webdav: fix unstrip_protocol

This is less error-prone than implementing these methods ourselves.

.github/workflows/benchmarks.yaml

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

.github/workflows/packages.yaml

@@ -10,11 +10,13 @@ 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:
@@ -39,11 +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:
@@ -68,11 +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:
@@ -100,11 +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]
@@ -128,7 +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:
@@ -136,7 +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

@@ -16,19 +16,21 @@ jobs:
     timeout-minutes: 10
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/[email protected]
+    - name: Cancel Previous Runs
+      uses: styfle/[email protected]
+      with:
+        access_token: ${{ github.token }}
+    - uses: actions/[email protected]
+      with:
+        fetch-depth: 0
     - name: Set up Python 3.8
       uses: actions/[email protected]
       with:
         python-version: 3.8
     - name: Install requirements
       run: |
-        pip install wheel
-        pip install ".[all,tests,terraform]" Pygments collective.checkdocs pre-commit
-    - name: install temporary dependencies
-      run: |
-        pip install git+https://github.com/isidentical/gdrivefs@service-account
-        pip install git+https://github.com/isidentical/pydata-google-auth@service-account
+        pip install --upgrade pip setuptools wheel
+        pip install ".[dev]" pre-commit
     - name: Check README
       run: python setup.py checkdocs
     - uses: pre-commit/[email protected]
@@ -38,19 +40,12 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        os: [ubuntu-18.04, windows-2019, macos-10.15]
-        pyv: ["3.6", "3.7", "3.8", "3.9"]
-        include:
-          - os: "macos-10.15"
-            pyv: "3.10.0-rc.1"
-        exclude:
-          - os: macos-10.15
-            pyv: "3.9"
-          - os: macos-10.15
-            pyv: "3.7"
-
+        os: [ubuntu-20.04, windows-latest, macos-latest]
+        pyv: ["3.7", "3.8", "3.9", "3.10"]
     steps:
-    - uses: actions/checkout@v2.3.4
+    - uses: actions/checkout@v2.4.0
+      with:
+        fetch-depth: 0
     - name: Set up Python
       uses: actions/[email protected]
       with:
@@ -61,56 +56,27 @@ jobs:
         echo "::set-output name=dir::$(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 libgit2 on Python3.10 in macOS
-      if: matrix.pyv == '3.10.0-rc.1' && matrix.os == 'macos-10.15'
-      run: |
-        brew install libgit2
     - name: install
       if: steps.cache.pip-cache-dir.cache-hit != 'true'
       run: |
         pip install --upgrade pip setuptools wheel
-        pip install -e ".[all,tests,terraform]"
-    - 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 -e ".[dev]"
     - 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
     - name: run tests
-      timeout-minutes: 30
-      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 }}
-        OSS_ACCESS_KEY_ID: ${{ secrets.OSS_ACCESS_KEY_ID}}
-        OSS_ACCESS_KEY_SECRET: ${{ secrets.OSS_ACCESS_KEY_SECRET}}
-        OSS_ENDPOINT: ${{ secrets.OSS_ENDPOINT}}
+      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/[email protected]
       with:

.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

@@ -1,24 +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: 21.8b0
+    rev: 21.12b0
+  - repo: https://github.com/pre-commit/pre-commit-hooks
+    rev: v4.0.1
+    hooks:
+      - id: check-toml
+      - id: check-yaml
   - hooks:
       - id: codespell
         args: 
           - --ignore-words-list
-          - ba,datas,fo,uptodate
+          - 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: 5.9.3
+    rev: 5.10.1
   - hooks:
       - id: flake8
         language_version: python3
@@ -29,19 +34,14 @@ repos:
           - flake8-string-format
     repo: https://gitlab.com/pycqa/flake8
     rev: 3.9.2
-  - hooks:
-      -  id: mypy
-         language_version: python3
-         files: ^dvc/
-         additional_dependencies:
-           - types-requests
-           - types-paramiko
-           - types-tabulate
-           - types-toml
-    repo: https://github.com/pre-commit/mirrors-mypy
-    rev: 'v0.910'
   - repo: local
     hooks:
+    - id: mypy
+      name: mypy
+      entry: mypy
+      files: ^dvc/
+      language: system
+      types: [python]
     - id: pylint
       name: pylint
       entry: pylint
@@ -54,7 +54,7 @@ repos:
         id: beautysh
         language_version: python3
     repo: https://github.com/lovesegfault/beautysh
-    rev: v6.1.0
+    rev: v6.2.1
   - hooks:
       - args:
           - git-hook

.pylintrc

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

LICENSE

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

MANIFEST.in

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

dvc/_debug.py

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

dvc/analytics.py

@@ -68,7 +68,7 @@ def send(path):
     url = "https://analytics.dvc.org"
     headers = {"content-type": "application/json"}
 
-    with open(path) as fobj:
+    with open(path, encoding="utf-8") as fobj:
         report = json.load(fobj)
 
     report.update(_runtime_info())
@@ -82,10 +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())
@@ -173,13 +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 +19,15 @@ 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)
+            metadata = _repo.repo_fs.metadata(fs_path)
 
         if not metadata.is_dvc:
             raise OutputNotFoundError(path, repo)
 
         cloud = metadata.repo.cloud
-        md5 = metadata.repo.dvcfs.info(path_info)["md5"]
+        md5 = metadata.repo.dvcfs.info(fs_path)["md5"]
         return cloud.get_url_for(remote, checksum=md5)
 
 
@@ -113,7 +112,7 @@ 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

@@ -1,351 +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.db.slow_link_detection import (  # type: ignore[attr-defined]
-    slow_link_guard,
-)
-from dvc.objects.errors import ObjectFormatError
-from dvc.objects.stage import stage
-from dvc.types import Optional
-
-logger = logging.getLogger(__name__)
-
-
-def _changed(path_info, fs, obj, cache, state=None):
-    logger.trace("checking if '%s'('%s') has changed.", path_info, obj)
-
-    try:
-        check(cache, obj)
-    except (FileNotFoundError, ObjectFormatError):
-        logger.debug(
-            "cache for '%s'('%s') has changed.", path_info, obj.hash_info
-        )
-        return True
-
-    actual = state.get(path_info, fs) if state else None
-    if actual is None:
-        try:
-            actual = stage(
-                cache, path_info, fs, obj.hash_info.name, dry_run=True
-            )[1].hash_info
-        except FileNotFoundError:
-            logger.debug("'%s' doesn't exist.", path_info)
-            return True
-
-    if obj.hash_info != actual:
-        logger.debug(
-            "hash value '%s' for '%s' has changed (actual '%s').",
-            obj.hash_info,
-            actual,
-            path_info,
-        )
-        return True
-
-    logger.trace("'%s' hasn't changed.", path_info)
-    return False
-
-
-def _remove(path_info, fs, cache, force=False):
-    if not fs.exists(path_info):
-        return
-
-    if force:
-        fs.remove(path_info)
-        return
-
-    current = stage(cache, path_info, fs, fs.PARAM_CHECKSUM, dry_run=True)[
-        1
-    ].hash_info
-    try:
-        obj = load(cache, current)
-        check(cache, obj)
-    except (FileNotFoundError, ObjectFormatError):
-        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):
-    cache.makedirs(to_info.parent)
-    try:
-        _try_links(cache, from_info, to_info, cache.cache_types)
-    except FileNotFoundError as exc:
-        raise CheckoutError([to_info]) from exc
-
-
-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, state=state):
-            modified = True
-            _remove(path_info, fs, cache, force=force)
-            _link(cache, cache_info, path_info)
-        else:
-            if fs.iscopy(path_info) and _cache_is_copy(cache, path_info):
-                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)
-
-    failed = []
-    for entry_key, entry_obj in obj:
-        try:
-            entry_modified = _checkout_file(
-                path_info.joinpath(*entry_key),
-                fs,
-                entry_obj,
-                cache,
-                force,
-                progress_callback,
-                relink,
-                state=None,
-            )
-            if entry_modified:
-                modified = True
-        except CheckoutError as exc:
-            failed.extend(exc.target_infos)
-
-    modified = (
-        _remove_redundant_files(
-            path_info, fs, obj, cache, force, dvcignore=dvcignore
-        )
-        or modified
-    )
-
-    if failed:
-        raise CheckoutError(failed)
-
-    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, state=state):
-        logger.trace("Data '%s' didn't change.", path_info)  # type: ignore
-        skip = True
-
-    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/command/dag.py

@@ -38,12 +38,14 @@ def _collect_targets(repo, target, outs):
     targets = []
 
     outs_trie = repo.index.outs_trie
-    for stage, info in pairs:
-        if not info:
+    for stage, path in pairs:
+        if not path:
             targets.extend([str(out) for out in stage.outs])
             continue
 
-        for out in 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
@@ -103,9 +105,8 @@ class CmdDAG(CmdBase):
         if self.args.dot:
             ui.write(_show_dot(G))
         else:
-            from dvc.utils.pager import pager
-
-            pager(_show_ascii(G))
+            with ui.pager():
+                ui.write(_show_ascii(G))
 
         return 0
 

dvc/command/data_sync.py

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

dvc/command/diff.py

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

dvc/command/experiments.py

@@ -1,14 +1,12 @@
 import argparse
 import logging
-import os
 from collections import Counter, OrderedDict, defaultdict
 from datetime import date, datetime
 from fnmatch import fnmatch
 from typing import TYPE_CHECKING, Dict, Iterable, Optional
 
-from funcy import lmap
+from funcy import compact, lmap
 
-from dvc import prompt
 from dvc.command import completion
 from dvc.command.base import CmdBase, append_doc_link, fix_subparsers
 from dvc.command.metrics import DEFAULT_PRECISION
@@ -20,9 +18,8 @@ from dvc.utils.flatten import flatten
 from dvc.utils.serialize import encode_exception
 
 if TYPE_CHECKING:
-    from rich.text import Text
-
     from dvc.compare import TabularData
+    from dvc.ui import RichText
 
 
 logger = logging.getLogger(__name__)
@@ -145,7 +142,7 @@ def _collect_rows(
     fill_value=FILL_VALUE,
     iso=False,
 ):
-    from dvc.scm.git import Git
+    from scmrepo.git import Git
 
     if sort_by:
         sort_path, sort_name, sort_type = _sort_column(
@@ -261,11 +258,12 @@ def _sort_column(sort_by, metric_names, param_names):
 def _sort_exp(experiments, sort_path, sort_name, typ, reverse):
     def _sort(item):
         rev, exp = item
-        tip = exp.get("checkpoint_tip")
+        exp_data = exp.get("data", {})
+        tip = exp_data.get("checkpoint_tip")
         if tip and tip != rev:
             # Sort checkpoint experiments by tip commit
             return _sort((tip, experiments[tip]))
-        data = exp.get(typ, {}).get(sort_path, {})
+        data = exp_data.get(typ, {}).get(sort_path, {}).get("data", {})
         val = flatten(data).get(sort_name)
         return val is None, val
 
@@ -292,8 +290,6 @@ def _format_time(datetime_obj, fill_value=FILL_VALUE, iso=False):
 
 
 def _extend_row(row, names, items, precision, fill_value=FILL_VALUE):
-    from rich.text import Text
-
     from dvc.compare import _format_field, with_value
 
     if not items:
@@ -308,10 +304,10 @@ def _extend_row(row, names, items, precision, fill_value=FILL_VALUE):
                 item.get(name),
                 FILL_VALUE_ERRORED if data.get("error", None) else fill_value,
             )
-            # wrap field data in rich.Text, otherwise rich may
+            # wrap field data in ui.rich_text, otherwise rich may
             # interpret unescaped braces from list/dict types as rich
             # markup tags
-            row.append(Text(str(_format_field(value, precision))))
+            row.append(ui.rich_text(str(_format_field(value, precision))))
 
 
 def _parse_filter_list(param_list):
@@ -362,17 +358,15 @@ def experiments_table(
     return td
 
 
-def prepare_exp_id(kwargs) -> "Text":
-    from rich.text import Text
-
+def prepare_exp_id(kwargs) -> "RichText":
     exp_name = kwargs["Experiment"]
     rev = kwargs["rev"]
     typ = kwargs.get("typ", "baseline")
 
     if typ == "baseline" or not exp_name:
-        text = Text(exp_name or rev)
+        text = ui.rich_text(exp_name or rev)
     else:
-        text = Text.assemble(rev, " [", (exp_name, "bold"), "]")
+        text = ui.rich_text.assemble(rev, " [", (exp_name, "bold"), "]")
 
     parent = kwargs.get("parent")
     suff = f" ({parent})" if parent else ""
@@ -380,7 +374,7 @@ def prepare_exp_id(kwargs) -> "Text":
 
     tree = experiment_types[typ]
     pref = f"{tree} " if tree else ""
-    return Text(pref) + text
+    return ui.rich_text(pref) + text
 
 
 def baseline_styler(typ):
@@ -388,7 +382,12 @@ def baseline_styler(typ):
 
 
 def show_experiments(
-    all_experiments, pager=True, no_timestamp=False, show_csv=False, **kwargs
+    all_experiments,
+    pager=True,
+    no_timestamp=False,
+    csv=False,
+    markdown=False,
+    **kwargs,
 ):
     from funcy.seqs import flatten as flatten_list
 
@@ -444,7 +443,7 @@ def show_experiments(
 
     row_styles = lmap(baseline_styler, td.column("typ"))
 
-    if not show_csv:
+    if not csv:
         merge_headers = ["Experiment", "rev", "typ", "parent"]
         td.column("Experiment")[:] = map(
             prepare_exp_id, td.as_dict(merge_headers)
@@ -472,13 +471,17 @@ def show_experiments(
         }
     )
 
+    if kwargs.get("only_changed", False):
+        td.drop_duplicates("cols")
+
     td.render(
         pager=pager,
         borders=True,
         rich_table=True,
         header_styles=styles,
         row_styles=row_styles,
-        show_csv=show_csv,
+        csv=csv,
+        markdown=markdown,
     )
 
 
@@ -511,18 +514,16 @@ class CmdExperimentsShow(CmdBase):
             logger.exception("failed to show experiments")
             return 1
 
-        if self.args.show_json:
-            import json
-
-            ui.write(json.dumps(all_experiments, default=_format_json))
+        if self.args.json:
+            ui.write_json(all_experiments, default=_format_json)
         else:
             precision = (
                 self.args.precision or None
-                if self.args.show_csv
+                if self.args.csv
                 else DEFAULT_PRECISION
             )
-            fill_value = "" if self.args.show_csv else FILL_VALUE
-            iso = True if self.args.show_csv else False
+            fill_value = "" if self.args.csv else FILL_VALUE
+            iso = True if self.args.csv else False
 
             show_experiments(
                 all_experiments,
@@ -537,7 +538,9 @@ class CmdExperimentsShow(CmdBase):
                 fill_value=fill_value,
                 iso=iso,
                 pager=not self.args.no_pager,
-                show_csv=self.args.show_csv,
+                csv=self.args.csv,
+                markdown=self.args.markdown,
+                only_changed=self.args.only_changed,
             )
         return 0
 
@@ -566,10 +569,8 @@ class CmdExperimentsDiff(CmdBase):
             logger.exception("failed to show experiments diff")
             return 1
 
-        if self.args.show_json:
-            import json
-
-            ui.write(json.dumps(diff))
+        if self.args.json:
+            ui.write_json(diff)
         else:
             from dvc.compare import show_diff
 
@@ -584,11 +585,12 @@ class CmdExperimentsDiff(CmdBase):
                 show_diff(
                     diff[key],
                     title=title,
-                    markdown=self.args.show_md,
+                    markdown=self.args.markdown,
                     no_path=self.args.no_path,
-                    old=self.args.old,
                     on_empty_diff="diff not supported",
                     precision=precision if key == "metrics" else None,
+                    a_rev=self.args.a_rev,
+                    b_rev=self.args.b_rev,
                 )
 
         return 0
@@ -669,7 +671,7 @@ class CmdExperimentsGC(CmdRepro):
         logger.warning(msg)
 
         msg = "Are you sure you want to proceed?"
-        if not self.args.force and not prompt.confirm(msg):
+        if not self.args.force and not ui.confirm(msg):
             return 1
 
         removed = self.repo.experiments.gc(
@@ -791,6 +793,7 @@ class CmdExperimentsRemove(CmdBase):
 
 
 class CmdExperimentsInit(CmdBase):
+    DEFAULT_NAME = "train"
     CODE = "src"
     DATA = "data"
     MODELS = "models"
@@ -798,51 +801,79 @@ class CmdExperimentsInit(CmdBase):
     DEFAULT_PARAMS = "params.yaml"
     PLOTS = "plots"
     DVCLIVE = "dvclive"
-    DEFAULT_NAME = "default"
+    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.command.stage import parse_cmd
 
-        cmd = parse_cmd(self.args.cmd)
-        if not cmd:
+        cmd = parse_cmd(self.args.command)
+        if not self.args.interactive and not cmd:
             raise InvalidArgumentError("command is not specified")
-        if self.args.interactive:
-            raise NotImplementedError(
-                "'-i/--interactive' is not implemented yet."
-            )
-        if self.args.explicit:
-            raise NotImplementedError("'--explicit' is not implemented yet.")
-        if self.args.template:
-            raise NotImplementedError("template is not supported yet.")
 
-        from dvc.utils.serialize import LOADERS
+        from dvc.repo.experiments.init import init
 
-        code = self.args.code or self.CODE
-        data = self.args.data or self.DATA
-        models = self.args.models or self.MODELS
-        metrics = self.args.metrics or self.DEFAULT_METRICS
-        params_path = self.args.params or self.DEFAULT_PARAMS
-        plots = self.args.plots or self.PLOTS
-        dvclive = self.args.live or self.DVCLIVE
+        defaults = {}
+        if not self.args.explicit:
+            config = self.repo.config["exp"]
+            defaults.update({**self.DEFAULTS, **config})
 
-        _, ext = os.path.splitext(params_path)
-        params = list(LOADERS[ext](params_path))
-
-        name = self.args.name or self.DEFAULT_NAME
-        stage = self.repo.stage.add(
-            name=name,
-            cmd=cmd,
-            deps=[code, data],
-            outs=[models],
-            params=[{params_path: params}],
-            metrics_no_cache=[metrics],
-            plots_no_cache=[plots],
-            live=dvclive,
-            force=True,
+        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=[stage.addressing])
+            return self.repo.experiments.run(
+                targets=[initialized_stage.addressing]
+            )
+
         return 0
 
 
@@ -951,7 +982,10 @@ def add_parser(subparsers, parent_parser):
     )
     experiments_show_parser.add_argument(
         "--sort-order",
-        help="Sort order to use with --sort-by.",
+        help=(
+            "Sort order to use with --sort-by."
+            " Defaults to ascending ('asc')."
+        ),
         choices=("asc", "desc"),
         default="asc",
     )
@@ -968,17 +1002,27 @@ def add_parser(subparsers, parent_parser):
         help="Always show git commit SHAs instead of branch/tag names.",
     )
     experiments_show_parser.add_argument(
+        "--json",
         "--show-json",
         action="store_true",
         default=False,
         help="Print output in JSON format instead of a human-readable table.",
     )
     experiments_show_parser.add_argument(
+        "--csv",
         "--show-csv",
         action="store_true",
         default=False,
         help="Print output in csv format instead of a human-readable table.",
     )
+    experiments_show_parser.add_argument(
+        "--md",
+        "--show-md",
+        action="store_true",
+        default=False,
+        dest="markdown",
+        help="Show tabulated output in the Markdown format (GFM).",
+    )
     experiments_show_parser.add_argument(
         "--precision",
         type=int,
@@ -988,6 +1032,15 @@ def add_parser(subparsers, parent_parser):
         ),
         metavar="<n>",
     )
+    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.set_defaults(func=CmdExperimentsShow)
 
     EXPERIMENTS_APPLY_HELP = (
@@ -1042,23 +1095,20 @@ def add_parser(subparsers, parent_parser):
         help="Show only params that are stage dependencies.",
     )
     experiments_diff_parser.add_argument(
+        "--json",
         "--show-json",
         action="store_true",
         default=False,
         help="Show output in JSON format.",
     )
     experiments_diff_parser.add_argument(
+        "--md",
         "--show-md",
         action="store_true",
         default=False,
+        dest="markdown",
         help="Show tabulated output in the Markdown format (GFM).",
     )
-    experiments_diff_parser.add_argument(
-        "--old",
-        action="store_true",
-        default=False,
-        help="Show old metric/param value.",
-    )
     experiments_diff_parser.add_argument(
         "--no-path",
         action="store_true",
@@ -1366,9 +1416,10 @@ def add_parser(subparsers, parent_parser):
         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(
-        "cmd",
+        "command",
         nargs=argparse.REMAINDER,
         help="Command to execute.",
         metavar="command",
@@ -1385,40 +1436,66 @@ def add_parser(subparsers, parent_parser):
         help="Prompt for values that are not provided",
     )
     experiments_init_parser.add_argument(
-        "--template", help="Stage template to use to fill with provided values"
+        "-f",
+        "--force",
+        action="store_true",
+        default=False,
+        help="Overwrite existing stage",
     )
     experiments_init_parser.add_argument(
-        "--explicit", help="Only use the path values explicitly provided"
+        "--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"
+        "--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",
+        "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",
+        "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",
+        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"
+        "--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"
+        "--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",
+        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"
+        "--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/command/ls/__init__.py

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

dvc/command/machine.py

@@ -2,7 +2,10 @@ import argparse
 
 from dvc.command.base import CmdBase, append_doc_link, fix_subparsers
 from dvc.command.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
 
@@ -70,18 +73,56 @@ class CmdMachineRemove(CmdMachineConfig):
 
 
 class CmdMachineList(CmdMachineConfig):
-    def run(self):
+    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, {})
-            prefix = self._config_file_prefix(
-                self.args.show_origin, self.config, level
-            )
+            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
 
 
@@ -104,6 +145,54 @@ class CmdMachineModify(CmdMachineConfig):
         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:
@@ -143,6 +232,59 @@ class CmdMachineCreate(CmdBase):
         return 0
 
