: setup gs creds
-      shell: bash
-      env:
-        GS_CREDS_JSON: ${{ secrets.GS_CREDS_JSON }}
-      if: env.GS_CREDS_JSON != ''
-      run: |
-        mkdir -p scripts/ci
-        echo "$GS_CREDS_JSON" > scripts/ci/gcp-creds.json
-    - name: disable gdrive tests
-      if: >-
-        github.event_name != 'schedule' &&
-        (matrix.pyv != '3.8' || matrix.os != 'ubuntu-18.04')
-      run: |
-        echo "extra_test_args=--disable-gdrive" >> $GITHUB_ENV
+        pip install --upgrade pip setuptools wheel
+        pip install -e ".[dev]"
     - name: run tests
-      env:
-        GDRIVE_CREDENTIALS_DATA: ${{ secrets.GDRIVE_CREDENTIALS_DATA }}
-        AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}
-        AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
+      timeout-minutes: 40
       run: >-
-        python -m tests -n=4
+        python -m tests -n=auto
         --cov-report=xml --cov-report=term
-        --all ${{ env.extra_test_args }}
+        ${{ env.extra_test_args }}
     - name: upload coverage report
-      uses: codecov/codecov-action@v1.5.2
+      uses: codecov/codecov-action@v2.1.0
       with:
         file: ./coverage.xml
         fail_ci_if_error: false
   notify:
-    if: github.ref == 'refs/heads/master' && failure()
+    if: github.ref == 'refs/heads/main' && failure()
     needs:
       - lint
       - tests
@@ -121,6 +90,6 @@ jobs:
       env:
         SLACK_WEBHOOK: ${{ secrets.SLACK_WEBHOOK }}
         SLACK_COLOR: ${{ job.status }}
-        SLACK_MESSAGE: 'CI Failed on master :boom:'
+        SLACK_MESSAGE: 'CI Failed on main :boom:'
         SLACK_TITLE: CI Status
         SLACK_USERNAME: DVC-CI

.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

@@ -1,17 +1,29 @@
 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: 21.6b0
+    rev: 22.1.0
+  - repo: https://github.com/pre-commit/pre-commit-hooks
+    rev: v4.1.0
+    hooks:
+      - id: check-toml
+      - id: check-yaml
+  - hooks:
+      - id: codespell
+        args: 
+          - --ignore-words-list
+          - ba,datas,fo,uptodate,cachable
+    repo: https://github.com/codespell-project/codespell
+    rev: v2.1.0
   - hooks:
       - id: isort
         language_version: python3
     repo: https://github.com/timothycrosley/isort
-    rev: 5.9.1
+    rev: 5.10.1
   - hooks:
       - id: flake8
         language_version: python3
@@ -22,19 +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: 'v0.902'
   - repo: local
     hooks:
+    - id: mypy
+      name: mypy
+      entry: mypy
+      files: ^dvc/
+      language: system
+      types: [python]
     - id: pylint
       name: pylint
       entry: pylint
@@ -47,7 +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.
@@ -29,9 +30,9 @@ learning projects. Key features:
 #. Helps manage experiments with Git tags/branches and **metrics** tracking.
 
 **DVC** aims to replace spreadsheet and document sharing tools (such as Excel or Google Docs)
-which are being used frequently as both knowledge repositories and team ledgers.
-DVC also replaces both ad-hoc scripts to track, move, and deploy different model versions;
-as well as ad-hoc data file suffixes and prefixes.
+frequently used as both knowledge repositories and team ledgers.
+DVC also replaces both ad-hoc scripts to track, move, and deploy different model versions
+and ad-hoc data file suffixes and prefixes.
 
 .. contents:: **Contents**
   :backlinks: none
@@ -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
 ============
@@ -102,7 +103,7 @@ Snap (Snapcraft/Linux)
 
 This corresponds to the latest tagged release.
 Add ``--beta`` for the latest tagged release candidate,
-or ``--edge`` for the latest ``master`` version.
+or ``--edge`` for the latest ``main`` version.
 
 Choco (Chocolatey/Windows)
 --------------------------
@@ -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 efficent 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
 ============
@@ -244,11 +254,7 @@ 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
+.. |CI| image:: https://github.com/iterative/dvc/workflows/Tests/badge.svg?branch=main
    :target: https://github.com/iterative/dvc/actions
    :alt: GHA Tests
 
@@ -256,7 +262,7 @@ Barrak, A., Eghan, E.E. and Adams, B. `On the Co-evolution of ML Pipelines and S
    :target: https://codeclimate.com/github/iterative/dvc
    :alt: Code Climate
 
-.. |Coverage| image:: https://codecov.io/gh/iterative/dvc/branch/master/graph/badge.svg
+.. |Coverage| image:: https://codecov.io/gh/iterative/dvc/branch/main/graph/badge.svg
    :target: https://codecov.io/gh/iterative/dvc
    :alt: Codecov
 
@@ -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/__main__.py

@@ -1,7 +1,7 @@
 """Main entry point for DVC command line tool."""
 import sys
 
-from dvc.main import main
+from dvc.cli import main
 
 if __name__ == "__main__":
     sys.exit(main(sys.argv[1:]))

dvc/_debug.py

@@ -1,10 +1,61 @@
 from contextlib import ExitStack, contextmanager
-from typing import TYPE_CHECKING
+from datetime import datetime
+from typing import TYPE_CHECKING, Callable, Union
 
 if TYPE_CHECKING:
     from argparse import Namespace
 
 
+@contextmanager
+def viztracer_profile(
+    path: Union[Callable[[], str], str],
+    max_stack_depth: int = -1,
+):
+    try:
+        import viztracer  # pylint: disable=import-error
+    except ImportError:
+        print("Failed to run profiler, viztracer is not installed")
+        yield
+        return
+
+    tracer = viztracer.VizTracer(max_stack_depth=max_stack_depth)
+
+    tracer.start()
+    yield
+    tracer.stop()
+
+    tracer.save(path() if callable(path) else path)
+
+
+@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 +121,19 @@ 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"):
+            output = "callgrind.dvc-{0:%Y%m%d}_{0:%H%M%S}.out"
+            stack.enter_context(
+                yappi_profile(path=lambda: output.format(datetime.now()))
+            )
+        if kw.get("viztracer") or kw.get("viztracer_depth"):
+            output = "viztracer.dvc-{0:%Y%m%d}_{0:%H%M%S}.json"
+            stack.enter_context(
+                viztracer_profile(
+                    path=lambda: output.format(datetime.now()),
+                    max_stack_depth=kw.get("viztracer_depth") or -1,
+                )
+            )
         yield
 
 
@@ -79,6 +143,13 @@ 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(
+        "--viztracer", action="store_true", default=False, help=SUPPRESS
+    )
+    parser.add_argument("--viztracer-depth", type=int, help=SUPPRESS)
     parser.add_argument("--cprofile-dump", help=SUPPRESS)
     parser.add_argument(
         "--pdb", action="store_true", default=False, help=SUPPRESS

dvc/analytics.py

@@ -2,6 +2,8 @@ import json
 import logging
 import os
 
+from .env import DVC_NO_ANALYTICS
+
 logger = logging.getLogger(__name__)
 
 
@@ -32,7 +34,7 @@ def collect_and_send_report(args=None, return_code=None):
 
     with tempfile.NamedTemporaryFile(delete=False, mode="w") as fobj:
         json.dump(report, fobj)
-        daemon(["analytics", fobj.name])
+    daemon(["analytics", fobj.name])
 
 
 def is_enabled():
@@ -42,9 +44,12 @@ def is_enabled():
     if env2bool("DVC_TEST"):
         return False
 
-    enabled = to_bool(
-        Config(validate=False).get("core", {}).get("analytics", "true")
-    )
+    enabled = not os.getenv(DVC_NO_ANALYTICS)
+    if enabled:
+        enabled = to_bool(
+            Config(validate=False).get("core", {}).get("analytics", "true")
+        )
+
     logger.debug("Analytics is {}abled.".format("en" if enabled else "dis"))
 
     return enabled
@@ -63,7 +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())
@@ -77,10 +82,12 @@ def send(path):
 
 
 def _scm_in_use():
+    from scmrepo.noscm import NoSCM
+
     from dvc.exceptions import NotDvcRepoError
     from dvc.repo import Repo
-    from dvc.scm import SCM, NoSCM
-    from dvc.scm.base import SCMError
+
+    from .scm import SCM, SCMError
 
     try:
         scm = SCM(root_dir=Repo.find_root())
@@ -168,13 +175,13 @@ 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

@@ -4,7 +4,6 @@ from contextlib import _GeneratorContextManager as GCM
 from funcy import reraise
 
 from dvc.exceptions import OutputNotFoundError, PathMissingError
-from dvc.path_info import PathInfo
 from dvc.repo import Repo
 
 
@@ -20,15 +19,16 @@ def get_url(path, repo=None, rev=None, remote=None):
     directory in the remote storage.
     """
     with Repo.open(repo, rev=rev, subrepos=True, uninitialized=True) as _repo:
-        path_info = PathInfo(_repo.root_dir) / path
+        fs_path = _repo.fs.path.join(_repo.root_dir, path)
         with reraise(FileNotFoundError, PathMissingError(path, repo)):
-            metadata = _repo.repo_fs.metadata(path_info)
+            info = _repo.repo_fs.info(fs_path)
 
-        if not metadata.is_dvc:
+        if not info["isdvc"]:
             raise OutputNotFoundError(path, repo)
 
-        cloud = metadata.repo.cloud
-        md5 = metadata.repo.dvcfs.info(path_info)["md5"]
+        cloud = info["repo"].cloud
+        dvc_path = _repo.fs.path.relpath(fs_path, info["repo"].root_dir)
+        md5 = info["repo"].dvcfs.info(dvc_path)["md5"]
         return cloud.get_url_for(remote, checksum=md5)
 
 
@@ -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/checkout.py

@@ -1,348 +0,0 @@
-import logging
-
-from shortuuid import uuid
-
-from dvc import prompt
-from dvc.exceptions import (
-    CacheLinkError,
-    CheckoutError,
-    ConfirmRemoveError,
-    DvcException,
-)
-from dvc.ignore import DvcIgnoreFilter
-from dvc.objects import check, load
-from dvc.objects.errors import ObjectFormatError
-from dvc.objects.stage import stage
-from dvc.remote.slow_link_detection import (  # type: ignore[attr-defined]
-    slow_link_guard,
-)
-from dvc.types import Optional
-
-logger = logging.getLogger(__name__)
-
-
-def _changed(path_info, fs, obj, cache):
-    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
-
-    try:
-        actual = stage(cache, path_info, fs, obj.hash_info.name).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):
-    if not fs.exists(path_info):
-        return
-
-    if force:
-        fs.remove(path_info)
-        return
-
-    current = stage(cache, path_info, fs, fs.PARAM_CHECKSUM).hash_info
-    try:
-        obj = load(cache, current)
-        check(cache, obj)
-    except (FileNotFoundError, ObjectFormatError):
-        msg = (
-            f"file/directory '{path_info}' is going to be removed. "
-            "Are you sure you want to proceed?"
-        )
-
-        if not prompt.confirm(msg):
-            raise ConfirmRemoveError(str(path_info))
-
-    fs.remove(path_info)
-
-
-def _verify_link(cache, path_info, link_type):
-    if link_type == "hardlink" and cache.fs.getsize(path_info) == 0:
-        return
-
-    if cache.cache_type_confirmed:
-        return
-
-    is_link = getattr(cache.fs, f"is_{link_type}", None)
-    if is_link and not is_link(path_info):
-        cache.fs.remove(path_info)
-        raise DvcException(f"failed to verify {link_type}")
-
-    cache.cache_type_confirmed = True
-
-
-def _do_link(cache, from_info, to_info, link_method):
-    if cache.fs.exists(to_info):
-        raise DvcException(f"Link '{to_info}' already exists!")
-
-    link_method(from_info, to_info)
-
-    logger.debug(
-        "Created '%s': %s -> %s", cache.cache_types[0], from_info, to_info
-    )
-
-
-@slow_link_guard
-def _try_links(cache, from_info, to_info, link_types):
-    while link_types:
-        link_method = getattr(cache.fs, link_types[0])
-        try:
-            _do_link(cache, from_info, to_info, link_method)
-            _verify_link(cache, to_info, link_types[0])
-            return
-
-        except DvcException as exc:
-            logger.debug(
-                "Cache type '%s' is not supported: %s", link_types[0], exc
-            )
-            del link_types[0]
-
-    raise CacheLinkError([to_info])
-
-
-def _link(cache, from_info, to_info):
-    assert cache.fs.isfile(from_info)
-    cache.makedirs(to_info.parent)
-    _try_links(cache, from_info, to_info, cache.cache_types)
-
-
-def _cache_is_copy(cache, path_info):
-    """Checks whether cache uses copies."""
-    if cache.cache_type_confirmed:
-        return cache.cache_types[0] == "copy"
-
-    if set(cache.cache_types) <= {"copy"}:
-        return True
-
-    workspace_file = path_info.with_name("." + uuid())
-    test_cache_file = cache.path_info / ".cache_type_test_file"
-    if not cache.fs.exists(test_cache_file):
-        cache.makedirs(test_cache_file.parent)
-        with cache.fs.open(test_cache_file, "wb") as fobj:
-            fobj.write(bytes(1))
-    try:
-        _link(cache, test_cache_file, workspace_file)
-    finally:
-        cache.fs.remove(workspace_file)
-        cache.fs.remove(test_cache_file)
-
-    cache.cache_type_confirmed = True
-    return cache.cache_types[0] == "copy"
-
-
-def _checkout_file(
-    path_info,
-    fs,
-    obj,
-    cache,
-    force,
-    progress_callback=None,
-    relink=False,
-    state=None,
-):
-    """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):
-            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):
-                cache.unprotect(path_info)
-            else:
-                _remove(path_info, fs, cache, force=force)
-                _link(cache, cache_info, path_info)
-    else:
-        _link(cache, cache_info, path_info)
-        modified = True
-
-    if state:
-        state.save(path_info, fs, obj.hash_info)
-
-    if progress_callback:
-        progress_callback(str(path_info))
-
-    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(
-    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)
-
-    for entry_key, entry_obj in obj:
-        entry_modified = _checkout_file(
-            path_info.joinpath(*entry_key),
-            fs,
-            entry_obj,
-            cache,
-            force,
-            progress_callback,
-            relink,
-            state=None,
-        )
-        if entry_modified:
-            modified = True
-
-    modified = (
-        _remove_redundant_files(
-            path_info, fs, obj, cache, force, dvcignore=dvcignore
-        )
-        or modified
-    )
-
-    if state:
-        state.save(path_info, fs, obj.hash_info)
-
-    # relink is not modified, assume it as nochange
-    return modified and not relink
-
-
-def _checkout(
-    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 state:
-        state.save_link(path_info, fs)
-
-    return ret
-
-
-def checkout(
-    path_info,
-    fs,
-    obj,
-    cache,
-    force=False,
-    progress_callback=None,
-    relink=False,
-    quiet=False,
-    dvcignore: Optional[DvcIgnoreFilter] = None,
-    state=None,
-):
-    if path_info.scheme not in ["local", cache.fs.scheme]:
-        raise NotImplementedError
-
-    failed = None
-    skip = False
-    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):
-        logger.trace("Data '%s' didn't change.", path_info)  # type: ignore
-        skip = True
-    else:
-        try:
-            check(cache, obj)
-        except (FileNotFoundError, ObjectFormatError):
-            if not quiet:
-                logger.warning(
-                    "Cache '%s' not found. File '%s' won't be created.",
-                    obj.hash_info,
-                    path_info,
-                )
-            _remove(path_info, fs, cache, force=force)
-            failed = path_info
-
-    if failed or skip:
-        if progress_callback and obj:
-            progress_callback(str(path_info), len(obj))
-        if 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,
-    )

dvc/cli/__init__.py

@@ -0,0 +1,138 @@
+"""This module provides an entrypoint to the dvc cli and parsing utils."""
+
+import logging
+
+# Workaround for CPython bug. See [1] and [2] for more info.
+# [1] https://github.com/aws/aws-cli/blob/1.16.277/awscli/clidriver.py#L55
+# [2] https://bugs.python.org/issue29288
+
+"".encode("idna")
+
+logger = logging.getLogger("dvc")
+
+
+class DvcParserError(Exception):
+    """Base class for CLI parser errors."""
+
+    def __init__(self):
+        super().__init__("parser error")
+
+
+def parse_args(argv=None):
+    """Parses CLI arguments.
+
+    Args:
+        argv: optional list of arguments to parse. sys.argv is used by default.
+
+    Raises:
+        DvcParserError: raised for argument parsing errors.
+    """
+    from .parser import get_main_parser
+
+    parser = get_main_parser()
+    args = parser.parse_args(argv)
+    return args
+
+
+def main(argv=None):  # noqa: C901
+    """Main entry point for dvc CLI.
+
+    Args:
+        argv: optional list of arguments to parse. sys.argv is used by default.
+
+    Returns:
+        int: command's return code.
+    """
+    from dvc._debug import debugtools
+    from dvc.config import ConfigError
+    from dvc.exceptions import DvcException, NotDvcRepoError
+    from dvc.logger import FOOTER, disable_other_loggers
+
+    args = None
+    disable_other_loggers()
+
+    outerLogLevel = logger.level
+    try:
+        args = parse_args(argv)
+
+        level = None
+        if args.quiet:
+            level = logging.CRITICAL
+        elif args.verbose == 1:
+            level = logging.DEBUG
+        elif args.verbose > 1:
+            level = logging.TRACE
+
+        if level is not None:
+            logger.setLevel(level)
+
+        logger.trace(args)
+
+        if not args.quiet:
+            from dvc.ui import ui
+
+            ui.enable()
+
+        with debugtools(args):
+            cmd = args.func(args)
+            ret = cmd.do_run()
+    except ConfigError:
+        logger.exception("configuration error")
+        ret = 251
+    except KeyboardInterrupt:
+        logger.exception("interrupted by the user")
+        ret = 252
+    except NotDvcRepoError:
+        logger.exception("")
+        ret = 253
+    except DvcException:
+        ret = 255
+        logger.exception("")
+    except DvcParserError:
+        ret = 254
+    except Exception as exc:  # noqa, pylint: disable=broad-except
+        # pylint: disable=no-member
+        import errno
+
+        if isinstance(exc, OSError) and exc.errno == errno.EMFILE:
+            from dvc.utils import error_link
+
+            logger.exception(
+                "too many open files, please visit "
+                "{} to see how to handle this "
+                "problem".format(error_link("many-files")),
+                extra={"tb_only": True},
+            )
+        else:
+            from dvc.info import get_dvc_info
+
+            logger.exception("unexpected error")
+
+            dvc_info = get_dvc_info()
+            logger.debug("Version info for developers:\n%s", dvc_info)
+
+            logger.info(FOOTER)
+        ret = 255
+
+    try:
+        from dvc import analytics
+
+        if analytics.is_enabled():
+            analytics.collect_and_send_report(args, ret)
+
+        return ret
+    finally:
+        logger.setLevel(outerLogLevel)
+
+        from dvc.fs.pool import close_pools
+
+        # Closing pools by-hand to prevent weird messages when closing SSH
+        # connections. See https://github.com/iterative/dvc/issues/3248 for
+        # more info.
+        close_pools()
+
+        from dvc.external_repo import clean_repos
+
+        # Remove cached repos in the end of the call, these are anonymous
+        # so won't be reused by any other subsequent run anyway.
+        clean_repos()

dvc/cli/command.py

@@ -0,0 +1,36 @@
+import logging
+import os
+from abc import ABC, abstractmethod
+
+logger = logging.getLogger(__name__)
+
+
+class CmdBase(ABC):
+    UNINITIALIZED = False
+
+    def __init__(self, args):
+        from dvc.repo import Repo
+
+        os.chdir(args.cd)
+
+        self.repo = Repo(uninitialized=self.UNINITIALIZED)
+        self.config = self.repo.config
+        self.args = args
+
+    def do_run(self):
+        with self.repo:
+            return self.run()
+
+    @abstractmethod
+    def run(self):
+        pass
+
+
+class CmdBaseNoRepo(CmdBase):
+    def __init__(self, args):  # pylint: disable=super-init-not-called
+        self.args = args
+
+        os.chdir(args.cd)
+
+    def do_run(self):
+        return self.run()

dvc/cli/parser.py

@@ -1,11 +1,10 @@
-"""DVC command line interface"""
+"""Main parser for the dvc cli."""
 import argparse
 import logging
 import os
 import sys
 
-from ._debug import add_debugging_flags
-from .command import (
+from dvc.commands import (
     add,
     cache,
     check_ignore,
@@ -30,6 +29,7 @@ from .command import (
     install,
     live,
     ls,
+    machine,
     metrics,
     move,
     params,
@@ -44,8 +44,8 @@ from .command import (
     update,
     version,
 )
-from .command.base import fix_subparsers
-from .exceptions import DvcParserError
+
+from . import DvcParserError
 
 logger = logging.getLogger(__name__)
 
@@ -87,6 +87,7 @@ COMMANDS = [
     experiments,
     check_ignore,
     live,
+    machine,
 ]
 
 
@@ -94,7 +95,7 @@ def _find_parser(parser, cmd_cls):
     defaults = parser._defaults  # pylint: disable=protected-access
     if not cmd_cls or cmd_cls == defaults.get("func"):
         parser.print_help()
-        raise DvcParserError()
+        raise DvcParserError
 
     actions = parser._actions  # pylint: disable=protected-access
     for action in actions:
@@ -141,6 +142,8 @@ def get_parent_parser():
     When overwriting `-q` or `-v`, you need to instantiate a new object
     in order to prevent some weird behavior.
     """
+    from dvc._debug import add_debugging_flags
+
     parent_parser = argparse.ArgumentParser(add_help=False)
     add_debugging_flags(parent_parser)
     log_level_group = parent_parser.add_mutually_exclusive_group()
@@ -205,23 +208,11 @@ def get_main_parser():
         help="Use `dvc COMMAND --help` for command-specific help.",
     )
 
+    from .utils import fix_subparsers
+
     fix_subparsers(subparsers)
 
     for cmd in COMMANDS:
         cmd.add_parser(subparsers, parent_parser)
 
     return parser
-
-
-def parse_args(argv=None):
-    """Parses CLI arguments.
-
-    Args:
-        argv: optional list of arguments to parse. sys.argv is used by default.
-
-    Raises:
-        dvc.exceptions.DvcParserError: raised for argument parsing errors.
-    """
-    parser = get_main_parser()
-    args = parser.parse_args(argv)
-    return args

dvc/cli/utils.py

@@ -0,0 +1,31 @@
+def fix_subparsers(subparsers):
+    """Workaround for bug in Python 3. See more info at:
+    https://bugs.python.org/issue16308
+    https://github.com/iterative/dvc/issues/769
+
+    Args:
+        subparsers: subparsers to fix.
+    """
+    subparsers.required = True
+    subparsers.dest = "cmd"
+
+
+def append_doc_link(help_message, path):
+    from dvc.utils import format_link
+
+    if not path:
+        return help_message
+    doc_base = "https://man.dvc.org/"
+    return "{message}\nDocumentation: {link}".format(
+        message=help_message, link=format_link(doc_base + path)
+    )
+
+
+def fix_plumbing_subparsers(subparsers):
+    # metavar needs to be explicitly set in order to hide plumbing subcommands
+    # from the 'positional arguments' choices list
+    # see: https://bugs.python.org/issue22848
+    cmds = [
+        cmd for cmd, parser in subparsers.choices.items() if parser.add_help
+    ]
+    subparsers.metavar = "{{{}}}".format(",".join(cmds))

dvc/command/base.py

@@ -1,64 +0,0 @@
-import logging
-import os
-from abc import ABC, abstractmethod
-
-logger = logging.getLogger(__name__)
-
-
-def fix_subparsers(subparsers):
-    """Workaround for bug in Python 3. See more info at:
-    https://bugs.python.org/issue16308
-    https://github.com/iterative/dvc/issues/769
-
-    Args:
-        subparsers: subparsers to fix.
-    """
-    subparsers.required = True
-    subparsers.dest = "cmd"
-
-
-def append_doc_link(help_message, path):
-    from dvc.utils import format_link
-
-    if not path:
-        return help_message
-    doc_base = "https://man.dvc.org/"
-    return "{message}\nDocumentation: {link}".format(
-        message=help_message, link=format_link(doc_base + path)
-    )
-
-
-class CmdBase(ABC):
-    UNINITIALIZED = False
-
-    def __init__(self, args):
-        from dvc.repo import Repo
-        from dvc.updater import Updater
-
-        os.chdir(args.cd)
-
-        self.repo = Repo(uninitialized=self.UNINITIALIZED)
-        self.config = self.repo.config
-        self.args = args
-        if self.repo.tmp_dir:
-            hardlink_lock = self.config["core"].get("hardlink_lock", False)
-            updater = Updater(self.repo.tmp_dir, hardlink_lock=hardlink_lock)
-            updater.check()
-
-    def do_run(self):
-        with self.repo:
-            return self.run()
-
-    @abstractmethod
-    def run(self):
-        pass
-
-
-class CmdBaseNoRepo(CmdBase):
-    def __init__(self, args):  # pylint: disable=super-init-not-called
-        self.args = args
-
-        os.chdir(args.cd)
-
-    def do_run(self):
-        return self.run()

dvc/commands/add.py

@@ -1,8 +1,9 @@
 import argparse
 import logging
 
-from dvc.command import completion
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.commands import completion
 
 logger = logging.getLogger(__name__)
 
@@ -101,7 +102,9 @@ def add_parser(subparsers, parent_parser):
         "--jobs",
         type=int,
         help=(
-            "Number of jobs to run simultaneously. "
+            "Only used along with '--to-remote'. "
+            "Number of jobs to run simultaneously "
+            "when pushing data to remote."
             "The default value is 4 * cpu_count(). "
             "For SSH remotes, the default is 4. "
         ),

dvc/commands/cache.py

@@ -1,8 +1,8 @@
 import argparse
 
-from dvc.command import completion
-from dvc.command.base import append_doc_link, fix_subparsers
-from dvc.command.config import CmdConfig
+from dvc.cli.utils import append_doc_link, fix_subparsers
+from dvc.commands import completion
+from dvc.commands.config import CmdConfig
 from dvc.ui import ui
 
 
@@ -28,7 +28,7 @@ class CmdCacheDir(CmdConfig):
 
 
 def add_parser(subparsers, parent_parser):
-    from dvc.command.config import parent_config_parser
+    from dvc.commands.config import parent_config_parser
 
     CACHE_HELP = "Manage cache settings."
 

dvc/commands/check_ignore.py

@@ -1,7 +1,8 @@
 import argparse
 
-from dvc.command import completion
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.commands import completion
 from dvc.ui import ui
 
 

dvc/commands/checkout.py

@@ -1,8 +1,9 @@
 import argparse
 import operator
 
-from dvc.command import completion
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.commands import completion
 from dvc.exceptions import CheckoutError
 from dvc.ui import ui
 

dvc/commands/commit.py

@@ -1,8 +1,9 @@
 import argparse
 import logging
 
-from dvc.command import completion
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.commands import completion
 
 logger = logging.getLogger(__name__)
 

dvc/commands/completion.py

@@ -3,7 +3,8 @@ import logging
 
 import shtab
 
-from dvc.command.base import CmdBaseNoRepo, append_doc_link
+from dvc.cli.command import CmdBaseNoRepo
+from dvc.cli.utils import append_doc_link
 from dvc.ui import ui
 
 logger = logging.getLogger(__name__)
@@ -82,7 +83,7 @@ EXPERIMENT = {"bash": "_dvc_compgen_exps", "zsh": "_dvc_compadd_exps"}
 
 class CmdCompletion(CmdBaseNoRepo):
     def run(self):
-        from dvc.cli import get_main_parser
+        from dvc.cli.parser import get_main_parser
 
         parser = get_main_parser()
         shell = self.args.shell

dvc/commands/config.py

@@ -2,7 +2,8 @@ import argparse
 import logging
 import os
 
-from dvc.command.base import CmdBaseNoRepo, append_doc_link
+from dvc.cli.command import CmdBaseNoRepo
+from dvc.cli.utils import append_doc_link
 from dvc.ui import ui
 
 logger = logging.getLogger(__name__)
@@ -66,7 +67,8 @@ 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 +83,7 @@ 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 +136,20 @@ 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/commands/daemon.py

@@ -1,5 +1,6 @@
-from dvc.command import completion
-from dvc.command.base import CmdBaseNoRepo, fix_subparsers
+from dvc.cli.command import CmdBaseNoRepo
+from dvc.cli.utils import fix_subparsers
+from dvc.commands import completion
 
 
 class CmdDaemonBase(CmdBaseNoRepo):

dvc/commands/dag.py

@@ -1,6 +1,7 @@
 import argparse
 
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
 from dvc.ui import ui
 
 
@@ -36,18 +37,22 @@ def _collect_targets(repo, target, outs):
         return [stage.addressing for stage, _ in pairs]
 
     targets = []
-    for stage, info in pairs:
-        if not info:
+
+    outs_trie = repo.index.outs_trie
+    for stage, path in pairs:
+        if not path:
             targets.extend([str(out) for out in stage.outs])
             continue
 
-        for out in repo.outs_trie.itervalues(prefix=info.parts):  # noqa: B301
+        for out in outs_trie.itervalues(
+            prefix=repo.fs.path.parts(path)
+        ):  # noqa: B301
             targets.extend(str(out))
 
     return targets
 
 
-def _transform(repo, outs):
+def _transform(index, outs):
     import networkx as nx
 
     from dvc.stage import Stage
@@ -55,7 +60,7 @@ def _transform(repo, outs):
     def _relabel(node) -> str:
         return node.addressing if isinstance(node, Stage) else str(node)
 
-    G = repo.outs_graph if outs else repo.graph
+    G = index.outs_graph if outs else index.graph
     return nx.relabel_nodes(G, _relabel, copy=True)
 
 
@@ -85,7 +90,7 @@ def _filter(G, targets, full):
 
 def _build(repo, target=None, full=False, outs=False):
     targets = _collect_targets(repo, target, outs)
-    G = _transform(repo, outs)
+    G = _transform(repo.index, outs)
     return _filter(G, targets, full)
 
 
@@ -101,9 +106,8 @@ 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/commands/data_sync.py

@@ -1,8 +1,9 @@
 import argparse
 import logging
 
-from dvc.command import completion
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.commands import completion
 
 logger = logging.getLogger(__name__)
 
@@ -18,7 +19,7 @@ class CmdDataBase(CmdBase):
 
 class CmdDataPull(CmdDataBase):
     def log_summary(self, stats):
-        from dvc.command.checkout import log_changes
+        from dvc.commands.checkout import log_changes
 
         log_changes(stats)
         super().log_summary(stats)
@@ -97,7 +98,7 @@ class CmdDataFetch(CmdDataBase):
 
 
 def shared_parent_parser():
-    from dvc.cli import get_parent_parser
+    from dvc.cli.parser import get_parent_parser
 
     # Parent parser used in pull/push/status
     parent_parser = argparse.ArgumentParser(
@@ -127,7 +128,7 @@ def shared_parent_parser():
 
 
 def add_parser(subparsers, _parent_parser):
-    from dvc.command.status import CmdDataStatus
+    from dvc.commands.status import CmdDataStatus
 
     # Pull
     PULL_HELP = "Download tracked files or directories from remote storage."
@@ -393,6 +394,7 @@ 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/commands/destroy.py

@@ -1,7 +1,8 @@
 import argparse
 import logging
 
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
 
 logger = logging.getLogger(__name__)
 

dvc/commands/diff.py

@@ -1,10 +1,10 @@
 import argparse
-import json
 import logging
 import os
 
-from dvc.command import completion
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.commands import completion
 from dvc.ui import ui
 
 logger = logging.getLogger(__name__)
@@ -16,7 +16,7 @@ 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"]
@@ -111,14 +111,12 @@ class CmdDiff(CmdBase):
         if not sum(summary.values()):
             return None
 
-        fmt = (
-            "files summary: {added} added, {deleted} deleted,"
-            " {renamed} renamed, {modified} modified"
+        states_summary = ", ".join(
+            f"{summary[state]} {state}"
+            for state in states
+            if summary[state] > 0
         )
-        if not hide_missing:
-            fmt += ", {not in cache} not in cache"
-
-        ui.write(fmt.format_map(summary))
+        ui.write("files summary:", states_summary)
 
     def run(self):
         from dvc.exceptions import DvcException
@@ -144,10 +142,10 @@ 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)
 
@@ -190,6 +188,7 @@ def add_parser(subparsers, parent_parser):
         nargs="?",
     )
     diff_parser.add_argument(
+        "--json",
         "--show-json",
         help="Format the output into a JSON",
         action="store_true",
@@ -202,9 +201,11 @@ 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/commands/experiments/__init__.py

@@ -0,0 +1,60 @@
+import argparse
+
+from dvc.cli.utils import (
+    append_doc_link,
+    fix_plumbing_subparsers,
+    fix_subparsers,
+)
+from dvc.commands.experiments import (
+    apply,
+    branch,
+    diff,
+    exec_run,
+    gc,
+    init,
+    ls,
+    pull,
+    push,
+    remove,
+    run,
+    show,
+)
+
+SUB_COMMANDS = [
+    apply,
+    branch,
+    diff,
+    exec_run,
+    gc,
+    init,
+    ls,
+    pull,
+    push,
+    remove,
+    run,
+    show,
+]
+
+
+def add_parser(subparsers, parent_parser):
+    EXPERIMENTS_HELP = "Commands to run and compare experiments."
+
+    experiments_parser = subparsers.add_parser(
+        "experiments",
+        parents=[parent_parser],
+        aliases=["exp"],
+        description=append_doc_link(EXPERIMENTS_HELP, "exp"),
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+        help=EXPERIMENTS_HELP,
+    )
+
+    experiments_subparsers = experiments_parser.add_subparsers(
+        dest="cmd",
+        help="Use `dvc experiments CMD --help` to display "
+        "command-specific help.",
+    )
+
+    fix_subparsers(experiments_subparsers)
+    for cmd in SUB_COMMANDS:
+        cmd.add_parser(experiments_subparsers, parent_parser)
+    fix_plumbing_subparsers(experiments_subparsers)

dvc/commands/experiments/apply.py

@@ -0,0 +1,42 @@
+import argparse
+import logging
+
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.commands import completion
+
+logger = logging.getLogger(__name__)
+
+
+class CmdExperimentsApply(CmdBase):
+    def run(self):
+
+        self.repo.experiments.apply(
+            self.args.experiment, force=self.args.force
+        )
+
+        return 0
+
+
+def add_parser(experiments_subparsers, parent_parser):
+
+    EXPERIMENTS_APPLY_HELP = (
+        "Apply the changes from an experiment to your workspace."
+    )
+    experiments_apply_parser = experiments_subparsers.add_parser(
+        "apply",
+        parents=[parent_parser],
+        description=append_doc_link(EXPERIMENTS_APPLY_HELP, "exp/apply"),
+        help=EXPERIMENTS_APPLY_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    experiments_apply_parser.add_argument(
+        "--no-force",
+        action="store_false",
+        dest="force",
+        help="Fail if this command would overwrite conflicting changes.",
+    )
+    experiments_apply_parser.add_argument(
+        "experiment", help="Experiment to be applied."
+    ).complete = completion.EXPERIMENT
+    experiments_apply_parser.set_defaults(func=CmdExperimentsApply)

dvc/commands/experiments/branch.py

@@ -0,0 +1,34 @@
+import argparse
+import logging
+
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+
+logger = logging.getLogger(__name__)
+
+
+class CmdExperimentsBranch(CmdBase):
+    def run(self):
+
+        self.repo.experiments.branch(self.args.experiment, self.args.branch)
+
+        return 0
+
+
+def add_parser(experiments_subparsers, parent_parser):
+
+    EXPERIMENTS_BRANCH_HELP = "Promote an experiment to a Git branch."
+    experiments_branch_parser = experiments_subparsers.add_parser(
+        "branch",
+        parents=[parent_parser],
+        description=append_doc_link(EXPERIMENTS_BRANCH_HELP, "exp/branch"),
+        help=EXPERIMENTS_BRANCH_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    experiments_branch_parser.add_argument(
+        "experiment", help="Experiment to be promoted."
+    )
+    experiments_branch_parser.add_argument(
+        "branch", help="Git branch name to use."
+    )
+    experiments_branch_parser.set_defaults(func=CmdExperimentsBranch)

dvc/commands/experiments/diff.py

@@ -0,0 +1,118 @@
+import argparse
+import logging
+
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.commands import completion
+from dvc.commands.metrics import DEFAULT_PRECISION
+from dvc.exceptions import DvcException
+from dvc.ui import ui
+
+logger = logging.getLogger(__name__)
+
+
+class CmdExperimentsDiff(CmdBase):
+    def run(self):
+
+        try:
+            diff = self.repo.experiments.diff(
+                a_rev=self.args.a_rev,
+                b_rev=self.args.b_rev,
+                all=self.args.all,
+                param_deps=self.args.param_deps,
+            )
+        except DvcException:
+            logger.exception("failed to show experiments diff")
+            return 1
+
+        if self.args.json:
+            ui.write_json(diff)
+        else:
+            from dvc.compare import show_diff
+
+            precision = self.args.precision or DEFAULT_PRECISION
+            diffs = [("metrics", "Metric"), ("params", "Param")]
+            for idx, (key, title) in enumerate(diffs):
+                if idx:
+                    # we are printing tables even in `--quiet` mode
+                    # so we should also be printing the "table" separator
+                    ui.write(force=True)
+
+                show_diff(
+                    diff[key],
+                    title=title,
+                    markdown=self.args.markdown,
+                    no_path=self.args.no_path,
+                    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
+
+
+def add_parser(experiments_subparsers, parent_parser):
+
+    EXPERIMENTS_DIFF_HELP = (
+        "Show changes between experiments in the DVC repository."
+    )
+
+    experiments_diff_parser = experiments_subparsers.add_parser(
+        "diff",
+        parents=[parent_parser],
+        description=append_doc_link(EXPERIMENTS_DIFF_HELP, "exp/diff"),
+        help=EXPERIMENTS_DIFF_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    experiments_diff_parser.add_argument(
+        "a_rev", nargs="?", help="Old experiment to compare (defaults to HEAD)"
+    ).complete = completion.EXPERIMENT
+    experiments_diff_parser.add_argument(
+        "b_rev",
+        nargs="?",
+        help="New experiment to compare (defaults to the current workspace)",
+    ).complete = completion.EXPERIMENT
+    experiments_diff_parser.add_argument(
+        "--all",
+        action="store_true",
+        default=False,
+        help="Show unchanged metrics/params as well.",
+    )
+    experiments_diff_parser.add_argument(
+        "--param-deps",
+        action="store_true",
+        default=False,
+        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(
+        "--no-path",
+        action="store_true",
+        default=False,
+        help="Don't show metric/param path.",
+    )
+    experiments_diff_parser.add_argument(
+        "--precision",
+        type=int,
+        help=(
+            "Round metrics/params to `n` digits precision after the decimal "
+            f"point. Rounds to {DEFAULT_PRECISION} digits by default."
+        ),
+        metavar="<n>",
+    )
+    experiments_diff_parser.set_defaults(func=CmdExperimentsDiff)

dvc/commands/experiments/exec_run.py

@@ -0,0 +1,42 @@
+import logging
+
+from dvc.cli.command import CmdBaseNoRepo
+
+logger = logging.getLogger(__name__)
+
+
+class CmdExecutorRun(CmdBaseNoRepo):
+    """Run an experiment executor."""
+
+    def run(self):
+        from dvc.repo.experiments.executor.base import (
+            BaseExecutor,
+            ExecutorInfo,
+        )
+        from dvc.utils.serialize import load_json
+
+        info = ExecutorInfo.from_dict(load_json(self.args.infofile))
+        BaseExecutor.reproduce(
+            info=info,
+            rev="",
+            queue=None,
+            log_level=logger.getEffectiveLevel(),
+            infofile=self.args.infofile,
+        )
+        return 0
+
+
+def add_parser(experiments_subparsers, parent_parser):
+    EXEC_RUN_HELP = "Run an experiment executor."
+    exec_run_parser = experiments_subparsers.add_parser(
+        "exec-run",
+        parents=[parent_parser],
+        description=EXEC_RUN_HELP,
+        add_help=False,
+    )
+    exec_run_parser.add_argument(
+        "--infofile",
+        help="Path to executor info file",
+        default=None,
+    )
+    exec_run_parser.set_defaults(func=CmdExecutorRun)

dvc/commands/experiments/gc.py

@@ -0,0 +1,129 @@
+import argparse
+import logging
+
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.exceptions import InvalidArgumentError
+from dvc.ui import ui
+
+logger = logging.getLogger(__name__)
+
+
+def _raise_error_if_all_disabled(**kwargs):
+    if not any(kwargs.values()):
+        raise InvalidArgumentError(
+            "Either of `-w|--workspace`, `-a|--all-branches`, `-T|--all-tags` "
+            "or `--all-commits` needs to be set."
+        )
+
+
+class CmdExperimentsGC(CmdBase):
+    def run(self):
+        _raise_error_if_all_disabled(
+            all_branches=self.args.all_branches,
+            all_tags=self.args.all_tags,
+            all_commits=self.args.all_commits,
+            workspace=self.args.workspace,
+        )
+
+        msg = "This will remove all experiments except those derived from "
+
+        msg += "the workspace"
+        if self.args.all_commits:
+            msg += " and all git commits"
+        elif self.args.all_branches and self.args.all_tags:
+            msg += " and all git branches and tags"
+        elif self.args.all_branches:
+            msg += " and all git branches"
+        elif self.args.all_tags:
+            msg += " and all git tags"
+        msg += " of the current repo."
+        if self.args.queued:
+            msg += " Run queued experiments will be preserved."
+        else:
+            msg += " Run queued experiments will be removed."
+
+        logger.warning(msg)
+
+        msg = "Are you sure you want to proceed?"
+        if not self.args.force and not ui.confirm(msg):
+            return 1
+
+        removed = self.repo.experiments.gc(
+            all_branches=self.args.all_branches,
+            all_tags=self.args.all_tags,
+            all_commits=self.args.all_commits,
+            workspace=self.args.workspace,
+            queued=self.args.queued,
+        )
+
+        if removed:
+            ui.write(
+                f"Removed {removed} experiments.",
+                "To remove unused cache files",
+                "use 'dvc gc'.",
+            )
+        else:
+            ui.write("No experiments to remove.")
+        return 0
+
+
+def add_parser(experiments_subparsers, parent_parser):
+
+    EXPERIMENTS_GC_HELP = "Garbage collect unneeded experiments."
+    EXPERIMENTS_GC_DESCRIPTION = (
+        "Removes all experiments which are not derived from the specified"
+        "Git revisions."
+    )
+    experiments_gc_parser = experiments_subparsers.add_parser(
+        "gc",
+        parents=[parent_parser],
+        description=append_doc_link(EXPERIMENTS_GC_DESCRIPTION, "exp/gc"),
+        help=EXPERIMENTS_GC_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    experiments_gc_parser.add_argument(
+        "-w",
+        "--workspace",
+        action="store_true",
+        default=False,
+        help="Keep experiments derived from the current workspace.",
+    )
+    experiments_gc_parser.add_argument(
+        "-a",
+        "--all-branches",
+        action="store_true",
+        default=False,
+        help="Keep experiments derived from the tips of all Git branches.",
+    )
+    experiments_gc_parser.add_argument(
+        "-T",
+        "--all-tags",
+        action="store_true",
+        default=False,
+        help="Keep experiments derived from all Git tags.",
+    )
+    experiments_gc_parser.add_argument(
+        "-A",
+        "--all-commits",
+        action="store_true",
+        default=False,
+        help="Keep experiments derived from all Git commits.",
+    )
+    experiments_gc_parser.add_argument(
+        "--queued",
+        action="store_true",
+        default=False,
+        help=(
+            "Keep queued experiments (experiments run queue will be cleared "
+            "by default)."
+        ),
+    )
+    experiments_gc_parser.add_argument(
+        "-f",
+        "--force",
+        action="store_true",
+        default=False,
+        help="Force garbage collection - automatically agree to all prompts.",
+    )
+    experiments_gc_parser.set_defaults(func=CmdExperimentsGC)

dvc/commands/experiments/init.py

@@ -0,0 +1,188 @@
+import argparse
+import logging
+
+from funcy import compact
+
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.exceptions import InvalidArgumentError
+from dvc.ui import ui
+
+logger = logging.getLogger(__name__)
+
+
+class CmdExperimentsInit(CmdBase):
+    DEFAULT_NAME = "train"
+    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://s.dvc.org/g/exp/run"
+
+    def run(self):
+        from dvc.commands.stage import parse_cmd
+
+        cmd = parse_cmd(self.args.command)
+        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,
+        )
+
+        text = ui.rich_text.assemble(
+            "\n" if self.args.interactive else "",
+            "Created ",
+            (self.args.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(experiments_subparsers, parent_parser):
+
+    EXPERIMENTS_INIT_HELP = "Quickly setup any project to use 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,
+        help=EXPERIMENTS_INIT_HELP,
+    )
+    experiments_init_parser.add_argument(
+        "command",
+        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 (default: %(default)s)",
+        default=CmdExperimentsInit.DEFAULT_NAME,
+    )
+    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", "dl"],
+        default="default",
+        help="Select type of stage to create (default: %(default)s)",
+    )
+    experiments_init_parser.set_defaults(func=CmdExperimentsInit)

dvc/commands/experiments/ls.py

@@ -0,0 +1,73 @@
+import argparse
+import logging
+
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+
+logger = logging.getLogger(__name__)
+
+
+class CmdExperimentsList(CmdBase):
+    def run(self):
+        names_only = self.args.names_only
+        exps = self.repo.experiments.ls(
+            rev=self.args.rev,
+            git_remote=self.args.git_remote,
+            all_=self.args.all,
+        )
+        for baseline in exps:
+            tag = self.repo.scm.describe(baseline)
+            if not tag:
+                branch = self.repo.scm.describe(baseline, base="refs/heads")
+                if branch:
+                    tag = branch.split("/")[-1]
+            name = tag if tag else baseline[:7]
+            if not names_only:
+                print(f"{name}:")
+            for exp_name in exps[baseline]:
+                indent = "" if names_only else "\t"
+                print(f"{indent}{exp_name}")
+
+        return 0
+
+
+def add_parser(experiments_subparsers, parent_parser):
+
+    EXPERIMENTS_LIST_HELP = "List local and remote experiments."
+    experiments_list_parser = experiments_subparsers.add_parser(
+        "list",
+        parents=[parent_parser],
+        description=append_doc_link(EXPERIMENTS_LIST_HELP, "exp/list"),
+        help=EXPERIMENTS_LIST_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    experiments_list_parser.add_argument(
+        "--rev",
+        type=str,
+        default=None,
+        help=(
+            "List experiments derived from the specified revision. "
+            "Defaults to HEAD if neither `--rev` nor `--all` are specified."
+        ),
+        metavar="<rev>",
+    )
+    experiments_list_parser.add_argument(
+        "--all", action="store_true", help="List all experiments."
+    )
+    experiments_list_parser.add_argument(
+        "--names-only",
+        action="store_true",
+        help="Only output experiment names (without parent commits).",
+    )
+    experiments_list_parser.add_argument(
+        "git_remote",
+        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 ones."
+        ),
+        metavar="[<git_remote>]",
+    )
+    experiments_list_parser.set_defaults(func=CmdExperimentsList)