 
+class CmdMachineStatus(CmdBase):
+    INSTANCE_FIELD = ["name", "instance", "status"]
+    SHOWN_FIELD = [
+        "cloud",
+        "instance_ip",
+        "instance_type",
+        "instance_hdd_size",
+        "instance_gpu",
+    ]
+
+    def _add_row(
+        self,
+        name: str,
+        all_status: List[Dict],
+        td: TabularData,
+    ):
+
+        if not all_status:
+            row = [name, None, "offline"]
+            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="-"
+        )
+
+        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:
@@ -280,6 +422,18 @@ def add_parser(subparsers, parent_parser):
     )
     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",
@@ -296,7 +450,7 @@ def add_parser(subparsers, parent_parser):
     machine_CREATE_HELP = "Create and start a machine instance."
     machine_create_parser = machine_subparsers.add_parser(
         "create",
-        parents=[parent_config_parser, parent_parser],
+        parents=[parent_parser],
         description=append_doc_link(machine_CREATE_HELP, "machine/create"),
         help=machine_CREATE_HELP,
         formatter_class=argparse.RawDescriptionHelpFormatter,
@@ -306,10 +460,25 @@ def add_parser(subparsers, parent_parser):
     )
     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_config_parser, parent_parser],
+        parents=[parent_parser],
         description=append_doc_link(machine_DESTROY_HELP, "machine/destroy"),
         help=machine_DESTROY_HELP,
         formatter_class=argparse.RawDescriptionHelpFormatter,
@@ -322,7 +491,7 @@ def add_parser(subparsers, parent_parser):
     machine_SSH_HELP = "Connect to a machine via SSH."
     machine_ssh_parser = machine_subparsers.add_parser(
         "ssh",
-        parents=[parent_config_parser, parent_parser],
+        parents=[parent_parser],
         description=append_doc_link(machine_SSH_HELP, "machine/ssh"),
         help=machine_SSH_HELP,
         formatter_class=argparse.RawDescriptionHelpFormatter,

dvc/command/metrics.py

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

dvc/command/params.py

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

dvc/command/plots.py

@@ -62,19 +62,11 @@ class CmdPlots(CmdBase):
                 html_template_path=self.args.html_template,
             )
 
-            assert index_path.is_absolute()
-            url = index_path.as_uri()
-            ui.write(url)
+            ui.write(index_path.as_uri())
 
             if self.args.open:
-                import webbrowser
+                return ui.open_browser(index_path)
 
-                opened = webbrowser.open(index_path)
-                if not opened:
-                    ui.error_write(
-                        "Failed to open. Please try opening it manually."
-                    )
-                    return 1
             return 0
 
         except DvcException:
@@ -183,7 +175,10 @@ def add_parser(subparsers, parent_parser):
     _add_output_arguments(plots_diff_parser)
     plots_diff_parser.set_defaults(func=CmdPlotsDiff)
 
-    PLOTS_MODIFY_HELP = "Modify display properties of plot metrics files."
+    PLOTS_MODIFY_HELP = (
+        "Modify display properties of data-series plots "
+        "(has no effect on image-type plots)."
+    )
     plots_modify_parser = plots_subparsers.add_parser(
         "modify",
         parents=[parent_parser],

dvc/command/remote.py

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

dvc/command/run.py

@@ -23,7 +23,7 @@ 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 +36,7 @@ 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"),

dvc/command/stage.py

@@ -129,7 +129,7 @@ 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 +281,7 @@ def _add_common_args(parser):
         ),
     )
     parser.add_argument(
-        "cmd",
+        "command",
         nargs=argparse.REMAINDER,
         help="Command to execute.",
         metavar="command",

dvc/command/status.py

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

dvc/compare.py

@@ -12,6 +12,7 @@ from typing import (
     Mapping,
     MutableSequence,
     Sequence,
+    Set,
     Tuple,
     Union,
     overload,
@@ -115,6 +116,8 @@ class TabularData(MutableSequence[Sequence["CellT"]]):
             del col[item]
 
     def __len__(self) -> int:
+        if not self._columns:
+            return 0
         return len(self.columns[0])
 
     @property
@@ -168,7 +171,7 @@ class TabularData(MutableSequence[Sequence["CellT"]]):
     def render(self, **kwargs: Any):
         from dvc.ui import ui
 
-        if kwargs.pop("show_csv", False):
+        if kwargs.pop("csv", False):
             ui.write(self.to_csv(), end="")
         else:
             ui.table(self, headers=self.keys(), **kwargs)
@@ -181,6 +184,86 @@ 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"):
+        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 = True
+        if how == "all":
+            match = False
+
+        for n_row, row in enumerate(self):
+            for n_col, col in enumerate(row):
+                if (col == self._fill_value) is match:
+                    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"):
+        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):
+                # Cast to str because Text is not hashable error
+                unique_vals = {str(x) for x in col if x != 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):
+                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}"
@@ -211,8 +294,12 @@ 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)
 
@@ -240,8 +327,8 @@ 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
 
@@ -256,6 +343,8 @@ 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,
@@ -266,6 +355,8 @@ def show_diff(
         precision=precision,
         round_digits=round_digits,
         on_empty_diff=on_empty_diff,
+        a_rev=a_rev,
+        b_rev=b_rev,
     )
     td.render(markdown=markdown)
 

dvc/config.py

@@ -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__)
 
@@ -141,7 +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):
@@ -229,6 +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
@@ -238,8 +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):

dvc/config_schema.py

@@ -156,6 +156,8 @@ SCHEMA = {
                     "grant_write_acp": str,
                     "grant_full_control": str,
                     "cache_regions": bool,
+                    "read_timeout": Coerce(int),
+                    "connect_timeout": Coerce(int),
                     **REMOTE_COMMON,
                 },
                 "gs": {
@@ -177,10 +179,12 @@ SCHEMA = {
                 },
                 "hdfs": {"user": str, "kerb_ticket": str, **REMOTE_COMMON},
                 "webhdfs": {
-                    "hdfscli_config": str,
-                    "webhdfs_token": str,
-                    "user": str,
-                    "webhdfs_alias": str,
+                    "kerberos": Bool,
+                    "kerberos_principal": str,
+                    "proxy_to": str,
+                    "ssl_verify": Any(Bool, str),
+                    "token": str,
+                    "use_https": Bool,
                     **REMOTE_COMMON,
                 },
                 "azure": {
@@ -223,9 +227,13 @@ 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")),
@@ -233,7 +241,6 @@ SCHEMA = {
                 Lower, Choices("us-west", "us-east", "eu-west", "eu-north")
             ),
             "image": str,
-            "name": str,
             "spot": Bool,
             "spot_price": Coerce(float),
             "instance_hdd_size": Coerce(int),
@@ -250,4 +257,13 @@ SCHEMA = {
         Optional("parametrization", default=True): Bool,
     },
     "plots": {"html_template": str},
+    "exp": {
+        "code": str,
+        "data": str,
+        "models": str,
+        "metrics": str,
+        "params": str,
+        "plots": str,
+        "live": str,
+    },
 }

dvc/daemon.py

@@ -12,8 +12,6 @@ from dvc.utils import fix_env, is_binary
 
 logger = logging.getLogger(__name__)
 
-CREATE_NO_WINDOW = 0x08000000  # only available since Python 3.7 in subprocess.
-
 
 def _popen(cmd, **kwargs):
     prefix = [sys.executable]
@@ -28,6 +26,7 @@ def _popen(cmd, **kwargs):
 def _spawn_windows(cmd, env):
     from subprocess import (
         CREATE_NEW_PROCESS_GROUP,
+        CREATE_NO_WINDOW,
         STARTF_USESHOWWINDOW,
         STARTUPINFO,
     )
@@ -49,17 +48,17 @@ def _spawn_posix(cmd, env):
     # with PyInstaller has trouble with SystemExit exception and throws
     # errors such as "[26338] Failed to execute script __main__"
     try:
-        pid = os.fork()
+        pid = os.fork()  # pylint: disable=no-member
         if pid > 0:
             return
     except OSError:
         logger.exception("failed at first fork")
         os._exit(1)  # pylint: disable=protected-access
 
-    os.setsid()
+    os.setsid()  # pylint: disable=no-member
 
     try:
-        pid = os.fork()
+        pid = os.fork()  # pylint: disable=no-member
         if pid > 0:
             os._exit(0)  # pylint: disable=protected-access
     except OSError:

dvc/data_cloud.py

@@ -58,9 +58,9 @@ class DataCloud:
         from dvc.fs import get_cloud_fs
         from dvc.objects.db import get_odb
 
-        cls, config, path_info = get_cloud_fs(self.repo, name=name)
-        config["tmp_dir"] = self.repo.tmp_dir
-        return get_odb(cls(**config), path_info, **config)
+        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,
@@ -156,5 +156,6 @@ class DataCloud:
         )
 
     def get_url_for(self, remote, checksum):
-        remote_odb = self.get_remote_odb(remote)
-        return str(remote_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 +1,4 @@
+import logging
 import os
 from collections import defaultdict
 
@@ -10,6 +11,8 @@ from dvc.utils.serialize import LOADERS, ParseError
 
 from .base import Dependency
 
+logger = logging.getLogger(__name__)
+
 
 class MissingParamsError(DvcException):
     pass
@@ -89,10 +92,10 @@ class ParamsDependency(Dependency):
         if not self.exists:
             return {}
 
-        suffix = self.path_info.suffix.lower()
+        suffix = self.repo.fs.path.suffix(self.fs_path).lower()
         loader = LOADERS[suffix]
         try:
-            return loader(self.path_info, fs=self.repo.fs)
+            return loader(self.fs_path, fs=self.repo.fs)
         except ParseError as exc:
             raise BadParamFileError(
                 f"Unable to read parameters from '{self}'"
@@ -133,3 +136,20 @@ class ParamsDependency(Dependency):
             )
 
         return HashInfo(self.PARAM_PARAMS, info)
+
+    def save(self):
+        if not self.exists:
+            raise self.DoesNotExistError(self)
+
+        if not self.isfile and not self.isdir:
+            raise self.IsNotFileOrDirError(self)
+
+        if self.is_empty:
+            logger.warning(f"'{self}' is empty.")
+
+        self.ignore()
+
+        if self.metric or self.plot:
+            self.verify_metric()
+
+        self.hash_info = self.get_hash()

dvc/dependency/repo.py

@@ -5,8 +5,6 @@ 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:
@@ -34,7 +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
@@ -63,14 +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.checkout import checkout
 
         for odb, objs in self.get_used_objs().items():
             self.repo.cloud.pull(objs, jobs=jobs, odb=odb)
 
         obj = self.get_obj()
         checkout(
-            to.path_info,
+            to.fs_path,
             to.fs,
             obj,
             self.repo.odb.local,
@@ -114,11 +112,11 @@ class RepoDependency(Dependency):
             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(
-                        [os.fspath(path_info)],
+                        [path],
                         force=True,
                         jobs=kwargs.get("jobs"),
                         recursive=True,
@@ -132,9 +130,9 @@ class RepoDependency(Dependency):
                     pass
 
             try:
-                staging, staged_obj = stage(
+                staging, _, staged_obj = stage(
                     local_odb,
-                    path_info,
+                    path,
                     repo.repo_fs,
                     local_odb.fs.PARAM_CHECKSUM,
                 )
@@ -148,9 +146,7 @@ class RepoDependency(Dependency):
             self._staged_objs[rev] = staged_obj
             used_obj_ids[staging].add(staged_obj.hash_info)
             if isinstance(staged_obj, Tree):
-                used_obj_ids[staging].update(
-                    entry.hash_info for _, entry in staged_obj
-                )
+                used_obj_ids[staging].update(oid for _, _, oid in staged_obj)
             return used_obj_ids, staged_obj
 
     def _check_circular_import(self, odb, obj_ids):
@@ -166,7 +162,7 @@ class RepoDependency(Dependency):
             for hash_info in obj_ids:
                 if hash_info.isdir:
                     tree = Tree.load(odb, hash_info)
-                    yield from (odb.get(entry.hash_info) for _, entry in tree)
+                    yield from (odb.get(oid) for _, _, oid in tree)
                 else:
                     yield odb.get(hash_info)
 

dvc/dvcfile.py

@@ -1,10 +1,7 @@
-import collections
 import contextlib
 import logging
 import os
-from typing import TYPE_CHECKING, Any, Union
-
-from voluptuous import MultipleInvalid
+from typing import TYPE_CHECKING, Any, Callable, Tuple, TypeVar, Union
 
 from dvc.exceptions import DvcException
 from dvc.parsing.versions import LOCKFILE_VERSION, SCHEMA_KWD
@@ -12,23 +9,18 @@ from dvc.stage import serialize
 from dvc.stage.exceptions import (
     StageFileBadNameError,
     StageFileDoesNotExistError,
-    StageFileFormatError,
     StageFileIsNotDvcFileError,
 )
 from dvc.types import AnyPath
 from dvc.utils import relpath
 from dvc.utils.collections import apply_diff
-from dvc.utils.serialize import (
-    dump_yaml,
-    modify_yaml,
-    parse_yaml,
-    parse_yaml_for_update,
-)
+from dvc.utils.serialize import dump_yaml, modify_yaml
 
 if TYPE_CHECKING:
     from dvc.repo import Repo
 
 logger = logging.getLogger(__name__)
+_T = TypeVar("_T")
 
 DVC_FILE = "Dvcfile"
 DVC_FILE_SUFFIX = ".dvc"
@@ -45,10 +37,6 @@ class FileIsGitIgnored(DvcException):
         )
 
 
-class LockfileCorruptedError(DvcException):
-    pass
-
-
 class ParametrizedDumpError(DvcException):
     pass
 
@@ -72,7 +60,7 @@ def is_lock_file(path):
 
 def is_git_ignored(repo, path):
     from dvc.fs.local import LocalFileSystem
-    from dvc.scm.base import NoSCMError
+    from dvc.scm import NoSCMError
 
     try:
         return isinstance(repo.fs, LocalFileSystem) and repo.scm.is_ignored(
@@ -96,7 +84,7 @@ def check_dvcfile_path(repo, path):
 
 
 class FileMixin:
-    SCHEMA = None
+    SCHEMA: Callable[[_T], _T]
 
     def __init__(self, repo, path, verify=True, **kwargs):
         self.repo = repo
@@ -138,7 +126,11 @@ class FileMixin:
         if self._is_git_ignored():
             raise FileIsGitIgnored(self.path)
 
-    def _load(self):
+    def load(self, **kwargs: Any) -> Any:
+        d, _ = self._load(**kwargs)
+        return d
+
+    def _load(self, **kwargs: Any) -> Tuple[Any, str]:
         # it raises the proper exceptions by priority:
         # 1. when the file doesn't exists
         # 2. filename is not a DVC file
@@ -155,18 +147,23 @@ class FileMixin:
             raise StageFileIsNotDvcFileError(self.path)
 
         self._check_gitignored()
-        with self.repo.fs.open(self.path, encoding="utf-8") as fd:
-            stage_text = fd.read()
-        d = parse_yaml(stage_text, self.path)
-        return self.validate(d, self.relpath), stage_text
+        return self._load_yaml(**kwargs)
 
     @classmethod
-    def validate(cls, d, fname=None):
-        assert isinstance(cls.SCHEMA, collections.abc.Callable)
-        try:
-            return cls.SCHEMA(d)  # pylint: disable=not-callable
-        except MultipleInvalid as exc:
-            raise StageFileFormatError(f"'{fname}' format error: {exc}")
+    def validate(cls, d: _T, fname: str = None) -> _T:
+        from dvc.utils.strictyaml import validate
+
+        return validate(d, cls.SCHEMA, path=fname)  # type: ignore[arg-type]
+
+    def _load_yaml(self, **kwargs: Any) -> Tuple[Any, str]:
+        from dvc.utils import strictyaml
+
+        return strictyaml.load(
+            self.path,
+            self.SCHEMA,  # type: ignore[arg-type]
+            self.repo.fs,
+            **kwargs,
+        )
 
     def remove(self, force=False):  # pylint: disable=unused-argument
         with contextlib.suppress(FileNotFoundError):
@@ -202,7 +199,7 @@ class SingleStageFile(FileMixin):
             check_dvcfile_path(self.repo, self.path)
         logger.debug(f"Saving information to '{relpath(self.path)}'.")
         dump_yaml(self.path, serialize.to_single_stage_file(stage))
-        self.repo.scm.track_file(self.relpath)
+        self.repo.scm_context.track_file(self.relpath)
 
     def remove_stage(self, stage):  # pylint: disable=unused-argument
         self.remove()
@@ -271,7 +268,7 @@ class PipelineFile(FileMixin):
             else:
                 data["stages"].update(stage_data)
 
-        self.repo.scm.track_file(self.relpath)
+        self.repo.scm_context.track_file(self.relpath)
 
     @property
     def stage(self):
@@ -298,10 +295,7 @@ class PipelineFile(FileMixin):
         if not self.exists():
             return
 
-        with open(self.path, encoding="utf-8") as f:
-            d = parse_yaml_for_update(f.read(), self.path)
-
-        self.validate(d, self.path)
+        d, _ = self._load_yaml(round_trip=True)
         if stage.name not in d.get("stages", {}):
             return
 
@@ -343,34 +337,25 @@ def migrate_lock_v1_to_v2(d, version_info):
     d["stages"] = stages
 
 
+def lockfile_schema(data: _T) -> _T:
+    schema = get_lockfile_schema(data)
+    return schema(data)
+
+
 class Lockfile(FileMixin):
-    @classmethod
-    def validate(cls, d, fname=None):
-        schema = get_lockfile_schema(d)
-        try:
-            return schema(d)
-        except MultipleInvalid as exc:
-            raise StageFileFormatError(f"'{fname}' format error: {exc}")
+    SCHEMA = staticmethod(lockfile_schema)  # type: ignore[assignment]
 
     def _verify_filename(self):
         pass  # lockfile path is hardcoded, so no need to verify here
 
-    def _load(self):
+    def _load(self, **kwargs: Any):
         try:
-            return super()._load()
+            return super()._load(**kwargs)
         except StageFileDoesNotExistError:
             # we still need to account for git-ignored dvc.lock file
             # even though it may not exist or have been .dvcignored
             self._check_gitignored()
             return {}, ""
-        except StageFileFormatError as exc:
-            raise LockfileCorruptedError(
-                f"Lockfile '{self.relpath}' is corrupted."
-            ) from exc
-
-    def load(self):
-        d, _ = self._load()
-        return d
 
     @property
     def latest_version_info(self):
@@ -403,16 +388,13 @@ class Lockfile(FileMixin):
             data["stages"].update(stage_data)
 
         if modified:
-            self.repo.scm.track_file(self.relpath)
+            self.repo.scm_context.track_file(self.relpath)
 
     def remove_stage(self, stage):
         if not self.exists():
             return
 
-        with open(self.path, encoding="utf-8") as f:
-            d = parse_yaml_for_update(f.read(), self.path)
-        self.validate(d, self.path)
-
+        d, _ = self._load_yaml(round_trip=True)
         version = LOCKFILE_VERSION.from_dict(d)
         data = d if version == LOCKFILE_VERSION.V1 else d.get("stages", {})
         if stage.name not in data:

dvc/exceptions.py

@@ -216,18 +216,6 @@ class DvcIgnoreInCollectedDirError(DvcException):
         )
 
 
-class GitHookAlreadyExistsError(DvcException):
-    def __init__(self, hook_name):
-        from dvc.utils import format_link
-
-        super().__init__(
-            "Hook '{}' already exists. Please refer to {} for more "
-            "info.".format(
-                hook_name, format_link("https://man.dvc.org/install")
-            )
-        )
-
-
 class FileTransferError(DvcException):
     _METHOD = "transfer"
 
@@ -305,7 +293,7 @@ class PathMissingError(DvcException):
 
 
 class RemoteCacheRequiredError(DvcException):
-    def __init__(self, path_info):
+    def __init__(self, scheme, fs_path):
         from dvc.utils import format_link
 
         super().__init__(
@@ -314,8 +302,8 @@ class RemoteCacheRequiredError(DvcException):
                 "existing cache on '{}' remote. See {} for information on how "
                 "to set up remote cache."
             ).format(
-                path_info,
-                path_info.scheme,
+                fs_path,
+                scheme,
                 format_link("https://man.dvc.org/config#cache"),
             )
         )
@@ -350,12 +338,12 @@ class CacheLinkError(DvcException):
         )
     )
 
-    def __init__(self, path_infos):
+    def __init__(self, fs_paths):
         msg = "No possible cache link types for '{}'. {}".format(
-            ", ".join([str(path) for path in path_infos]), self.SUPPORT_LINK
+            ", ".join(fs_paths), self.SUPPORT_LINK
         )
         super().__init__(msg)
-        self.path_infos = path_infos
+        self.fs_paths = fs_paths
 
 
 class CircularImportError(DvcException):
@@ -364,3 +352,8 @@ class CircularImportError(DvcException):
             f"'{dep}' contains invalid circular import. "
             f"DVC repo '{a}' already imports from '{b}'."
         )
+
+
+class PrettyDvcException(DvcException):
+    def __pretty_exc__(self, **kwargs):
+        """Print prettier exception message."""

dvc/external_repo.py

@@ -16,17 +16,20 @@ from dvc.exceptions import (
     PathMissingError,
 )
 from dvc.repo import Repo
+from dvc.scm import CloneError, map_scm_exception
 from dvc.utils import relpath
 
 logger = logging.getLogger(__name__)
 
 
 @contextmanager
+@map_scm_exception()
 def external_repo(
     url, rev=None, for_write=False, cache_dir=None, cache_types=None, **kwargs
 ):
+    from scmrepo.git import Git
+
     from dvc.config import NoRemoteError
-    from dvc.scm.git import Git
 
     logger.debug("Creating external repo %s@%s", url, rev)
     path = _cached_clone(url, rev, for_write=for_write)
@@ -173,7 +176,7 @@ def _clone_default_branch(url, rev, for_write=False):
 
     The cloned is reactualized with git pull unless rev is a known sha.
     """
-    from dvc.scm.git import Git
+    from scmrepo.git import Git
 
     clone_path, shallow = CLONES.get(url, (None, False))
 