dvc/commands/experiments/pull.py

@@ -0,0 +1,96 @@
+import argparse
+import logging
+
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.ui import ui
+
+logger = logging.getLogger(__name__)
+
+
+class CmdExperimentsPull(CmdBase):
+    def run(self):
+
+        self.repo.experiments.pull(
+            self.args.git_remote,
+            self.args.experiment,
+            force=self.args.force,
+            pull_cache=self.args.pull_cache,
+            dvc_remote=self.args.dvc_remote,
+            jobs=self.args.jobs,
+            run_cache=self.args.run_cache,
+        )
+
+        ui.write(
+            f"Pulled experiment '{self.args.experiment}'",
+            f"from Git remote '{self.args.git_remote}'.",
+        )
+        if not self.args.pull_cache:
+            ui.write(
+                "To pull cached outputs for this experiment"
+                "from DVC remote storage,"
+                "re-run this command without '--no-cache'."
+            )
+
+        return 0
+
+
+def add_parser(experiments_subparsers, parent_parser):
+    EXPERIMENTS_PULL_HELP = "Pull an experiment from a Git remote."
+    experiments_pull_parser = experiments_subparsers.add_parser(
+        "pull",
+        parents=[parent_parser],
+        description=append_doc_link(EXPERIMENTS_PULL_HELP, "exp/pull"),
+        help=EXPERIMENTS_PULL_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    experiments_pull_parser.add_argument(
+        "-f",
+        "--force",
+        action="store_true",
+        help="Replace local experiment already exists.",
+    )
+    experiments_pull_parser.add_argument(
+        "--no-cache",
+        action="store_false",
+        dest="pull_cache",
+        help=(
+            "Do not pull cached outputs for this experiment from DVC remote "
+            "storage."
+        ),
+    )
+    experiments_pull_parser.add_argument(
+        "-r",
+        "--remote",
+        dest="dvc_remote",
+        metavar="<name>",
+        help="Name of the DVC remote to use when pulling cached outputs.",
+    )
+    experiments_pull_parser.add_argument(
+        "-j",
+        "--jobs",
+        type=int,
+        metavar="<number>",
+        help=(
+            "Number of jobs to run simultaneously when pulling from DVC "
+            "remote storage."
+        ),
+    )
+    experiments_pull_parser.add_argument(
+        "--run-cache",
+        action="store_true",
+        default=False,
+        help="Pull run history for all stages.",
+    )
+    experiments_pull_parser.add_argument(
+        "git_remote",
+        help="Git remote name or Git URL.",
+        metavar="<git_remote>",
+    )
+    experiments_pull_parser.add_argument(
+        "experiment",
+        nargs="+",
+        help="Experiments to pull.",
+        metavar="<experiment>",
+    )
+    experiments_pull_parser.set_defaults(func=CmdExperimentsPull)

dvc/commands/experiments/push.py

@@ -0,0 +1,97 @@
+import argparse
+import logging
+
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.commands import completion
+from dvc.ui import ui
+
+logger = logging.getLogger(__name__)
+
+
+class CmdExperimentsPush(CmdBase):
+    def run(self):
+
+        self.repo.experiments.push(
+            self.args.git_remote,
+            self.args.experiment,
+            force=self.args.force,
+            push_cache=self.args.push_cache,
+            dvc_remote=self.args.dvc_remote,
+            jobs=self.args.jobs,
+            run_cache=self.args.run_cache,
+        )
+
+        ui.write(
+            f"Pushed experiment '{self.args.experiment}'"
+            f"to Git remote '{self.args.git_remote}'."
+        )
+        if not self.args.push_cache:
+            ui.write(
+                "To push cached outputs",
+                "for this experiment to DVC remote storage,"
+                "re-run this command without '--no-cache'.",
+            )
+
+        return 0
+
+
+def add_parser(experiments_subparsers, parent_parser):
+    EXPERIMENTS_PUSH_HELP = "Push a local experiment to a Git remote."
+    experiments_push_parser = experiments_subparsers.add_parser(
+        "push",
+        parents=[parent_parser],
+        description=append_doc_link(EXPERIMENTS_PUSH_HELP, "exp/push"),
+        help=EXPERIMENTS_PUSH_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    experiments_push_parser.add_argument(
+        "-f",
+        "--force",
+        action="store_true",
+        help="Replace experiment in the Git remote if it already exists.",
+    )
+    experiments_push_parser.add_argument(
+        "--no-cache",
+        action="store_false",
+        dest="push_cache",
+        help=(
+            "Do not push cached outputs for this experiment to DVC remote "
+            "storage."
+        ),
+    )
+    experiments_push_parser.add_argument(
+        "-r",
+        "--remote",
+        dest="dvc_remote",
+        metavar="<name>",
+        help="Name of the DVC remote to use when pushing cached outputs.",
+    )
+    experiments_push_parser.add_argument(
+        "-j",
+        "--jobs",
+        type=int,
+        metavar="<number>",
+        help=(
+            "Number of jobs to run simultaneously when pushing to DVC remote "
+            "storage."
+        ),
+    )
+    experiments_push_parser.add_argument(
+        "--run-cache",
+        action="store_true",
+        default=False,
+        help="Push run history for all stages.",
+    )
+    experiments_push_parser.add_argument(
+        "git_remote",
+        help="Git remote name or Git URL.",
+        metavar="<git_remote>",
+    )
+    experiments_push_parser.add_argument(
+        "experiment",
+        nargs="+",
+        help="Experiments to push.",
+        metavar="<experiment>",
+    ).complete = completion.EXPERIMENT
+    experiments_push_parser.set_defaults(func=CmdExperimentsPush)

dvc/commands/experiments/remove.py

@@ -0,0 +1,55 @@
+import argparse
+import logging
+
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+
+logger = logging.getLogger(__name__)
+
+
+class CmdExperimentsRemove(CmdBase):
+    def run(self):
+
+        self.repo.experiments.remove(
+            exp_names=self.args.experiment,
+            queue=self.args.queue,
+            clear_all=self.args.all,
+            remote=self.args.git_remote,
+        )
+
+        return 0
+
+
+def add_parser(experiments_subparsers, parent_parser):
+
+    EXPERIMENTS_REMOVE_HELP = "Remove experiments."
+    experiments_remove_parser = experiments_subparsers.add_parser(
+        "remove",
+        parents=[parent_parser],
+        description=append_doc_link(EXPERIMENTS_REMOVE_HELP, "exp/remove"),
+        help=EXPERIMENTS_REMOVE_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    remove_group = experiments_remove_parser.add_mutually_exclusive_group()
+    remove_group.add_argument(
+        "--queue", action="store_true", help="Remove all queued experiments."
+    )
+    remove_group.add_argument(
+        "-A",
+        "--all",
+        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="*",
+        help="Experiments to remove.",
+        metavar="<experiment>",
+    )
+    experiments_remove_parser.set_defaults(func=CmdExperimentsRemove)

dvc/commands/experiments/run.py

@@ -0,0 +1,142 @@
+import argparse
+import logging
+
+from dvc.cli.utils import append_doc_link
+from dvc.commands import completion
+from dvc.commands.repro import CmdRepro
+from dvc.commands.repro import add_arguments as add_repro_arguments
+from dvc.exceptions import InvalidArgumentError
+from dvc.ui import ui
+
+logger = logging.getLogger(__name__)
+
+
+class CmdExperimentsRun(CmdRepro):
+    def run(self):
+        from dvc.compare import show_metrics
+
+        if self.args.checkpoint_resume:
+            if self.args.reset:
+                raise InvalidArgumentError(
+                    "--reset and --rev are mutually exclusive."
+                )
+            if not (self.args.queue or self.args.tmp_dir):
+                raise InvalidArgumentError(
+                    "--rev can only be used in conjunction with "
+                    "--queue or --temp."
+                )
+
+        if self.args.reset:
+            ui.write("Any existing checkpoints will be reset and re-run.")
+
+        results = self.repo.experiments.run(
+            name=self.args.name,
+            queue=self.args.queue,
+            run_all=self.args.run_all,
+            jobs=self.args.jobs,
+            params=self.args.set_param,
+            checkpoint_resume=self.args.checkpoint_resume,
+            reset=self.args.reset,
+            tmp_dir=self.args.tmp_dir,
+            machine=self.args.machine,
+            **self._repro_kwargs,
+        )
+
+        if self.args.metrics and results:
+            metrics = self.repo.metrics.show(revs=list(results))
+            metrics.pop("workspace", None)
+            show_metrics(metrics)
+
+        return 0
+
+
+def add_parser(experiments_subparsers, parent_parser):
+
+    EXPERIMENTS_RUN_HELP = "Run or resume an experiment."
+    experiments_run_parser = experiments_subparsers.add_parser(
+        "run",
+        parents=[parent_parser],
+        description=append_doc_link(EXPERIMENTS_RUN_HELP, "exp/run"),
+        help=EXPERIMENTS_RUN_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    _add_run_common(experiments_run_parser)
+    experiments_run_parser.add_argument(
+        "-r",
+        "--rev",
+        type=str,
+        dest="checkpoint_resume",
+        help=(
+            "Continue the specified checkpoint experiment. Can only be used "
+            "in conjunction with --queue or --temp."
+        ),
+        metavar="<experiment_rev>",
+    ).complete = completion.EXPERIMENT
+    experiments_run_parser.add_argument(
+        "--reset",
+        action="store_true",
+        help="Reset existing checkpoints and restart the experiment.",
+    )
+    experiments_run_parser.set_defaults(func=CmdExperimentsRun)
+
+
+def _add_run_common(parser):
+    """Add common args for 'exp run' and 'exp resume'."""
+    # inherit arguments from `dvc repro`
+    add_repro_arguments(parser)
+    parser.add_argument(
+        "-n",
+        "--name",
+        default=None,
+        help=(
+            "Human-readable experiment name. If not specified, a name will "
+            "be auto-generated."
+        ),
+        metavar="<name>",
+    )
+    parser.add_argument(
+        "-S",
+        "--set-param",
+        action="append",
+        default=[],
+        help="Use the specified param value when reproducing pipelines.",
+        metavar="[<filename>:]<param_name>=<param_value>",
+    )
+    parser.add_argument(
+        "--queue",
+        action="store_true",
+        default=False,
+        help="Stage this experiment in the run queue for future execution.",
+    )
+    parser.add_argument(
+        "--run-all",
+        action="store_true",
+        default=False,
+        help="Execute all experiments in the run queue. Implies --temp.",
+    )
+    parser.add_argument(
+        "-j",
+        "--jobs",
+        type=int,
+        default=1,
+        help="Run the specified number of experiments at a time in parallel.",
+        metavar="<number>",
+    )
+    parser.add_argument(
+        "--temp",
+        action="store_true",
+        dest="tmp_dir",
+        help=(
+            "Run this experiment in a separate temporary directory instead of "
+            "your workspace."
+        ),
+    )
+    parser.add_argument(
+        "--machine",
+        default=None,
+        help=argparse.SUPPRESS,
+        # help=(
+        #     "Run this experiment on the specified 'dvc machine' instance."
+        # )
+        # metavar="<name>",
+    )

dvc/commands/experiments/show.py

@@ -0,0 +1,669 @@
+import argparse
+import logging
+import os
+import re
+from collections import Counter, OrderedDict, defaultdict
+from datetime import date, datetime
+from typing import TYPE_CHECKING
+
+from funcy import lmap
+
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.commands import completion
+from dvc.commands.metrics import DEFAULT_PRECISION
+from dvc.exceptions import DvcException, InvalidArgumentError
+from dvc.ui import ui
+from dvc.utils.flatten import flatten
+from dvc.utils.serialize import encode_exception
+
+if TYPE_CHECKING:
+    from dvc.compare import TabularData
+    from dvc.ui import RichText
+
+FILL_VALUE = "-"
+FILL_VALUE_ERRORED = "!"
+
+
+logger = logging.getLogger(__name__)
+
+
+def _update_names(names, items):
+    for name, item in items:
+        item = item.get("data", {})
+        if isinstance(item, dict):
+            item = flatten(item)
+            names[name].update({key: None for key in item})
+
+
+def _collect_names(all_experiments, **kwargs):
+    metric_names = defaultdict(dict)
+    param_names = defaultdict(dict)
+    deps_names = set()
+
+    for _, experiments in all_experiments.items():
+        for exp_data in experiments.values():
+            exp = exp_data.get("data", {})
+            _update_names(metric_names, exp.get("metrics", {}).items())
+            _update_names(param_names, exp.get("params", {}).items())
+            for dep_name in exp.get("deps", {}):
+                deps_names.add(dep_name)
+
+    return metric_names, param_names, sorted(deps_names)
+
+
+experiment_types = {
+    "checkpoint_tip": "│ ╓",
+    "checkpoint_commit": "│ ╟",
+    "checkpoint_base": "├─╨",
+    "branch_commit": "├──",
+    "branch_base": "└──",
+    "baseline": "",
+}
+
+
+def _collect_rows(
+    base_rev,
+    experiments,
+    all_headers,
+    metric_headers,
+    param_headers,
+    metric_names,
+    param_names,
+    deps_names,
+    precision=DEFAULT_PRECISION,
+    sort_by=None,
+    sort_order=None,
+    fill_value=FILL_VALUE,
+    iso=False,
+):
+    from scmrepo.git import Git
+
+    if sort_by:
+        sort_path, sort_name, sort_type = _sort_column(
+            sort_by, metric_names, param_names
+        )
+        reverse = sort_order == "desc"
+        experiments = _sort_exp(
+            experiments, sort_path, sort_name, sort_type, reverse
+        )
+
+    new_checkpoint = True
+    for i, (rev, results) in enumerate(experiments.items()):
+        fill_value = FILL_VALUE_ERRORED if results.get("error") else fill_value
+        row_dict = {k: fill_value for k in all_headers}
+
+        exp = results.get("data", {})
+
+        if exp.get("running"):
+            state = "Running"
+        elif exp.get("queued"):
+            state = "Queued"
+        else:
+            state = fill_value
+
+        is_baseline = rev == "baseline"
+
+        if is_baseline:
+            name_rev = base_rev[:7] if Git.is_sha(base_rev) else base_rev
+        else:
+            name_rev = rev[:7]
+
+        tip = exp.get("checkpoint_tip")
+        parent_rev = exp.get("checkpoint_parent", "")
+        parent_exp = experiments.get(parent_rev, {}).get("data", {})
+        parent_tip = parent_exp.get("checkpoint_tip")
+
+        parent = ""
+        if is_baseline:
+            typ = "baseline"
+        elif tip:
+            if tip == parent_tip:
+                typ = (
+                    "checkpoint_tip" if new_checkpoint else "checkpoint_commit"
+                )
+            elif parent_rev == base_rev:
+                typ = "checkpoint_base"
+            else:
+                typ = "checkpoint_commit"
+                parent = parent_rev[:7]
+        elif i < len(experiments) - 1:
+            typ = "branch_commit"
+        else:
+            typ = "branch_base"
+
+        if not is_baseline:
+            new_checkpoint = not (tip and tip == parent_tip)
+
+        row_dict["Experiment"] = exp.get("name", "")
+        row_dict["rev"] = name_rev
+        row_dict["typ"] = typ
+        row_dict["Created"] = _format_time(
+            exp.get("timestamp"), fill_value, iso
+        )
+        row_dict["parent"] = parent
+        row_dict["State"] = state
+        row_dict["Executor"] = exp.get("executor", fill_value)
+
+        _extend_row(
+            row_dict,
+            metric_names,
+            metric_headers,
+            exp.get("metrics", {}).items(),
+            precision,
+            fill_value=fill_value,
+        )
+        _extend_row(
+            row_dict,
+            param_names,
+            param_headers,
+            exp.get("params", {}).items(),
+            precision,
+            fill_value=fill_value,
+        )
+        for dep in deps_names:
+            hash_info = exp.get("deps", {}).get(dep, {}).get("hash")
+            if hash_info is not None:
+                hash_info = hash_info[:7]
+            row_dict[dep] = hash_info
+        yield list(row_dict.values())
+
+
+def _sort_column(sort_by, metric_names, param_names):
+    path, _, sort_name = sort_by.rpartition(":")
+    matches = set()
+
+    if path:
+        if path in metric_names and sort_name in metric_names[path]:
+            matches.add((path, sort_name, "metrics"))
+        if path in param_names and sort_name in param_names[path]:
+            matches.add((path, sort_name, "params"))
+    else:
+        for path in metric_names:
+            if sort_name in metric_names[path]:
+                matches.add((path, sort_name, "metrics"))
+        for path in param_names:
+            if sort_name in param_names[path]:
+                matches.add((path, sort_name, "params"))
+
+    if len(matches) == 1:
+        return matches.pop()
+    if len(matches) > 1:
+        raise InvalidArgumentError(
+            "Ambiguous sort column '{}' matched '{}'".format(
+                sort_by,
+                ", ".join([f"{path}:{name}" for path, name, _ in matches]),
+            )
+        )
+    raise InvalidArgumentError(f"Unknown sort column '{sort_by}'")
+
+
+def _sort_exp(experiments, sort_path, sort_name, typ, reverse):
+    def _sort(item):
+        rev, exp = item
+        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_data.get(typ, {}).get(sort_path, {}).get("data", {})
+        val = flatten(data).get(sort_name)
+        return val is None, val
+
+    ret = OrderedDict()
+    if "baseline" in experiments:
+        ret["baseline"] = experiments.pop("baseline")
+
+    ret.update(sorted(experiments.items(), key=_sort, reverse=reverse))
+    return ret
+
+
+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 datetime_obj.strftime(fmt)
+
+
+def _extend_row(row, names, headers, items, precision, fill_value=FILL_VALUE):
+    from dvc.compare import _format_field, with_value
+
+    for fname, data in items:
+        item = data.get("data", {})
+        item = flatten(item) if isinstance(item, dict) else {fname: item}
+        for name in names[fname]:
+            value = with_value(
+                item.get(name),
+                FILL_VALUE_ERRORED if data.get("error", None) else fill_value,
+            )
+            # wrap field data in ui.rich_text, otherwise rich may
+            # interpret unescaped braces from list/dict types as rich
+            # markup tags
+            value = ui.rich_text(str(_format_field(value, precision)))
+            if name in headers:
+                row[name] = value
+            else:
+                row[f"{fname}:{name}"] = value
+
+
+def experiments_table(
+    all_experiments,
+    headers,
+    metric_headers,
+    metric_names,
+    param_headers,
+    param_names,
+    deps_names,
+    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
+
+    all_headers = lconcat(headers, metric_headers, param_headers, deps_names)
+    td = TabularData(all_headers, fill_value=fill_value)
+    for base_rev, experiments in all_experiments.items():
+        rows = _collect_rows(
+            base_rev,
+            experiments,
+            all_headers,
+            metric_headers,
+            param_headers,
+            metric_names,
+            param_names,
+            deps_names,
+            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) -> "RichText":
+    exp_name = kwargs["Experiment"]
+    rev = kwargs["rev"]
+    typ = kwargs.get("typ", "baseline")
+
+    if typ == "baseline" or not exp_name:
+        text = ui.rich_text(exp_name or rev)
+    else:
+        text = ui.rich_text.assemble(rev, " [", (exp_name, "bold"), "]")
+
+    parent = kwargs.get("parent")
+    suff = f" ({parent})" if parent else ""
+    text.append(suff)
+
+    tree = experiment_types[typ]
+    pref = f"{tree} " if tree else ""
+    return ui.rich_text(pref) + text
+
+
+def baseline_styler(typ):
+    return {"style": "bold"} if typ == "baseline" else {}
+
+
+def show_experiments(
+    all_experiments,
+    keep=None,
+    drop=None,
+    pager=True,
+    csv=False,
+    markdown=False,
+    pcp=False,
+    **kwargs,
+):
+    from funcy.seqs import flatten as flatten_list
+
+    metric_names, param_names, deps_names = _collect_names(all_experiments)
+
+    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,
+        param_names,
+        deps_names,
+        kwargs.get("sort_by"),
+        kwargs.get("sort_order"),
+        kwargs.get("precision"),
+        kwargs.get("fill_value"),
+        kwargs.get("iso"),
+    )
+    if keep:
+        for col in td.keys():
+            if re.match(keep, col):
+                td.protect(col)
+
+    for col in ("State", "Executor"):
+        if td.is_empty(col):
+            td.drop(col)
+
+    row_styles = lmap(baseline_styler, td.column("typ"))
+
+    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,
+        "deps": deps_names,
+    }
+    styles = {
+        "Experiment": {"no_wrap": True, "header_style": "black on grey93"},
+        "Created": {"header_style": "black on grey93"},
+        "State": {"header_style": "black on grey93"},
+        "Executor": {"header_style": "black on grey93"},
+    }
+    header_bg_colors = {
+        "metrics": "cornsilk1",
+        "params": "light_cyan1",
+        "deps": "plum2",
+    }
+    styles.update(
+        {
+            header: {
+                "justify": "right" if typ == "metrics" else "left",
+                "header_style": f"black on {header_bg_colors[typ]}",
+                "collapse": idx != 0,
+                "no_wrap": typ == "metrics",
+            }
+            for typ, hs in headers.items()
+            for idx, header in enumerate(hs)
+        }
+    )
+
+    if kwargs.get("only_changed", False) or pcp:
+        td.drop_duplicates("cols", ignore_empty=False)
+
+    cols_to_drop = set()
+    if drop is not None:
+        cols_to_drop = {col for col in td.keys() if re.match(drop, col)}
+    if pcp:
+        cols_to_drop.add("Created")
+    td.drop(*cols_to_drop)
+
+    if pcp:
+        subset = {x for x in td.keys() if x != "Experiment"}
+        td.dropna(
+            "rows",
+            how="all",
+            subset=subset,
+        )
+        td.drop_duplicates("rows", subset=subset)
+        td.column("Experiment")[:] = [
+            # remove tree characters
+            str(x).encode("ascii", "ignore").strip().decode()
+            for x in td.column("Experiment")
+        ]
+        out = kwargs.get("out") or "dvc_plots"
+        ui.write(
+            td.to_parallel_coordinates(
+                output_path=os.path.abspath(out),
+                color_by=kwargs.get("sort_by") or "Experiment",
+            )
+        )
+        if kwargs.get("open"):
+            return ui.open_browser(os.path.join(out, "index.html"))
+
+    else:
+        td.render(
+            pager=pager,
+            borders="horizontals",
+            rich_table=True,
+            header_styles=styles,
+            row_styles=row_styles,
+            csv=csv,
+            markdown=markdown,
+        )
+
+
+def _normalize_headers(names, count):
+    return [
+        name if count[name] == 1 else f"{path}:{name}"
+        for path in names
+        for name in names[path]
+    ]
+
+
+def _format_json(item):
+    if isinstance(item, (date, datetime)):
+        return item.isoformat()
+    return encode_exception(item)
+
+
+class CmdExperimentsShow(CmdBase):
+    def run(self):
+        try:
+            all_experiments = self.repo.experiments.show(
+                all_branches=self.args.all_branches,
+                all_tags=self.args.all_tags,
+                all_commits=self.args.all_commits,
+                revs=self.args.rev,
+                num=self.args.num,
+                sha_only=self.args.sha,
+                param_deps=self.args.param_deps,
+            )
+        except DvcException:
+            logger.exception("failed to show experiments")
+            return 1
+
+        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,
+                keep=self.args.keep,
+                drop=self.args.drop,
+                sort_by=self.args.sort_by,
+                sort_order=self.args.sort_order,
+                precision=precision,
+                fill_value=fill_value,
+                iso=iso,
+                pager=not self.args.no_pager,
+                csv=self.args.csv,
+                markdown=self.args.markdown,
+                only_changed=self.args.only_changed,
+                pcp=self.args.pcp,
+                out=self.args.out,
+                open=self.args.open,
+            )
+        return 0
+
+
+def add_parser(experiments_subparsers, parent_parser):
+    EXPERIMENTS_SHOW_HELP = "Print experiments."
+    experiments_show_parser = experiments_subparsers.add_parser(
+        "show",
+        parents=[parent_parser],
+        description=append_doc_link(EXPERIMENTS_SHOW_HELP, "exp/show"),
+        help=EXPERIMENTS_SHOW_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    experiments_show_parser.add_argument(
+        "-a",
+        "--all-branches",
+        action="store_true",
+        default=False,
+        help="Show experiments derived from the tip of all Git branches.",
+    )
+    experiments_show_parser.add_argument(
+        "-T",
+        "--all-tags",
+        action="store_true",
+        default=False,
+        help="Show experiments derived from all Git tags.",
+    )
+    experiments_show_parser.add_argument(
+        "-A",
+        "--all-commits",
+        action="store_true",
+        default=False,
+        help="Show experiments derived from all Git commits.",
+    )
+    experiments_show_parser.add_argument(
+        "--rev",
+        type=str,
+        default=None,
+        help=(
+            "Show experiments derived from the specified revision. "
+            "Defaults to HEAD if none of `--rev`,`-a`,`-A`,`-T` is specified."
+        ),
+        metavar="<rev>",
+    )
+    experiments_show_parser.add_argument(
+        "-n",
+        "--num",
+        type=int,
+        default=1,
+        dest="num",
+        metavar="<num>",
+        help="Show the last `num` commits from <rev>.",
+    )
+    experiments_show_parser.add_argument(
+        "--no-pager",
+        action="store_true",
+        default=False,
+        help="Do not pipe output into a pager.",
+    )
+    experiments_show_parser.add_argument(
+        "--only-changed",
+        action="store_true",
+        default=False,
+        help=(
+            "Only show metrics/params with values varying "
+            "across the selected experiments."
+        ),
+    )
+    experiments_show_parser.add_argument(
+        "--drop",
+        help="Remove the columns matching the specified regex pattern.",
+        metavar="<regex_pattern>",
+    )
+    experiments_show_parser.add_argument(
+        "--keep",
+        help="Preserve the columns matching the specified regex pattern.",
+        metavar="<regex_pattern>",
+    )
+    experiments_show_parser.add_argument(
+        "--param-deps",
+        action="store_true",
+        default=False,
+        help="Show only params that are stage dependencies.",
+    )
+    experiments_show_parser.add_argument(
+        "--sort-by",
+        help="Sort related experiments by the specified metric or param.",
+        metavar="<metric/param>",
+    )
+    experiments_show_parser.add_argument(
+        "--sort-order",
+        help=(
+            "Sort order to use with --sort-by."
+            " Defaults to ascending ('asc')."
+        ),
+        choices=("asc", "desc"),
+        default="asc",
+    )
+    experiments_show_parser.add_argument(
+        "--sha",
+        action="store_true",
+        default=False,
+        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,
+        help=(
+            "Round metrics/params to `n` digits precision after the decimal "
+            f"point. Rounds to {DEFAULT_PRECISION} digits by default."
+        ),
+        metavar="<n>",
+    )
+    experiments_show_parser.add_argument(
+        "--parallel-coordinates-plot",
+        "--pcp",
+        dest="pcp",
+        action="store_true",
+        default=False,
+        help="Generate a Parallel Coordinates Plot from the tabulated output.",
+    )
+    experiments_show_parser.add_argument(
+        "-o",
+        "--out",
+        default=None,
+        help="Destination folder to save the Parallel Coordinates Plot to",
+        metavar="<path>",
+    ).complete = completion.DIR
+    experiments_show_parser.add_argument(
+        "--open",
+        action="store_true",
+        default=False,
+        help="Open the Parallel Coordinates Plot directly in the browser.",
+    )
+    experiments_show_parser.set_defaults(func=CmdExperimentsShow)

dvc/commands/freeze.py

@@ -1,8 +1,9 @@
 import argparse
 import logging
 
-from dvc.command import completion
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.commands import completion
 from dvc.exceptions import DvcException
 
 logger = logging.getLogger(__name__)

dvc/commands/gc.py

@@ -2,7 +2,8 @@ import argparse
 import logging
 import os
 
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
 from dvc.ui import ui
 
 logger = logging.getLogger(__name__)

dvc/commands/get.py

@@ -1,10 +1,11 @@
 import argparse
 import logging
 
+from dvc.cli.command import CmdBaseNoRepo
 from dvc.exceptions import DvcException
 
+from ..cli.utils import append_doc_link
 from . import completion
-from .base import CmdBaseNoRepo, append_doc_link
 
 logger = logging.getLogger(__name__)
 

dvc/commands/get_url.py

@@ -1,10 +1,11 @@
 import argparse
 import logging
 
+from dvc.cli.command import CmdBaseNoRepo
 from dvc.exceptions import DvcException
 
+from ..cli.utils import append_doc_link
 from . import completion
-from .base import CmdBaseNoRepo, append_doc_link
 
 logger = logging.getLogger(__name__)
 

dvc/commands/git_hook.py

@@ -1,7 +1,8 @@
 import logging
 import os
 
-from dvc.command.base import CmdBaseNoRepo, fix_subparsers
+from dvc.cli.command import CmdBaseNoRepo
+from dvc.cli.utils import fix_subparsers
 from dvc.exceptions import NotDvcRepoError
 
 logger = logging.getLogger(__name__)
@@ -25,7 +26,7 @@ class CmdHookBase(CmdBaseNoRepo):
 
 class CmdPreCommit(CmdHookBase):
     def _run(self):
-        from dvc.main import main
+        from dvc.cli import main
 
         return main(["status"])
 
@@ -49,14 +50,14 @@ class CmdPostCheckout(CmdHookBase):
         if os.path.isdir(os.path.join(".git", "rebase-merge")):
             return 0
 
-        from dvc.main import main
+        from dvc.cli import main
 
         return main(["checkout"])
 
 
 class CmdPrePush(CmdHookBase):
     def _run(self):
-        from dvc.main import main
+        from dvc.cli import main
 
         return main(["push"])
 

dvc/commands/imp.py

@@ -1,8 +1,9 @@
 import argparse
 import logging
 
-from dvc.command import completion
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.commands import completion
 from dvc.exceptions import DvcException
 
 logger = logging.getLogger(__name__)

dvc/commands/imp_url.py

@@ -1,8 +1,9 @@
 import argparse
 import logging
 
-from dvc.command import completion
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.commands import completion
 from dvc.exceptions import DvcException
 
 logger = logging.getLogger(__name__)

dvc/commands/init.py

@@ -4,7 +4,8 @@ import logging
 import colorama
 
 from dvc import analytics
-from dvc.command.base import CmdBaseNoRepo, append_doc_link
+from dvc.cli.command import CmdBaseNoRepo
+from dvc.cli.utils import append_doc_link
 from dvc.utils import boxify
 from dvc.utils import format_link as fmt_link
 

dvc/commands/install.py

@@ -1,7 +1,8 @@
 import argparse
 import logging
 
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
 from dvc.exceptions import DvcException
 
 logger = logging.getLogger(__name__)

dvc/commands/live.py

@@ -1,8 +1,8 @@
 import argparse
-import os
 
-from dvc.command import completion
-from dvc.command.base import CmdBase, fix_subparsers
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import fix_subparsers
+from dvc.commands import completion
 from dvc.ui import ui
 
 
@@ -10,14 +10,20 @@ class CmdLive(CmdBase):
     UNINITIALIZED = True
 
     def _run(self, target, revs=None):
+        from dvc.render.utils import match_renderers, render
+
         metrics, plots = self.repo.live.show(target=target, revs=revs)
 
-        html_path = self.args.target + ".html"
-        self.repo.plots.write_html(html_path, plots, metrics)
+        if plots:
+            from pathlib import Path
 
-        ui.write("\nfile://", os.path.abspath(html_path), sep="")
+            html_path = Path.cwd() / (self.args.target + "_html")
 
-        return 0
+            renderers = match_renderers(plots, self.repo.plots.templates)
+            index_path = render(self.repo, renderers, metrics, html_path)
+            ui.write(index_path.as_uri())
+            return 0
+        return 1
 
 
 class CmdLiveShow(CmdLive):

dvc/commands/ls/__init__.py

@@ -1,9 +1,10 @@
 import argparse
 import logging
 
-from dvc.command import completion
-from dvc.command.base import CmdBaseNoRepo, append_doc_link
-from dvc.command.ls.ls_colors import LsColors
+from dvc.cli.command import CmdBaseNoRepo
+from dvc.cli.utils import append_doc_link
+from dvc.commands import completion
+from dvc.commands.ls.ls_colors import LsColors
 from dvc.exceptions import DvcException
 from dvc.ui import ui
 
@@ -34,10 +35,8 @@ 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 +53,7 @@ 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 +70,10 @@ 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/commands/machine.py

@@ -0,0 +1,508 @@
+import argparse
+
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link, fix_subparsers
+from dvc.commands.config import CmdConfig
+from dvc.compare import TabularData
+from dvc.config import ConfigError
+from dvc.exceptions import DvcException
+from dvc.types import Dict, List
+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):
+    TABLE_COLUMNS = [
+        "name",
+        "cloud",
+        "region",
+        "image",
+        "spot",
+        "spot_price",
+        "instance_hdd_size",
+        "instance_type",
+        "ssh_private",
+        "startup_script",
+    ]
+
+    PRIVATE_COLUMNS = ["ssh_private", "startup_script"]
+
+    def _hide_private(self, conf):
+        for machine in conf:
+            for column in self.PRIVATE_COLUMNS:
+                if column in conf[machine]:
+                    conf[machine][column] = "***"
+
+    def _show_origin(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, {})
+            self._hide_private(conf)
+            prefix = self._config_file_prefix(True, self.config, level)
+            configs = list(self._format_config(conf, prefix))
+            if configs:
+                ui.write("\n".join(configs))
+
+    def _show_table(self):
+        td = TabularData(self.TABLE_COLUMNS, fill_value="-")
+        conf = self.config.read()["machine"]
+        if self.args.name:
+            conf = {self.args.name: conf.get(self.args.name, {})}
+        self._hide_private(conf)
+        for machine, machine_config in conf.items():
+            machine_config["name"] = machine
+            td.row_from_dict(machine_config)
+        td.dropna("cols", "all")
+        td.render()
+
+    def run(self):
+        if self.args.show_origin:
+            self._show_origin()
+        else:
+            self._show_table()
+        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):
+    INSTANCE_FIELD = ["name", "instance", "status"]
+    SHOWN_FIELD = [
+        "cloud",
+        "instance_ip",
+        "instance_type",
+        "instance_hdd_size",
+        "instance_gpu",
+    ]
+    FILL_VALUE = "-"
+
+    def _add_row(
+        self,
+        name: str,
+        all_status: List[Dict],
+        td: TabularData,
+    ):
+
+        if not all_status:
+            row = [
+                name,
+                self.FILL_VALUE,
+                "offline",
+            ]  # back to `None` after #7167
+            td.append(row)
+        for i, status in enumerate(all_status, start=1):
+            row = [name, f"num_{i}", "running" if status else "offline"]
+            for field in self.SHOWN_FIELD:
+                value = str(status.get(field, ""))
+                row.append(value)
+            td.append(row)
+
+    def run(self):
+        if self.repo.machine is None:
+            raise MachineDisabledError
+
+        td = TabularData(
+            self.INSTANCE_FIELD + self.SHOWN_FIELD, fill_value=self.FILL_VALUE
+        )
+
+        if self.args.name:
+            all_status = list(self.repo.machine.status(self.args.name))
+            self._add_row(self.args.name, all_status, td)
+        else:
+            name_set = set()
+            for level in self.repo.config.LEVELS:
+                conf = self.repo.config.read(level)["machine"]
+                name_set.update(conf.keys())
+            name_list = list(name_set)
+            for name in sorted(name_list):
+                all_status = list(self.repo.machine.status(name))
+                self._add_row(name, all_status, td)
+
+        td.dropna("cols", "all")
+        td.render()
+        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.commands.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_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 running instances for one/all machines."
+    )
+    machine_status_parser = machine_subparsers.add_parser(
+        "status",
+        parents=[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", nargs="?", help="(optional) Name of the 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_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_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/commands/metrics.py

@@ -1,10 +1,12 @@
 import argparse
 import logging
 
-from dvc.command import completion
-from dvc.command.base import CmdBase, append_doc_link, fix_subparsers
-from dvc.exceptions import BadMetricError, DvcException
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link, fix_subparsers
+from dvc.commands import completion
+from dvc.exceptions import DvcException
 from dvc.ui import ui
+from dvc.utils.serialize import encode_exception
 
 logger = logging.getLogger(__name__)
 
@@ -30,23 +32,14 @@ class CmdMetricsShow(CmdMetricsBase):
             logger.exception("")
             return 1
 
-        if self.args.show_json:
-            import json
-
-            ui.write(json.dumps(metrics))
+        if self.args.json:
+            ui.write_json(metrics, default=encode_exception)
         else:
             from dvc.compare import show_metrics
 
-            # When `metrics` contains a `None` key, it means that some files
-            # specified as `targets` in `repo.metrics.show` didn't contain any
-            # metrics.
-            missing = metrics.pop(None, None)
-            if missing:
-                raise BadMetricError(missing)
-
             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,
@@ -71,20 +64,20 @@ 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
@@ -146,15 +139,18 @@ 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(
@@ -226,15 +222,18 @@ 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/commands/move.py

@@ -1,8 +1,9 @@
 import argparse
 import logging
 
-from dvc.command import completion
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.commands import completion
 from dvc.exceptions import DvcException
 
 logger = logging.getLogger(__name__)

dvc/commands/params.py

@@ -1,8 +1,9 @@
 import argparse
 import logging
 
-from dvc.command import completion
-from dvc.command.base import CmdBase, append_doc_link, fix_subparsers
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link, fix_subparsers
+from dvc.commands import completion
 from dvc.exceptions import DvcException
 from dvc.ui import ui
 
@@ -25,19 +26,19 @@ 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 +104,18 @@ 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/commands/plots.py

@@ -1,8 +1,12 @@
 import argparse
+import json
 import logging
 
-from dvc.command import completion
-from dvc.command.base import CmdBase, append_doc_link, fix_subparsers
+from funcy import first
+
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link, fix_subparsers
+from dvc.commands import completion
 from dvc.exceptions import DvcException
 from dvc.ui import ui
 from dvc.utils import format_link
@@ -10,6 +14,18 @@ from dvc.utils import format_link
 logger = logging.getLogger(__name__)
 
 
+def _show_json(renderers, path: None):
+    if any(r.needs_output_path for r in renderers) and not path:
+        raise DvcException("Output path ('-o') is required!")
+
+    result = {
+        renderer.filename: json.loads(renderer.as_json(path=path))
+        for renderer in renderers
+    }
+    if result:
+        ui.write_json(result)
+
+
 class CmdPlots(CmdBase):
     def _func(self, *args, **kwargs):
         raise NotImplementedError
@@ -24,6 +40,9 @@ class CmdPlots(CmdBase):
     def run(self):
         from pathlib import Path
 
+        from dvc.render.utils import match_renderers, render
+        from dvc.render.vega import VegaRenderer
+
         if self.args.show_vega:
             if not self.args.targets:
                 logger.error("please specify a target for `--show-vega`")
@@ -33,40 +52,66 @@ class CmdPlots(CmdBase):
                     "you can only specify one target for `--show-vega`"
                 )
                 return 1
+            if self.args.json:
+                logger.error(
+                    "'--show-vega' and '--json' are mutually exclusive "
+                    "options."
+                )
+                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."
+                )
+
+            renderers = match_renderers(
+                plots_data=plots_data, templates=self.repo.plots.templates
+            )
 
             if self.args.show_vega:
-                target = self.args.targets[0]
-                ui.write(plots[target])
+                renderer = first(
+                    filter(lambda r: isinstance(r, VegaRenderer), renderers)
+                )
+                if renderer:
+                    ui.write_json(renderer.asdict())
+                return 0
+            if self.args.json:
+                _show_json(renderers, self.args.out)
                 return 0
 
-            rel: str = self.args.out or "plots.html"
-            path = (Path.cwd() / rel).resolve()
-            self.repo.plots.write_html(
-                path, plots=plots, html_template_path=self.args.html_template
+            rel: str = self.args.out or "dvc_plots"
+            path: Path = (Path.cwd() / rel).resolve()
+            index_path = render(
+                self.repo,
+                renderers,
+                path=path,
+                html_template_path=self.args.html_template,
             )
 
+            ui.write(index_path.as_uri())
+            auto_open = self.repo.config["plots"].get("auto_open", False)
+            if self.args.open or auto_open:
+                if not auto_open:
+                    ui.write(
+                        "To enable auto opening, you can run:\n"
+                        "\n"
+                        "\tdvc config plots.auto_open true"
+                    )
+                return ui.open_browser(index_path)
+
+            return 0
+
         except DvcException:
             logger.exception("")
             return 1
 
-        assert path.is_absolute()  # as_uri throws ValueError if not absolute
-        url = path.as_uri()
-        ui.write(url)
-        if self.args.open:
-            import webbrowser
-
-            opened = webbrowser.open(rel)
-            if not opened:
-                ui.error_write(
-                    "Failed to open. Please try opening it manually."
-                )
-                return 1
-
-        return 0
-
 
 class CmdPlotsShow(CmdPlots):
     UNINITIALIZED = True
@@ -95,6 +140,37 @@ class CmdPlotsModify(CmdPlots):
         return 0
 
 
+class CmdPlotsTemplates(CmdBase):
+    TEMPLATES_CHOICES = [
+        "simple",
+        "linear",
+        "confusion",
+        "confusion_normalized",
+        "scatter",
+        "smooth",
+    ]
+
+    def run(self):
+        import os
+
+        try:
+            out = (
+                os.path.join(os.getcwd(), self.args.out)
+                if self.args.out
+                else self.repo.plots.templates.templates_dir
+            )
+
+            targets = [self.args.target] if self.args.target else None
+            self.repo.plots.templates.init(output=out, targets=targets)
+            templates_path = os.path.relpath(out, os.getcwd())
+            ui.write(f"Templates have been written into '{templates_path}'.")
+
+            return 0
+        except DvcException:
+            logger.exception("")
+            return 1
+
+
 def add_parser(subparsers, parent_parser):
     PLOTS_HELP = (
         "Commands to visualize and compare plot metrics in structured files "
@@ -131,7 +207,8 @@ def add_parser(subparsers, parent_parser):
         "Shows all plots by default.",
     ).complete = completion.FILE
     _add_props_arguments(plots_show_parser)
-    _add_output_arguments(plots_show_parser)
+    _add_output_argument(plots_show_parser)
+    _add_ui_arguments(plots_show_parser)
     plots_show_parser.set_defaults(func=CmdPlotsShow)
 
     PLOTS_DIFF_HELP = (
@@ -166,10 +243,14 @@ def add_parser(subparsers, parent_parser):
         "revisions", nargs="*", default=None, help="Git commits to plot from"
     )
     _add_props_arguments(plots_diff_parser)
-    _add_output_arguments(plots_diff_parser)
+    _add_output_argument(plots_diff_parser)
+    _add_ui_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],
@@ -189,6 +270,27 @@ def add_parser(subparsers, parent_parser):
     )
     plots_modify_parser.set_defaults(func=CmdPlotsModify)
 
+    TEMPLATES_HELP = (
+        "Write built-in plots templates to a directory "
+        "(.dvc/plots by default)."
+    )
+    plots_templates_parser = plots_subparsers.add_parser(
+        "templates",
+        parents=[parent_parser],
+        description=append_doc_link(TEMPLATES_HELP, "plots/templates"),
+        help=TEMPLATES_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    plots_templates_parser.add_argument(
+        "target",
+        default=None,
+        nargs="?",
+        choices=CmdPlotsTemplates.TEMPLATES_CHOICES,
+        help="Template to write. Writes all templates by default.",
+    )
+    _add_output_argument(plots_templates_parser, typ="templates")
+    plots_templates_parser.set_defaults(func=CmdPlotsTemplates)
+
 
 def _add_props_arguments(parser):
     parser.add_argument(
@@ -228,20 +330,30 @@ def _add_props_arguments(parser):
     )
 
 
-def _add_output_arguments(parser):
+def _add_output_argument(parser, typ="plots"):
     parser.add_argument(
         "-o",
         "--out",
         default=None,
-        help="Destination path to save plots to",
+        help=f"Directory to save {typ} to.",
         metavar="<path>",
     ).complete = completion.DIR
+
+
+def _add_ui_arguments(parser):
     parser.add_argument(
         "--show-vega",
         action="store_true",
         default=False,
         help="Show output in Vega format.",
     )
+    parser.add_argument(
+        "--json",
+        "--show-json",
+        action="store_true",
+        default=False,
+        help=argparse.SUPPRESS,
+    )
     parser.add_argument(
         "--open",
         action="store_true",

dvc/commands/remote.py

@@ -1,7 +1,7 @@
 import argparse
 
-from dvc.command.base import append_doc_link, fix_subparsers
-from dvc.command.config import CmdConfig
+from dvc.cli.utils import append_doc_link, fix_subparsers
+from dvc.commands.config import CmdConfig
 from dvc.ui import ui
 from dvc.utils import format_link
 
@@ -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
@@ -151,7 +151,7 @@ class CmdRemoteRename(CmdRemote):
 
 
 def add_parser(subparsers, parent_parser):
-    from dvc.command.config import parent_config_parser
+    from dvc.commands.config import parent_config_parser
 