@@ -198,14 +201,15 @@ def _clone_default_branch(url, rev, for_write=False):
                     logger.debug("erepo: git pull '%s'", url)
                     git.pull()
         else:
+            from dvc.scm import clone
+
             logger.debug("erepo: git clone '%s' to a temporary dir", url)
             clone_path = tempfile.mkdtemp("dvc-clone")
             if not for_write and rev and not Git.is_sha(rev):
                 # If rev is a tag or branch name try shallow clone first
-                from dvc.scm.base import CloneError
 
                 try:
-                    git = Git.clone(url, clone_path, shallow_branch=rev)
+                    git = clone(url, clone_path, shallow_branch=rev)
                     shallow = True
                     logger.debug(
                         "erepo: using shallow clone for branch '%s'", rev
@@ -213,7 +217,7 @@ def _clone_default_branch(url, rev, for_write=False):
                 except CloneError:
                     pass
             if not git:
-                git = Git.clone(url, clone_path)
+                git = clone(url, clone_path)
                 shallow = False
             CLONES[url] = (clone_path, shallow)
     finally:
@@ -237,7 +241,7 @@ def _unshallow(git):
 
 
 def _git_checkout(repo_path, rev):
-    from dvc.scm.git import Git
+    from scmrepo.git import Git
 
     logger.debug("erepo: git checkout %s@%s", repo_path, rev)
     git = Git(repo_path)

dvc/fs/__init__.py

@@ -1,4 +1,3 @@
-import posixpath
 from urllib.parse import urlparse
 
 from ..scheme import Schemes
@@ -39,7 +38,7 @@ def get_fs_cls(remote_conf, scheme=None):
     return FS_MAP.get(scheme, LocalFileSystem)
 
 
-def get_fs_config(config, **kwargs):
+def get_fs_config(repo, config, **kwargs):
     name = kwargs.get("name")
     if name:
         try:
@@ -50,10 +49,10 @@ def get_fs_config(config, **kwargs):
             raise RemoteNotFoundError(f"remote '{name}' doesn't exist")
     else:
         remote_conf = kwargs
-    return _resolve_remote_refs(config, remote_conf)
+    return _resolve_remote_refs(repo, config, remote_conf)
 
 
-def _resolve_remote_refs(config, remote_conf):
+def _resolve_remote_refs(repo, config, remote_conf):
     # Support for cross referenced remotes.
     # This will merge the settings, shadowing base ref with remote_conf.
     # For example, having:
@@ -79,8 +78,10 @@ def _resolve_remote_refs(config, remote_conf):
     if parsed.scheme != "remote":
         return remote_conf
 
-    base = get_fs_config(config, name=parsed.netloc)
-    url = posixpath.join(base["url"], parsed.path.lstrip("/"))
+    base = get_fs_config(repo, config, name=parsed.netloc)
+    cls, _, _ = get_cloud_fs(repo, **base)
+    relpath = parsed.path.lstrip("/").replace("/", cls.sep)
+    url = cls.sep.join((base["url"], relpath))
     return {**base, **remote_conf, "url": url}
 
 
@@ -91,7 +92,7 @@ def get_cloud_fs(repo, **kwargs):
     repo_config = repo.config if repo else {}
     core_config = repo_config.get("core", {})
 
-    remote_conf = get_fs_config(repo_config, **kwargs)
+    remote_conf = get_fs_config(repo, repo_config, **kwargs)
     try:
         remote_conf = SCHEMA["remote"][str](remote_conf)
     except Invalid as exc:
@@ -113,10 +114,8 @@ def get_cloud_fs(repo, **kwargs):
         remote_conf["gdrive_credentials_tmp_dir"] = repo.tmp_dir
 
     url = remote_conf.pop("url")
-    path_info = cls.PATH_CLS(
-        cls._strip_protocol(url)  # pylint:disable=protected-access
-    )
+    fs_path = cls._strip_protocol(url)  # pylint:disable=protected-access
 
     extras = cls._get_kwargs_from_urls(url)  # pylint:disable=protected-access
     conf = {**extras, **remote_conf}  # remote config takes priority
-    return cls, conf, path_info
+    return cls, conf, fs_path

dvc/fs/_metadata.py

@@ -2,7 +2,6 @@ from dataclasses import dataclass, field
 from typing import TYPE_CHECKING, List
 
 from dvc.output import Output
-from dvc.path_info import PathInfo
 
 if TYPE_CHECKING:
     from dvc.repo import Repo
@@ -16,7 +15,7 @@ class Metadata:
     """
 
     # required field
-    path_info: PathInfo
+    fs_path: str
     repo: "Repo"
 
     # computed fields
@@ -47,11 +46,13 @@ class Metadata:
         else:
             out = self.outs[0]
             # or, the path itself could be an output
-            self.is_output = self.path_info == out.path_info
+            self.is_output = self.fs_path == out.fs_path
 
             # or, a directory must have an output somewhere deep inside it
             if not self.is_output:
-                self.contains_outputs = out.path_info.isin(self.path_info)
+                self.contains_outputs = self.repo.fs.path.isin(
+                    out.fs_path, self.fs_path
+                )
 
         # or, the path could be a part of an output, i.e. inside of an output
         self.part_of_output = not self.is_output and not self.contains_outputs
@@ -69,4 +70,4 @@ class Metadata:
         return not self.isdir
 
     def __str__(self):
-        return f"Metadata: {self.path_info}"
+        return f"Metadata: {self.fs_path}"

dvc/fs/azure.py

@@ -7,7 +7,6 @@ from fsspec.utils import infer_storage_options
 from funcy import cached_property, memoize, wrap_prop
 
 from dvc.exceptions import DvcException
-from dvc.path_info import CloudURLInfo
 from dvc.scheme import Schemes
 from dvc.utils import format_link
 
@@ -41,7 +40,6 @@ def _az_config():
 # pylint:disable=abstract-method
 class AzureFileSystem(CallbackMixin, ObjectFSWrapper):
     scheme = Schemes.AZURE
-    PATH_CLS = CloudURLInfo
     PARAM_CHECKSUM = "etag"
     DETAIL_FIELDS = frozenset(("etag", "size"))
     REQUIRES = {
@@ -52,12 +50,14 @@ class AzureFileSystem(CallbackMixin, ObjectFSWrapper):
 
     @classmethod
     def _strip_protocol(cls, path: str):
-        bucket = infer_storage_options(path).get("host")
-        if bucket:
-            return path
+        opts = infer_storage_options(path)
+        if opts.get("host"):
+            return "{host}{path}".format(**opts)
 
-        bucket = _az_config().get("storage", "container_name", None)
-        return f"azure://{bucket}"
+        return _az_config().get("storage", "container_name", None)
+
+    def unstrip_protocol(self, path: str) -> str:
+        return "azure://" + path.lstrip("/")
 
     @staticmethod
     def _get_kwargs_from_urls(urlpath):

dvc/fs/base.py

@@ -4,13 +4,12 @@ import os
 from concurrent.futures import ThreadPoolExecutor, as_completed
 from functools import partialmethod
 from multiprocessing import cpu_count
-from typing import Any, ClassVar, Dict, FrozenSet, Optional
+from typing import ClassVar, Dict, FrozenSet, Optional
 
 from funcy import cached_property
 from tqdm.utils import CallbackIOWrapper
 
 from dvc.exceptions import DvcException
-from dvc.path_info import URLInfo
 from dvc.progress import DEFAULT_CALLBACK, FsspecCallback
 from dvc.ui import ui
 from dvc.utils import tmp_fname
@@ -37,12 +36,11 @@ class RemoteMissingDepsError(DvcException):
     pass
 
 
-class BaseFileSystem:
+class FileSystem:
     sep = "/"
 
     scheme = "base"
     REQUIRES: ClassVar[Dict[str, str]] = {}
-    PATH_CLS = URLInfo  # type: Any
     _JOBS = 4 * cpu_count()
 
     CHECKSUM_DIR_SUFFIX = ".dir"
@@ -70,10 +68,19 @@ class BaseFileSystem:
     def config(self):
         return self._config
 
+    @cached_property
+    def path(self):
+        from .path import Path
+
+        return Path(self.sep)
+
     @classmethod
     def _strip_protocol(cls, path: str):
         return path
 
+    def unstrip_protocol(self, path):
+        return path
+
     @staticmethod
     def _get_kwargs_from_urls(urlpath):  # pylint:disable=unused-argument
         return {}
@@ -130,36 +137,36 @@ class BaseFileSystem:
             f"dependencies: {missing}. {hint}"
         )
 
-    def checksum(self, path_info) -> str:
+    def checksum(self, path) -> str:
         raise NotImplementedError
 
-    def open(self, path_info, mode: str = "r", encoding: str = None, **kwargs):
+    def open(self, path, mode: str = "r", encoding: str = None, **kwargs):
         raise RemoteActionNotImplemented("open", self.scheme)
 
-    def exists(self, path_info) -> bool:
+    def exists(self, path) -> bool:
         raise NotImplementedError
 
     # pylint: disable=unused-argument
 
-    def isdir(self, path_info):
+    def isdir(self, path):
         """Optional: Overwrite only if the remote has a way to distinguish
         between a directory and a file.
         """
         return False
 
-    def isfile(self, path_info):
+    def isfile(self, path):
         """Optional: Overwrite only if the remote has a way to distinguish
         between a directory and a file.
         """
         return True
 
-    def isexec(self, path_info):
+    def isexec(self, path):
         """Optional: Overwrite only if the remote has a way to distinguish
         between executable and non-executable file.
         """
         return False
 
-    def iscopy(self, path_info):
+    def iscopy(self, path):
         """Check if this file is an independent copy."""
         return False  # We can't be sure by default
 
@@ -167,34 +174,28 @@ class BaseFileSystem:
         """Return a generator with (root, dirs, files)."""
         raise NotImplementedError
 
-    def walk_files(self, path_info, **kwargs):
-        """Return a generator with `PathInfo`s to all the files.
+    def find(self, path, prefix=None):
+        """Return a generator with `str`s to all the files.
 
         Optional kwargs:
-            prefix (bool): If true `path_info` will be treated as a prefix
+            prefix (bool): If true `path` will be treated as a prefix
                 rather than directory path.
         """
         raise NotImplementedError
 
-    def ls(self, path_info, detail=False):
-        raise RemoteActionNotImplemented("ls", self.scheme)
-
-    def find(self, path_info, detail=False, prefix=None):
-        raise RemoteActionNotImplemented("find", self.scheme)
-
-    def is_empty(self, path_info):
+    def is_empty(self, path):
         return False
 
-    def info(self, path_info):
+    def info(self, path):
         raise NotImplementedError
 
-    def getsize(self, path_info):
-        return self.info(path_info).get("size")
+    def getsize(self, path):
+        return self.info(path).get("size")
 
-    def remove(self, path_info):
+    def remove(self, path):
         raise RemoteActionNotImplemented("remove", self.scheme)
 
-    def makedirs(self, path_info, **kwargs):
+    def makedirs(self, path, **kwargs):
         """Optional: Implement only if the remote needs to create
         directories before copying/linking/moving data
         """
@@ -223,12 +224,6 @@ class BaseFileSystem:
             return False
         return hash_.endswith(cls.CHECKSUM_DIR_SUFFIX)
 
-    @cached_property
-    def _local_fs(self):
-        from dvc.fs import LocalFileSystem
-
-        return LocalFileSystem()
-
     def upload(
         self,
         from_info,
@@ -244,11 +239,10 @@ class BaseFileSystem:
         if not hasattr(self, method):
             raise RemoteActionNotImplemented(method, self.scheme)
 
-        if to_info.scheme != self.scheme:
-            raise NotImplementedError
-
         if not is_file_obj:
-            desc = desc or from_info.name
+            from .local import localfs
+
+            desc = desc or localfs.path.name(from_info)
 
         stack = contextlib.ExitStack()
         if not callback:
@@ -260,7 +254,7 @@ class BaseFileSystem:
                 **pbar_args,
             )
             stack.enter_context(pbar)
-            callback = pbar.as_callback(self._local_fs, from_info)
+            callback = pbar.as_callback()
             if total:
                 callback.set_size(total)
 
@@ -274,9 +268,6 @@ class BaseFileSystem:
                 # pylint: disable=no-member
                 return self.upload_fobj(wrapped, to_info, size=total)
 
-            if from_info.scheme != "local":
-                raise NotImplementedError
-
             logger.debug("Uploading '%s' to '%s'", from_info, to_info)
             # pylint: disable=no-member
             return self.put_file(
@@ -294,18 +285,14 @@ class BaseFileSystem:
         _only_file=False,
         **kwargs,
     ):
+        from .local import localfs
+
         if not hasattr(self, "get_file"):
             raise RemoteActionNotImplemented("get_file", self.scheme)
 
-        if from_info.scheme != self.scheme:
-            raise NotImplementedError
-
-        if to_info.scheme != "local":
-            raise NotImplementedError
-
         download_dir = not _only_file and self.isdir(from_info)
 
-        desc = name or to_info.name
+        desc = name or localfs.path.name(to_info)
         stack = contextlib.ExitStack()
         if not callback:
             pbar_kwargs = {"unit": "Files"} if download_dir else {}
@@ -317,7 +304,7 @@ class BaseFileSystem:
                 **pbar_kwargs,
             )
             stack.enter_context(pbar)
-            callback = pbar.as_callback(self, from_info)
+            callback = pbar.as_callback()
 
         with stack:
             if download_dir:
@@ -336,13 +323,16 @@ class BaseFileSystem:
         jobs=None,
         **kwargs,
     ):
-        from_infos = list(self.walk_files(from_info, **kwargs))
+        from .local import localfs
+
+        from_infos = list(self.find(from_info, **kwargs))
         if not from_infos:
             makedirs(to_info, exist_ok=True)
             return None
 
         to_infos = (
-            to_info / info.relative_to(from_info) for info in from_infos
+            localfs.path.join(to_info, *self.path.relparts(info, from_info))
+            for info in from_infos
         )
         callback.set_size(len(from_infos))
 
@@ -375,7 +365,9 @@ class BaseFileSystem:
         to_info,
         callback=DEFAULT_CALLBACK,
     ):
-        makedirs(to_info.parent, exist_ok=True)
+        from .local import localfs
+
+        makedirs(localfs.path.parent(to_info), exist_ok=True)
         tmp_file = tmp_fname(to_info)
 
         logger.debug("Downloading '%s' to '%s'", from_info, to_info)

dvc/fs/dvc.py

@@ -3,21 +3,20 @@ import os
 import typing
 
 from dvc.exceptions import OutputNotFoundError
-from dvc.path_info import PathInfo
 from dvc.utils import relpath
 
 from ..progress import DEFAULT_CALLBACK
 from ._metadata import Metadata
-from .base import BaseFileSystem
+from .base import FileSystem
 
 if typing.TYPE_CHECKING:
     from dvc.output import Output
-
+    from dvc.types import AnyPath
 
 logger = logging.getLogger(__name__)
 
 
-class DvcFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
+class DvcFileSystem(FileSystem):  # pylint:disable=abstract-method
     """DVC repo fs.
 
     Args:
@@ -49,28 +48,25 @@ class DvcFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
 
         return outs
 
-    def _get_granular_hash(
-        self, path_info: PathInfo, out: "Output", remote=None
-    ):
-        assert isinstance(path_info, PathInfo)
+    def _get_granular_hash(self, path: "AnyPath", out: "Output", remote=None):
         # NOTE: use string paths here for performance reasons
-        key = tuple(relpath(path_info, out.path_info).split(os.sep))
+        key = tuple(relpath(path, out.fs_path).split(os.sep))
         out.get_dir_cache(remote=remote)
         if out.obj is None:
             raise FileNotFoundError
-        obj = out.obj.trie.get(key)
-        if obj:
-            return obj.hash_info
+        (_, oid) = out.obj.trie.get(key) or (None, None)
+        if oid:
+            return oid
         raise FileNotFoundError
 
-    def _get_fs_path(self, path: PathInfo, remote=None):
+    def _get_fs_path(self, path: "AnyPath", remote=None):
         try:
             outs = self._find_outs(path, strict=False)
         except OutputNotFoundError as exc:
             raise FileNotFoundError from exc
 
         if len(outs) != 1 or (
-            outs[0].is_dir_checksum and path == outs[0].path_info
+            outs[0].is_dir_checksum and path == outs[0].fs_path
         ):
             raise IsADirectoryError
 
@@ -90,18 +86,20 @@ class DvcFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
                 checksum = self._get_granular_hash(path, out).value
             else:
                 checksum = out.hash_info.value
-            remote_info = remote_odb.hash_to_path_info(checksum)
-            return remote_odb.fs, remote_info
+            remote_fs_path = remote_odb.hash_to_path(checksum)
+            return remote_odb.fs, remote_fs_path
 
         if out.is_dir_checksum:
             checksum = self._get_granular_hash(path, out).value
-            cache_path = out.odb.hash_to_path_info(checksum).url
+            cache_path = out.odb.fs.unstrip_protocol(
+                out.odb.hash_to_path(checksum)
+            )
         else:
             cache_path = out.cache_path
         return out.odb.fs, cache_path
 
     def open(  # type: ignore
-        self, path: PathInfo, mode="r", encoding=None, **kwargs
+        self, path: str, mode="r", encoding=None, **kwargs
     ):  # pylint: disable=arguments-renamed
         fs, fspath = self._get_fs_path(path, **kwargs)
         return fs.open(fspath, mode=mode, encoding=encoding)
@@ -120,18 +118,18 @@ class DvcFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
         except FileNotFoundError:
             return False
 
-    def check_isdir(self, path_info, outs):
+    def check_isdir(self, path, outs):
         if len(outs) != 1:
             return True
 
         out = outs[0]
         if not out.is_dir_checksum:
-            return out.path_info != path_info
-        if out.path_info == path_info:
+            return out.fs_path != path
+        if out.fs_path == path:
             return True
 
         try:
-            self._get_granular_hash(path_info, out)
+            self._get_granular_hash(path, out)
             return False
         except FileNotFoundError:
             return True
@@ -153,42 +151,46 @@ class DvcFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
     def _add_dir(self, trie, out, **kwargs):
         self._fetch_dir(out, **kwargs)
 
-        base = out.path_info.parts
-        for key, _ in out.obj:  # noqa: B301
+        base = out.fs.path.parts(out.fs_path)
+        for key, _, _ in out.obj:  # noqa: B301
             trie[base + key] = None
 
     def _walk(self, root, trie, topdown=True, **kwargs):
         dirs = set()
         files = []
 
-        out = trie.get(root.parts)
+        root_parts = self.path.parts(root)
+        out = trie.get(root_parts)
         if out and out.is_dir_checksum:
             self._add_dir(trie, out, **kwargs)
 
-        root_len = len(root.parts)
-        for key, out in trie.iteritems(prefix=root.parts):  # noqa: B301
-            if key == root.parts:
-                continue
+        root_len = len(root_parts)
+        try:
+            for key, out in trie.iteritems(prefix=root_parts):  # noqa: B301
+                if key == root_parts:
+                    continue
 
-            name = key[root_len]
-            if len(key) > root_len + 1 or (out and out.is_dir_checksum):
-                dirs.add(name)
-                continue
+                name = key[root_len]
+                if len(key) > root_len + 1 or (out and out.is_dir_checksum):
+                    dirs.add(name)
+                    continue
 
-            files.append(name)
+                files.append(name)
+        except KeyError:
+            pass
 
         assert topdown
         dirs = list(dirs)
-        yield root.fspath, dirs, files
+        yield root, dirs, files
 
         for dname in dirs:
-            yield from self._walk(root / dname, trie)
+            yield from self._walk(self.path.join(root, dname), trie)
 
     def walk(self, top, topdown=True, onerror=None, **kwargs):
         from pygtrie import Trie
 
         assert topdown
-        root = PathInfo(os.path.abspath(top))
+        root = os.path.abspath(top)
         try:
             meta = self.metadata(root)
         except FileNotFoundError:
@@ -203,18 +205,18 @@ class DvcFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
 
         trie = Trie()
         for out in meta.outs:
-            trie[out.path_info.parts] = out
+            trie[out.fs.path.parts(out.fs_path)] = out
 
-            if out.is_dir_checksum and root.isin_or_eq(out.path_info):
+            if out.is_dir_checksum and self.path.isin_or_eq(root, out.fs_path):
                 self._add_dir(trie, out, **kwargs)
 
         yield from self._walk(root, trie, topdown=topdown, **kwargs)
 
-    def walk_files(self, path_info, **kwargs):
-        for root, _, files in self.walk(path_info):
+    def find(self, path, prefix=None):
+        for root, _, files in self.walk(path):
             for fname in files:
                 # NOTE: os.path.join is ~5.5 times slower
-                yield PathInfo(f"{root}{os.sep}{fname}")
+                yield f"{root}{os.sep}{fname}"
 
     def isdvc(self, path, recursive=False, strict=True):
         try:
@@ -225,35 +227,35 @@ class DvcFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
         recurse = recursive or not strict
         return meta.output_exists if recurse else meta.is_output
 
-    def isexec(self, path_info):  # pylint: disable=unused-argument
+    def isexec(self, path):  # pylint: disable=unused-argument
         return False
 
-    def metadata(self, path_info):
-        path_info = PathInfo(os.path.abspath(path_info))
+    def metadata(self, fs_path):
+        abspath = os.path.abspath(fs_path)
 
         try:
-            outs = self._find_outs(path_info, strict=False, recursive=True)
+            outs = self._find_outs(abspath, strict=False, recursive=True)
         except OutputNotFoundError as exc:
             raise FileNotFoundError from exc
 
-        meta = Metadata(path_info=path_info, outs=outs, repo=self.repo)
-        meta.isdir = meta.isdir or self.check_isdir(meta.path_info, meta.outs)
+        meta = Metadata(fs_path=abspath, outs=outs, repo=self.repo)
+        meta.isdir = meta.isdir or self.check_isdir(meta.fs_path, meta.outs)
         return meta
 