     REMOTE_HELP = "Set up and manage data remotes."
     remote_parser = subparsers.add_parser(

dvc/commands/remove.py

@@ -1,8 +1,9 @@
 import argparse
 import logging
 
-from dvc.command import completion
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.commands import completion
 from dvc.exceptions import DvcException
 
 logger = logging.getLogger(__name__)

dvc/commands/repro.py

@@ -1,8 +1,9 @@
 import argparse
 
-from dvc.command import completion
-from dvc.command.base import CmdBase, append_doc_link
-from dvc.command.status import CmdDataStatus
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.commands import completion
+from dvc.commands.status import CmdDataStatus
 
 
 class CmdRepro(CmdBase):

dvc/commands/root.py

@@ -1,7 +1,8 @@
 import argparse
 import logging
 
-from dvc.command.base import CmdBaseNoRepo, append_doc_link
+from dvc.cli.command import CmdBaseNoRepo
+from dvc.cli.utils import append_doc_link
 from dvc.utils import relpath
 
 logger = logging.getLogger(__name__)

dvc/commands/run.py

@@ -1,8 +1,9 @@
 import argparse
 import logging
 
-from dvc.command.base import CmdBase, append_doc_link
-from dvc.command.stage import parse_cmd
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.commands.stage import parse_cmd
 from dvc.exceptions import DvcException
 
 logger = logging.getLogger(__name__)
@@ -23,7 +24,7 @@ class CmdRun(CmdBase):
                 self.args.outs_persist_no_cache,
                 self.args.checkpoints,
                 self.args.params,
-                self.args.cmd,
+                self.args.command,
             ]
         ):  # pragma: no cover
             logger.error(
@@ -36,7 +37,7 @@ class CmdRun(CmdBase):
         kwargs = vars(self.args)
         kwargs.update(
             {
-                "cmd": parse_cmd(self.args.cmd),
+                "cmd": parse_cmd(self.args.command),
                 "fname": kwargs.pop("file"),
                 "no_exec": (self.args.no_exec or bool(self.args.checkpoints)),
                 "run_cache": not kwargs.pop("no_run_cache"),
@@ -52,7 +53,7 @@ class CmdRun(CmdBase):
 
 
 def add_parser(subparsers, parent_parser):
-    from dvc.command.stage import _add_common_args
+    from dvc.commands.stage import _add_common_args
 
     RUN_HELP = (
         "Generate a dvc.yaml file from a command and execute the command."

dvc/commands/stage.py

@@ -3,8 +3,9 @@ import logging
 from itertools import chain, filterfalse
 from typing import TYPE_CHECKING, Dict, Iterable, List
 
-from dvc.command import completion
-from dvc.command.base import CmdBase, append_doc_link, fix_subparsers
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link, fix_subparsers
+from dvc.commands import completion
 from dvc.utils.cli_parse import parse_params
 from dvc.utils.humanize import truncate_text
 
@@ -66,7 +67,7 @@ def prepare_stages_data(
 class CmdStageList(CmdBase):
     def _get_stages(self) -> Iterable["Stage"]:
         if self.args.all:
-            stages: List["Stage"] = self.repo.stages  # type: ignore
+            stages: List["Stage"] = self.repo.index.stages  # type: ignore
             logger.trace(  # type: ignore[attr-defined]
                 "%d no. of stages found", len(stages)
             )
@@ -129,7 +130,7 @@ class CmdStageAdd(CmdBase):
         kwargs = vars(self.args)
         kwargs.update(
             {
-                "cmd": parse_cmd(kwargs.pop("cmd")),
+                "cmd": parse_cmd(kwargs.pop("command")),
                 "params": parse_params(self.args.params),
             }
         )
@@ -281,7 +282,7 @@ def _add_common_args(parser):
         ),
     )
     parser.add_argument(
-        "cmd",
+        "command",
         nargs=argparse.REMAINDER,
         help="Command to execute.",
         metavar="command",

dvc/commands/status.py

@@ -1,6 +1,6 @@
 import logging
 
-from dvc.command.data_sync import CmdDataBase
+from dvc.commands.data_sync import CmdDataBase
 from dvc.exceptions import DvcException
 from dvc.ui import ui
 from dvc.utils import format_link
@@ -39,7 +39,7 @@ class CmdDataStatus(CmdDataBase):
 
         for key, value in status.items():
             if isinstance(value, str):
-                ui.write("{}{}{}".format(ind, self._normalize(value), key))
+                ui.write(f"{ind}{self._normalize(value)}{key}")
             elif value:
                 ui.write(f"{ind}{key}:")
                 self._show(value, indent + 1)
@@ -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:
@@ -80,7 +78,7 @@ class CmdDataStatus(CmdDataBase):
                 return 0
 
             # additional hints for the user
-            if not self.repo.stages:
+            if not self.repo.index.stages:
                 ui.write(self.EMPTY_PROJECT_MSG)
             elif self.args.cloud or self.args.remote:
                 remote = self.args.remote or self.repo.config["core"].get(

dvc/commands/unprotect.py

@@ -1,8 +1,9 @@
 import argparse
 import logging
 
-from dvc.command import completion
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.commands import completion
 from dvc.exceptions import DvcException
 
 logger = logging.getLogger(__name__)

dvc/commands/update.py

@@ -1,8 +1,9 @@
 import argparse
 import logging
 
-from dvc.command import completion
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.commands import completion
 from dvc.exceptions import DvcException
 
 logger = logging.getLogger(__name__)

dvc/commands/version.py

@@ -1,7 +1,8 @@
 import argparse
 import logging
 
-from dvc.command.base import CmdBaseNoRepo, append_doc_link
+from dvc.cli.command import CmdBaseNoRepo
+from dvc.cli.utils import append_doc_link
 from dvc.ui import ui
 
 logger = logging.getLogger(__name__)
@@ -10,9 +11,12 @@ logger = logging.getLogger(__name__)
 class CmdVersion(CmdBaseNoRepo):
     def run(self):
         from dvc.info import get_dvc_info
+        from dvc.updater import notify_updates
 
         dvc_info = get_dvc_info()
         ui.write(dvc_info, force=True)
+
+        notify_updates()
         return 0
 
 

dvc/compare.py

@@ -11,7 +11,9 @@ from typing import (
     List,
     Mapping,
     MutableSequence,
+    Optional,
     Sequence,
+    Set,
     Tuple,
     Union,
     overload,
@@ -20,6 +22,7 @@ from typing import (
 from funcy import reraise
 
 if TYPE_CHECKING:
+    from dvc.types import StrPath
     from dvc.ui.table import CellT
 
 
@@ -36,11 +39,21 @@ class TabularData(MutableSequence[Sequence["CellT"]]):
         self._columns: Dict[str, Column] = {name: Column() for name in columns}
         self._keys: List[str] = list(columns)
         self._fill_value = fill_value
+        self._protected: Set[str] = set()
 
     @property
     def columns(self) -> List[Column]:
         return list(map(self.column, self.keys()))
 
+    def is_protected(self, col_name) -> bool:
+        return col_name in self._protected
+
+    def protect(self, *col_names: str):
+        self._protected.update(col_names)
+
+    def unprotect(self, *col_names: str):
+        self._protected = self._protected.difference(col_names)
+
     def column(self, name: str) -> Column:
         return self._columns[name]
 
@@ -115,6 +128,8 @@ class TabularData(MutableSequence[Sequence["CellT"]]):
             del col[item]
 
     def __len__(self) -> int:
+        if not self._columns:
+            return 0
         return len(self.columns[0])
 
     @property
@@ -122,9 +137,10 @@ class TabularData(MutableSequence[Sequence["CellT"]]):
         return len(self.columns), len(self)
 
     def drop(self, *col_names: str) -> None:
-        for col_name in col_names:
-            self._keys.remove(col_name)
-            self._columns.pop(col_name)
+        for col in col_names:
+            if not self.is_protected(col):
+                self._keys.remove(col)
+                self._columns.pop(col)
 
     def rename(self, from_col_name: str, to_col_name: str) -> None:
         self._columns[to_col_name] = self._columns.pop(from_col_name)
@@ -134,6 +150,10 @@ class TabularData(MutableSequence[Sequence["CellT"]]):
         self.drop(*(set(self._keys) - set(col_names)))
         self._keys = list(col_names)
 
+    def is_empty(self, col_name: str) -> bool:
+        col = self.column(col_name)
+        return not any(item != self._fill_value for item in col)
+
     def to_csv(self) -> str:
         import csv
         from io import StringIO
@@ -146,6 +166,20 @@ class TabularData(MutableSequence[Sequence["CellT"]]):
             writer.writerow(row)
         return buff.getvalue()
 
+    def to_parallel_coordinates(
+        self, output_path: "StrPath", color_by: str = None
+    ) -> str:
+        from dvc.render.html import write
+        from dvc.render.plotly import ParallelCoordinatesRenderer
+
+        index_path = write(
+            output_path,
+            renderers=[
+                ParallelCoordinatesRenderer(self, color_by, self._fill_value)
+            ],
+        )
+        return index_path.as_uri()
+
     def add_column(self, name: str) -> None:
         self._columns[name] = Column([self._fill_value] * len(self))
         self._keys.append(name)
@@ -164,7 +198,10 @@ 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
@@ -174,6 +211,109 @@ 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",
+        how="any",
+        subset: Optional[Iterable[str]] = None,
+    ):
+        if axis not in ["rows", "cols"]:
+            raise ValueError(
+                f"Invalid 'axis' value {axis}."
+                "Choose one of ['rows', 'cols']"
+            )
+        if how not in ["any", "all"]:
+            raise ValueError(
+                f"Invalid 'how' value {how}." "Choose one of ['any', 'all']"
+            )
+
+        match_line: Set = set()
+        match_any = True
+        if how == "all":
+            match_any = False
+
+        for n_row, row in enumerate(self):
+            for n_col, col in enumerate(row):
+                if subset and self.keys()[n_col] not in subset:
+                    continue
+                if (col == self._fill_value) is match_any:
+                    if axis == "rows":
+                        match_line.add(n_row)
+                        break
+                    else:
+                        match_line.add(self.keys()[n_col])
+
+        to_drop = match_line
+        if how == "all":
+            if axis == "rows":
+                to_drop = set(range(len(self)))
+            else:
+                to_drop = set(self.keys())
+            to_drop -= match_line
+
+        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 drop_duplicates(
+        self,
+        axis: str = "rows",
+        subset: Optional[Iterable[str]] = None,
+        ignore_empty: bool = True,
+    ):
+        if axis not in ["rows", "cols"]:
+            raise ValueError(
+                f"Invalid 'axis' value {axis}."
+                "Choose one of ['rows', 'cols']"
+            )
+
+        if axis == "cols":
+            cols_to_drop: List[str] = []
+            for n_col, col in enumerate(self.columns):
+                if subset and self.keys()[n_col] not in subset:
+                    continue
+                # Cast to str because Text is not hashable error
+                unique_vals = {str(x) for x in col}
+                if ignore_empty and self._fill_value in unique_vals:
+                    unique_vals -= {self._fill_value}
+                if len(unique_vals) == 1:
+                    cols_to_drop.append(self.keys()[n_col])
+            self.drop(*cols_to_drop)
+
+        elif axis == "rows":
+            unique_rows = []
+            rows_to_drop: List[int] = []
+            for n_row, row in enumerate(self):
+                if subset:
+                    row = [
+                        col
+                        for n_col, col in enumerate(row)
+                        if self.keys()[n_col] in subset
+                    ]
+
+                tuple_row = tuple(row)
+                if tuple_row in unique_rows:
+                    rows_to_drop.append(n_row)
+                else:
+                    unique_rows.append(tuple_row)
+
+            for name in self.keys():
+                self._columns[name] = Column(
+                    [
+                        x
+                        for n, x in enumerate(self._columns[name])
+                        if n not in rows_to_drop
+                    ]
+                )
+
 
 def _normalize_float(val: float, precision: int):
     return f"{val:.{precision}g}"
@@ -204,8 +344,12 @@ 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)
 
@@ -233,8 +377,8 @@ 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
 
@@ -249,6 +393,8 @@ 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,
@@ -259,6 +405,8 @@ 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)
 
@@ -277,8 +425,9 @@ def metrics_table(
     td = TabularData(["Revision", "Path"], fill_value="-")
 
     for branch, val in metrics.items():
-        for fname, metric in val.items():
+        for fname, metric in val.get("data", {}).items():
             row_data: Dict[str, str] = {"Revision": branch, "Path": fname}
+            metric = metric.get("data", {})
             flattened = (
                 flatten(format_dict(metric))
                 if isinstance(metric, dict)

dvc/config.py

@@ -8,7 +8,6 @@ from functools import partial
 from funcy import cached_property, compact, memoize, re_find
 
 from dvc.exceptions import DvcException, NotDvcRepoError
-from dvc.path_info import PathInfo
 
 logger = logging.getLogger(__name__)
 
@@ -20,7 +19,27 @@ class ConfigError(DvcException):
         super().__init__(f"config file error: {msg}")
 
 
-class NoRemoteError(ConfigError):
+class RemoteConfigError(ConfigError):
+    pass
+
+
+class NoRemoteError(RemoteConfigError):
+    pass
+
+
+class RemoteNotFoundError(RemoteConfigError):
+    pass
+
+
+class MachineConfigError(ConfigError):
+    pass
+
+
+class NoMachineError(MachineConfigError):
+    pass
+
+
+class MachineNotFoundError(MachineConfigError):
     pass
 
 
@@ -56,8 +75,10 @@ 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"
@@ -73,7 +94,7 @@ class Config(dict):
             try:
                 from dvc.repo import Repo
 
-                self.dvc_dir = os.path.join(Repo.find_dvc_dir())
+                self.dvc_dir = Repo.find_dvc_dir()
             except NotDvcRepoError:
                 self.dvc_dir = None
         else:
@@ -119,7 +140,7 @@ 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):
@@ -159,7 +180,7 @@ 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
@@ -171,7 +192,7 @@ 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
@@ -207,6 +228,7 @@ class Config(dict):
 
     @staticmethod
     def _to_relpath(conf_dir, path):
+        from dvc.fs.local import localfs
         from dvc.utils import relpath
 
         from .config_schema import RelPath
@@ -216,8 +238,9 @@ class Config(dict):
 
         if isinstance(path, RelPath) or not os.path.isabs(path):
             path = relpath(path, conf_dir)
+            return localfs.path.as_posix(path)
 
-        return PathInfo(path).as_posix()
+        return path
 
     @staticmethod
     def _save_paths(conf, filename):
@@ -243,6 +266,12 @@ class Config(dict):
                     "key_path": func,
                 }
             },
+            "machine": {
+                str: {
+                    "startup_script": func,
+                    "setup_script": func,
+                }
+            },
         }
         return Schema(dirs_schema, extra=ALLOW_EXTRA)(conf)
 
@@ -265,7 +294,7 @@ 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)
@@ -292,27 +321,29 @@ 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

@@ -79,13 +79,13 @@ REMOTE_COMMON = {
     "checksum_jobs": All(Coerce(int), Range(1)),
     "jobs": All(Coerce(int), Range(1)),
     Optional("no_traverse"): Bool,  # obsoleted
-    "verify": Bool,
 }
 LOCAL_COMMON = {
     "type": supported_cache_type,
     Optional("protected", default=False): Bool,  # obsoleted
     "shared": All(Lower, Choices("group")),
     Optional("slow_link_warning", default=True): Bool,
+    Optional("verify", default=False): Bool,
 }
 HTTP_COMMON = {
     "auth": All(Lower, Choices("basic", "digest", "custom")),
@@ -95,6 +95,7 @@ HTTP_COMMON = {
     "ask_password": Bool,
     "ssl_verify": Any(Bool, str),
     "method": str,
+    Optional("verify", default=False): Bool,
 }
 WEBDAV_COMMON = {
     "user": str,
@@ -105,6 +106,7 @@ WEBDAV_COMMON = {
     "key_path": str,
     "timeout": Coerce(int),
     "ssl_verify": Any(Bool, str),
+    Optional("verify", default=False): Bool,
 }
 
 SCHEMA = {
@@ -118,6 +120,7 @@ SCHEMA = {
         Optional("autostage", default=False): Bool,
         Optional("experiments"): Bool,  # obsoleted
         Optional("check_update", default=True): Bool,
+        "machine": Lower,
     },
     "cache": {
         "local": str,
@@ -154,11 +157,16 @@ SCHEMA = {
                     "grant_read_acp": str,
                     "grant_write_acp": str,
                     "grant_full_control": str,
+                    "cache_regions": bool,
+                    "read_timeout": Coerce(int),
+                    "connect_timeout": Coerce(int),
+                    Optional("verify", default=False): Bool,
                     **REMOTE_COMMON,
                 },
                 "gs": {
                     "projectname": str,
                     "credentialpath": str,
+                    Optional("verify", default=False): Bool,
                     **REMOTE_COMMON,
                 },
                 "ssh": {
@@ -171,14 +179,18 @@ SCHEMA = {
                     "timeout": Coerce(int),
                     "gss_auth": Bool,
                     "allow_agent": Bool,
+                    Optional("verify", default=False): Bool,
                     **REMOTE_COMMON,
                 },
                 "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,
+                    Optional("verify", default=False): Bool,
                     **REMOTE_COMMON,
                 },
                 "azure": {
@@ -189,13 +201,19 @@ SCHEMA = {
                     "tenant_id": str,
                     "client_id": str,
                     "client_secret": str,
-                    "allow_anonymous_login": bool,
+                    "allow_anonymous_login": Bool,
+                    "exclude_environment_credential": Bool,
+                    "exclude_visual_studio_code_credential": Bool,
+                    "exclude_shared_token_cache_credential": Bool,
+                    "exclude_managed_identity_credential": Bool,
+                    Optional("verify", default=False): Bool,
                     **REMOTE_COMMON,
                 },
                 "oss": {
                     "oss_key_id": str,
                     "oss_key_secret": str,
                     "oss_endpoint": str,
+                    Optional("verify", default=True): Bool,
                     **REMOTE_COMMON,
                 },
                 "gdrive": {
@@ -206,6 +224,7 @@ SCHEMA = {
                     "gdrive_service_account_user_email": str,
                     "gdrive_service_account_json_file_path": str,
                     Optional("gdrive_trash_only", default=False): Bool,
+                    Optional("verify", default=True): Bool,
                     **REMOTE_COMMON,
                 },
                 "http": {**HTTP_COMMON, **REMOTE_COMMON},
@@ -217,13 +236,47 @@ 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,
+            "spot": Bool,
+            "spot_price": Coerce(float),
+            "instance_hdd_size": Coerce(int),
+            "instance_type": Lower,
+            "instance_gpu": Lower,
+            "ssh_private": str,
+            "startup_script": str,
+            "setup_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,
+        Optional("auto_open", default=False): Bool,
+    },
+    "exp": {
+        "code": str,
+        "data": str,
+        "models": str,
+        "metrics": str,
+        "params": str,
+        "plots": str,
+        "live": str,
     },
-    "plots": {"html_template": str},
 }

dvc/daemon.py

@@ -12,49 +12,53 @@ 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):
-    from dvc.main import main
+    from dvc.cli import main
 
     # NOTE: using os._exit instead of sys.exit, because dvc built
     # 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/dagascii.py

@@ -163,7 +163,7 @@ class AsciiCanvas:
         self.point(x0 + width, y0 + height, "+")
 
 
-def _build_sugiyama_layout(vertexes, edges):
+def _build_sugiyama_layout(vertices, edges):
     #
     # Just a reminder about naming conventions:
     # +------------X
@@ -174,17 +174,17 @@ def _build_sugiyama_layout(vertexes, edges):
     # Y
     #
 
-    vertexes = {v: Vertex(f" {v} ") for v in vertexes}
+    vertices = {v: Vertex(f" {v} ") for v in vertices}
     # NOTE: reverting edges to correctly orientate the graph
-    edges = [Edge(vertexes[e], vertexes[s]) for s, e in edges]
-    vertexes = vertexes.values()
-    graph = Graph(vertexes, edges)
+    edges = [Edge(vertices[e], vertices[s]) for s, e in edges]
+    vertices = vertices.values()
+    graph = Graph(vertices, edges)
 
-    for vertex in vertexes:
+    for vertex in vertices:
         vertex.view = VertexViewer(vertex.data)
 
     # NOTE: determine min box length to create the best layout
-    minw = min(v.view.w for v in vertexes)
+    minw = min(v.view.w for v in vertices)
 
     for edge in edges:
         edge.view = EdgeViewer()
@@ -204,11 +204,11 @@ def _build_sugiyama_layout(vertexes, edges):
     return sug
 
 
-def draw(vertexes, edges):
+def draw(vertices, edges):
     """Build a DAG and draw it in ASCII.
 
     Args:
-        vertexes (list): list of graph vertexes.
+        vertices (list): list of graph vertices.
         edges (list): list of graph edges.
     """
     # pylint: disable=too-many-locals
@@ -217,7 +217,7 @@ def draw(vertexes, edges):
     Xs = []  # pylint: disable=invalid-name
     Ys = []  # pylint: disable=invalid-name
 
-    sug = _build_sugiyama_layout(vertexes, edges)
+    sug = _build_sugiyama_layout(vertices, edges)
 
     for vertex in sug.g.sV:
         # NOTE: moving boxes w/2 to the left

dvc/data/__init__.py

@@ -0,0 +1,33 @@
+import logging
+from typing import TYPE_CHECKING, Iterator, Union
+
+from .tree import Tree
+
+if TYPE_CHECKING:
+    from dvc.hash_info import HashInfo
+    from dvc.objects.db import ObjectDB
+    from dvc.objects.file import HashFile
+
+logger = logging.getLogger(__name__)
+
+
+def check(odb: "ObjectDB", obj: "HashFile", **kwargs):
+    if isinstance(obj, Tree):
+        for _, _, oid in obj:
+            odb.check(oid, **kwargs)
+
+    odb.check(obj.hash_info, **kwargs)
+
+
+def load(odb: "ObjectDB", hash_info: "HashInfo") -> "HashFile":
+    if hash_info.isdir:
+        return Tree.load(odb, hash_info)
+    return odb.get(hash_info)
+
+
+def iterobjs(
+    obj: Union["Tree", "HashFile"]
+) -> Iterator[Union["Tree", "HashFile"]]:
+    if isinstance(obj, Tree):
+        yield from (entry_obj for _, entry_obj in obj)
+    yield obj

dvc/data/checkout.py

@@ -0,0 +1,265 @@
+import logging
+from itertools import chain
+
+from dvc import prompt
+from dvc.exceptions import CacheLinkError, CheckoutError, ConfirmRemoveError
+from dvc.fs.utils import test_links, transfer
+from dvc.ignore import DvcIgnoreFilter
+from dvc.types import Optional
+
+from .diff import ROOT
+from .diff import diff as odiff
+from .slow_link_detection import slow_link_guard  # type: ignore[attr-defined]
+from .stage import stage
+
+logger = logging.getLogger(__name__)
+
+
+def _remove(fs_path, fs, in_cache, force=False):
+    if not fs.exists(fs_path):
+        return
+
+    if force:
+        fs.remove(fs_path)
+        return
+
+    if not in_cache:
+        msg = (
+            f"file/directory '{fs_path}' is going to be removed. "
+            "Are you sure you want to proceed?"
+        )
+
+        if not prompt.confirm(msg):
+            raise ConfirmRemoveError(fs_path)
+
+    fs.remove(fs_path)
+
+
+def _relink(link, cache, cache_info, fs, path, in_cache, force):
+    _remove(path, fs, in_cache, force=force)
+    link(cache, cache_info, fs, path)
+    # 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(
+    link,
+    fs_path,
+    fs,
+    change,
+    cache,
+    force,
+    progress_callback=None,
+    relink=False,
+    state=None,
+):
+    """The file is changed we need to checkout a new copy"""
+    modified = False
+
+    cache_fs_path = cache.hash_to_path(change.new.oid.value)
+    if change.old.oid:
+        if relink:
+            if fs.iscopy(fs_path) and cache.cache_types[0] == "copy":
+                cache.unprotect(fs_path)
+            else:
+                _relink(
+                    link,
+                    cache,
+                    cache_fs_path,
+                    fs,
+                    fs_path,
+                    change.old.in_cache,
+                    force=force,
+                )
+        else:
+            modified = True
+            _relink(
+                link,
+                cache,
+                cache_fs_path,
+                fs,
+                fs_path,
+                change.old.in_cache,
+                force=force,
+            )
+    else:
+        link(cache, cache_fs_path, fs, fs_path)
+        modified = True
+
+    if state:
+        state.save(fs_path, fs, change.new.oid)
+
+    if progress_callback:
+        progress_callback(fs_path)
+
+    return modified
+
+
+def _diff(
+    fs_path,
+    fs,
+    obj,
+    cache,
+    relink=False,
+    dvcignore: Optional[DvcIgnoreFilter] = None,
+):
+    old = None
+    try:
+        _, _, old = stage(
+            cache,
+            fs_path,
+            fs,
+            obj.hash_info.name if obj else cache.fs.PARAM_CHECKSUM,
+            dry_run=True,
+            dvcignore=dvcignore,
+        )
+    except FileNotFoundError:
+        pass
+
+    diff = odiff(old, obj, cache)
+
+    if relink:
+        diff.modified.extend(diff.unchanged)
+
+    return diff
+
+
+class Link:
+    def __init__(self, links):
+        self._links = links
+
+    @slow_link_guard
+    def __call__(self, cache, from_path, to_fs, to_path):
+        if to_fs.exists(to_path):
+            to_fs.remove(to_path)  # broken symlink
+
+        cache.makedirs(cache.fs.path.parent(to_path))
+        try:
+            transfer(cache.fs, from_path, to_fs, to_path, links=self._links)
+        except FileNotFoundError as exc:
+            raise CheckoutError([to_path]) from exc
+        except OSError as exc:
+            raise CacheLinkError([to_path]) from exc
+
+
+def _checkout(
+    diff,
+    fs_path,
+    fs,
+    cache,
+    force=False,
+    progress_callback=None,
+    relink=False,
+    state=None,
+):
+    if not diff:
+        return
+
+    links = test_links(cache.cache_types, cache.fs, cache.fs_path, fs, fs_path)
+    if not links:
+        raise CacheLinkError([fs_path])
+    link = Link(links)
+
+    for change in diff.deleted:
+        entry_path = (
+            fs.path.join(fs_path, *change.old.key)
+            if change.old.key != ROOT
+            else fs_path
+        )
+        _remove(entry_path, fs, change.old.in_cache, force=force)
+
+    failed = []
+    for change in chain(diff.added, diff.modified):
+        entry_path = (
+            fs.path.join(fs_path, *change.new.key)
+            if change.new.key != ROOT
+            else fs_path
+        )
+        if change.new.oid.isdir:
+            fs.makedirs(entry_path)
+            continue
+
+        try:
+            _checkout_file(
+                link,
+                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(
+    fs_path,
+    fs,
+    obj,
+    cache,
+    force=False,
+    progress_callback=None,
+    relink=False,
+    quiet=False,
+    dvcignore: Optional[DvcIgnoreFilter] = None,
+    state=None,
+):
+    # if scheme(fs_path) not in ["local", cache.fs.scheme]:
+    #    raise NotImplementedError
+
+    diff = _diff(
+        fs_path,
+        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.",
+                fs_path,
+            )
+        failed.append(fs_path)
+    elif not diff:
+        logger.trace("Data '%s' didn't change.", fs_path)  # type: ignore
+
+    try:
+        _checkout(
+            diff,
+            fs_path,
+            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(fs_path, fs)
+        if not failed:
+            state.save(fs_path, fs, obj.hash_info)
+
+    if failed or not diff:
+        if progress_callback and obj:
+            progress_callback(fs_path, len(obj))
+        if failed:
+            raise CheckoutError(failed)
+        return
+
+    return bool(diff) and not relink

dvc/data/db/__init__.py

@@ -1,22 +1,20 @@
+from typing import TYPE_CHECKING
+
 from dvc.scheme import Schemes
 
+if TYPE_CHECKING:
+    from .index import ObjectDBIndexBase
+
+
+def get_odb(fs, fs_path, **config):
+    from dvc.objects.db import ObjectDB
 
-def get_odb(fs, path_info, **config):
-    from .base import ObjectDB
-    from .gdrive import GDriveObjectDB
     from .local import LocalObjectDB
-    from .ssh import SSHObjectDB
 
     if fs.scheme == Schemes.LOCAL:
-        return LocalObjectDB(fs, path_info, **config)
+        return LocalObjectDB(fs, fs_path, **config)
 
-    if fs.scheme == Schemes.SSH:
-        return SSHObjectDB(fs, path_info, **config)
-
-    if fs.scheme == Schemes.GDRIVE:
-        return GDriveObjectDB(fs, path_info, **config)
-
-    return ObjectDB(fs, path_info, **config)
+    return ObjectDB(fs, fs_path, **config)
 
 
 def _get_odb(repo, settings):
@@ -25,8 +23,23 @@ def _get_odb(repo, settings):
     if not settings:
         return None
 
-    cls, config, path_info = get_cloud_fs(repo, **settings)
-    return get_odb(cls(**config), path_info, state=repo.state, **config)
+    cls, config, fs_path = get_cloud_fs(repo, **settings)
+    config["tmp_dir"] = repo.tmp_dir
+    return get_odb(cls(**config), fs_path, state=repo.state, **config)
+
+
+def get_index(odb) -> "ObjectDBIndexBase":
+    import hashlib
+
+    from .index import ObjectDBIndex, ObjectDBIndexNoop
+
+    cls = ObjectDBIndex if odb.tmp_dir else ObjectDBIndexNoop
+    return cls(
+        odb.tmp_dir,
+        hashlib.sha256(
+            odb.fs.unstrip_protocol(odb.fs_path).encode("utf-8")
+        ).hexdigest(),
+    )
 
 
 class ODBManager:

dvc/data/db/index.py

@@ -0,0 +1,149 @@
+import logging
+import os
+from abc import ABC, abstractmethod
+from typing import TYPE_CHECKING, Iterable, Set
+
+from dvc.objects.errors import ObjectDBError
+from dvc.utils.decorators import with_diskcache
+
+if TYPE_CHECKING:
+    from dvc.types import StrPath
+
+logger = logging.getLogger(__name__)
+
+
+class ObjectDBIndexBase(ABC):
+    @abstractmethod
+    def __init__(
+        self,
+        tmp_dir: "StrPath",
+        name: str,
+    ):
+        pass
+
+    @abstractmethod
+    def __iter__(self):
+        pass
+
+    @abstractmethod
+    def __contains__(self, hash_):
+        pass
+
+    def hashes(self):
+        return iter(self)
+
+    @abstractmethod
+    def dir_hashes(self):
+        pass
+
+    @abstractmethod
+    def clear(self):
+        pass
+
+    @abstractmethod
+    def update(self, dir_hashes: Iterable[str], file_hashes: Iterable[str]):
+        pass
+
+    @abstractmethod
+    def intersection(self, hashes: Set[str]):
+        pass
+
+
+class ObjectDBIndexNoop(ObjectDBIndexBase):
+    """No-op class for ODBs which are not indexed."""
+
+    def __init__(
+        self,
+        tmp_dir: "StrPath",
+        name: str,
+    ):  # pylint: disable=super-init-not-called
+        pass
+
+    def __iter__(self):
+        return iter([])
+
+    def __contains__(self, hash_):
+        return False
+
+    def dir_hashes(self):
+        return []
+
+    def clear(self):
+        pass
+
+    def update(self, dir_hashes: Iterable[str], file_hashes: Iterable[str]):
+        pass
+
+    def intersection(self, hashes: Set[str]):
+        return []
+
+
+class ObjectDBIndex(ObjectDBIndexBase):
+    """Class for indexing hashes in an ODB."""
+
+    INDEX_SUFFIX = ".idx"
+    INDEX_DIR = "index"
+
+    def __init__(
+        self,
+        tmp_dir: "StrPath",
+        name: str,
+    ):  # pylint: disable=super-init-not-called
+        from diskcache import Cache, Index
+
+        from dvc.fs.local import LocalFileSystem
+        from dvc.utils.fs import makedirs
+
+        self.index_dir = os.path.join(tmp_dir, self.INDEX_DIR, name)
+        makedirs(self.index_dir, exist_ok=True)
+        self.fs = LocalFileSystem()
+        self.index = Index.fromcache(
+            Cache(
+                self.index_dir,
+                disk_pickle_protocol=4,
+                eviction_policy="none",
+            )
+        )
+
+    @with_diskcache(name="index")
+    def __iter__(self):
+        return iter(self.index)
+
+    @with_diskcache(name="index")
+    def __contains__(self, hash_):
+        return hash_ in self.index
+
+    @with_diskcache(name="index")
+    def dir_hashes(self):
+        """Iterate over .dir hashes stored in the index."""
+        yield from (hash_ for hash_, is_dir in self.index.items() if is_dir)
+
+    @with_diskcache(name="index")
+    def clear(self):
+        """Clear this index (to force re-indexing later)."""
+        from diskcache import Timeout
+
+        try:
+            self.index.clear()
+        except Timeout as exc:
+            raise ObjectDBError("Failed to clear ODB index") from exc
+
+    @with_diskcache(name="index")
+    def update(self, dir_hashes: Iterable[str], file_hashes: Iterable[str]):
+        """Update this index, adding the specified hashes."""
+        from diskcache import Timeout
+
+        try:
+            with self.index.transact():
+                for hash_ in dir_hashes:
+                    self.index[hash_] = True
+            with self.index.transact():
+                for hash_ in file_hashes:
+                    self.index[hash_] = False
+        except Timeout as exc:
+            raise ObjectDBError("Failed to update ODB index") from exc
+
+    @with_diskcache(name="index")
+    def intersection(self, hashes: Set[str]):
+        """Iterate over values from `hashes` which exist in the index."""
+        yield from hashes.intersection(self.index.keys())

dvc/data/db/local.py

@@ -6,14 +6,12 @@ from funcy import cached_property
 from shortuuid import uuid
 
 from dvc.hash_info import HashInfo
+from dvc.objects.db import ObjectDB
 from dvc.objects.errors import ObjectFormatError
-from dvc.path_info import PathInfo
 from dvc.progress import Tqdm
 from dvc.utils import relpath
 from dvc.utils.fs import copyfile, remove, umask, walk_files
 
-from .base import ObjectDB
-
 logger = logging.getLogger(__name__)
 
 
@@ -22,9 +20,9 @@ class LocalObjectDB(ObjectDB):
     CACHE_MODE = 0o444
     UNPACKED_DIR_SUFFIX = ".unpacked"
 
-    def __init__(self, fs, path_info, **config):
-        super().__init__(fs, path_info, **config)
-        self.cache_dir = path_info.fspath
+    def __init__(self, fs, fs_path, **config):
+        super().__init__(fs, fs_path, **config)
+        self.cache_dir = fs_path
 
         shared = config.get("shared")
         if shared:
@@ -36,11 +34,11 @@ class LocalObjectDB(ObjectDB):
 
     @property
     def cache_dir(self):
-        return self.path_info.fspath if self.path_info else None
+        return self.fs_path if self.fs_path else None
 
     @cache_dir.setter
     def cache_dir(self, value):
-        self.path_info = PathInfo(value) if value else None
+        self.fs_path = value
 
     @cached_property
     def cache_path(self):
@@ -50,10 +48,10 @@ class LocalObjectDB(ObjectDB):
         super().move(from_info, to_info)
         os.chmod(to_info, self._file_mode)
 
-    def makedirs(self, path_info):
+    def makedirs(self, fs_path):
         from dvc.utils.fs import makedirs
 
-        makedirs(path_info, exist_ok=True, mode=self._dir_mode)
+        makedirs(fs_path, exist_ok=True, mode=self._dir_mode)
 
     def hash_to_path(self, hash_):
         # NOTE: `self.cache_path` is already normalized so we can simply use
@@ -81,26 +79,30 @@ class LocalObjectDB(ObjectDB):
         return ret
 
     def _list_paths(self, prefix=None, progress_callback=None):
-        assert self.path_info is not None
+        assert self.fs_path is not None
         if prefix:
-            path_info = self.path_info / prefix[:2]
-            if not self.fs.exists(path_info):
+            fs_path = self.fs.path.join(self.fs_path, prefix[:2])
+            if not self.fs.exists(fs_path):
                 return
         else:
-            path_info = self.path_info
+            fs_path = self.fs_path
+
         # NOTE: use utils.fs walk_files since fs.walk_files will not follow
         # symlinks
         if progress_callback:
-            for path in walk_files(path_info):
+            for path in walk_files(fs_path):
                 progress_callback()
                 yield path
         else:
-            yield from walk_files(path_info)
+            yield from walk_files(fs_path)
 
     def _remove_unpacked_dir(self, hash_):
-        info = self.hash_to_path_info(hash_)
-        path_info = info.with_name(info.name + self.UNPACKED_DIR_SUFFIX)
-        self.fs.remove(path_info)
+        hash_fs_path = self.hash_to_path(hash_)
+        fs_path = self.fs.path.with_name(
+            hash_fs_path,
+            self.fs.path.name(hash_fs_path) + self.UNPACKED_DIR_SUFFIX,
+        )
+        self.fs.remove(fs_path)
 
     def _unprotect_file(self, path):
         if self.fs.is_symlink(path) or self.fs.is_hardlink(path):
@@ -112,7 +114,7 @@ class LocalObjectDB(ObjectDB):
             # would get only the part of file. So, at first, the file should be
             # copied with the temporary name, and then original file should be
             # replaced by new.
-            copyfile(path, tmp, name="Unprotecting '{}'".format(relpath(path)))
+            copyfile(path, tmp, name=f"Unprotecting '{relpath(path)}'")
             remove(path)
             os.rename(tmp, path)
 
@@ -126,47 +128,47 @@ class LocalObjectDB(ObjectDB):
         os.chmod(path, self._file_mode)
 
     def _unprotect_dir(self, path):
-        for fname in self.fs.walk_files(path):
+        for fname in self.fs.find(path):
             self._unprotect_file(fname)
 
-    def unprotect(self, path_info):
-        if not os.path.exists(path_info):
+    def unprotect(self, fs_path):
+        if not os.path.exists(fs_path):
             from dvc.exceptions import DvcException
 
             raise DvcException(
-                f"can't unprotect non-existing data '{path_info}'"
+                f"can't unprotect non-existing data '{fs_path}'"
             )
 
-        if os.path.isdir(path_info):
-            self._unprotect_dir(path_info)
+        if os.path.isdir(fs_path):
+            self._unprotect_dir(fs_path)
         else:
-            self._unprotect_file(path_info)
+            self._unprotect_file(fs_path)
 
-    def protect(self, path_info):
+    def protect(self, fs_path):
         try:
-            os.chmod(path_info, self.CACHE_MODE)
+            os.chmod(fs_path, self.CACHE_MODE)
         except OSError:
             # NOTE: not being able to protect cache file is not fatal, it
             # might happen on funky filesystems (e.g. Samba, see #5255),
             # read-only filesystems or in a shared cache scenario.
-            logger.trace("failed to protect '%s'", path_info, exc_info=True)
+            logger.trace("failed to protect '%s'", fs_path, exc_info=True)
 
-    def is_protected(self, path_info):
+    def is_protected(self, fs_path):
         try:
-            mode = os.stat(path_info).st_mode
+            mode = os.stat(fs_path).st_mode
         except FileNotFoundError:
             return False
 
         return stat.S_IMODE(mode) == self.CACHE_MODE
 
-    def set_exec(self, path_info):
-        mode = os.stat(path_info).st_mode | stat.S_IEXEC
+    def set_exec(self, fs_path):
+        mode = os.stat(fs_path).st_mode | stat.S_IEXEC
         try:
-            os.chmod(path_info, mode)
+            os.chmod(fs_path, mode)
         except OSError:
             logger.trace(
                 "failed to chmod '%s' '%s'",
                 oct(mode),
-                path_info,
+                fs_path,
                 exc_info=True,
             )

dvc/data/db/reference.py

@@ -0,0 +1,88 @@
+import io
+import logging
+import os
+from typing import TYPE_CHECKING, Dict
+
+from dvc.objects.db import ObjectDB
+from dvc.objects.errors import ObjectFormatError
+from dvc.scheme import Schemes
+
+from ..reference import ReferenceHashFile
+
+if TYPE_CHECKING:
+    from dvc.fs.base import FileSystem
+    from dvc.hash_info import HashInfo
+    from dvc.types import AnyPath
+
+logger = logging.getLogger(__name__)
+
+
+class ReferenceObjectDB(ObjectDB):
+    """Reference ODB.
+
+    File objects are stored as ReferenceHashFiles which reference paths outside
+    of the staging ODB fs. Tree objects are stored natively.
+    """
+
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        self._fs_cache: Dict[tuple, "FileSystem"] = {}
+        self._obj_cache: Dict["HashInfo", "ReferenceHashFile"] = {}
+
+    def get(self, hash_info: "HashInfo"):
+        if hash_info.isdir:
+            return super().get(hash_info)
+        try:
+            return self._obj_cache[hash_info]
+        except KeyError:
+            pass
+        fs_path = self.hash_to_path(hash_info.value)
+        try:
+            with self.fs.open(fs_path, "rb") as fobj:
+                ref_file = ReferenceHashFile.from_bytes(
+                    fobj.read(), fs_cache=self._fs_cache
+                )
+        except OSError:
+            raise FileNotFoundError
+        try:
+            ref_file.check(self, check_hash=False)
+        except ObjectFormatError:
+            self.fs.remove(fs_path)
+            raise
+        self._obj_cache[hash_info] = ref_file
+        return ref_file
+
+    def _add_file(
+        self,
+        from_fs: "FileSystem",
+        from_info: "AnyPath",
+        to_info: "AnyPath",
+        hash_info: "HashInfo",
+        hardlink: bool = False,
+    ):
+        self.makedirs(self.fs.path.parent(to_info))
+        if hash_info.isdir:
+            return super()._add_file(
+                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(ref_fobj, to_info)
+        except OSError as exc:
+            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)
+            else:
+                raise
+        if from_fs.scheme != Schemes.LOCAL:
+            self._fs_cache[ReferenceHashFile.config_tuple(from_fs)] = from_fs

dvc/data/diff.py

@@ -0,0 +1,133 @@
+from dataclasses import dataclass, field
+from typing import TYPE_CHECKING, List, Optional, Tuple
+
+if TYPE_CHECKING:
+    from dvc.hash_info import HashInfo
+    from dvc.objects.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 dvc.objects.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/data/gc.py

@@ -0,0 +1,40 @@
+def gc(odb, used, jobs=None, cache_odb=None, shallow=True):
+    from dvc.data.tree import Tree
+    from dvc.objects.errors import ObjectDBPermissionError
+
+    if odb.read_only:
+        raise ObjectDBPermissionError("Cannot gc read-only ODB")
+    if not cache_odb:
+        cache_odb = odb
+    used_hashes = set()
+    for hash_info in used:
+        used_hashes.add(hash_info.value)
+        if hash_info.isdir and not shallow:
+            tree = Tree.load(cache_odb, hash_info)
+            used_hashes.update(
+                entry_obj.hash_info.value for _, entry_obj in tree
+            )
+
+    def _is_dir_hash(_hash):
+        from dvc.hash_info import HASH_DIR_SUFFIX
+
+        return _hash.endswith(HASH_DIR_SUFFIX)
+
+    removed = False
+    # hashes must be sorted to ensure we always remove .dir files first
+    for hash_ in sorted(
+        odb.all(jobs, odb.fs_path),
+        key=_is_dir_hash,
+        reverse=True,
+    ):
+        if hash_ in used_hashes:
+            continue
+        fs_path = odb.hash_to_path(hash_)
+        if _is_dir_hash(hash_):
+            # backward compatibility
+            # pylint: disable=protected-access
+            odb._remove_unpacked_dir(hash_)
+        odb.fs.remove(fs_path)
+        removed = True
+
+    return removed

dvc/data/meta.py

@@ -0,0 +1,47 @@
+from collections import OrderedDict
+from dataclasses import dataclass, field
+from typing import TYPE_CHECKING, Optional
+
+if TYPE_CHECKING:
+    from dvc.objects.db import ObjectDB
+    from dvc.objects.file import HashFile
+
+
+@dataclass
+class Meta:
+    PARAM_SIZE = "size"
+    PARAM_NFILES = "nfiles"
+    PARAM_ISEXEC = "isexec"
+
+    size: Optional[int] = field(default=None)
+    nfiles: Optional[int] = field(default=None)
+    isexec: Optional[bool] = field(default=False)
+
+    obj: Optional["HashFile"] = field(default=None)
+    odb: Optional["ObjectDB"] = field(default=None)
+    remote: Optional[str] = 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)
+        isexec = d.pop(cls.PARAM_ISEXEC, False)
+
+        return cls(size=size, nfiles=nfiles, isexec=isexec)
+
+    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
+
+        if self.isexec:
+            ret[self.PARAM_ISEXEC] = self.isexec
+
+        return ret

dvc/data/reference.py

@@ -0,0 +1,120 @@
+import errno
+import logging
+import os
+import pickle
+from typing import TYPE_CHECKING, Optional
+
+from dvc.objects.errors import ObjectFormatError
+from dvc.objects.file import HashFile
+
+if TYPE_CHECKING:
+    from dvc.fs.base import FileSystem
+    from dvc.hash_info import HashInfo
+    from dvc.objects.db import ObjectDB
+    from dvc.types import AnyPath
+
+logger = logging.getLogger(__name__)
+
+
+class ReferenceHashFile(HashFile):
+    PARAM_PATH = "path"
+    PARAM_HASH = "hash"
+    PARAM_CHECKSUM = "checksum"
+    PARAM_FS_CONFIG = "fs_config"
+
+    def __init__(
+        self,
+        fs_path: "AnyPath",
+        fs: "FileSystem",
+        hash_info: "HashInfo",
+        checksum: Optional[str] = None,
+        **kwargs,
+    ):
+        super().__init__(fs_path, fs, hash_info, **kwargs)
+        self.checksum = checksum or fs.checksum(fs_path)
+
+    def __str__(self):
+        return f"ref object {self.hash_info} -> {self.fs_path}"
+
+    def check(self, odb: "ObjectDB", check_hash: bool = True):
+        assert self.fs
+        if not self.fs.exists(self.fs_path):
+            raise FileNotFoundError(
+                errno.ENOENT, os.strerror(errno.ENOENT), self.fs_path
+            )
+        if self.checksum != self._get_checksum():
+            raise ObjectFormatError(f"{self} is changed")
+        if check_hash:
+            self._check_hash(odb)
+
+    def _get_checksum(self) -> str:
+        assert self.fs
+        return self.fs.checksum(self.fs_path)
+
+    def to_bytes(self):
+        from dvc.fs.repo import RepoFileSystem
+
+        # NOTE: dumping reference FS's this way is insecure, as the
+        # fully parsed remote FS config will include credentials
+        #
+        # ReferenceHashFiles should currently only be serialized in
+        # memory and not to disk
+        fs_path = self.fs_path
+        if isinstance(self.fs, RepoFileSystem):
+            fs_path = self.fs.path.relpath(fs_path, self.fs.root_dir)
+
+        dict_ = {
+            self.PARAM_PATH: fs_path,
+            self.PARAM_HASH: self.hash_info,
+            self.PARAM_CHECKSUM: self.checksum,
+            self.PARAM_FS_CONFIG: self.config_tuple(self.fs),
+        }
+        try:
+            return pickle.dumps(dict_)
+        except pickle.PickleError as exc:
+            raise ObjectFormatError(f"Could not pickle {self}") from exc
+
+    @classmethod
+    def from_bytes(cls, data: bytes, fs_cache: Optional[dict] = None):
+        from dvc.fs import get_fs_cls
+        from dvc.fs.repo import RepoFileSystem
+
+        try:
+            dict_ = pickle.loads(data)
+        except pickle.PickleError as exc:
+            raise ObjectFormatError("ReferenceHashFile is corrupted") from exc
+
+        try:
+            fs_path = dict_[cls.PARAM_PATH]
+            hash_info = dict_[cls.PARAM_HASH]
+        except KeyError as exc:
+            raise ObjectFormatError("ReferenceHashFile is corrupted") from exc
+
+        scheme, config_pairs = dict_.get(cls.PARAM_FS_CONFIG)
+        fs = fs_cache.get((scheme, config_pairs)) if fs_cache else None
+        if not fs:
+            config = dict(config_pairs)
+            if RepoFileSystem.PARAM_REPO_URL in config:
+                fs = RepoFileSystem(**config)
+                fs_path = fs.path.join(fs.root_dir, fs_path)
+            else:
+                fs_cls = get_fs_cls(config, scheme=scheme)
+                fs = fs_cls(**config)
+        return ReferenceHashFile(
+            fs_path,
+            fs,
+            hash_info,
+            checksum=dict_.get(cls.PARAM_CHECKSUM),
+        )
+
+    @staticmethod
+    def config_tuple(fs: "FileSystem"):
+        return (
+            fs.scheme,
+            tuple(
+                (key, value)
+                for key, value in sorted(
+                    fs.config.items(), key=lambda item: item[0]
+                )
+            ),
+        )

dvc/data/slow_link_detection.py

@@ -25,15 +25,15 @@ this.message = (
 
 def slow_link_guard(f):
     @wraps(f)
-    def wrapper(remote, *args, **kwargs):
+    def wrapper(self, odb, *args, **kwargs):
         if this.already_displayed:
-            return f(remote, *args, **kwargs)
+            return f(self, odb, *args, **kwargs)
 