-    def info(self, path_info):
-        meta = self.metadata(path_info)
+    def info(self, path):
+        meta = self.metadata(path)
         ret = {"type": "directory" if meta.isdir else "file"}
         if meta.is_output and len(meta.outs) == 1 and meta.outs[0].hash_info:
-            hash_info = meta.outs[0].hash_info
-            ret["size"] = hash_info.size
-            ret[hash_info.name] = hash_info.value
+            out = meta.outs[0]
+            ret["size"] = out.meta.size
+            ret[out.hash_info.name] = out.hash_info.value
         elif meta.part_of_output:
             (out,) = meta.outs
-            key = path_info.relative_to(out.path_info).parts
-            obj = out.obj.trie.get(key)
-            if obj:
-                ret["size"] = obj.size
-                ret[obj.hash_info.name] = obj.hash_info.value
+            key = self.path.parts(self.path.relpath(path, out.fs_path))
+            (obj_meta, oid) = out.obj.trie.get(key) or (None, None)
+            if oid:
+                ret["size"] = obj_meta.size if obj_meta else 0
+                ret[oid.name] = oid.value
 
         return ret
 
@@ -265,8 +267,8 @@ class DvcFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
             path, to_file, callback=callback, **kwargs
         )
 
-    def checksum(self, path_info):
-        info = self.info(path_info)
+    def checksum(self, path):
+        info = self.info(path)
         md5 = info.get("md5")
         if md5:
             return md5

dvc/fs/fsspec_wrapper.py

@@ -1,18 +1,28 @@
 import os
 import shutil
-from functools import lru_cache
+from typing import IO, TYPE_CHECKING, Any, Dict, Iterator, Optional, overload
 
 from funcy import cached_property
 from tqdm.utils import CallbackIOWrapper
 
 from dvc.progress import DEFAULT_CALLBACK
 
-from .base import BaseFileSystem
-from .local import LocalFileSystem
+from .base import FileSystem
+
+FSPath = str
+AnyFSPath = str
+
+if TYPE_CHECKING:
+    from typing_extensions import Literal
+
+
+# An info() entry, might evolve to a TypedDict
+# in the future (e.g for properly type 'size' etc).
+Entry = Dict[str, Any]
 
 
 # pylint: disable=no-member
-class FSSpecWrapper(BaseFileSystem):
+class FSSpecWrapper(FileSystem):
     TRAVERSE_PREFIX_LEN = 2
 
     def __init__(self, **kwargs):
@@ -20,18 +30,8 @@ class FSSpecWrapper(BaseFileSystem):
         self.fs_args = {"skip_instance_cache": True}
         self.fs_args.update(self._prepare_credentials(**kwargs))
 
-    @cached_property
-    def fs(self):
-        raise NotImplementedError
-
-    def _with_bucket(self, path):
-        return str(path)
-
-    def _strip_bucket(self, entry):
-        return entry
-
     @staticmethod
-    def _get_kwargs_from_urls(urlpath):
+    def _get_kwargs_from_urls(urlpath: str) -> "Dict[str, Any]":
         from fsspec.utils import infer_storage_options
 
         options = infer_storage_options(urlpath)
@@ -39,139 +39,117 @@ class FSSpecWrapper(BaseFileSystem):
         options.pop("protocol", None)
         return options
 
-    def _strip_buckets(self, entries, detail=False):
-        for entry in entries:
-            if detail:
-                entry = self._entry_hook(entry.copy())
-                entry["name"] = self._strip_bucket(entry["name"])
-            else:
-                entry = self._strip_bucket(entry)
-            yield entry
-
-    def _entry_hook(self, entry):
-        """Simple hook method to be overridden when wanted to process
-        entries within info() and ls(detail=True) calls"""
-        return entry
+    @cached_property
+    def fs(self):
+        raise NotImplementedError
 
     def _prepare_credentials(
-        self, **config
-    ):  # pylint: disable=unused-argument
+        self, **config: Dict[str, Any]  # pylint: disable=unused-argument
+    ) -> Dict[str, Any]:
         """Prepare the arguments for authentication to the
         host filesystem"""
         return {}
 
-    def _isdir(self, path_info):
-        return self.fs.isdir(self._with_bucket(path_info))
+    def _isdir(self, path: AnyFSPath) -> bool:
+        return self.fs.isdir(path)
 
-    def isdir(self, path_info):
+    def isdir(self, path: AnyFSPath) -> bool:
         try:
-            return self._isdir(path_info)
+            return self._isdir(path)
         except FileNotFoundError:
             return False
 
-    def isfile(self, path_info):
+    def isfile(self, path: AnyFSPath) -> bool:
         try:
-            return not self._isdir(path_info)
+            return not self._isdir(path)
         except FileNotFoundError:
             return False
 
-    def is_empty(self, path_info):
-        entry = self.info(path_info)
+    def is_empty(self, path: AnyFSPath) -> bool:
+        entry = self.info(path)
         if entry["type"] == "directory":
-            return not self.fs.ls(self._with_bucket(path_info))
+            return not self.fs.ls(path)
         return entry["size"] == 0
 
     def open(
-        self, path_info, mode="r", **kwargs
-    ):  # pylint: disable=arguments-differ
-        return self.fs.open(self._with_bucket(path_info), mode=mode)
+        self,
+        path: AnyFSPath,
+        mode: str = "r",
+        encoding: Optional[str] = None,
+        **kwargs,
+    ) -> "IO":  # pylint: disable=arguments-differ
+        return self.fs.open(path, mode=mode)
 
-    def checksum(self, path_info):
-        return self.fs.checksum(self._with_bucket(path_info))
+    def checksum(self, path: AnyFSPath) -> str:
+        return self.fs.checksum(path)
 
-    def copy(self, from_info, to_info):
-        self.makedirs(to_info.parent)
-        self.fs.copy(self._with_bucket(from_info), self._with_bucket(to_info))
+    def copy(self, from_info: AnyFSPath, to_info: AnyFSPath) -> None:
+        self.makedirs(self.path.parent(to_info))
+        self.fs.copy(from_info, to_info)
 
-    def exists(self, path_info) -> bool:
-        return self.fs.exists(self._with_bucket(path_info))
+    def exists(self, path: AnyFSPath) -> bool:
+        return self.fs.exists(path)
 
-    def ls(self, path_info, detail=False):
-        path = self._with_bucket(path_info)
-        files = self.fs.ls(path, detail=detail)
-        yield from self._strip_buckets(files, detail=detail)
+    @overload
+    def ls(
+        self, path: AnyFSPath, detail: "Literal[True]"
+    ) -> "Iterator[Entry]":
+        ...
 
-    # pylint: disable=unused-argument
-    def find(self, path_info, detail=False, prefix=None):
-        path = self._with_bucket(path_info)
-        files = self.fs.find(path, detail=detail)
-        if detail:
-            files = files.values()
+    @overload
+    def ls(self, path: AnyFSPath, detail: "Literal[False]") -> Iterator[str]:
+        ...
 
-        yield from self._strip_buckets(files, detail=detail)
+    def ls(self, path, detail=False):
+        yield from self.fs.ls(path, detail=detail)
 
-    def walk_files(self, path_info, **kwargs):
-        for file in self.find(path_info, **kwargs):
-            yield path_info.replace(path=file)
+    def find(self, path, prefix=None):
+        yield from self.fs.find(path)
 
-    def move(self, from_info, to_info):
-        self.fs.move(self._with_bucket(from_info), self._with_bucket(to_info))
+    def move(self, from_info: AnyFSPath, to_info: AnyFSPath) -> None:
+        self.fs.move(from_info, to_info)
 
-    def remove(self, path_info):
-        self.fs.rm_file(self._with_bucket(path_info))
+    def remove(self, path: AnyFSPath) -> None:
+        self.fs.rm_file(path)
 
-    def info(self, path_info):
-        info = self.fs.info(self._with_bucket(path_info)).copy()
-        info = self._entry_hook(info)
-        info["name"] = self._strip_bucket(info["name"])
-        return info
+    def info(self, path: AnyFSPath) -> "Entry":
+        return self.fs.info(path)
 
-    def makedirs(self, path_info, **kwargs):
-        self.fs.makedirs(
-            self._with_bucket(path_info), exist_ok=kwargs.pop("exist_ok", True)
-        )
+    def makedirs(self, path: AnyFSPath, **kwargs) -> None:
+        self.fs.makedirs(path, exist_ok=kwargs.pop("exist_ok", True))
 
     def put_file(
-        self, from_file, to_info, callback=DEFAULT_CALLBACK, **kwargs
-    ):
-        self.fs.put_file(
-            from_file, self._with_bucket(to_info), callback=callback, **kwargs
-        )
-        self.fs.invalidate_cache(self._with_bucket(to_info.parent))
+        self,
+        from_file: AnyFSPath,
+        to_info: AnyFSPath,
+        callback: Any = DEFAULT_CALLBACK,
+        **kwargs,
+    ) -> None:
+        self.fs.put_file(from_file, to_info, callback=callback, **kwargs)
+        self.fs.invalidate_cache(self.path.parent(to_info))
 
     def get_file(
-        self, from_info, to_info, callback=DEFAULT_CALLBACK, **kwargs
-    ):
-        self.fs.get_file(
-            self._with_bucket(from_info), to_info, callback=callback, **kwargs
-        )
+        self,
+        from_info: AnyFSPath,
+        to_info: AnyFSPath,
+        callback: Any = DEFAULT_CALLBACK,
+        **kwargs,
+    ) -> None:
+        self.fs.get_file(from_info, to_info, callback=callback, **kwargs)
 
-    def upload_fobj(self, fobj, to_info, **kwargs):
-        self.makedirs(to_info.parent)
+    def upload_fobj(self, fobj: IO, to_info: AnyFSPath, **kwargs) -> None:
+        self.makedirs(self.path.parent(to_info))
         with self.open(to_info, "wb") as fdest:
-            shutil.copyfileobj(fobj, fdest, length=fdest.blocksize)
+            shutil.copyfileobj(
+                fobj, fdest, length=getattr(fdest, "blocksize", None)
+            )
 
 
 # pylint: disable=abstract-method
 class ObjectFSWrapper(FSSpecWrapper):
     TRAVERSE_PREFIX_LEN = 3
 
-    @lru_cache(512)
-    def _with_bucket(self, path):
-        if isinstance(path, self.PATH_CLS):
-            return f"{path.bucket}/{path.path}"
-        return path
-
-    def _strip_bucket(self, entry):
-        try:
-            bucket, path = entry.split("/", 1)
-        except ValueError:
-            # If there is no path attached, only returns
-            # the bucket (top-level).
-            bucket, path = entry, None
-        return path or bucket
-
-    def makedirs(self, path_info, **kwargs):
+    def makedirs(self, path: AnyFSPath, **kwargs) -> None:
         # For object storages make this method a no-op. The original
         # fs.makedirs() method will only check if the bucket exists
         # and create if it doesn't though we don't want to support
@@ -179,42 +157,37 @@ class ObjectFSWrapper(FSSpecWrapper):
         # simply ignore all mkdir()/makedirs() calls.
         return None
 
-    def _isdir(self, path_info):
+    def _isdir(self, path: AnyFSPath) -> bool:
         # Directory in object storages are interpreted differently
         # among different fsspec providers, so this logic is a temporary
         # measure for us to adapt as of now. It checks whether it is a
         # directory (as in a prefix with contents) or whether it is an empty
         # file where it's name ends with a forward slash
 
-        entry = self.info(path_info)
+        entry = self.info(path)
         return entry["type"] == "directory" or (
             entry["size"] == 0
             and entry["type"] == "file"
             and entry["name"].endswith("/")
         )
 
-    def find(self, path_info, detail=False, prefix=None):
+    def find(self, path, prefix=None):
         if prefix:
-            path = self._with_bucket(path_info.parent)
-            files = self.fs.find(
-                path, detail=detail, prefix=path_info.parts[-1]
-            )
+            with_prefix = self.path.parent(path)
+            files = self.fs.find(with_prefix, prefix=self.path.parts(path)[-1])
         else:
-            path = self._with_bucket(path_info)
-            files = self.fs.find(path, detail=detail)
-
-        if detail:
-            files = list(files.values())
+            with_prefix = path
+            files = self.fs.find(path)
 
         # When calling find() on a file, it returns the same file in a list.
         # For object-based storages, the same behavior applies to empty
         # directories since they are represented as files. This condition
         # checks whether we should yield an empty list (if it is an empty
         # directory) or just yield the file itself.
-        if len(files) == 1 and files[0] == path and self.isdir(path_info):
+        if len(files) == 1 and files[0] == with_prefix and self.isdir(path):
             return None
 
-        yield from self._strip_buckets(files, detail=detail)
+        yield from files
 
 
 # pylint: disable=arguments-differ
@@ -231,16 +204,10 @@ class NoDirectoriesMixin:
     def walk(self, *args, **kwargs):
         raise NotImplementedError
 
-    def walk_files(self, *args, **kwargs):
-        raise NotImplementedError
-
     def ls(self, *args, **kwargs):
         raise NotImplementedError
 
 
-_LOCAL_FS = LocalFileSystem()
-
-
 class CallbackMixin:
     """Provides callback support for the filesystem that don't support yet."""
 
@@ -253,13 +220,13 @@ class CallbackMixin:
     ):
         """Add compatibility support for Callback."""
         # pylint: disable=protected-access
-        self.makedirs(to_info.parent)
+        self.makedirs(self.path.parent(to_info))
         size = os.path.getsize(from_file)
         with open(from_file, "rb") as fobj:
             callback.set_size(size)
             wrapped = CallbackIOWrapper(callback.relative_update, fobj)
             self.upload_fobj(wrapped, to_info)
-            self.fs.invalidate_cache(self._with_bucket(to_info.parent))
+            self.fs.invalidate_cache(self.path.parent(to_info))
 
     def get_file(
         self,

dvc/fs/gdrive.py

@@ -1,14 +1,10 @@
 import logging
 import os
-import posixpath
-import re
 import threading
-from urllib.parse import urlparse
 
-from funcy import cached_property, retry, wrap_prop
+from funcy import cached_property, wrap_prop
 
 from dvc.exceptions import DvcException
-from dvc.path_info import CloudURLInfo
 from dvc.scheme import Schemes
 from dvc.utils import format_link, tmp_fname
 
@@ -34,54 +30,8 @@ class GDriveAuthError(DvcException):
         super().__init__(message)
 
 
-def _gdrive_retry(func):
-    def should_retry(exc):
-        from pydrive2.files import ApiRequestError
-
-        if not isinstance(exc, ApiRequestError):
-            return False
-
-        error_code = exc.error.get("code", 0)
-        result = False
-        if 500 <= error_code < 600:
-            result = True
-
-        if error_code == 403:
-            result = exc.GetField("reason") in [
-                "userRateLimitExceeded",
-                "rateLimitExceeded",
-            ]
-        if result:
-            logger.debug(f"Retrying GDrive API call, error: {exc}.")
-
-        return result
-
-    # 16 tries, start at 0.5s, multiply by golden ratio, cap at 20s
-    return retry(
-        16,
-        timeout=lambda a: min(0.5 * 1.618 ** a, 20),
-        filter_errors=should_retry,
-    )(func)
-
-
-class GDriveURLInfo(CloudURLInfo):
-    def __init__(self, url):
-        super().__init__(url)
-
-        # GDrive URL host part is case sensitive,
-        # we are restoring it here.
-        p = urlparse(url)
-        self.host = p.netloc
-        assert self.netloc == self.host
-
-        # Normalize path. Important since we have a cache (path to ID)
-        # and don't want to deal with different variations of path in it.
-        self._spath = re.sub("/{2,}", "/", self._spath.rstrip("/"))
-
-
 class GDriveFileSystem(FSSpecWrapper):  # pylint:disable=abstract-method
     scheme = Schemes.GDRIVE
-    PATH_CLS = GDriveURLInfo
     PARAM_CHECKSUM = "checksum"
     REQUIRES = {"pydrive2": "pydrive2"}
     # Always prefer traverse for GDrive since API usage quotas are a concern.
@@ -95,11 +45,14 @@ class GDriveFileSystem(FSSpecWrapper):  # pylint:disable=abstract-method
     DEFAULT_GDRIVE_CLIENT_SECRET = "a1Fz59uTpVNeG_VGuSKDLJXv"
 
     def __init__(self, **config):
+        from fsspec.utils import infer_storage_options
+
         super().__init__(**config)
 
-        self.path_info = self.PATH_CLS(config["url"])
+        self.url = config["url"]
+        opts = infer_storage_options(self.url)
 
-        if not self.path_info.bucket:
+        if not opts["host"]:
             raise DvcException(
                 "Empty GDrive URL '{}'. Learn more at {}".format(
                     config["url"],
@@ -107,8 +60,8 @@ class GDriveFileSystem(FSSpecWrapper):  # pylint:disable=abstract-method
                 )
             )
 
-        self._bucket = self.path_info.bucket
-        self._path = self.path_info.path
+        self._bucket = opts["host"]
+        self._path = opts["path"].lstrip("/")
         self._trash_only = config.get("gdrive_trash_only")
         self._use_service_account = config.get("gdrive_use_service_account")
         self._service_account_user_email = config.get(
@@ -136,6 +89,17 @@ class GDriveFileSystem(FSSpecWrapper):  # pylint:disable=abstract-method
             )
         )
 
+    @classmethod
+    def _strip_protocol(cls, path):
+        from fsspec.utils import infer_storage_options
+
+        opts = infer_storage_options(path)
+
+        return "{host}{path}".format(**opts)
+
+    def unstrip_protocol(self, path):
+        return f"gdrive://{path}"
+
     @staticmethod
     def _get_kwargs_from_urls(urlpath):
         return {"url": urlpath}
@@ -219,7 +183,7 @@ class GDriveFileSystem(FSSpecWrapper):  # pylint:disable=abstract-method
             temporary_save_path = self._gdrive_service_credentials_path
 
         if is_credentials_temp:
-            with open(temporary_save_path, "w") as cred_file:
+            with open(temporary_save_path, "w", encoding="utf-8") as cred_file:
                 cred_file.write(
                     os.getenv(GDriveFileSystem.GDRIVE_CREDENTIALS_DATA)
                 )
@@ -285,27 +249,11 @@ class GDriveFileSystem(FSSpecWrapper):  # pylint:disable=abstract-method
                 os.remove(temporary_save_path)
 
         return _GDriveFileSystem(
-            self._with_bucket(self.path_info),
+            f"{self._bucket}/{self._path}",
             gauth,
             trash_only=self._trash_only,
         )
 
-    def _with_bucket(self, path):
-        if isinstance(path, self.PATH_CLS):
-            return posixpath.join(path.bucket, path.path)
-
-        return super()._with_bucket(path)
-
-    def _strip_bucket(self, entry):
-        try:
-            bucket, path = entry.split("/", 1)
-        except ValueError:
-            # If there is no path attached, only returns
-            # the bucket (top-level).
-            bucket, path = entry, None
-        return path or bucket
-
     def upload_fobj(self, fobj, to_info, **kwargs):
-        rpath = self._with_bucket(to_info)
-        self.makedirs(os.path.dirname(rpath))
-        return self.fs.upload_fobj(fobj, rpath, **kwargs)
+        self.makedirs(os.path.dirname(to_info))
+        return self.fs.upload_fobj(fobj, to_info, **kwargs)

dvc/fs/git.py

@@ -1,136 +1,72 @@
-import errno
 import os
+import threading
+from typing import TYPE_CHECKING, Any, Callable
 
-from tqdm.utils import CallbackIOWrapper
+from funcy import cached_property, wrap_prop
 
-from dvc.utils import is_exec, relpath
+from .fsspec_wrapper import AnyFSPath, FSSpecWrapper
 
-from ..progress import DEFAULT_CALLBACK
-from .base import BaseFileSystem
+if TYPE_CHECKING:
+    from scmrepo.fs import GitFileSystem as FsspecGitFileSystem
+    from scmrepo.git import Git
+    from scmrepo.git.objects import GitTrie
 
 
-class GitFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
+class GitFileSystem(FSSpecWrapper):  # pylint:disable=abstract-method
     """Proxies the repo file access methods to Git objects"""
 
     sep = os.sep
-
     scheme = "local"
 
-    def __init__(self, root_dir, trie):
+    def __init__(
+        self,
+        path: str = None,
+        rev: str = None,
+        scm: "Git" = None,
+        trie: "GitTrie" = None,
+        **kwargs: Any,
+    ) -> None:
+        from dvc.scm import resolve_rev
+
         super().__init__()
-        self._root = root_dir
-        self.trie = trie
+        self.fs_args.update(
+            {
+                "path": path,
+                "rev": rev,
+                "scm": scm,
+                "trie": trie,
+                "rev_resolver": resolve_rev,
+                **kwargs,
+            }
+        )
+
+    @wrap_prop(threading.Lock())
+    @cached_property
+    def fs(self) -> "FsspecGitFileSystem":
+        from scmrepo.fs import GitFileSystem as FsspecGitFileSystem
+
+        return FsspecGitFileSystem(**self.fs_args)
 
     @property
-    def rev(self):
-        return self.trie.rev
+    def rev(self) -> str:
+        return self.fs.rev
 
-    def _get_key(self, path):
-        if isinstance(path, str):
-            if not os.path.isabs(path):
-                relparts = path.split(os.sep)
-            else:
-                relparts = relpath(path, self._root).split(os.sep)
-        else:
-            relparts = path.relative_to(self._root).parts
-        if relparts == ["."]:
-            return ()
-        return tuple(relparts)
+    def isexec(self, path: AnyFSPath) -> bool:
+        from dvc.utils import is_exec
 