-        if not remote.slow_link_warning or remote.cache_types:
-            return f(remote, *args, **kwargs)
+        if not odb.slow_link_warning or odb.cache_types:
+            return f(self, odb, *args, **kwargs)
 
         start = time.time()
-        result = f(remote, *args, **kwargs)
+        result = f(self, odb, *args, **kwargs)
         delta = time.time() - start
 
         if delta >= this.timeout_seconds:

dvc/data/stage.py

@@ -0,0 +1,348 @@
+import hashlib
+import logging
+import os
+from concurrent.futures import ThreadPoolExecutor
+from functools import partial
+from typing import TYPE_CHECKING, Dict, Optional, Tuple
+
+from dvc.exceptions import DvcIgnoreInCollectedDirError
+from dvc.hash_info import HashInfo
+from dvc.ignore import DvcIgnore
+from dvc.objects.file import HashFile
+from dvc.progress import Tqdm
+from dvc.utils import file_md5, is_exec
+
+from .db.reference import ReferenceObjectDB
+from .meta import Meta
+
+if TYPE_CHECKING:
+    from dvc.fs.base import FileSystem
+    from dvc.objects.db import ObjectDB
+    from dvc.types import AnyPath
+
+logger = logging.getLogger(__name__)
+
+
+_STAGING_MEMFS_PATH = "dvc-staging"
+
+
+def _upload_file(from_fs_path, fs, odb, upload_odb):
+    from dvc.utils import tmp_fname
+    from dvc.utils.stream import HashedStreamReader
+
+    fs_path = upload_odb.fs.path
+    tmp_info = fs_path.join(upload_odb.fs_path, tmp_fname())
+    with fs.open(from_fs_path, mode="rb", chunk_size=fs.CHUNK_SIZE) as stream:
+        stream = HashedStreamReader(stream)
+        size = fs.getsize(from_fs_path)
+        upload_odb.fs.upload(
+            stream, tmp_info, desc=fs_path.name(from_fs_path), total=size
+        )
+
+    odb.add(tmp_info, upload_odb.fs, stream.hash_info)
+    meta = Meta(size=size)
+    return from_fs_path, meta, odb.get(stream.hash_info)
+
+
+def _adapt_info(info, scheme):
+    if scheme == "s3" and "ETag" in info:
+        info["etag"] = info["ETag"].strip('"')
+    elif scheme == "gs" and "etag" in info:
+        import base64
+
+        info["etag"] = base64.b64decode(info["etag"]).hex()
+    elif scheme.startswith("http") and (
+        "ETag" in info or "Content-MD5" in info
+    ):
+        info["checksum"] = info.get("ETag") or info.get("Content-MD5")
+    return info
+
+
+def _get_file_hash(fs_path, fs, name):
+    info = _adapt_info(fs.info(fs_path), fs.scheme)
+
+    if name in info:
+        assert not info[name].endswith(".dir")
+        hash_value = info[name]
+    elif hasattr(fs, name):
+        func = getattr(fs, name)
+        hash_value = func(fs_path)
+    elif name == "md5":
+        hash_value = file_md5(fs_path, fs)
+    else:
+        raise NotImplementedError
+
+    meta = Meta(size=info["size"], isexec=is_exec(info.get("mode", 0)))
+    hash_info = HashInfo(name, hash_value)
+    return meta, hash_info
+
+
+def get_file_hash(fs_path, fs, name, state=None):
+    if state:
+        meta, hash_info = state.get(  # pylint: disable=assignment-from-none
+            fs_path, fs
+        )
+        if hash_info:
+            return meta, hash_info
+
+    meta, hash_info = _get_file_hash(fs_path, fs, name)
+
+    if state:
+        assert ".dir" not in hash_info.value
+        state.save(fs_path, fs, hash_info)
+
+    return meta, hash_info
+
+
+def _stage_file(fs_path, fs, name, odb=None, upload_odb=None, dry_run=False):
+    state = odb.state if odb else None
+    meta, hash_info = get_file_hash(fs_path, fs, name, state=state)
+    if upload_odb and not dry_run:
+        assert odb and name == "md5"
+        return _upload_file(fs_path, fs, odb, upload_odb)
+
+    if dry_run:
+        obj = HashFile(fs_path, fs, hash_info)
+    else:
+        odb.add(fs_path, fs, hash_info, hardlink=False)
+        obj = odb.get(hash_info)
+
+    return fs_path, meta, obj
+
+
+def _build_objects(
+    fs_path,
+    fs,
+    name,
+    dvcignore=None,
+    jobs=None,
+    no_progress_bar=False,
+    **kwargs,
+):
+    if dvcignore:
+        walk_iterator = dvcignore.find(fs, fs_path)
+    else:
+        walk_iterator = fs.find(fs_path)
+    with Tqdm(
+        unit="md5",
+        desc="Computing file/dir hashes (only done once)",
+        disable=no_progress_bar,
+    ) as pbar:
+        worker = pbar.wrap_fn(
+            partial(
+                _stage_file,
+                fs=fs,
+                name=name,
+                **kwargs,
+            )
+        )
+        with ThreadPoolExecutor(
+            max_workers=jobs if jobs is not None else fs.hash_jobs
+        ) as executor:
+            yield from executor.map(worker, walk_iterator)
+
+
+def _iter_objects(fs_path, fs, name, **kwargs):
+    yield from _build_objects(fs_path, fs, name, **kwargs)
+
+
+def _build_tree(fs_path, fs, name, **kwargs):
+    from .tree import Tree
+
+    tree_meta = Meta(size=0, nfiles=0)
+    tree = Tree(None, None, None)
+    for file_fs_path, meta, obj in _iter_objects(fs_path, fs, name, **kwargs):
+        if DvcIgnore.DVCIGNORE_FILE == fs.path.name(file_fs_path):
+            raise DvcIgnoreInCollectedDirError(fs.path.parent(file_fs_path))
+
+        # NOTE: this is lossy transformation:
+        #   "hey\there" -> "hey/there"
+        #   "hey/there" -> "hey/there"
+        # The latter is fine filename on Windows, which
+        # will transform to dir/file on back transform.
+        #
+        # Yes, this is a BUG, as long as we permit "/" in
+        # filenames on Windows and "\" on Unix
+
+        key = fs.path.relparts(file_fs_path, fs_path)
+        assert key
+        tree.add(key, meta, obj.hash_info)
+
+        tree_meta.size += meta.size or 0
+        tree_meta.nfiles += 1
+
+    return tree_meta, tree
+
+
+def _stage_tree(fs_path, fs, fs_info, name, odb=None, **kwargs):
+    from .tree import Tree
+
+    value = fs_info.get(name)
+    if odb and value:
+        hash_info = HashInfo(name, value)
+        try:
+            tree = Tree.load(odb, hash_info)
+            return Meta(nfiles=len(tree)), tree
+        except FileNotFoundError:
+            pass
+
+    meta, tree = _build_tree(fs_path, 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
+            fs_path, fs
+        )
+    tree.digest(hash_info=hash_info)
+    odb.add(tree.fs_path, 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
+    if odb.fs != tree.fs:
+        tree.fs.remove(tree.fs_path)
+    tree.fs = raw.fs
+    tree.fs_path = raw.fs_path
+    return meta, tree
+
+
+_url_cache: Dict[str, str] = {}
+
+
+def _make_staging_url(
+    fs: "FileSystem", odb: "ObjectDB", fs_path: Optional[str]
+):
+    from dvc.scheme import Schemes
+
+    url = f"{Schemes.MEMORY}://{_STAGING_MEMFS_PATH}"
+
+    if fs_path is not None:
+        if odb.fs.scheme == Schemes.LOCAL:
+            fs_path = os.path.abspath(fs_path)
+
+        if fs_path not in _url_cache:
+            _url_cache[fs_path] = hashlib.sha256(
+                fs_path.encode("utf-8")
+            ).hexdigest()
+
+        url = fs.path.join(url, _url_cache[fs_path])
+
+    return url
+
+
+def _get_staging(odb: "ObjectDB") -> "ObjectDB":
+    """Return an ODB that can be used for staging objects.
+
+    Staging will be a reference ODB stored in the the global memfs.
+    """
+
+    from dvc.fs.memory import MemoryFileSystem
+
+    fs = MemoryFileSystem()
+    fs_path = _make_staging_url(fs, odb, odb.fs_path)
+    state = odb.state
+    return ReferenceObjectDB(fs, fs_path, state=state)
+
+
+def _load_from_state(odb, staging, fs_path, fs, name):
+    from dvc.objects.errors import ObjectFormatError
+
+    from . import check, load
+    from .tree import Tree
+
+    state = odb.state
+    meta, hash_info = state.get(fs_path, fs)
+    if hash_info:
+        for odb_ in (odb, staging):
+            if odb_.exists(hash_info):
+                try:
+                    obj = load(odb_, hash_info)
+                    check(odb_, obj, check_hash=False)
+                    if isinstance(obj, Tree):
+                        meta.nfiles = len(obj)
+                    assert obj.hash_info.name == name
+                    return odb_, meta, obj
+                except (ObjectFormatError, FileNotFoundError):
+                    pass
+    raise FileNotFoundError
+
+
+def _stage_external_tree_info(odb, tree, name):
+    # NOTE: used only for external outputs. Initial reasoning was to be
+    # able to validate .dir files right in the workspace (e.g. check s3
+    # etag), but could be dropped for manual validation with regular md5,
+    # that would be universal for all clouds.
+    assert odb and name != "md5"
+
+    odb.add(tree.fs_path, tree.fs, tree.hash_info)
+    raw = odb.get(tree.hash_info)
+    _, hash_info = get_file_hash(raw.fs_path, raw.fs, name, state=odb.state)
+    tree.fs_path = raw.fs_path
+    tree.fs = raw.fs
+    tree.hash_info.name = hash_info.name
+    tree.hash_info.value = hash_info.value
+    if not tree.hash_info.value.endswith(".dir"):
+        tree.hash_info.value += ".dir"
+    return tree
+
+
+def stage(
+    odb: "ObjectDB",
+    fs_path: "AnyPath",
+    fs: "FileSystem",
+    name: str,
+    upload: bool = False,
+    dry_run: bool = False,
+    **kwargs,
+) -> 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
+    be completed by transferring the object from staging to the dest ODB.
+
+    If dry_run is True, object hashes will be computed and returned, but file
+    objects themselves will not be added to the staging ODB (i.e. the resulting
+    file objects cannot transferred from staging to another ODB).
+
+    If upload is True, files will be uploaded to a temporary path on the dest
+    ODB filesystem, and staged objects will reference the uploaded path rather
+    than the original source path.
+    """
+    assert fs_path
+    # assert scheme(fs_path) == fs.scheme
+
+    details = fs.info(fs_path)
+    staging = _get_staging(odb)
+    if odb:
+        try:
+            return _load_from_state(odb, staging, fs_path, fs, name)
+        except FileNotFoundError:
+            pass
+
+    if details["type"] == "directory":
+        meta, obj = _stage_tree(
+            fs_path,
+            fs,
+            details,
+            name,
+            odb=staging,
+            upload_odb=odb if upload else None,
+            dry_run=dry_run,
+            **kwargs,
+        )
+        logger.debug("staged tree '%s'", obj)
+        if name != "md5":
+            obj = _stage_external_tree_info(odb, obj, name)
+    else:
+        _, meta, obj = _stage_file(
+            fs_path,
+            fs,
+            name,
+            odb=staging,
+            upload_odb=odb if upload else None,
+            dry_run=dry_run,
+        )
+
+    if odb and odb.state and obj.hash_info:
+        odb.state.save(fs_path, fs, obj.hash_info)
+
+    return staging, meta, obj

dvc/data/status.py

@@ -0,0 +1,186 @@
+import logging
+from typing import TYPE_CHECKING, Dict, Iterable, NamedTuple, Optional, Set
+
+from dvc.hash_info import HashInfo
+from dvc.scheme import Schemes
+
+from .tree import Tree
+
+if TYPE_CHECKING:
+    from dvc.objects.db import ObjectDB
+    from dvc.objects.file import HashFile
+
+    from .db.index import ObjectDBIndexBase
+
+logger = logging.getLogger(__name__)
+
+
+class StatusResult(NamedTuple):
+    exists: Set["HashInfo"]
+    missing: Set["HashInfo"]
+
+
+class CompareStatusResult(NamedTuple):
+    ok: Set["HashInfo"]
+    missing: Set["HashInfo"]
+    new: Set["HashInfo"]
+    deleted: Set["HashInfo"]
+
+
+def _indexed_dir_hashes(odb, index, dir_objs, name, cache_odb):
+    # Validate our index by verifying all indexed .dir hashes
+    # still exist on the remote
+    dir_hashes = set(dir_objs.keys())
+    indexed_dirs = set(index.dir_hashes())
+    indexed_dir_exists = set()
+    if indexed_dirs:
+        indexed_dir_exists.update(odb.list_hashes_exists(indexed_dirs))
+        missing_dirs = indexed_dirs.difference(indexed_dir_exists)
+        if missing_dirs:
+            logger.debug(
+                "Remote cache missing indexed .dir hashes '%s', "
+                "clearing remote index",
+                ", ".join(missing_dirs),
+            )
+            index.clear()
+
+    # Check if non-indexed (new) dir hashes exist on remote
+    dir_exists = dir_hashes.intersection(indexed_dir_exists)
+    dir_exists.update(odb.list_hashes_exists(dir_hashes - dir_exists))
+
+    # If .dir hash exists in the ODB, assume directory contents
+    # also exists
+    for dir_hash in dir_exists:
+        tree = dir_objs.get(dir_hash)
+        if not tree:
+            try:
+                tree = Tree.load(cache_odb, HashInfo(name, dir_hash))
+            except FileNotFoundError:
+                continue
+        file_hashes = [oid.value for _, _, oid in tree]
+        if dir_hash not in index:
+            logger.debug(
+                "Indexing new .dir '%s' with '%s' nested files",
+                dir_hash,
+                len(file_hashes),
+            )
+            index.update([dir_hash], file_hashes)
+        yield from file_hashes
+        yield tree.hash_info.value
+
+
+def status(
+    odb: "ObjectDB",
+    obj_ids: Iterable["HashInfo"],
+    name: Optional[str] = None,
+    index: Optional["ObjectDBIndexBase"] = None,
+    cache_odb: Optional["ObjectDB"] = None,
+    shallow: bool = True,
+    **kwargs,
+) -> "StatusResult":
+    """Return status of whether or not the specified objects exist odb.
+
+    If cache_odb is set, trees will be loaded from cache_odb instead of odb
+    when needed.
+
+    Status is returned as a tuple of:
+        exists: objs that exist in odb
+        missing: objs that do not exist in ODB
+    """
+    logger.debug("Preparing to collect status from '%s'", odb.fs_path)
+    if not name:
+        name = odb.fs.PARAM_CHECKSUM
+
+    if cache_odb is None:
+        cache_odb = odb
+
+    hash_infos: Dict[str, "HashInfo"] = {}
+    dir_objs: Dict[str, Optional["HashFile"]] = {}
+    for hash_info in obj_ids:
+        assert hash_info.value
+        if hash_info.isdir:
+            if shallow:
+                tree = None
+            else:
+                tree = Tree.load(cache_odb, 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
+
+    if odb.fs.scheme == Schemes.MEMORY:
+        # assume memfs staged objects already exist
+        return StatusResult(set(hash_infos.values()), set())
+
+    hashes: Set[str] = set(hash_infos.keys())
+    exists: Set[str] = set()
+
+    logger.debug("Collecting status from '%s'", odb.fs_path)
+    if index and hashes:
+        if dir_objs:
+            exists = hashes.intersection(
+                _indexed_dir_hashes(odb, index, dir_objs, name, cache_odb)
+            )
+            hashes.difference_update(exists)
+        if hashes:
+            exists.update(index.intersection(hashes))
+            hashes.difference_update(exists)
+
+    if hashes:
+        exists.update(odb.hashes_exist(hashes, name=odb.fs_path, **kwargs))
+    return StatusResult(
+        {hash_infos[hash_] for hash_ in exists},
+        {hash_infos[hash_] for hash_ in (hashes - exists)},
+    )
+
+
+def compare_status(
+    src: "ObjectDB",
+    dest: "ObjectDB",
+    obj_ids: Iterable["HashInfo"],
+    log_missing: bool = True,
+    check_deleted: bool = True,
+    src_index: Optional["ObjectDBIndexBase"] = None,
+    dest_index: Optional["ObjectDBIndexBase"] = None,
+    **kwargs,
+) -> "CompareStatusResult":
+    """Compare status for the specified objects between two ODBs.
+
+    Status is returned as a tuple of:
+        ok: hashes that exist in both src and dest
+        missing: hashes that do not exist in neither src nor dest
+        new: hashes that only exist in src
+        deleted: hashes that only exist in dest
+    """
+    if "cache_odb" not in kwargs:
+        kwargs["cache_odb"] = src
+    dest_exists, dest_missing = status(
+        dest, obj_ids, index=dest_index, **kwargs
+    )
+    # for transfer operations we can skip src status check when all objects
+    # already exist in dest
+    if dest_missing or check_deleted:
+        src_exists, src_missing = status(
+            src, obj_ids, index=src_index, **kwargs
+        )
+    else:
+        src_exists = dest_exists
+        src_missing = set()
+    result = CompareStatusResult(
+        src_exists & dest_exists,
+        src_missing & dest_missing,
+        src_exists - dest_exists,
+        dest_exists - src_exists,
+    )
+    if log_missing and result.missing:
+        missing_desc = "\n".join(
+            f"name: {hash_info.obj_name}, {hash_info}"
+            for hash_info in result.missing
+        )
+        logger.warning(
+            "Some of the cache files do not exist neither locally "
+            f"nor on remote. Missing cache files:\n{missing_desc}"
+        )
+    return result

dvc/data/transfer.py

@@ -0,0 +1,177 @@
+import errno
+import logging
+from functools import partial, wraps
+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 dvc.objects.db import ObjectDB
+
+    from .db.index import ObjectDBIndexBase
+    from .tree import Tree
+
+logger = logging.getLogger(__name__)
+
+
+def _log_exceptions(func):
+    @wraps(func)
+    def wrapper(fs_path, *args, **kwargs):
+        try:
+            func(fs_path, *args, **kwargs)
+            return 0
+        except Exception as exc:  # pylint: disable=broad-except
+            # NOTE: this means we ran out of file descriptors and there is no
+            # reason to try to proceed, as we will hit this error anyways.
+            # pylint: disable=no-member
+            if isinstance(exc, OSError) and exc.errno == errno.EMFILE:
+                raise
+
+            logger.exception("failed to transfer '%s'", fs_path)
+            return 1
+
+    return wrapper
+
+
+def find_tree_by_obj_id(
+    odbs: Iterable[Optional["ObjectDB"]], obj_id: "HashInfo"
+) -> Optional["Tree"]:
+    from dvc.objects.errors import ObjectFormatError
+
+    from .tree import Tree
+
+    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",
+    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: Any,
+):
+    from dvc.exceptions import FileTransferError
+
+    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:
+        dir_obj = find_tree_by_obj_id([cache_odb, src], dir_hash)
+        assert dir_obj
+
+        entry_ids = {oid for _, _, oid in dir_obj}
+        bound_file_ids = all_file_ids & entry_ids
+        all_file_ids -= entry_ids
+
+        dir_fails = sum(processor(bound_file_ids))
+        if dir_fails:
+            logger.debug(
+                "failed to upload full contents of '%s', "
+                "aborting .dir file upload",
+                dir_obj.name,
+            )
+            logger.error(
+                "failed to upload '%s' to '%s'",
+                src.get(dir_obj.hash_info).fs_path,
+                dest.get(dir_obj.hash_info).fs_path,
+            )
+            total_fails += dir_fails + 1
+        elif entry_ids.intersection(missing_ids):
+            # if for some reason a file contained in this dir is
+            # missing both locally and in the remote, we want to
+            # push whatever file content we have, but should not
+            # push .dir file
+            logger.debug(
+                "directory '%s' contains missing files,"
+                "skipping .dir file upload",
+                dir_obj.name,
+            )
+        else:
+            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 += sum(processor(all_file_ids))
+    if total_fails:
+        if src_index:
+            src_index.clear()
+        raise FileTransferError(total_fails)
+
+    # index successfully pushed dirs
+    if dest_index:
+        for dir_obj in succeeded_dir_objs:
+            file_hashes = {oid.value for _, _, oid in dir_obj}
+            logger.debug(
+                "Indexing pushed dir '%s' with '%s' nested files",
+                dir_obj.hash_info,
+                len(file_hashes),
+            )
+            assert dir_obj.hash_info and dir_obj.hash_info.value
+            dest_index.update([dir_obj.hash_info.value], file_hashes)
+
+
+def transfer(
+    src: "ObjectDB",
+    dest: "ObjectDB",
+    obj_ids: Iterable["HashInfo"],
+    jobs: Optional[int] = None,
+    verify: bool = False,
+    hardlink: bool = False,
+    **kwargs,
+) -> int:
+    """Transfer (copy) the specified objects from one ODB to another.
+
+    Returns the number of successfully transferred objects
+    """
+    from .status import compare_status
+
+    logger.debug(
+        "Preparing to transfer data from '%s' to '%s'",
+        src.fs_path,
+        dest.fs_path,
+    )
+    if src == dest:
+        return 0
+
+    status = compare_status(src, dest, obj_ids, check_deleted=False, **kwargs)
+    if not status.new:
+        return 0
+
+    def func(hash_info: "HashInfo") -> None:
+        obj = src.get(hash_info)
+        return dest.add(
+            obj.fs_path,
+            obj.fs,
+            obj.hash_info,
+            verify=verify,
+            hardlink=hardlink,
+        )
+
+    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/data/tree.py