-    def open(
-        self, path, mode="r", encoding=None, **kwargs
-    ):  # pylint: disable=arguments-renamed
-        # NOTE: this is incorrect, we need to use locale to determine default
-        # encoding.
-        encoding = encoding or "utf-8"
-
-        key = self._get_key(path)
         try:
-            return self.trie.open(key, mode=mode, encoding=encoding)
-        except KeyError as exc:
-            msg = os.strerror(errno.ENOENT) + f"in branch '{self.rev}'"
-            raise FileNotFoundError(errno.ENOENT, msg, path) from exc
-        except IsADirectoryError as exc:
-            raise IsADirectoryError(
-                errno.EISDIR, os.strerror(errno.EISDIR), path
-            ) from exc
-
-    def exists(self, path_info) -> bool:
-        key = self._get_key(path_info)
-        return self.trie.exists(key)
-
-    def isdir(self, path_info) -> bool:
-        key = self._get_key(path_info)
-        return self.trie.isdir(key)
-
-    def isfile(self, path_info) -> bool:
-        key = self._get_key(path_info)
-        return self.trie.isfile(key)
-
-    def walk(self, top, topdown=True, onerror=None, **kwargs):
-        """Directory tree generator.
-
-        See `os.walk` for the docs. Differences:
-        - no support for symlinks
-        """
-        if not self.isdir(top):
-            if onerror:
-                if self.exists(top):
-                    exc = NotADirectoryError(
-                        errno.ENOTDIR, os.strerror(errno.ENOTDIR), top
-                    )
-                else:
-                    exc = FileNotFoundError(
-                        errno.ENOENT, os.strerror(errno.ENOENT), top
-                    )
-                onerror(exc)
-            return []
-
-        key = self._get_key(top)
-        for prefix, dirs, files in self.trie.walk(key, topdown=topdown):
-            if prefix:
-                root = os.path.join(self._root, os.sep.join(prefix))
-            else:
-                root = self._root
-            yield root, dirs, files
-
-    def isexec(self, path_info):
-        if not self.exists(path_info):
+            return is_exec(self.info(path)["mode"])
+        except FileNotFoundError:
             return False
 
-        mode = self.info(path_info)["mode"]
-        return is_exec(mode)
+    def isfile(self, path: AnyFSPath) -> bool:
+        return self.fs.isfile(path)
 
-    def info(self, path_info):
-        key = self._get_key(path_info)
-        try:
-            return self.trie.info(key)
-        except KeyError:
-            raise FileNotFoundError(
-                errno.ENOENT, os.strerror(errno.ENOENT), path_info
-            )
-
-    def checksum(self, path_info):
-        return self.info(path_info)["sha"]
-
-    def walk_files(self, path_info, **kwargs):
-        for root, _, files in self.walk(path_info, **kwargs):
-            for file in files:
-                # NOTE: os.path.join is ~5.5 times slower
-                yield f"{root}{os.sep}{file}"
-
-    def get_file(
-        self, from_info, to_file, callback=DEFAULT_CALLBACK, **kwargs
+    def walk(
+        self,
+        top: AnyFSPath,
+        topdown: bool = True,
+        onerror: Callable[[OSError], None] = None,
+        **kwargs: Any,
     ):
-        import shutil
-
-        total = self.getsize(from_info)
-        if total:
-            callback.set_size(total)
-
-        with self.open(from_info, "rb", **kwargs) as from_fobj:
-            with open(to_file, "wb+") as to_fobj:
-                wrapped = CallbackIOWrapper(
-                    callback.relative_update, from_fobj
-                )
-                shutil.copyfileobj(wrapped, to_fobj)
+        return self.fs.walk(top, topdown=topdown, onerror=onerror, **kwargs)

dvc/fs/gs.py

@@ -1,9 +1,7 @@
-import base64
 import threading
 
 from funcy import cached_property, wrap_prop
 
-from dvc.path_info import CloudURLInfo
 from dvc.scheme import Schemes
 
 # pylint:disable=abstract-method
@@ -12,11 +10,9 @@ from .fsspec_wrapper import CallbackMixin, ObjectFSWrapper
 
 class GSFileSystem(CallbackMixin, ObjectFSWrapper):
     scheme = Schemes.GS
-    PATH_CLS = CloudURLInfo
     REQUIRES = {"gcsfs": "gcsfs"}
     PARAM_CHECKSUM = "etag"
     DETAIL_FIELDS = frozenset(("etag", "size"))
-    TRAVERSE_PREFIX_LEN = 2
 
     def _prepare_credentials(self, **config):
         login_info = {"consistency": None}
@@ -24,11 +20,6 @@ class GSFileSystem(CallbackMixin, ObjectFSWrapper):
         login_info["token"] = config.get("credentialpath")
         return login_info
 
-    def _entry_hook(self, entry):
-        if "etag" in entry:
-            entry["etag"] = base64.b64decode(entry["etag"]).hex()
-        return entry
-
     @wrap_prop(threading.Lock())
     @cached_property
     def fs(self):

dvc/fs/hdfs.py

@@ -1,282 +1,99 @@
-import errno
-import io
-import logging
 import os
 import re
-import shutil
 import subprocess
-from collections import deque
-from contextlib import closing, contextmanager
+import threading
 
-from tqdm.utils import CallbackIOWrapper
+from funcy import cached_property, wrap_prop
 
-from dvc.hash_info import HashInfo
-from dvc.progress import DEFAULT_CALLBACK
 from dvc.scheme import Schemes
-from dvc.utils import fix_env, tmp_fname
+from dvc.utils import fix_env
 
-from .base import BaseFileSystem, RemoteCmdError
-from .pool import get_connection
+from .fsspec_wrapper import CallbackMixin, FSSpecWrapper
 
-logger = logging.getLogger(__name__)
+CHECKSUM_REGEX = re.compile(r".*\t.*\t(?P<checksum>.*)")
 
 
-def _hadoop_fs(cmd, user=None):
-    cmd = "hadoop fs -" + cmd
-    if user:
-        cmd = f"HADOOP_USER_NAME={user} " + cmd
-
-    # NOTE: close_fds doesn't work with redirected stdin/stdout/stderr.
-    # See https://github.com/iterative/dvc/issues/1197.
-    close_fds = os.name != "nt"
-
-    executable = os.getenv("SHELL") if os.name != "nt" else None
-    p = subprocess.Popen(
-        cmd,
-        shell=True,
-        close_fds=close_fds,
-        executable=executable,
-        env=fix_env(os.environ),
-        stdin=subprocess.PIPE,
-        stdout=subprocess.PIPE,
-        stderr=subprocess.PIPE,
-    )
-    out, err = p.communicate()
-    if p.returncode != 0:
-        raise RemoteCmdError("hdfs", cmd, p.returncode, err)
-    return out.decode("utf-8")
-
-
-def _group(regex, s, gname):
-    match = re.match(regex, s)
-    assert match is not None
-    return match.group(gname)
-
-
-def _hadoop_fs_checksum(path_info):
-    # NOTE: pyarrow doesn't support checksum, so we need to use hadoop
-    regex = r".*\t.*\t(?P<checksum>.*)"
-    stdout = _hadoop_fs(f"checksum {path_info.url}", user=path_info.user)
-    return _group(regex, stdout, "checksum")
-
-
-class HDFSFileSystem(BaseFileSystem):
+# pylint: disable=abstract-method
+class HDFSFileSystem(CallbackMixin, FSSpecWrapper):
     scheme = Schemes.HDFS
-    REQUIRES = {"pyarrow": "pyarrow"}
-    REGEX = r"^hdfs://((?P<user>.*)@)?.*$"
+    REQUIRES = {"fsspec": "fsspec", "pyarrow": "pyarrow"}
     PARAM_CHECKSUM = "checksum"
-    TRAVERSE_PREFIX_LEN = 2
 
-    BLOCK_SIZE = 2 ** 26  # 128MB
+    @classmethod
+    def _strip_protocol(cls, path: str) -> str:
+        from fsspec.utils import infer_storage_options
 
-    def __init__(self, **config):
-        super().__init__(**config)
+        return infer_storage_options(path)["path"].lstrip("/")
 
-        self.host = config["host"]
-        self.user = config.get("user")
-        self.port = config.get("port")
-
-        self.kerb_ticket = config.get("kerb_ticket")
+    def unstrip_protocol(self, path: str) -> str:
+        host = self.fs_args["host"]
+        port = self.fs_args["port"]
+        path = path.lstrip("/")
+        return f"hdfs://{host}:{port}/{path}"
 
     @staticmethod
     def _get_kwargs_from_urls(urlpath):
-        from fsspec.implementations.hdfs import PyArrowHDFS
+        from fsspec.implementations.arrow import HadoopFileSystem
 
         # pylint:disable=protected-access
-        return PyArrowHDFS._get_kwargs_from_urls(urlpath)
+        return HadoopFileSystem._get_kwargs_from_urls(urlpath)
 
-    def hdfs(self, path_info):
-        import pyarrow.fs
+    def _prepare_credentials(self, **config):
+        return config
 
-        # NOTE: HadoopFileSystem is not meant to be closed by us and doesn't
-        # have close() method or alternative, so we add a noop one to satisfy
-        # our connection pool.
-        class HDFSConnection(pyarrow.fs.HadoopFileSystem):
-            def close(self):
-                pass
+    @wrap_prop(threading.Lock())
+    @cached_property
+    def fs(self):
+        from fsspec.implementations.arrow import HadoopFileSystem
 
-        return get_connection(
-            HDFSConnection,
-            path_info.host,
-            path_info.port,
-            user=path_info.user,
-            kerb_ticket=self.kerb_ticket,
+        return HadoopFileSystem(**self.fs_args)
+
+    def checksum(self, path):
+        return self._checksum(path)
+
+    def _checksum(self, path, **kwargs):
+        # PyArrow doesn't natively support retrieving the
+        # checksum, so we have to use hadoop fs
+
+        url = self.unstrip_protocol(path)
+
+        result = self._run_command(
+            f"checksum {url}",
+            env=fix_env(os.environ),
+            user=self.fs_args["user"],
         )
+        if result is None:
+            return None
 
-    @contextmanager
-    def open(self, path_info, mode="r", encoding=None, **kwargs):
-        assert mode in {"r", "rt", "rb"}
+        match = CHECKSUM_REGEX.match(result)
+        if match is None:
+            return None
 
-        try:
-            with self.hdfs(path_info) as hdfs, closing(
-                hdfs.open_input_stream(path_info.path)
-            ) as fd:
-                if mode == "rb":
-                    yield fd
-                else:
-                    yield io.TextIOWrapper(fd, encoding=encoding)
-        except OSError as e:
-            # Empty .errno and not specific enough error class in pyarrow,
-            # see https://issues.apache.org/jira/browse/ARROW-6248
-            if "file does not exist" in str(e):
-                raise FileNotFoundError(*e.args)
-            raise
+        return match.group("checksum")
 
-    def exists(self, path_info) -> bool:
-        assert not isinstance(path_info, list)
-        assert path_info.scheme == "hdfs"
-        with self.hdfs(path_info) as hdfs:
-            import pyarrow.fs
+    def _run_command(self, cmd, env=None, user=None):
+        cmd = "hadoop fs -" + cmd
+        if user:
+            cmd = f"HADOOP_USER_NAME={user} " + cmd
 
-            file_info = hdfs.get_file_info(path_info.path)
-            return file_info.type != pyarrow.fs.FileType.NotFound
+        # NOTE: close_fds doesn't work with redirected stdin/stdout/stderr.
+        # See https://github.com/iterative/dvc/issues/1197.
+        close_fds = os.name != "nt"
 
-    def _walk(self, hdfs, root, topdown=True):
-        import posixpath
-
-        from pyarrow.fs import FileSelector, FileType
-
-        dirs = deque()
-        nondirs = deque()
-
-        selector = FileSelector(root)
-        for entry in hdfs.get_file_info(selector):
-            if entry.type == FileType.Directory:
-                dirs.append(entry.base_name)
-            else:
-                nondirs.append(entry.base_name)
-
-        if topdown:
-            yield root, dirs, nondirs
-
-        for dname in dirs:
-            # NOTE: posixpath.join() is slower
-            yield from self._walk(
-                hdfs, f"{root}{posixpath.sep}{dname}", topdown=topdown
-            )
-
-        if not topdown:
-            yield root, dirs, nondirs
-
-    def walk(self, top, topdown=True, onerror=None, **kwargs):
-        if not self.isdir(top):
-            return
-
-        with self.hdfs(top) as hdfs:
-            for root, dnames, fnames in self._walk(
-                hdfs, top.path, topdown=topdown
-            ):
-                yield top.replace(path=root), dnames, fnames
-
-    def walk_files(self, path_info, **kwargs):
-        for root, _, fnames in self.walk(path_info):
-            yield from (root / fname for fname in fnames)
-
-    def remove(self, path_info):
-        if path_info.scheme != "hdfs":
-            raise NotImplementedError
-
-        if self.exists(path_info):
-            logger.debug(f"Removing {path_info.path}")
-            with self.hdfs(path_info) as hdfs:
-                import pyarrow.fs
-
-                file_info = hdfs.get_file_info(path_info.path)
-                if file_info.type == pyarrow.fs.FileType.Directory:
-                    hdfs.delete_dir(path_info.path)
-                else:
-                    hdfs.delete_file(path_info.path)
-
-    def makedirs(self, path_info, **kwargs):
-        with self.hdfs(path_info) as hdfs:
-            # NOTE: fs.create_dir creates parents by default
-            hdfs.create_dir(path_info.path)
-
-    def copy(self, from_info, to_info, **_kwargs):
-        # NOTE: hdfs.copy_file is not supported yet in pyarrow
-        with self.hdfs(to_info) as hdfs:
-            # NOTE: this is how `hadoop fs -cp` works too: it copies through
-            # your local machine.
-            with closing(hdfs.open_input_stream(from_info.path)) as from_fobj:
-                tmp_info = to_info.parent / tmp_fname(to_info.name)
-                try:
-                    with closing(
-                        hdfs.open_output_stream(tmp_info.path)
-                    ) as tmp_fobj:
-                        shutil.copyfileobj(from_fobj, tmp_fobj)
-                    hdfs.move(tmp_info.path, to_info.path)
-                except Exception:
-                    self.remove(tmp_info)
-                    raise
-
-    def isfile(self, path_info):
-        with self.hdfs(path_info) as hdfs:
-            import pyarrow.fs
-
-            file_info = hdfs.get_file_info(path_info.path)
-            return file_info.type == pyarrow.fs.FileType.File
-
-    def isdir(self, path_info):
-        with self.hdfs(path_info) as hdfs:
-            import pyarrow.fs
-
-            file_info = hdfs.get_file_info(path_info.path)
-            return file_info.type == pyarrow.fs.FileType.Directory
-
-    def info(self, path_info):
-        with self.hdfs(path_info) as hdfs:
-            from pyarrow.fs import FileType
-
-            file_info = hdfs.get_file_info(path_info.path)
-            if file_info.type is FileType.Directory:
-                kind = "directory"
-            elif file_info.type is FileType.File:
-                kind = "file"
-            else:
-                raise FileNotFoundError(
-                    errno.ENOENT, os.strerror(errno.ENOENT), path_info
-                )
-
-            return {"size": file_info.size, "type": kind}
-
-    def checksum(self, path_info):
-        return HashInfo(
-            "checksum",
-            _hadoop_fs_checksum(path_info),
-            size=self.getsize(path_info),
+        executable = os.getenv("SHELL") if os.name != "nt" else None
+        p = subprocess.Popen(
+            cmd,
+            shell=True,
+            close_fds=close_fds,
+            executable=executable,
+            env=env or os.environ,
+            stdin=subprocess.PIPE,
+            stdout=subprocess.PIPE,
+            stderr=subprocess.PIPE,
         )
+        out, err = p.communicate()
 
-    def upload_fobj(self, fobj, to_info, **kwargs):
-        with self.hdfs(to_info) as hdfs:
-            with hdfs.open_output_stream(to_info.path) as fdest:
-                shutil.copyfileobj(fobj, fdest, self.BLOCK_SIZE)
-
-    def put_file(
-        self, from_file, to_info, callback=DEFAULT_CALLBACK, **kwargs
-    ):
-        with self.hdfs(to_info) as hdfs:
-            hdfs.create_dir(to_info.parent.path)
-
-            tmp_file = tmp_fname(to_info.path)
-            total: int = os.path.getsize(from_file)
-            callback.set_size(total)
-
-            with open(from_file, "rb") as fobj:
-                wrapped = CallbackIOWrapper(callback.relative_update, fobj)
-                with hdfs.open_output_stream(tmp_file) as sobj:
-                    shutil.copyfileobj(wrapped, sobj, self.BLOCK_SIZE)
-            hdfs.move(tmp_file, to_info.path)
-
-    def get_file(
-        self, from_info, to_file, callback=DEFAULT_CALLBACK, **kwargs
-    ):
-        with self.hdfs(from_info) as hdfs:
-            file_info = hdfs.get_file_info(from_info.path)
-            total = file_info.size
-            if total:
-                callback.set_size(total)
-
-            with hdfs.open_input_stream(from_info.path) as sobj:
-                with open(to_file, "wb+") as fobj:
-                    wrapped = CallbackIOWrapper(callback.relative_update, sobj)
-                    shutil.copyfileobj(wrapped, fobj, self.BLOCK_SIZE)
+        if p.returncode != 0:
+            raise subprocess.CalledProcessError(p.returncode, cmd, out, err)
+        else:
+            return out.decode("utf-8")

dvc/fs/http.py

@@ -1,12 +1,13 @@
 import threading
+from typing import Any
 
 from funcy import cached_property, memoize, wrap_with
 
 from dvc import prompt
-from dvc.path_info import HTTPURLInfo
+from dvc.progress import DEFAULT_CALLBACK
 from dvc.scheme import Schemes
 
-from .fsspec_wrapper import FSSpecWrapper, NoDirectoriesMixin
+from .fsspec_wrapper import AnyFSPath, FSSpecWrapper, NoDirectoriesMixin
 
 
 @wrap_with(threading.Lock())
@@ -34,7 +35,6 @@ def make_context(ssl_verify):
 # pylint: disable=abstract-method
 class HTTPFileSystem(NoDirectoriesMixin, FSSpecWrapper):
     scheme = Schemes.HTTP
-    PATH_CLS = HTTPURLInfo
     PARAM_CHECKSUM = "checksum"
     REQUIRES = {"aiohttp": "aiohttp", "aiohttp-retry": "aiohttp_retry"}
     CAN_TRAVERSE = False
@@ -87,6 +87,9 @@ class HTTPFileSystem(NoDirectoriesMixin, FSSpecWrapper):
                     ssl=make_context(config["ssl_verify"])
                 )
 
+        # Allow reading proxy configurations from the environment.
+        client_kwargs["trust_env"] = True
+
         credentials["get_client"] = self.get_client
         self.upload_method = config.get("method", "POST")
         return credentials
@@ -99,6 +102,7 @@ class HTTPFileSystem(NoDirectoriesMixin, FSSpecWrapper):
             attempts=self.SESSION_RETRIES,
             factor=self.SESSION_BACKOFF_FACTOR,
             max_timeout=self.REQUEST_TIMEOUT,
+            exceptions=[aiohttp.ClientError],
         )
 
         # The default timeout for the aiohttp is 300 seconds
@@ -123,6 +127,15 @@ class HTTPFileSystem(NoDirectoriesMixin, FSSpecWrapper):
 
         return _HTTPFileSystem(**self.fs_args)
 
-    def _entry_hook(self, entry):
-        entry["checksum"] = entry.get("ETag") or entry.get("Content-MD5")
-        return entry
+    def unstrip_protocol(self, path: str) -> str:
+        return path
+
+    def put_file(
+        self,
+        from_file: AnyFSPath,
+        to_info: AnyFSPath,
+        callback: Any = DEFAULT_CALLBACK,
+        **kwargs,
+    ) -> None:
+        kwargs.setdefault("method", self.upload_method)
+        super().put_file(from_file, to_info, callback=callback, **kwargs)

dvc/fs/local.py

@@ -1,23 +1,21 @@
 import logging
 import os
 
-from dvc.path_info import PathInfo
 from dvc.scheme import Schemes
 from dvc.system import System
 from dvc.utils import is_exec, tmp_fname
 from dvc.utils.fs import copy_fobj_to_file, copyfile, makedirs, move, remove
 
 from ..progress import DEFAULT_CALLBACK
-from .base import BaseFileSystem
+from .base import FileSystem
 
 logger = logging.getLogger(__name__)
 
 
-class LocalFileSystem(BaseFileSystem):
+class LocalFileSystem(FileSystem):
     sep = os.sep
 
     scheme = Schemes.LOCAL
-    PATH_CLS = PathInfo
     PARAM_CHECKSUM = "md5"
     PARAM_PATH = "path"
     TRAVERSE_PREFIX_LEN = 2
@@ -29,28 +27,29 @@ class LocalFileSystem(BaseFileSystem):
         self.fs = LocalFS()
 
     @staticmethod
-    def open(path_info, mode="r", encoding=None, **kwargs):
-        return open(path_info, mode=mode, encoding=encoding)
+    def open(path, mode="r", encoding=None, **kwargs):
+        return open(path, mode=mode, encoding=encoding)
 
-    def exists(self, path_info) -> bool:
-        assert isinstance(path_info, str) or path_info.scheme == "local"
+    def exists(self, path) -> bool:
         # TODO: replace this with os.path.exists once the problem is fixed on
         # the fsspec https://github.com/intake/filesystem_spec/issues/742
-        return os.path.lexists(path_info)
+        return os.path.lexists(path)
 
-    def checksum(self, path_info) -> str:
-        return self.fs.checksum(path_info)
+    def checksum(self, path) -> str:
+        from fsspec.utils import tokenize
 
-    def isfile(self, path_info) -> bool:
-        return os.path.isfile(path_info)
+        st = os.stat(path)
 
-    def isdir(self, path_info) -> bool:
-        return os.path.isdir(path_info)
+        return str(int(tokenize([st.st_ino, st.st_mtime, st.st_size]), 16))
 
-    def iscopy(self, path_info):
-        return not (
-            System.is_symlink(path_info) or System.is_hardlink(path_info)
-        )
+    def isfile(self, path) -> bool:
+        return os.path.isfile(path)
+
+    def isdir(self, path) -> bool:
+        return os.path.isdir(path)
+
+    def iscopy(self, path):
+        return not (System.is_symlink(path) or System.is_hardlink(path))
 
     def walk(self, top, topdown=True, onerror=None, **kwargs):
         """Directory fs generator.
@@ -63,45 +62,37 @@ class LocalFileSystem(BaseFileSystem):
         ):
             yield os.path.normpath(root), dirs, files
 
-    def walk_files(self, path_info, **kwargs):
-        for root, _, files in self.walk(path_info):
+    def find(self, path, prefix=None):
+        for root, _, files in self.walk(path):
             for file in files:
                 # NOTE: os.path.join is ~5.5 times slower
-                yield PathInfo(f"{root}{os.sep}{file}")
+                yield f"{root}{os.sep}{file}"
 
-    def is_empty(self, path_info):
-        if self.isfile(path_info) and os.path.getsize(path_info) == 0:
+    def is_empty(self, path):
+        if self.isfile(path) and os.path.getsize(path) == 0:
             return True
 
-        if self.isdir(path_info) and len(os.listdir(path_info)) == 0:
+        if self.isdir(path) and len(os.listdir(path)) == 0:
             return True
 
         return False
 
-    def remove(self, path_info):
-        if isinstance(path_info, PathInfo):
-            if path_info.scheme != "local":
-                raise NotImplementedError
-        remove(path_info)
+    def remove(self, path):
+        remove(path)
 
-    def makedirs(self, path_info, **kwargs):
-        makedirs(path_info, exist_ok=kwargs.pop("exist_ok", True))
+    def makedirs(self, path, **kwargs):
+        makedirs(path, exist_ok=kwargs.pop("exist_ok", True))
 
-    def isexec(self, path_info):
-        mode = self.info(path_info)["mode"]
+    def isexec(self, path):
+        mode = self.info(path)["mode"]
         return is_exec(mode)
 
     def move(self, from_info, to_info):
-        if (
-            isinstance(from_info, PathInfo) and from_info.scheme != "local"
-        ) or (isinstance(to_info, PathInfo) and to_info.scheme != "local"):
-            raise NotImplementedError
-
-        self.makedirs(to_info.parent)
+        self.makedirs(self.path.parent(to_info))
         move(from_info, to_info)
 
     def copy(self, from_info, to_info):
-        tmp_info = to_info.parent / tmp_fname("")
+        tmp_info = self.path.join(self.path.parent(to_info), tmp_fname(""))
         try:
             copyfile(from_info, tmp_info)
             os.rename(tmp_info, to_info)
@@ -110,8 +101,8 @@ class LocalFileSystem(BaseFileSystem):
             raise
 
     def upload_fobj(self, fobj, to_info, **kwargs):
-        self.makedirs(to_info.parent)
-        tmp_info = to_info.parent / tmp_fname("")
+        self.makedirs(self.path.parent(to_info))
+        tmp_info = self.path.join(self.path.parent(to_info), tmp_fname(""))
         try:
             copy_fobj_to_file(fobj, tmp_info)
             os.rename(tmp_info, to_info)
@@ -124,8 +115,8 @@ class LocalFileSystem(BaseFileSystem):
         System.symlink(from_info, to_info)
 
     @staticmethod
-    def is_symlink(path_info):
-        return System.is_symlink(path_info)
+    def is_symlink(path):
+        return System.is_symlink(path)
 
     def hardlink(self, from_info, to_info):
         # If there are a lot of empty files (which happens a lot in datasets),
@@ -150,20 +141,21 @@ class LocalFileSystem(BaseFileSystem):
         System.hardlink(from_info, to_info)
 
     @staticmethod
-    def is_hardlink(path_info):
-        return System.is_hardlink(path_info)
+    def is_hardlink(path):
+        return System.is_hardlink(path)
 
     def reflink(self, from_info, to_info):
         System.reflink(from_info, to_info)
 
-    def info(self, path_info):
-        return self.fs.info(path_info)
+    def info(self, path):
+        return self.fs.info(path)
 
     def put_file(
         self, from_file, to_info, callback=DEFAULT_CALLBACK, **kwargs
     ):
-        makedirs(to_info.parent, exist_ok=True)
-        tmp_file = tmp_fname(to_info)
+        parent = self.path.parent(to_info)
+        makedirs(parent, exist_ok=True)
+        tmp_file = self.path.join(parent, tmp_fname())
         copyfile(from_file, tmp_file, callback=callback)
         os.replace(tmp_file, to_info)
 
@@ -171,3 +163,6 @@ class LocalFileSystem(BaseFileSystem):
         self, from_info, to_file, callback=DEFAULT_CALLBACK, **kwargs
     ):
         copyfile(from_info, to_file, callback=callback)
+
+
+localfs = LocalFileSystem()

dvc/fs/memory.py

@@ -2,7 +2,6 @@ import threading
 
 from funcy import cached_property, wrap_prop
 
-from dvc.path_info import CloudURLInfo
 from dvc.scheme import Schemes
 
 from .fsspec_wrapper import FSSpecWrapper
@@ -11,7 +10,6 @@ from .fsspec_wrapper import FSSpecWrapper
 class MemoryFileSystem(FSSpecWrapper):  # pylint:disable=abstract-method
     scheme = Schemes.MEMORY
     PARAM_CHECKSUM = "md5"
-    PATH_CLS = CloudURLInfo
     TRAVERSE_PREFIX_LEN = 2
     DEFAULT_BLOCKSIZE = 4096
 

dvc/fs/oss.py

@@ -4,7 +4,6 @@ import threading
 
 from funcy import cached_property, wrap_prop
 
-from dvc.path_info import CloudURLInfo
 from dvc.scheme import Schemes
 
 from .fsspec_wrapper import ObjectFSWrapper
@@ -15,7 +14,6 @@ logger = logging.getLogger(__name__)
 # pylint:disable=abstract-method
 class OSSFileSystem(ObjectFSWrapper):
     scheme = Schemes.OSS
-    PATH_CLS = CloudURLInfo
     REQUIRES = {"ossfs": "ossfs"}
     PARAM_CHECKSUM = "etag"
     COPY_POLL_SECONDS = 5

dvc/fs/path.py

@@ -0,0 +1,94 @@
+import ntpath
+import posixpath
+
+
+class Path:
+    def __init__(self, sep):
+        if sep == posixpath.sep:
+            self.flavour = posixpath
+        elif sep == ntpath.sep:
+            self.flavour = ntpath
+        else:
+            raise ValueError(f"unsupported separator '{sep}'")
+
+    def join(self, *parts):
+        return self.flavour.join(*parts)
+
+    def parts(self, path):
+        drive, path = self.flavour.splitdrive(path)
+
+        ret = []
+        while True:
+            path, part = self.flavour.split(path)
+
+            if part:
+                ret.append(part)
+                continue
+
+            if path:
+                ret.append(path)
+
+            break
+
+        ret.reverse()
+
+        if drive:
+            ret = [drive] + ret
+
+        return tuple(ret)
+
+    def parent(self, path):
+        return self.flavour.dirname(path)
+
+    def parents(self, path):
+        parts = self.parts(path)
+        return tuple(
+            self.join(*parts[:length])
+            for length in range(len(parts) - 1, 0, -1)
+        )
+
+    def name(self, path):
+        return self.parts(path)[-1]
+
+    def suffix(self, path):
+        name = self.name(path)
+        _, dot, suffix = name.partition(".")
+        return dot + suffix
+
+    def with_name(self, path, name):
+        parts = list(self.parts(path))
+        parts[-1] = name
+        return self.join(*parts)
+
+    def with_suffix(self, path, suffix):
+        parts = list(self.parts(path))
+        real_path, _, _ = parts[-1].partition(".")
+        parts[-1] = real_path + suffix
+        return self.join(*parts)
+
+    def isin(self, left, right):
+        left_parts = self.parts(left)
+        right_parts = self.parts(right)
+        left_len = len(left_parts)
+        right_len = len(right_parts)
+        return left_len > right_len and left_parts[:right_len] == right_parts
+
+    def isin_or_eq(self, left, right):
+        return left == right or self.isin(left, right)
+
+    def overlaps(self, left, right):
+        # pylint: disable=arguments-out-of-order
+        return self.isin_or_eq(left, right) or self.isin(right, left)
+
+    def relpath(self, path, base):
+        assert len(path) > len(base)
+        assert path.startswith(base)
+        normpath = path.rstrip(self.flavour.sep)
+        normbase = base.rstrip(self.flavour.sep)
+        return normpath[len(normbase) + 1 :]
+
+    def relparts(self, path, base):
+        return self.parts(self.relpath(path, base))
+
+    def as_posix(self, path):
+        return path.replace(self.flavour.sep, posixpath.sep)

dvc/fs/repo.py

@@ -7,10 +7,8 @@ from typing import TYPE_CHECKING, Callable, Optional, Tuple, Type, Union
 
 from funcy import lfilter, wrap_with
 
-from dvc.path_info import PathInfo
-
 from ..progress import DEFAULT_CALLBACK
-from .base import BaseFileSystem
+from .base import FileSystem
 from .dvc import DvcFileSystem
 
 if TYPE_CHECKING:
@@ -21,7 +19,7 @@ logger = logging.getLogger(__name__)
 RepoFactory = Union[Callable[[str], "Repo"], Type["Repo"]]
 
 
-class RepoFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
+class RepoFileSystem(FileSystem):  # pylint:disable=abstract-method
     """DVC + git-tracked files fs.
 
     Args:
@@ -164,7 +162,7 @@ class RepoFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
         if not prefix:
             return None
 
-        parents = (parent.fspath for parent in PathInfo(path).parents)
+        parents = (parent for parent in self.path.parents(path))
         dirs = [path] + list(takewhile(lambda p: p != prefix, parents))
         dirs.reverse()
         self._update(dirs, starting_repo=repo)
@@ -195,9 +193,7 @@ class RepoFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
         repo_path = os.path.join(dir_path, Repo.DVC_DIR)
         return self._main_repo.fs.isdir(repo_path)
 
-    def _get_fs_pair(
-        self, path
-    ) -> Tuple[BaseFileSystem, Optional[DvcFileSystem]]:
+    def _get_fs_pair(self, path) -> Tuple[FileSystem, Optional[DvcFileSystem]]:
         """
         Returns a pair of fss based on repo the path falls in, using prefix.
         """
@@ -217,34 +213,33 @@ class RepoFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
             encoding = None
 
         fs, dvc_fs = self._get_fs_pair(path)
-        path_info = PathInfo(path)
         try:
-            return fs.open(path_info, mode=mode, encoding=encoding)
+            return fs.open(path, mode=mode, encoding=encoding)
         except FileNotFoundError:
             if not dvc_fs:
                 raise
 
-        return dvc_fs.open(path_info, mode=mode, encoding=encoding, **kwargs)
+        return dvc_fs.open(path, mode=mode, encoding=encoding, **kwargs)
 
-    def exists(self, path_info) -> bool:
-        fs, dvc_fs = self._get_fs_pair(path_info)
+    def exists(self, path) -> bool:
+        fs, dvc_fs = self._get_fs_pair(path)
 
         if not dvc_fs:
-            return fs.exists(path_info)
+            return fs.exists(path)
 
-        if dvc_fs.repo.dvcignore.is_ignored(fs, path_info):
+        if dvc_fs.repo.dvcignore.is_ignored(fs, path):
             return False
 
-        if fs.exists(path_info):
+        if fs.exists(path):
             return True
 
         try:
-            meta = dvc_fs.metadata(path_info)
+            meta = dvc_fs.metadata(path)
         except FileNotFoundError:
             return False
 
         for out in meta.outs:
-            if fs.exists(out.path_info):
+            if fs.exists(out.fs_path):
                 return False
 
         return True
@@ -271,7 +266,7 @@ class RepoFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
             return False
 
         for out in meta.outs:
-            if fs.exists(out.path_info):
+            if fs.exists(out.fs_path):
                 return False
 
         return meta.isdir
@@ -303,15 +298,15 @@ class RepoFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
 
         (out,) = meta.outs
         assert len(meta.outs) == 1
-        if fs.exists(out.path_info):
+        if fs.exists(out.fs_path):
             return False
         return meta.isfile
 
-    def isexec(self, path_info):
-        fs, dvc_fs = self._get_fs_pair(path_info)
-        if dvc_fs and dvc_fs.exists(path_info):
-            return dvc_fs.isexec(path_info)
-        return fs.isexec(path_info)
+    def isexec(self, path):
+        fs, dvc_fs = self._get_fs_pair(path)
+        if dvc_fs and dvc_fs.exists(path):
+            return dvc_fs.isexec(path)
+        return fs.isexec(path)
 
     def _dvc_walk(self, walk):
         try:
@@ -444,10 +439,10 @@ class RepoFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
 
         yield from self._walk(repo_walk, dvc_walk, dvcfiles=dvcfiles)
 
-    def walk_files(self, path_info, **kwargs):
-        for root, _, files in self.walk(path_info, **kwargs):
+    def find(self, path, prefix=None):
+        for root, _, files in self.walk(path):
             for fname in files:
-                yield PathInfo(root) / fname
+                yield self.path.join(root, fname)
 
     def get_file(
         self, from_info, to_file, callback=DEFAULT_CALLBACK, **kwargs
@@ -467,18 +462,17 @@ class RepoFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
         )
 
     def metadata(self, path):
-        abspath = os.path.abspath(path)
-        path_info = PathInfo(abspath)
-        fs, dvc_fs = self._get_fs_pair(path_info)
+        fs_path = os.path.abspath(path)
+        fs, dvc_fs = self._get_fs_pair(fs_path)
 
         dvc_meta = None
         if dvc_fs:
             with suppress(FileNotFoundError):
-                dvc_meta = dvc_fs.metadata(path_info)
+                dvc_meta = dvc_fs.metadata(fs_path)
 
         info_result = None
         with suppress(FileNotFoundError):
-            info_result = fs.info(path_info)
+            info_result = fs.info(fs_path)
 
         if not info_result and not dvc_meta:
             raise FileNotFoundError
@@ -486,8 +480,8 @@ class RepoFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
         from ._metadata import Metadata
 
         meta = dvc_meta or Metadata(
-            path_info=path_info,
-            repo=self._get_repo(abspath) or self._main_repo,
+            fs_path=fs_path,
+            repo=self._get_repo(fs_path) or self._main_repo,
         )
 
         isdir = bool(info_result) and info_result["type"] == "directory"
@@ -499,18 +493,18 @@ class RepoFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
             meta.is_exec = bool(info_result) and is_exec(info_result["mode"])
         return meta
 
-    def info(self, path_info):
-        fs, dvc_fs = self._get_fs_pair(path_info)
+    def info(self, path):
+        fs, dvc_fs = self._get_fs_pair(path)
 
         try:
-            return fs.info(path_info)
+            return fs.info(path)
         except FileNotFoundError:
-            return dvc_fs.info(path_info)
+            return dvc_fs.info(path)
 
-    def checksum(self, path_info):
-        fs, dvc_fs = self._get_fs_pair(path_info)
+    def checksum(self, path):
+        fs, dvc_fs = self._get_fs_pair(path)
 
         try:
-            return fs.checksum(path_info)
+            return fs.checksum(path)
         except FileNotFoundError:
-            return dvc_fs.checksum(path_info)
+            return dvc_fs.checksum(path)

dvc/fs/s3.py

@@ -5,7 +5,6 @@ from collections import defaultdict
 
 from funcy import cached_property, wrap_prop
 
-from dvc.path_info import CloudURLInfo
 from dvc.progress import DEFAULT_CALLBACK
 from dvc.scheme import Schemes
 
@@ -17,7 +16,6 @@ _AWS_CONFIG_PATH = os.path.join(os.path.expanduser("~"), ".aws", "config")
 # pylint:disable=abstract-method
 class BaseS3FileSystem(ObjectFSWrapper):
     scheme = Schemes.S3
-    PATH_CLS = CloudURLInfo
     REQUIRES = {"s3fs": "s3fs", "boto3": "boto3"}
     PARAM_CHECKSUM = "etag"
     DETAIL_FIELDS = frozenset(("etag", "size"))
@@ -99,6 +97,11 @@ class BaseS3FileSystem(ObjectFSWrapper):
         client["endpoint_url"] = config.get("endpointurl")
         client["verify"] = config.get("ssl_verify")
 
+        # timeout configuration
+        config_kwargs = login_info["config_kwargs"]
+        config_kwargs["read_timeout"] = config.get("read_timeout")
+        config_kwargs["connect_timeout"] = config.get("connect_timeout")
+
         # encryptions
         additional = login_info["s3_additional_kwargs"]
         additional["ServerSideEncryption"] = config.get("sse")
@@ -143,12 +146,6 @@ class BaseS3FileSystem(ObjectFSWrapper):
             }
         )
 
-    def _entry_hook(self, entry):
-        entry = entry.copy()
-        if "ETag" in entry:
-            entry["etag"] = entry["ETag"].strip('"')
-        return entry
-
     @wrap_prop(threading.Lock())
     @cached_property
     def fs(self):
@@ -156,6 +153,15 @@ class BaseS3FileSystem(ObjectFSWrapper):
 
         return _S3FileSystem(**self.fs_args)
 
+    @classmethod
+    def _strip_protocol(cls, path: str) -> str:
+        from fsspec.utils import infer_storage_options
+
+        return infer_storage_options(path)["path"]
+
+    def unstrip_protocol(self, path):
+        return "s3://" + path.lstrip("/")
+
 
 def _translate_exceptions(func):
     @functools.wraps(func)
@@ -199,9 +205,10 @@ class S3FileSystem(BaseS3FileSystem):  # pylint:disable=abstract-method
             verify=client_kwargs.get("verify"),
         )
 
-    def _get_obj(self, path_info):
-        bucket = self.s3.Bucket(path_info.bucket)
-        return bucket.Object(path_info.path)
+    def _get_obj(self, path):
+        bucket_name, key, _ = self.fs.split_path(path)
+        bucket = self.s3.Bucket(bucket_name)
+        return bucket.Object(key)
 
     @_translate_exceptions
     def put_file(
@@ -215,7 +222,7 @@ class S3FileSystem(BaseS3FileSystem):  # pylint:disable=abstract-method
             ExtraArgs=self.fs_args.get("s3_additional_kwargs"),
             Config=self._transfer_config,
         )
-        self.fs.invalidate_cache(self._with_bucket(to_info.parent))
+        self.fs.invalidate_cache(self.path.parent(to_info))
 
     @_translate_exceptions
     def get_file(

dvc/fs/ssh.py

@@ -34,12 +34,17 @@ class SSHFileSystem(FSSpecWrapper):
     DEFAULT_PORT = 22
     PARAM_CHECKSUM = "md5"
 
-    def _with_bucket(self, path):
-        if isinstance(path, self.PATH_CLS):
-            path_str = path.path
-        else:
-            path_str = super()._with_bucket(path)
-        return os.path.normpath(path_str)
+    @classmethod
+    def _strip_protocol(cls, path: str) -> str:
+        from fsspec.utils import infer_storage_options
+
+        return infer_storage_options(path)["path"]
+
+    def unstrip_protocol(self, path: str) -> str:
+        host = self.fs_args["host"]
+        port = self.fs_args["port"]
+        path = path.lstrip("/")
+        return f"ssh://{host}:{port}/{path}"
 
     def _prepare_credentials(self, **config):
         self.CAN_TRAVERSE = True

dvc/fs/utils.py

@@ -1,40 +1,99 @@
+import errno
 import logging
 import os
-from typing import TYPE_CHECKING
+from typing import TYPE_CHECKING, List, Optional
 
+from .base import RemoteActionNotImplemented
 from .local import LocalFileSystem
 
 if TYPE_CHECKING:
-    from dvc.types import AnyPath, DvcPath
-
-    from .base import BaseFileSystem
+    from .base import FileSystem
+    from .types import AnyPath
 
 logger = logging.getLogger(__name__)
 
 
-def transfer(
-    from_fs: "BaseFileSystem",
-    from_info: "AnyPath",
-    to_fs: "BaseFileSystem",
-    to_info: "DvcPath",
-    move: bool = False,
+def _link(
+    link: "AnyPath",
+    from_fs: "FileSystem",
+    from_path: "AnyPath",
+    to_fs: "FileSystem",
+    to_path: "AnyPath",
 ) -> None:
-    use_move = isinstance(from_fs, type(to_fs)) and move
+    if not isinstance(from_fs, type(to_fs)):
+        raise OSError(errno.EXDEV, "can't link across filesystems")
+
     try:
-        if use_move:
-            return to_fs.move(from_info, to_info)
+        func = getattr(to_fs, link)
+        func(from_path, to_path)
+    except (OSError, AttributeError, RemoteActionNotImplemented) as exc:
+        # NOTE: there are so many potential error codes that a link call can
+        # throw, that is safer to just catch them all and try another link.
+        raise OSError(
+            errno.ENOTSUP, f"'{link}' is not supported by {type(from_fs)}"
+        ) from exc
 
-        if isinstance(from_fs, LocalFileSystem):
-            if not isinstance(from_info, from_fs.PATH_CLS):
-                from_info = from_fs.PATH_CLS(from_info)
-            return to_fs.upload(from_info, to_info)
 
-        if isinstance(to_fs, LocalFileSystem):
-            return from_fs.download_file(from_info, to_info)
+def _copy(
+    from_fs: "FileSystem",
+    from_path: "AnyPath",
+    to_fs: "FileSystem",
+    to_path: "AnyPath",
+) -> None:
+    if isinstance(from_fs, LocalFileSystem):
+        return to_fs.upload(from_path, to_path)
 