@@ -1,65 +1,105 @@
 import json
 import logging
 import posixpath
-from typing import TYPE_CHECKING, Optional, Tuple
+from typing import TYPE_CHECKING, Dict, Iterable, Optional, Tuple
 
 from funcy import cached_property
 
-from .errors import ObjectFormatError
-from .file import HashFile
+from dvc.objects.errors import ObjectFormatError
+from dvc.objects.file import HashFile
+
 from .stage import get_file_hash
 
 if TYPE_CHECKING:
-    from .db.base import ObjectDB
+    from dvc.hash_info import HashInfo
+    from dvc.objects.db import ObjectDB
+
+    from .meta import Meta
 
 logger = logging.getLogger(__name__)
 
 
+def _try_load(
+    odbs: Iterable["ObjectDB"],
+    hash_info: "HashInfo",
+) -> Optional["HashFile"]:
+    for odb in odbs:
+        if not odb:
+            continue
+
+        try:
+            return Tree.load(odb, hash_info)
+        except (FileNotFoundError, ObjectFormatError):
+            pass
+
+    return None
+
+
 class Tree(HashFile):
     PARAM_RELPATH = "relpath"
 
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
-        self._dict = {}
+        self._dict: Dict[Tuple[str], Tuple["Meta", "HashInfo"]] = {}
 
     @cached_property
-    def trie(self):
+    def _trie(self):
         from pygtrie import Trie
 
         return Trie(self._dict)
 
-    @property
-    def size(self):
-        try:
-            return sum(obj.size for _, obj in self)
-        except TypeError:
-            return None
-
-    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):
+    def digest(self, hash_info: Optional["HashInfo"] = None):
         from dvc.fs.memory import MemoryFileSystem
-        from dvc.path_info import PathInfo
         from dvc.utils import tmp_fname
 
         memfs = MemoryFileSystem()
-        path_info = PathInfo(tmp_fname(""))
-        with memfs.open(path_info, "wb") as fobj:
+        fs_path = "memory://{}".format(tmp_fname(""))
+        with memfs.open(fs_path, "wb") as fobj:
             fobj.write(self.as_bytes())
         self.fs = memfs
-        self.path_info = path_info
-        self.hash_info = get_file_hash(path_info, memfs, "md5")
-        self.hash_info.value += ".dir"
-        self.hash_info.size = self.size
-        self.hash_info.nfiles = len(self)
+        self.fs_path = fs_path
+        if hash_info:
+            self.hash_info = hash_info
+        else:
+            _, self.hash_info = get_file_hash(fs_path, memfs, "md5")
+            assert self.hash_info.value
+            self.hash_info.value += ".dir"
+
+    def _load(self, key, meta, hash_info):
+        if hash_info and hash_info.isdir and not meta.obj:
+            meta.obj = _try_load([meta.odb, meta.remote], hash_info)
+            if meta.obj:
+                for ikey, value in meta.obj.iteritems():
+                    self._trie[key + ikey] = value
+                    self._dict[key + ikey] = value
+
+    def iteritems(self, prefix=None):
+        kwargs = {}
+        if prefix:
+            kwargs = {"prefix": prefix}
+            item = self._trie.longest_prefix(prefix)
+            if item:
+                key, (meta, hash_info) = item
+                self._load(key, meta, hash_info)
+
+        for key, (meta, hash_info) in self._trie.iteritems(**kwargs):
+            self._load(key, meta, hash_info)
+            yield key, (meta, hash_info)
+
+    def shortest_prefix(self, *args, **kwargs):
+        return self._trie.shortest_prefix(*args, **kwargs)
 
     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 +113,10 @@ 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),
         )
@@ -94,17 +134,15 @@ 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
     def load(cls, odb, hash_info):
-
         obj = odb.get(hash_info)
 
         try:
-            with obj.fs.open(obj.path_info, "r") as fobj:
+            with obj.fs.open(obj.fs_path, "r") as fobj:
                 raw = json.load(fobj)
         except ValueError as exc:
             raise ObjectFormatError(f"{obj} is corrupted") from exc
@@ -112,60 +150,58 @@ class Tree(HashFile):
         if not isinstance(raw, list):
             logger.error(
                 "dir cache file format error '%s' [skipping the file]",
-                obj.path_info,
+                obj.fs_path,
             )
             raise ObjectFormatError(f"{obj} is corrupted")
 
         tree = cls.from_list(raw)
-        tree.path_info = obj.path_info
+        tree.fs_path = obj.fs_path
         tree.fs = obj.fs
-        for _, entry_obj in tree:
-            entry_obj.fs = obj.fs
         tree.hash_info = hash_info
 
         return tree
 
-    def filter(
-        self, odb: "ObjectDB", prefix: Tuple[str], copy: bool = False
-    ) -> Optional[HashFile]:
-        """Return filter object(s) for this tree.
+    def filter(self, prefix: Tuple[str]) -> Optional["Tree"]:
+        """Return a filtered copy of this tree that only contains entries
+        inside prefix.
 
-        If copy is True, returned object will be a Tree containing
-        filtered entries, but with hash_info copied from the original tree.
+        The returned tree will contain the original tree's hash_info and
+        fs_path.
 
-        If copy is False, returned object will be a raw HashFile or Tree with
-        newly computed hash_info for the filtered object.
+        Returns an empty tree if no object exists at the specified prefix.
         """
-        obj = self._dict.get(prefix)
-        if obj:
-            if copy:
-                tree = Tree(self.path_info, self.fs, self.hash_info)
-                tree.add(prefix, obj)
-                return tree
-            return obj
-
-        if copy:
-            tree = Tree(self.path_info, self.fs, self.hash_info)
-            depth = 0
-        else:
-            tree = Tree(None, None, None)
-            depth = len(prefix)
+        tree = Tree(self.fs_path, self.fs, self.hash_info)
         try:
-            for key, obj in self.trie.items(prefix):
-                tree.add(key[depth:], obj)
+            for key, (meta, oid) in self._trie.items(prefix):
+                tree.add(key, meta, oid)
+        except KeyError:
+            pass
+        return tree
+
+    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.
+        """
+        _, 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, (meta, entry_oid) in self._trie.items(prefix):
+                tree.add(key[depth:], meta, entry_oid)
         except KeyError:
             return None
-        if not copy:
-            tree.digest()
-            odb.add(tree.path_info, tree.fs, tree.hash_info)
+        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(oid.value)) for _, _, oid in tree
         )
     except FileNotFoundError:
         return None
@@ -226,11 +262,8 @@ 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/data_cloud.py

@@ -3,9 +3,11 @@
 import logging
 from typing import TYPE_CHECKING, Iterable, Optional
 
+from dvc.data.db import get_index
+
 if TYPE_CHECKING:
-    from dvc.objects.file import HashFile
-    from dvc.remote.base import Remote
+    from dvc.hash_info import HashInfo
+    from dvc.objects.db import ObjectDB
 
 logger = logging.getLogger(__name__)
 
@@ -24,18 +26,18 @@ class DataCloud:
     def __init__(self, repo):
         self.repo = repo
 
-    def get_remote(
+    def get_remote_odb(
         self,
         name: Optional[str] = None,
         command: str = "<command>",
-    ) -> "Remote":
+    ) -> "ObjectDB":
         from dvc.config import NoRemoteError
 
         if not name:
             name = self.repo.config["core"].get("remote")
 
         if name:
-            return self._init_remote(name)
+            return self._init_odb(name)
 
         if bool(self.repo.config["remote"]):
             error_msg = (
@@ -52,69 +54,79 @@ class DataCloud:
 
         raise NoRemoteError(error_msg)
 
-    def _init_remote(self, name):
-        from dvc.remote import get_remote
+    def _init_odb(self, name):
+        from dvc.data.db import get_odb
+        from dvc.fs import get_cloud_fs
 
-        return get_remote(self.repo, name=name)
+        cls, config, fs_path = get_cloud_fs(self.repo, name=name)
+        config["tmp_dir"] = self.repo.index_db_dir
+        return get_odb(cls(**config), fs_path, **config)
 
     def push(
         self,
-        objs: Iterable["HashFile"],
+        objs: Iterable["HashInfo"],
         jobs: Optional[int] = None,
         remote: Optional[str] = None,
-        show_checksums: bool = False,
+        odb: Optional["ObjectDB"] = None,
     ):
         """Push data items in a cloud-agnostic way.
 
         Args:
             objs: objects to push to the cloud.
             jobs: number of jobs that can be running simultaneously.
-            remote: optional remote to push to.
+            remote: optional name of remote to push to.
                 By default remote from core.remote config option is used.
-            show_checksums: show checksums instead of file names in
-                information messages.
+            odb: optional ODB to push to. Overrides remote.
         """
-        remote_obj = self.get_remote(remote, "push")
+        from dvc.data.transfer import transfer
 
-        return remote_obj.push(
+        if not odb:
+            odb = self.get_remote_odb(remote, "push")
+        return transfer(
             self.repo.odb.local,
+            odb,
             objs,
             jobs=jobs,
-            show_checksums=show_checksums,
+            dest_index=get_index(odb),
+            cache_odb=self.repo.odb.local,
         )
 
     def pull(
         self,
-        objs: Iterable["HashFile"],
+        objs: Iterable["HashInfo"],
         jobs: Optional[int] = None,
         remote: Optional[str] = None,
-        show_checksums: bool = False,
+        odb: Optional["ObjectDB"] = None,
     ):
         """Pull data items in a cloud-agnostic way.
 
         Args:
             objs: objects to pull from the cloud.
             jobs: number of jobs that can be running simultaneously.
-            remote: optional remote to pull from.
+            remote: optional name of remote to pull from.
                 By default remote from core.remote config option is used.
-            show_checksums: show checksums instead of file names in
-                information messages.
+            odb: optional ODB to pull from. Overrides remote.
         """
-        remote_obj = self.get_remote(remote, "pull")
+        from dvc.data.transfer import transfer
 
-        return remote_obj.pull(
+        if not odb:
+            odb = self.get_remote_odb(remote, "pull")
+        return transfer(
+            odb,
             self.repo.odb.local,
             objs,
             jobs=jobs,
-            show_checksums=show_checksums,
+            src_index=get_index(odb),
+            cache_odb=self.repo.odb.local,
+            verify=odb.verify,
         )
 
     def status(
         self,
-        objs: Iterable["HashFile"],
+        objs: Iterable["HashInfo"],
         jobs: Optional[int] = None,
         remote: Optional[str] = None,
-        show_checksums: bool = False,
+        odb: Optional["ObjectDB"] = None,
         log_missing: bool = True,
     ):
         """Check status of data items in a cloud-agnostic way.
@@ -125,20 +137,25 @@ class DataCloud:
             remote: optional remote to compare
                 cache to. By default remote from core.remote config option
                 is used.
-            show_checksums: show checksums instead of file names in
-                information messages.
+            odb: optional ODB to check status from. Overrides remote.
             log_missing: log warning messages if file doesn't exist
                 neither in cache, neither in cloud.
         """
-        remote_obj = self.get_remote(remote, "status")
-        return remote_obj.status(
+        from dvc.data.status import compare_status
+
+        if not odb:
+            odb = self.get_remote_odb(remote, "status")
+        return compare_status(
             self.repo.odb.local,
+            odb,
             objs,
             jobs=jobs,
-            show_checksums=show_checksums,
             log_missing=log_missing,
+            dest_index=get_index(odb),
+            cache_odb=self.repo.odb.local,
         )
 
     def get_url_for(self, remote, checksum):
-        remote_obj = self.get_remote(remote)
-        return str(remote_obj.odb.hash_to_path_info(checksum))
+        odb = self.get_remote_odb(remote)
+        path = odb.hash_to_path(checksum)
+        return odb.fs.unstrip_protocol(path)

dvc/dependency/param.py

@@ -1,3 +1,5 @@
+import errno
+import logging
 import os
 from collections import defaultdict
 
@@ -10,11 +12,21 @@ from dvc.utils.serialize import LOADERS, ParseError
 
 from .base import Dependency
 
+logger = logging.getLogger(__name__)
+
 
 class MissingParamsError(DvcException):
     pass
 
 
+class MissingParamsFile(DvcException):
+    pass
+
+
+class ParamsIsADirectoryError(DvcException):
+    pass
+
+
 class BadParamFileError(DvcException):
     pass
 
@@ -24,9 +36,9 @@ class ParamsDependency(Dependency):
     PARAM_SCHEMA = {PARAM_PARAMS: Any(dict, list, None)}
     DEFAULT_PARAMS_FILE = "params.yaml"
 
-    def __init__(self, stage, path, params):
+    def __init__(self, stage, path, params=None, repo=None):
         info = {}
-        self.params = []
+        self.params = params or []
         if params:
             if isinstance(params, list):
                 self.params = params
@@ -40,6 +52,7 @@ class ParamsDependency(Dependency):
             path
             or os.path.join(stage.repo.root_dir, self.DEFAULT_PARAMS_FILE),
             info=info,
+            repo=repo,
         )
 
     def dumpd(self):
@@ -85,20 +98,45 @@ class ParamsDependency(Dependency):
     def status(self):
         return self.workspace_status()
 
-    def _read(self):
+    def validate_filepath(self):
         if not self.exists:
-            return {}
+            raise FileNotFoundError(
+                errno.ENOENT, os.strerror(errno.ENOENT), str(self)
+            )
+        if self.isdir():
+            raise IsADirectoryError(
+                errno.EISDIR, os.strerror(errno.EISDIR), str(self)
+            )
+
+    def read_file(self):
+        _, ext = os.path.splitext(self.fs_path)
+        loader = LOADERS[ext]
 
-        suffix = self.path_info.suffix.lower()
-        loader = LOADERS[suffix]
         try:
-            return loader(self.path_info, fs=self.repo.fs)
+            self.validate_filepath()
+        except FileNotFoundError as exc:
+            raise MissingParamsFile(
+                f"Parameters file '{self}' does not exist"
+            ) from exc
+        except IsADirectoryError as exc:
+            raise ParamsIsADirectoryError(
+                f"'{self}' is a directory, expected a parameters file"
+            ) from exc
+
+        try:
+            return loader(self.fs_path, fs=self.repo.fs)
         except ParseError as exc:
             raise BadParamFileError(
                 f"Unable to read parameters from '{self}'"
             ) from exc
 
-    def read_params_d(self):
+    def _read(self):
+        try:
+            return self.read_file()
+        except MissingParamsFile:
+            return {}
+
+    def read_params_d(self, **kwargs):
         config = self._read()
 
         ret = {}
@@ -133,3 +171,20 @@ 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

@@ -1,15 +1,15 @@
 import os
 from collections import defaultdict
-from typing import TYPE_CHECKING, Dict, Optional, Set
+from copy import copy
+from typing import TYPE_CHECKING, Dict, Optional, Set, Tuple
 
 from voluptuous import Required
 
-from dvc.path_info import PathInfo
-
 from .base import Dependency
 
 if TYPE_CHECKING:
-    from dvc.objects.db.base import ObjectDB
+    from dvc.hash_info import HashInfo
+    from dvc.objects.db import ObjectDB
     from dvc.objects.file import HashFile
 
 
@@ -32,7 +32,7 @@ class RepoDependency(Dependency):
         self._staged_objs: Dict[str, "HashFile"] = {}
         super().__init__(stage, *args, **kwargs)
 
-    def _parse_path(self, fs, path_info):
+    def _parse_path(self, fs, fs_path):
         return None
 
     @property
@@ -40,7 +40,7 @@ class RepoDependency(Dependency):
         return False
 
     def __str__(self):
-        return "{} ({})".format(self.def_path, self.def_repo[self.PARAM_URL])
+        return f"{self.def_path} ({self.def_repo[self.PARAM_URL]})"
 
     def workspace_status(self):
         current = self.get_obj(locked=True).hash_info
@@ -61,19 +61,14 @@ 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 import save
-        from dvc.objects.db.git import GitObjectDB
-        from dvc.repo.fetch import fetch_from_odb
+        from dvc.data.checkout import checkout
 
         for odb, objs in self.get_used_objs().items():
-            if not isinstance(odb, GitObjectDB):
-                fetch_from_odb(self.repo, odb, objs, jobs=jobs)
+            self.repo.cloud.pull(objs, jobs=jobs, odb=odb)
 
         obj = self.get_obj()
-        save(self.repo.odb.local, obj, jobs=jobs)
         checkout(
-            to.path_info,
+            to.fs_path,
             to.fs,
             obj,
             self.repo.odb.local,
@@ -95,81 +90,109 @@ class RepoDependency(Dependency):
 
     def get_used_objs(
         self, **kwargs
-    ) -> Dict[Optional["ObjectDB"], Set["HashFile"]]:
+    ) -> Dict[Optional["ObjectDB"], Set["HashInfo"]]:
+        used, _ = self._get_used_and_obj(**kwargs)
+        return used
+
+    def _get_used_and_obj(
+        self, obj_only=False, **kwargs
+    ) -> Tuple[Dict[Optional["ObjectDB"], Set["HashInfo"]], "HashFile"]:
         from dvc.config import NoRemoteError
-        from dvc.exceptions import NoOutputOrStageError
-        from dvc.objects.db.git import GitObjectDB
-        from dvc.objects.stage import stage
+        from dvc.data.stage import stage
+        from dvc.data.tree import Tree
+        from dvc.exceptions import NoOutputOrStageError, PathMissingError
 
         local_odb = self.repo.odb.local
         locked = kwargs.pop("locked", True)
         with self._make_repo(
             locked=locked, cache_dir=local_odb.cache_dir
         ) as repo:
-            used_objs = defaultdict(set)
+            used_obj_ids = defaultdict(set)
             rev = repo.get_rev()
             if locked and self.def_repo.get(self.PARAM_REV_LOCK) is None:
                 self.def_repo[self.PARAM_REV_LOCK] = rev
 
-            path_info = PathInfo(repo.root_dir) / str(self.def_path)
+            path = os.path.abspath(os.path.join(repo.root_dir, self.def_path))
+            if not obj_only:
+                try:
+                    for odb, obj_ids in repo.used_objs(
+                        [path],
+                        force=True,
+                        jobs=kwargs.get("jobs"),
+                        recursive=True,
+                    ).items():
+                        if odb is None:
+                            odb = repo.cloud.get_remote_odb()
+                            odb.read_only = True
+                        self._check_circular_import(odb, obj_ids)
+                        used_obj_ids[odb].update(obj_ids)
+                except (NoRemoteError, NoOutputOrStageError):
+                    pass
+
             try:
-                for odb, objs in repo.used_objs(
-                    [os.fspath(path_info)],
-                    force=True,
-                    jobs=kwargs.get("jobs"),
-                    recursive=True,
-                ).items():
-                    if odb is None:
-                        odb = repo.cloud.get_remote().odb
-                    self._check_circular_import(odb)
-                    used_objs[odb].update(objs)
-            except (NoRemoteError, NoOutputOrStageError):
-                pass
+                staging, _, staged_obj = stage(
+                    local_odb,
+                    path,
+                    repo.repo_fs,
+                    local_odb.fs.PARAM_CHECKSUM,
+                )
+            except FileNotFoundError as exc:
+                raise PathMissingError(
+                    self.def_path, self.def_repo[self.PARAM_URL]
+                ) from exc
+            staging = copy(staging)
+            staging.read_only = True
 
-            staged_obj = stage(
-                local_odb,
-                path_info,
-                repo.repo_fs,
-                local_odb.fs.PARAM_CHECKSUM,
-            )
             self._staged_objs[rev] = staged_obj
-            git_odb = GitObjectDB(repo.repo_fs, repo.root_dir)
-            used_objs[git_odb].add(staged_obj)
-            return used_objs
+            used_obj_ids[staging].add(staged_obj.hash_info)
+            if isinstance(staged_obj, Tree):
+                used_obj_ids[staging].update(oid for _, _, oid in staged_obj)
+            return used_obj_ids, staged_obj
 
-    def _check_circular_import(self, odb):
+    def _check_circular_import(self, odb, obj_ids):
+        from dvc.data.db.reference import ReferenceObjectDB
+        from dvc.data.tree import Tree
         from dvc.exceptions import CircularImportError
         from dvc.fs.repo import RepoFileSystem
 
-        if not odb or not isinstance(odb.fs, RepoFileSystem):
+        if not isinstance(odb, ReferenceObjectDB):
             return
 
-        self_url = self.repo.url or self.repo.root_dir
-        if odb.fs.repo_url is not None and odb.fs.repo_url == self_url:
-            raise CircularImportError(self, odb.fs.repo_url, self_url)
+        def iter_objs():
+            for hash_info in obj_ids:
+                if hash_info.isdir:
+                    tree = Tree.load(odb, hash_info)
+                    yield from (odb.get(oid) for _, _, oid in tree)
+                else:
+                    yield odb.get(hash_info)
+
+        checked_urls = set()
+        for obj in iter_objs():
+            if not isinstance(obj.fs, RepoFileSystem):
+                continue
+            if (
+                obj.fs.repo_url in checked_urls
+                or obj.fs.root_dir in checked_urls
+            ):
+                continue
+            self_url = self.repo.url or self.repo.root_dir
+            if (
+                obj.fs.repo_url is not None
+                and obj.fs.repo_url == self_url
+                or obj.fs.root_dir == self.repo.root_dir
+            ):
+                raise CircularImportError(self, obj.fs.repo_url, self_url)
+            checked_urls.update([obj.fs.repo_url, obj.fs.root_dir])
 
     def get_obj(self, filter_info=None, **kwargs):
-        from dvc.objects.stage import stage
-
-        odb = self.repo.odb.local
-        locked = kwargs.pop("locked", True)
-        with self._make_repo(locked=locked, cache_dir=odb.cache_dir) as repo:
-            rev = repo.get_rev()
-            if locked and self.def_repo.get(self.PARAM_REV_LOCK) is None:
-                self.def_repo[self.PARAM_REV_LOCK] = rev
-            obj = self._staged_objs.get(rev)
-            if obj is not None:
-                return obj
-
-            path_info = PathInfo(repo.root_dir) / str(self.def_path)
-            obj = stage(
-                odb,
-                path_info,
-                repo.repo_fs,
-                odb.fs.PARAM_CHECKSUM,
-            )
-            self._staged_objs[rev] = obj
-            return obj
+        locked = kwargs.get("locked", True)
+        rev = self._get_rev(locked=locked)
+        if rev in self._staged_objs:
+            return self._staged_objs[rev]
+        _, obj = self._get_used_and_obj(
+            obj_only=True, filter_info=filter_info, **kwargs
+        )
+        return obj
 
     def _make_repo(self, locked=True, **kwargs):
         from dvc.external_repo import external_repo