-        with from_fs.open(from_info, mode="rb") as fobj:
-            size = from_fs.getsize(from_info)
-            return to_fs.upload(fobj, to_info, total=size)
+    if isinstance(to_fs, LocalFileSystem):
+        return from_fs.download_file(from_path, to_path)
+
+    with from_fs.open(from_path, mode="rb") as fobj:
+        size = from_fs.getsize(from_path)
+        return to_fs.upload(fobj, to_path, total=size)
+
+
+def _try_links(
+    links: List["AnyPath"],
+    from_fs: "FileSystem",
+    from_path: "AnyPath",
+    to_fs: "FileSystem",
+    to_path: "AnyPath",
+) -> None:
+    error = None
+    while links:
+        link = links[0]
+
+        if link == "copy":
+            return _copy(from_fs, from_path, to_fs, to_path)
+
+        try:
+            return _link(link, from_fs, from_path, to_fs, to_path)
+        except OSError as exc:
+            if exc.errno not in [errno.ENOTSUP, errno.EXDEV]:
+                raise
+            error = exc
+
+        del links[0]
+
+    raise OSError(
+        errno.ENOTSUP, "no more link types left to try out"
+    ) from error
+
+
+def transfer(
+    from_fs: "FileSystem",
+    from_path: "AnyPath",
+    to_fs: "FileSystem",
+    to_path: "AnyPath",
+    hardlink: bool = False,
+    links: Optional[List["AnyPath"]] = None,
+) -> None:
+    try:
+        assert not (hardlink and links)
+        if hardlink:
+            links = links or ["reflink", "hardlink", "copy"]
+        else:
+            links = links or ["reflink", "copy"]
+        _try_links(links, from_fs, from_path, to_fs, to_path)
     except OSError as exc:
         # If the target file already exists, we are going to simply
         # ignore the exception (#4992).
@@ -48,9 +107,62 @@ def transfer(
             and exc.__context__
             and isinstance(exc.__context__, FileExistsError)
         ):
-            logger.debug("'%s' file already exists, skipping", to_info)
-            if use_move:
-                from_fs.remove(from_info)
+            logger.debug("'%s' file already exists, skipping", to_path)
             return None
 
         raise
+
+
+def _test_link(
+    link: "AnyPath",
+    from_fs: "FileSystem",
+    from_file: "AnyPath",
+    to_fs: "FileSystem",
+    to_file: "AnyPath",
+) -> bool:
+    try:
+        _try_links([link], from_fs, from_file, to_fs, to_file)
+    except OSError:
+        logger.debug("", exc_info=True)
+        return False
+
+    try:
+        _is_link_func = getattr(to_fs, f"is_{link}")
+        return _is_link_func(to_file)
+    except AttributeError:
+        pass
+
+    return True
+
+
+def test_links(
+    links: List["AnyPath"],
+    from_fs: "FileSystem",
+    from_path: "AnyPath",
+    to_fs: "FileSystem",
+    to_path: "AnyPath",
+) -> List["AnyPath"]:
+    from dvc.utils import tmp_fname
+
+    from_file = from_fs.path.join(from_path, tmp_fname())
+    to_file = to_fs.path.join(
+        to_fs.path.parent(to_path),
+        tmp_fname(),
+    )
+
+    from_fs.makedirs(from_fs.path.parent(from_file))
+    with from_fs.open(from_file, "wb") as fobj:
+        fobj.write(b"test")
+
+    ret = []
+    try:
+        for link in links:
+            try:
+                if _test_link(link, from_fs, from_file, to_fs, to_file):
+                    ret.append(link)
+            finally:
+                to_fs.remove(to_file)
+    finally:
+        from_fs.remove(from_file)
+
+    return ret

dvc/fs/webdav.py

@@ -1,10 +1,8 @@
 import logging
 import threading
-from functools import lru_cache
 
 from funcy import cached_property, wrap_prop
 
-from dvc.path_info import WebDAVURLInfo
 from dvc.scheme import Schemes
 
 from .fsspec_wrapper import FSSpecWrapper
@@ -15,7 +13,6 @@ logger = logging.getLogger(__name__)
 
 class WebDAVFileSystem(FSSpecWrapper):  # pylint:disable=abstract-method
     scheme = Schemes.WEBDAV
-    PATH_CLS = WebDAVURLInfo
     CAN_TRAVERSE = True
     TRAVERSE_PREFIX_LEN = 2
     REQUIRES = {"webdav4": "webdav4"}
@@ -37,14 +34,35 @@ class WebDAVFileSystem(FSSpecWrapper):  # pylint:disable=abstract-method
             }
         )
 
+    @classmethod
+    def _strip_protocol(cls, path: str) -> str:
+        from fsspec.utils import infer_storage_options
+
+        return infer_storage_options(path)["path"].lstrip("/")
+
+    def unstrip_protocol(self, path: str) -> str:
+        return self.fs_args["base_url"] + "/" + path
+
     @staticmethod
     def _get_kwargs_from_urls(urlpath):
-        path_info = WebDAVURLInfo(urlpath)
+        from urllib.parse import urlparse, urlunparse
+
+        parsed = urlparse(urlpath)
+        scheme = parsed.scheme.replace("webdav", "http")
         return {
-            "prefix": path_info.path.rstrip("/"),
-            "host": path_info.replace(path="").url,
-            "url": path_info.url.rstrip("/"),
-            "user": path_info.user,
+            "prefix": parsed.path.rstrip("/"),
+            "host": urlunparse((scheme, parsed.netloc, "", None, None, None)),
+            "url": urlunparse(
+                (
+                    scheme,
+                    parsed.netloc,
+                    parsed.path.rstrip("/"),
+                    None,
+                    None,
+                    None,
+                )
+            ),
+            "user": parsed.username,
         }
 
     def _prepare_credentials(self, **config):
@@ -71,20 +89,11 @@ class WebDAVFileSystem(FSSpecWrapper):  # pylint:disable=abstract-method
         return WebdavFileSystem(**self.fs_args)
 
     def upload_fobj(self, fobj, to_info, **kwargs):
-        rpath = self.translate_path_info(to_info)
         size = kwargs.get("size")
         # using upload_fileobj to directly upload fileobj rather than buffering
         # and using overwrite=True to avoid check for an extra exists call,
         # as caller should ensure that the file does not exist beforehand.
-        return self.fs.upload_fileobj(fobj, rpath, overwrite=True, size=size)
-
-    @lru_cache(512)
-    def translate_path_info(self, path):
-        if isinstance(path, self.PATH_CLS):
-            return path.path[len(self.prefix) :].lstrip("/")
-        return path
-
-    _with_bucket = translate_path_info
+        return self.fs.upload_fileobj(fobj, to_info, overwrite=True, size=size)
 
 
 class WebDAVSFileSystem(WebDAVFileSystem):  # pylint:disable=abstract-method

dvc/fs/webhdfs.py

@@ -1,60 +1,29 @@
-import logging
-import os
-import posixpath
-import shutil
 import threading
-from contextlib import contextmanager
 
 from funcy import cached_property, wrap_prop
 
-from dvc.hash_info import HashInfo
-from dvc.path_info import CloudURLInfo
-from dvc.progress import DEFAULT_CALLBACK
 from dvc.scheme import Schemes
 
-from .base import BaseFileSystem
-
-logger = logging.getLogger(__name__)
+# pylint:disable=abstract-method
+from .fsspec_wrapper import CallbackMixin, FSSpecWrapper
 
 
-def update_pbar(pbar, total):
-    """Update pbar to accept the two arguments passed by hdfs"""
-
-    def update(_, bytes_transfered):
-        if bytes_transfered == -1:
-            pbar.update_to(total)
-            return
-        pbar.update_to(bytes_transfered)
-
-    return update
-
-
-def update_callback(callback, total):
-    def update(_, bytes_transfered):
-        if bytes_transfered == -1:
-            return callback.absolute_update(total)
-        return callback.relative_update(bytes_transfered)
-
-    return update
-
-
-class WebHDFSFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
+class WebHDFSFileSystem(CallbackMixin, FSSpecWrapper):
     scheme = Schemes.WEBHDFS
-    PATH_CLS = CloudURLInfo
-    REQUIRES = {"hdfs": "hdfs"}
+    REQUIRES = {"fsspec": "fsspec"}
     PARAM_CHECKSUM = "checksum"
-    TRAVERSE_PREFIX_LEN = 2
 
-    def __init__(self, **config):
-        super().__init__(**config)
+    @classmethod
+    def _strip_protocol(cls, path: str) -> str:
+        from fsspec.utils import infer_storage_options
 
-        self.host = config["host"]
-        self.user = config.get("user")
-        self.port = config.get("port")
+        return infer_storage_options(path)["path"]
 
-        self.hdfscli_config = config.get("hdfscli_config")
-        self.token = config.get("webhdfs_token")
-        self.alias = config.get("webhdfs_alias")
+    def unstrip_protocol(self, path: str) -> str:
+        host = self.fs_args["host"]
+        port = self.fs_args["port"]
+        path = path.lstrip("/")
+        return f"webhdfs://{host}:{port}/{path}"
 
     @staticmethod
     def _get_kwargs_from_urls(urlpath):
@@ -66,115 +35,22 @@ class WebHDFSFileSystem(BaseFileSystem):  # pylint:disable=abstract-method
             )
         )
 
+    def _prepare_credentials(self, **config):
+        self._ssl_verify = config.pop("ssl_verify", True)
+        principal = config.pop("kerberos_principal", None)
+        if principal:
+            config["kerb_kwargs"] = {"principal": principal}
+        return config
+
     @wrap_prop(threading.Lock())
     @cached_property
-    def hdfs_client(self):
-        import hdfs
+    def fs(self):
+        from fsspec.implementations.webhdfs import WebHDFS
 
-        logger.debug("HDFSConfig: %s", self.hdfscli_config)
+        fs = WebHDFS(**self.fs_args)
+        fs.session.verify = self._ssl_verify
+        return fs
 
-        try:
-            return hdfs.config.Config(self.hdfscli_config).get_client(
-                self.alias
-            )
-        except hdfs.util.HdfsError as exc:
-            exc_msg = str(exc)
-            errors = (
-                "No alias specified",
-                "Invalid configuration file",
-                f"Alias {self.alias} not found",
-            )
-            if not any(err in exc_msg for err in errors):
-                raise
-
-            http_url = f"http://{self.host}:{self.port}"
-            logger.debug("URL: %s", http_url)
-
-            if self.token is not None:
-                client = hdfs.TokenClient(http_url, token=self.token, root="/")
-            else:
-                client = hdfs.InsecureClient(
-                    http_url, user=self.user, root="/"
-                )
-
-        return client
-
-    @contextmanager
-    def open(self, path_info, mode="r", encoding=None, **kwargs):
-        assert mode in {"r", "rt", "rb"}
-
-        with self.hdfs_client.read(
-            path_info.path, encoding=encoding
-        ) as reader:
-            yield reader
-
-    def walk_files(self, path_info, **kwargs):
-        if not self.exists(path_info):
-            return
-
-        root = path_info.path
-        for path, _, fnames in self.hdfs_client.walk(root):
-            for fname in fnames:
-                yield path_info.replace(path=posixpath.join(path, fname))
-
-    def remove(self, path_info):
-        if path_info.scheme != self.scheme:
-            raise NotImplementedError
-
-        self.hdfs_client.delete(path_info.path)
-
-    def exists(self, path_info) -> bool:
-        assert not isinstance(path_info, list)
-        assert path_info.scheme == "webhdfs"
-
-        status = self.hdfs_client.status(path_info.path, strict=False)
-        return status is not None
-
-    def info(self, path_info):
-        st = self.hdfs_client.status(path_info.path)
-        return {"size": st["length"], "type": "file"}
-
-    def checksum(self, path_info):
-        return HashInfo(
-            "checksum",
-            self.hdfs_client.checksum(path_info.path)["bytes"],
-            size=self.hdfs_client.status(path_info.path)["length"],
-        )
-
-    def copy(self, from_info, to_info, **_kwargs):
-        with self.hdfs_client.read(from_info.path) as reader:
-            with self.hdfs_client.write(to_info.path) as writer:
-                shutil.copyfileobj(reader, writer)
-
-    def move(self, from_info, to_info):
-        self.hdfs_client.makedirs(to_info.parent.path)
-        self.hdfs_client.rename(from_info.path, to_info.path)
-
-    def upload_fobj(self, fobj, to_info, **kwargs):
-        with self.hdfs_client.write(to_info.path) as fdest:
-            shutil.copyfileobj(fobj, fdest)
-
-    def put_file(
-        self, from_file, to_info, callback=DEFAULT_CALLBACK, **kwargs
-    ):
-        total = os.path.getsize(from_file)
-        callback.set_size(total)
-
-        self.hdfs_client.makedirs(to_info.parent.path)
-        return self.hdfs_client.upload(
-            to_info.path,
-            from_file,
-            overwrite=True,
-            progress=update_callback(callback, total),
-        )
-
-    def get_file(
-        self, from_info, to_file, callback=DEFAULT_CALLBACK, **kwargs
-    ):
-        total = self.getsize(from_info)
-        if total:
-            callback.set_size(total)
-
-        self.hdfs_client.download(
-            from_info.path, to_file, progress=update_callback(callback, total)
-        )
+    def checksum(self, path):
+        ukey = self.fs.ukey(path)
+        return ukey["bytes"]

dvc/hash_info.py

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

dvc/ignore.py

@@ -7,11 +7,10 @@ from itertools import groupby, takewhile
 from pathspec.patterns import GitWildMatchPattern
 from pathspec.util import normalize_file
 
-from dvc.fs.base import BaseFileSystem
-from dvc.path_info import PathInfo
+from dvc.fs.base import FileSystem
+from dvc.fs.local import localfs
 from dvc.pathspec_math import PatternInfo, merge_patterns
 from dvc.scheme import Schemes
-from dvc.system import System
 from dvc.types import AnyPath, List, Optional
 from dvc.utils import relpath
 from dvc.utils.collections import PathStringTrie
@@ -84,7 +83,7 @@ class DvcIgnorePatterns(DvcIgnore):
         else:
             return False
 
-        if not System.is_unix():
+        if os.name == "nt":
             path = normalize_file(path)
         return path
 
@@ -268,25 +267,25 @@ class DvcIgnoreFilter:
             dirs, files = ignore_pattern(abs_root, dirs, files)
         return dirs, files
 
-    def walk(self, fs: BaseFileSystem, path_info: AnyPath, **kwargs):
+    def walk(self, fs: FileSystem, path: AnyPath, **kwargs):
         ignore_subrepos = kwargs.pop("ignore_subrepos", True)
         if fs.scheme == Schemes.LOCAL:
-            for root, dirs, files in fs.walk(path_info, **kwargs):
+            for root, dirs, files in fs.walk(path, **kwargs):
                 dirs[:], files[:] = self(
                     root, dirs, files, ignore_subrepos=ignore_subrepos
                 )
                 yield root, dirs, files
         else:
-            yield from fs.walk(path_info, **kwargs)
+            yield from fs.walk(path, **kwargs)
 
-    def walk_files(self, fs: BaseFileSystem, path_info: AnyPath, **kwargs):
+    def find(self, fs: FileSystem, path: AnyPath, **kwargs):
         if fs.scheme == Schemes.LOCAL:
-            for root, _, files in self.walk(fs, path_info, **kwargs):
+            for root, _, files in self.walk(fs, path, **kwargs):
                 for file in files:
                     # NOTE: os.path.join is ~5.5 times slower
-                    yield PathInfo(f"{root}{os.sep}{file}")
+                    yield f"{root}{os.sep}{file}"
         else:
-            yield from fs.walk_files(path_info)
+            yield from fs.find(path)
 
     def _get_trie_pattern(
         self, dirname, dnames: Optional["List"] = None, ignore_subrepos=True
@@ -308,7 +307,7 @@ class DvcIgnoreFilter:
         dirs = list(
             takewhile(
                 lambda path: path != prefix,
-                (parent.fspath for parent in PathInfo(dirname).parents),
+                (parent for parent in localfs.path.parents(dirname)),
             )
         )
         dirs.reverse()
@@ -344,8 +343,6 @@ class DvcIgnoreFilter:
         return self._is_ignored(path, False)
 
     def _outside_repo(self, path):
-        path = PathInfo(path)
-
         # paths outside of the repo should be ignored
         path = relpath(path, self.root_dir)
         if path.startswith("..") or (
@@ -374,7 +371,7 @@ class DvcIgnoreFilter:
         return _no_match(target)
 
     def is_ignored(
-        self, fs: BaseFileSystem, path: str, ignore_subrepos: bool = True
+        self, fs: FileSystem, path: str, ignore_subrepos: bool = True
     ) -> bool:
         # NOTE: can't use self.check_ignore(path).match for now, see
         # https://github.com/iterative/dvc/issues/4555
@@ -394,7 +391,7 @@ def init(path):
     if os.path.exists(dvcignore):
         return dvcignore
 
-    with open(dvcignore, "w") as fobj:
+    with open(dvcignore, "w", encoding="utf-8") as fobj:
         fobj.write(
             "# Add patterns of files dvc should ignore, which could improve\n"
             "# the performance. Learn more at\n"

dvc/info.py

@@ -2,23 +2,17 @@ import itertools
 import os
 import pathlib
 import platform
-import uuid
 
 import psutil
 
-from dvc.exceptions import DvcException, NotDvcRepoError
+from dvc import __version__
+from dvc.exceptions import NotDvcRepoError
 from dvc.fs import FS_MAP, get_fs_cls, get_fs_config
+from dvc.fs.utils import test_links
 from dvc.repo import Repo
-from dvc.scm.base import SCMError
-from dvc.system import System
+from dvc.scm import SCMError
 from dvc.utils import error_link
 from dvc.utils.pkg import PKG
-from dvc.version import __version__
-
-if PKG is None:
-    package = ""
-else:
-    package = f"({PKG})"
 
 try:
     import importlib.metadata as importlib_metadata
@@ -26,6 +20,9 @@ except ImportError:  # < 3.8
     import importlib_metadata  # type: ignore[no-redef]
 
 
+package = "" if PKG is None else f"({PKG})"
+
+
 def get_dvc_info():
     info = [
         f"DVC version: {__version__} {package}",
@@ -76,7 +73,7 @@ def _get_caches(cache):
 
 def _get_remotes(config):
     schemes = (
-        get_fs_cls(get_fs_config(config, name=remote)).scheme
+        get_fs_cls(get_fs_config(None, config, name=remote)).scheme
         for remote in config["remote"]
     )
 
@@ -84,35 +81,17 @@ def _get_remotes(config):
 
 
 def _get_linktype_support_info(repo):
+    odb = repo.odb.local
 
-    links = {
-        "reflink": (System.reflink, None),
-        "hardlink": (System.hardlink, System.is_hardlink),
-        "symlink": (System.symlink, System.is_symlink),
-    }
+    links = test_links(
+        ["reflink", "hardlink", "symlink"],
+        odb.fs,
+        odb.fs_path,
+        repo.fs,
+        repo.root_dir,
+    )
 
-    fname = "." + str(uuid.uuid4())
-    src = os.path.join(repo.odb.local.cache_dir, fname)
-    open(src, "w").close()
-    dst = os.path.join(repo.root_dir, fname)
-
-    cache = []
-
-    for name, (link, is_link) in links.items():
-        try:
-            link(src, dst)
-            status = "supported"
-            if is_link and not is_link(dst):
-                status = "broken"
-            os.unlink(dst)
-        except DvcException:
-            status = "not supported"
-
-        if status == "supported":
-            cache.append(name)
-    os.remove(src)
-
-    return ", ".join(cache)
+    return ", ".join(links)
 
 
 def _get_supported_remotes():
@@ -136,11 +115,14 @@ def _get_supported_remotes():
 
 
 def get_fs_type(path):
-
-    partition = {
-        pathlib.Path(part.mountpoint): (part.fstype + " on " + part.device)
-        for part in psutil.disk_partitions(all=True)
-    }
+    partition = {}
+    for part in psutil.disk_partitions(all=True):
+        if part.fstype != "":
+            try:
+                mountpoint = pathlib.Path(part.mountpoint).resolve()
+                partition[mountpoint] = part.fstype + " on " + part.device
+            except PermissionError:
+                pass
 
     # need to follow the symlink: https://github.com/iterative/dvc/issues/5065
     path = pathlib.Path(path).resolve()

dvc/lock.py

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

dvc/logger.py

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

dvc/machine/__init__.py

@@ -2,8 +2,10 @@ import logging
 import os
 from typing import (
     TYPE_CHECKING,
+    Any,
     Dict,
     Iterable,
+    Iterator,
     Mapping,
     Optional,
     Tuple,
@@ -181,7 +183,7 @@ class MachineManager:
         """Create and start the specified machine instance."""
         config, backend = self.get_config_and_backend(name)
         if "startup_script" in config:
-            with open(config["startup_script"]) as fobj:
+            with open(config["startup_script"], encoding="utf-8") as fobj:
                 startup_script = fobj.read()
         else:
             startup_script = DEFAULT_STARTUP_SCRIPT
@@ -200,3 +202,16 @@ class MachineManager:
     def run_shell(self, name: Optional[str]):
         config, backend = self.get_config_and_backend(name)
         return backend.run_shell(**config)
+
+    def status(self, name: str) -> Iterator[Dict[Any, Any]]:
+        config, backend = self.get_config_and_backend(name)
+        return backend.instances(**config)
+
+    def rename(self, name: str, new: str):
+        """move configuration to a new location if the machine is running."""
+        config, backend = self.get_config_and_backend(name)
+        return backend.rename(new=new, **config)
+
+    def get_executor_kwargs(self, name: Optional[str]):
+        config, backend = self.get_config_and_backend(name)
+        return backend.get_executor_kwargs(**config)

dvc/machine/backend/base.py

@@ -4,13 +4,12 @@ from typing import TYPE_CHECKING, Iterator, Optional
 
 if TYPE_CHECKING:
     from dvc.fs.ssh import SSHFileSystem
-    from dvc.repo.experiments.executor.base import BaseExecutor
     from dvc.types import StrPath
 
 
 class BaseMachineBackend(ABC):
     def __init__(self, tmp_dir: "StrPath", **kwargs):
-        raise NotImplementedError
+        self.tmp_dir = tmp_dir
 
     @abstractmethod
     def create(self, name: Optional[str] = None, **config):
@@ -34,10 +33,8 @@ class BaseMachineBackend(ABC):
         """Spawn an interactive SSH shell for the specified machine."""
 
     @abstractmethod
-    def get_executor(
-        self, name: Optional[str] = None, **config
-    ) -> "BaseExecutor":
-        """Return an executor instance which can be used for DVC
+    def get_executor_kwargs(self, name: str, **config) -> dict:
+        """Return SSHExecutor kwargs which can be used for DVC
         experiment/pipeline execution on the specified machine.
         """
 
@@ -48,3 +45,7 @@ class BaseMachineBackend(ABC):
     ) -> Iterator["SSHFileSystem"]:
         """Return an sshfs instance for the default directory on the
         specified machine."""
+
+    @abstractmethod
+    def rename(self, name: str, new: str, **config):
+        """Rename a machine instance."""

dvc/machine/backend/terraform.py

@@ -1,35 +1,103 @@
+import os
 from contextlib import contextmanager
+from functools import partial, partialmethod
 from typing import TYPE_CHECKING, Iterator, Optional
 
-from tpi import TerraformProviderIterative, terraform
-
+from dvc.exceptions import DvcException
 from dvc.fs.ssh import SSHFileSystem
+from dvc.utils.fs import makedirs
 
 from .base import BaseMachineBackend
 
 if TYPE_CHECKING:
-    from dvc.repo.experiments.executor.base import BaseExecutor
     from dvc.types import StrPath
 
 
-class TerraformBackend(terraform.TerraformBackend, BaseMachineBackend):
-    def __init__(self, tmp_dir: "StrPath", **kwargs):
-        super().__init__(tmp_dir)
+@contextmanager
+def _sshfs(resource: dict):
+    from tpi import TerraformProviderIterative
 
-    def get_executor(
-        self, name: Optional[str] = None, **config
-    ) -> "BaseExecutor":
-        raise NotImplementedError
+    with TerraformProviderIterative.pemfile(resource) as pem:
+        fs = SSHFileSystem(
+            host=resource["instance_ip"],
+            user="ubuntu",
+            keyfile=pem,
+        )
+        yield fs
+
+
+class TerraformBackend(BaseMachineBackend):
+    def __init__(self, tmp_dir: "StrPath", **kwargs):
+        super().__init__(tmp_dir, **kwargs)
+        makedirs(tmp_dir, exist_ok=True)
+
+    @contextmanager
+    def make_tpi(self, name: str):
+        from tpi import TPIError
+        from tpi.terraform import TerraformBackend as TPIBackend
+
+        try:
+            working_dir = os.path.join(self.tmp_dir, name)
+            makedirs(working_dir, exist_ok=True)
+            yield TPIBackend(working_dir=working_dir)
+        except TPIError as exc:
+            raise DvcException("TPI operation failed") from exc
+
+    def _tpi_func(self, fname, name: Optional[str] = None, **config):
+        from tpi import TPIError
+
+        assert name
+        with self.make_tpi(name) as tpi:
+            func = getattr(tpi, fname)
+            try:
+                return func(name=name, **config)
+            except TPIError as exc:
+                raise DvcException(f"TPI {fname} failed") from exc
+
+    create = partialmethod(_tpi_func, "create")  # type: ignore[assignment]
+    destroy = partialmethod(_tpi_func, "destroy")  # type: ignore[assignment]
+    instances = partialmethod(
+        _tpi_func, "instances"
+    )  # type: ignore[assignment]
+    run_shell = partialmethod(
+        _tpi_func, "run_shell"
+    )  # type: ignore[assignment]
+
+    def get_executor_kwargs(self, name: str, **config) -> dict:
+        with self.make_tpi(name) as tpi:
+            resource = tpi.default_resource(name)
+        return {
+            "host": resource["instance_ip"],
+            "port": SSHFileSystem.DEFAULT_PORT,
+            "username": "ubuntu",
+            "fs_factory": partial(_sshfs, dict(resource)),
+        }
 
     @contextmanager
     def get_sshfs(  # pylint: disable=unused-argument
         self, name: Optional[str] = None, **config
     ) -> Iterator["SSHFileSystem"]:
-        resource = self._default_resource(name)
-        with TerraformProviderIterative.pemfile(resource) as pem:
-            fs = SSHFileSystem(
-                host=resource["instance_ip"],
-                user="ubuntu",
-                keyfile=pem,
-            )
+        assert name
+        with self.make_tpi(name) as tpi:
+            resource = tpi.default_resource(name)
+        with _sshfs(resource) as fs:
             yield fs
+
+    def rename(self, name: str, new: str, **config):
+        """rename a dvc machine instance to a new name"""
+        import shutil
+
+        mtype = "iterative_machine"
+
+        new_dir = os.path.join(self.tmp_dir, new)
+        old_dir = os.path.join(self.tmp_dir, name)
+        if os.path.exists(new_dir):
+            raise DvcException(f"rename failed: path {new_dir} already exists")
+
+        if not os.path.exists(old_dir):
+            return
+
+        with self.make_tpi(name) as tpi:
+            tpi.state_mv(f"{mtype}.{name}", f"{mtype}.{new}", **config)
+
+        shutil.move(old_dir, new_dir)

dvc/objects/__init__.py

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

dvc/objects/checkout.py

@@ -0,0 +1,266 @@
+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.objects.db.slow_link_detection import (  # type: ignore[attr-defined]
+    slow_link_guard,
+)
+from dvc.objects.diff import ROOT
+from dvc.objects.diff import diff as odiff
+from dvc.objects.stage import stage
+from dvc.types import Optional
+
+logger = logging.getLogger(__name__)
+
+
+def _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/objects/db/__init__.py

@@ -6,22 +6,22 @@ if TYPE_CHECKING:
     from .index import ObjectDBIndexBase
 
 
-def get_odb(fs, path_info, **config):
+def get_odb(fs, fs_path, **config):
     from .base import ObjectDB
     from .gdrive import GDriveObjectDB
     from .local import LocalObjectDB
     from .oss import OSSObjectDB
 
     if fs.scheme == Schemes.LOCAL:
-        return LocalObjectDB(fs, path_info, **config)
+        return LocalObjectDB(fs, fs_path, **config)
 
     if fs.scheme == Schemes.GDRIVE:
-        return GDriveObjectDB(fs, path_info, **config)
+        return GDriveObjectDB(fs, fs_path, **config)
 
     if fs.scheme == Schemes.OSS:
-        return OSSObjectDB(fs, path_info, **config)
+        return OSSObjectDB(fs, fs_path, **config)
 
-    return ObjectDB(fs, path_info, **config)
+    return ObjectDB(fs, fs_path, **config)
 
 
 def _get_odb(repo, settings):
@@ -30,9 +30,9 @@ def _get_odb(repo, settings):
     if not settings:
         return None
 
-    cls, config, path_info = get_cloud_fs(repo, **settings)
+    cls, config, fs_path = get_cloud_fs(repo, **settings)
     config["tmp_dir"] = repo.tmp_dir
-    return get_odb(cls(**config), path_info, state=repo.state, **config)
+    return get_odb(cls(**config), fs_path, state=repo.state, **config)
 
 
 def get_index(odb) -> "ObjectDBIndexBase":
@@ -43,7 +43,9 @@ def get_index(odb) -> "ObjectDBIndexBase":
     cls = ObjectDBIndex if odb.tmp_dir else ObjectDBIndexNoop
     return cls(
         odb.tmp_dir,
-        hashlib.sha256(odb.path_info.url.encode("utf-8")).hexdigest(),
+        hashlib.sha256(
+            odb.fs.unstrip_protocol(odb.fs_path).encode("utf-8")
+        ).hexdigest(),
         odb.fs.CHECKSUM_DIR_SUFFIX,
     )
 

dvc/objects/db/base.py

@@ -10,9 +10,9 @@ from dvc.objects.file import HashFile
 from dvc.progress import Tqdm
 
 if TYPE_CHECKING:
-    from dvc.fs.base import BaseFileSystem
+    from dvc.fs.base import FileSystem
     from dvc.hash_info import HashInfo
-    from dvc.types import AnyPath, DvcPath
+    from dvc.types import AnyPath
 
 logger = logging.getLogger(__name__)
 
@@ -23,15 +23,14 @@ class ObjectDB:
     DEFAULT_CACHE_TYPES = ["copy"]
     CACHE_MODE: Optional[int] = None
 
-    def __init__(self, fs: "BaseFileSystem", path_info: "AnyPath", **config):
+    def __init__(self, fs: "FileSystem", path: str, **config):
         from dvc.state import StateNoop
 
         self.fs = fs
-        self.path_info = path_info
+        self.fs_path = path
         self.state = config.get("state", StateNoop())
         self.verify = config.get("verify", self.DEFAULT_VERIFY)
         self.cache_types = config.get("type") or copy(self.DEFAULT_CACHE_TYPES)
-        self.cache_type_confirmed = False
         self.slow_link_warning = config.get("slow_link_warning", True)
         self.tmp_dir = config.get("tmp_dir")
         self.read_only = config.get("read_only", False)
@@ -50,26 +49,25 @@ class ObjectDB:
     def __eq__(self, other):
         return (
             self.fs == other.fs
-            and self.path_info == other.path_info
+            and self.fs_path == other.fs_path
             and self.read_only == other.read_only
         )
 
     def __hash__(self):
-        return hash((self.fs.scheme, self.path_info))
+        return hash((self.fs.scheme, self.fs_path))
 
     def exists(self, hash_info: "HashInfo"):
-        return self.fs.exists(self.hash_to_path_info(hash_info.value))
+        return self.fs.exists(self.hash_to_path(hash_info.value))
 
     def move(self, from_info, to_info):
         self.fs.move(from_info, to_info)
 
-    def makedirs(self, path_info):
-        self.fs.makedirs(path_info)
+    def makedirs(self, fs_path):
+        self.fs.makedirs(fs_path)
 
     def get(self, hash_info: "HashInfo"):
         """get raw object"""
         return HashFile(
-            # Prefer string path over PathInfo when possible due to performance
             self.hash_to_path(hash_info.value),
             self.fs,
             hash_info,
@@ -77,25 +75,29 @@ class ObjectDB:
 
     def _add_file(
         self,
-        from_fs: "BaseFileSystem",
+        from_fs: "FileSystem",
         from_info: "AnyPath",
-        to_info: "DvcPath",
+        to_info: "AnyPath",
         _hash_info: "HashInfo",
-        move: bool = False,
+        hardlink: bool = False,
     ):
         from dvc import fs
 
-        self.makedirs(to_info.parent)
+        self.makedirs(self.fs.path.parent(to_info))
         return fs.utils.transfer(
-            from_fs, from_info, self.fs, to_info, move=move
+            from_fs,
+            from_info,
+            self.fs,
+            to_info,
+            hardlink=hardlink,
         )
 
     def add(
         self,
-        path_info: "AnyPath",
-        fs: "BaseFileSystem",
+        fs_path: "AnyPath",
+        fs: "FileSystem",
         hash_info: "HashInfo",
-        move: bool = True,
+        hardlink: bool = False,
         verify: Optional[bool] = None,
     ):
         if self.read_only:
@@ -109,35 +111,32 @@ class ObjectDB:
         except (ObjectFormatError, FileNotFoundError):
             pass
 
-        cache_info = self.hash_to_path_info(hash_info.value)
-        self._add_file(fs, path_info, cache_info, hash_info, move=move)
+        cache_fs_path = self.hash_to_path(hash_info.value)
+        self._add_file(
+            fs, fs_path, cache_fs_path, hash_info, hardlink=hardlink
+        )
 
         try:
             if verify:
                 self.check(hash_info, check_hash=True)
-            self.protect(cache_info)
-            self.state.save(cache_info, self.fs, hash_info)
+            self.protect(cache_fs_path)
+            self.state.save(cache_fs_path, self.fs, hash_info)
         except (ObjectFormatError, FileNotFoundError):
             pass
 
-    def hash_to_path_info(self, hash_) -> "DvcPath":
-        return self.path_info / hash_[0:2] / hash_[2:]
-
-    # Override to return path as a string instead of PathInfo for clouds
-    # which support string paths (see local)
     def hash_to_path(self, hash_):
-        return self.hash_to_path_info(hash_)
+        return self.fs.path.join(self.fs_path, hash_[0:2], hash_[2:])
 
-    def protect(self, path_info):  # pylint: disable=unused-argument
+    def protect(self, fs_path):  # pylint: disable=unused-argument
         pass
 
-    def is_protected(self, path_info):  # pylint: disable=unused-argument
+    def is_protected(self, fs_path):  # pylint: disable=unused-argument
         return False
 
-    def unprotect(self, path_info):  # pylint: disable=unused-argument
+    def unprotect(self, fs_path):  # pylint: disable=unused-argument
         pass
 
-    def set_exec(self, path_info):  # pylint: disable=unused-argument
+    def set_exec(self, fs_path):  # pylint: disable=unused-argument
         pass
 
     def check(
@@ -160,52 +159,54 @@ class ObjectDB:
         """
 
         obj = self.get(hash_info)
-        if self.is_protected(obj.path_info):
+        if self.is_protected(obj.fs_path):
             logger.trace(  # type: ignore[attr-defined]
                 "Assuming '%s' is unchanged since it is read-only",
-                obj.path_info,
+                obj.fs_path,
             )
             return
 
         try:
             obj.check(self, check_hash=check_hash)
         except ObjectFormatError:
-            logger.warning("corrupted cache file '%s'.", obj.path_info)
+            logger.warning("corrupted cache file '%s'.", obj.fs_path)
             with suppress(FileNotFoundError):
-                self.fs.remove(obj.path_info)
+                self.fs.remove(obj.fs_path)
             raise
 
         if check_hash:
             # making cache file read-only so we don't need to check it
             # next time
-            self.protect(obj.path_info)
+            self.protect(obj.fs_path)
 
     def _list_paths(self, prefix=None, progress_callback=None):
         if prefix:
             if len(prefix) > 2:
-                path_info = self.path_info / prefix[:2] / prefix[2:]
+                fs_path = self.fs.path.join(
+                    self.fs_path, prefix[:2], prefix[2:]
+                )
             else:
-                path_info = self.path_info / prefix[:2]
+                fs_path = self.fs.path.join(self.fs_path, prefix[:2])
             prefix = True
         else:
-            path_info = self.path_info
+            fs_path = self.fs_path
             prefix = False
         if progress_callback:
-            for file_info in self.fs.walk_files(path_info, prefix=prefix):
+            for file_info in self.fs.find(fs_path, prefix=prefix):
                 progress_callback()
-                yield file_info.path
+                yield file_info
         else:
-            yield from self.fs.walk_files(path_info, prefix=prefix)
+            yield from self.fs.find(fs_path, prefix=prefix)
 
     def _path_to_hash(self, path):
-        parts = self.fs.PATH_CLS(path).parts[-2:]
+        parts = self.fs.path.parts(path)[-2:]
 
         if not (len(parts) == 2 and parts[0] and len(parts[0]) == 2):
             raise ValueError(f"Bad cache file path '{path}'")
 
         return "".join(parts)
 
-    def list_hashes(self, prefix=None, progress_callback=None):
+    def _list_hashes(self, prefix=None, progress_callback=None):
         """Iterate over hashes in this fs.
 
         If `prefix` is specified, only hashes which begin with `prefix`
@@ -221,12 +222,12 @@ class ObjectDB:
 
     def _hashes_with_limit(self, limit, prefix=None, progress_callback=None):
         count = 0
-        for hash_ in self.list_hashes(prefix, progress_callback):
+        for hash_ in self._list_hashes(prefix, progress_callback):
             yield hash_
             count += 1
             if count > limit:
                 logger.debug(
-                    "`list_hashes()` returned max '{}' hashes, "
+                    "`_list_hashes()` returned max '{}' hashes, "
                     "skipping remaining results".format(limit)
                 )
                 return
@@ -265,7 +266,7 @@ class ObjectDB:
                     max_hashes / total_prefixes, prefix, update
                 )
             else:
-                hashes = self.list_hashes(prefix, update)
+                hashes = self._list_hashes(prefix, update)
 
             remote_hashes = set(hashes)
             if remote_hashes:
@@ -275,7 +276,7 @@ class ObjectDB:
             logger.debug(f"Estimated remote size: {remote_size} files")
         return remote_size, remote_hashes
 
-    def list_hashes_traverse(
+    def _list_hashes_traverse(
         self, remote_size, remote_hashes, jobs=None, name=None
     ):
         """Iterate over all hashes found in this fs.
@@ -321,7 +322,7 @@ class ObjectDB:
 
             def list_with_update(prefix):
                 return list(
-                    self.list_hashes(
+                    self._list_hashes(
                         prefix=prefix, progress_callback=pbar.update
                     )
                 )
@@ -345,10 +346,10 @@ class ObjectDB:
         )
 
         if not self.fs.CAN_TRAVERSE:
-            return self.list_hashes()
+            return self._list_hashes()
 
         remote_size, remote_hashes = self._estimate_remote_size(name=name)
-        return self.list_hashes_traverse(
+        return self._list_hashes_traverse(
             remote_size, remote_hashes, jobs, name
         )
 
@@ -374,18 +375,18 @@ class ObjectDB:
         removed = False
         # hashes must be sorted to ensure we always remove .dir files first
         for hash_ in sorted(
-            self.all(jobs, str(self.path_info)),
+            self.all(jobs, self.fs_path),
             key=self.fs.is_dir_hash,
             reverse=True,
         ):
             if hash_ in used_hashes:
                 continue
-            path_info = self.hash_to_path_info(hash_)
+            fs_path = self.hash_to_path(hash_)
             if self.fs.is_dir_hash(hash_):
                 # backward compatibility
                 # pylint: disable=protected-access
                 self._remove_unpacked_dir(hash_)
-            self.fs.remove(path_info)
+            self.fs.remove(fs_path)
             removed = True
 
         return removed
@@ -402,16 +403,16 @@ class ObjectDB:
             unit="file",
         ) as pbar:
 
-            def exists_with_progress(path_info):
-                ret = self.fs.exists(path_info)
-                pbar.update_msg(str(path_info))
+            def exists_with_progress(fs_path):
+                ret = self.fs.exists(fs_path)
+                pbar.update_msg(fs_path)
                 return ret
 
             with ThreadPoolExecutor(
                 max_workers=jobs or self.fs.jobs
             ) as executor:
-                path_infos = map(self.hash_to_path_info, hashes)
-                in_remote = executor.map(exists_with_progress, path_infos)
+                fs_paths = map(self.hash_to_path, hashes)
+                in_remote = executor.map(exists_with_progress, fs_paths)
                 ret = list(itertools.compress(hashes, in_remote))
                 return ret
 
@@ -454,7 +455,7 @@ class ObjectDB:
 
         # During the tests, for ensuring that the traverse behavior
         # is working we turn on this option. It will ensure the
-        # list_hashes_traverse() is called.
+        # _list_hashes_traverse() is called.
         always_traverse = getattr(self.fs, "_ALWAYS_TRAVERSE", False)
 
         hashes = set(hashes)
@@ -491,6 +492,6 @@ class ObjectDB:
 
         logger.debug(f"Querying '{len(hashes)}' hashes via traverse")
         remote_hashes = set(
-            self.list_hashes_traverse(remote_size, remote_hashes, jobs, name)
+            self._list_hashes_traverse(remote_size, remote_hashes, jobs, name)
         )
         return list(hashes & set(remote_hashes))

dvc/objects/db/local.py

@@ -7,7 +7,6 @@ from shortuuid import uuid
 
 from dvc.hash_info import HashInfo
 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
@@ -22,9 +21,9 @@ 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 +35,11 @@ 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 +49,10 @@ 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 +80,30 @@ 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):
@@ -126,47 +129,47 @@ 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/objects/db/reference.py

@@ -10,9 +10,9 @@ from ..reference import ReferenceHashFile
 from .base import ObjectDB
 
 if TYPE_CHECKING:
-    from dvc.fs.base import BaseFileSystem
+    from dvc.fs.base import FileSystem
     from dvc.hash_info import HashInfo
-    from dvc.types import AnyPath, DvcPath
+    from dvc.types import AnyPath
 
 logger = logging.getLogger(__name__)
 
@@ -26,7 +26,7 @@ class ReferenceObjectDB(ObjectDB):
 
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
-        self._fs_cache: Dict[tuple, "BaseFileSystem"] = {}
+        self._fs_cache: Dict[tuple, "FileSystem"] = {}
         self._obj_cache: Dict["HashInfo", "ReferenceHashFile"] = {}
 
     def get(self, hash_info: "HashInfo"):
@@ -36,9 +36,9 @@ class ReferenceObjectDB(ObjectDB):
             return self._obj_cache[hash_info]
         except KeyError:
             pass
-        path_info = self.hash_to_path(hash_info.value)
+        fs_path = self.hash_to_path(hash_info.value)
         try:
-            with self.fs.open(path_info, "rb") as fobj:
+            with self.fs.open(fs_path, "rb") as fobj:
                 ref_file = ReferenceHashFile.from_bytes(
                     fobj.read(), fs_cache=self._fs_cache
                 )
@@ -47,23 +47,27 @@ class ReferenceObjectDB(ObjectDB):
         try:
             ref_file.check(self, check_hash=False)
         except ObjectFormatError:
-            self.fs.remove(path_info)
+            self.fs.remove(fs_path)
             raise
         self._obj_cache[hash_info] = ref_file
         return ref_file
 
     def _add_file(
         self,
-        from_fs: "BaseFileSystem",
+        from_fs: "FileSystem",
         from_info: "AnyPath",
-        to_info: "DvcPath",
+        to_info: "AnyPath",
         hash_info: "HashInfo",
-        move: bool = False,
+        hardlink: bool = False,
     ):
-        self.makedirs(to_info.parent)
+        self.makedirs(self.fs.path.parent(to_info))
         if hash_info.isdir:
             return super()._add_file(
-                from_fs, from_info, to_info, hash_info, move
+                from_fs,
+                from_info,
+                to_info,
+                hash_info,
+                hardlink=hardlink,
             )
         ref_file = ReferenceHashFile(from_info, from_fs, hash_info)
         self._obj_cache[hash_info] = ref_file

dvc/objects/db/slow_link_detection.py

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

dvc/objects/diff.py

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

dvc/objects/file.py

@@ -6,7 +6,7 @@ from typing import TYPE_CHECKING, Optional
 from .errors import ObjectFormatError
 
 if TYPE_CHECKING:
-    from dvc.fs.base import BaseFileSystem
+    from dvc.fs.base import FileSystem
     from dvc.hash_info import HashInfo
     from dvc.types import AnyPath
 
@@ -18,20 +18,16 @@ logger = logging.getLogger(__name__)
 class HashFile:
     def __init__(
         self,
-        path_info: Optional["AnyPath"],
-        fs: Optional["BaseFileSystem"],
+        fs_path: Optional["AnyPath"],
+        fs: Optional["FileSystem"],
         hash_info: "HashInfo",
         name: Optional[str] = None,
     ):
-        self.path_info = path_info
+        self.fs_path = fs_path
         self.fs = fs
         self.hash_info = hash_info
         self.name = name
 
-    @property
-    def size(self):
-        return self.hash_info.size
-
     def __len__(self):
         return 1
 
@@ -45,7 +41,7 @@ class HashFile:
         if not isinstance(other, HashFile):
             return False
         return (
-            self.path_info == other.path_info
+            self.fs_path == other.fs_path
             and self.fs == other.fs
             and self.hash_info == other.hash_info
         )
@@ -54,7 +50,7 @@ class HashFile:
         return hash(
             (
                 self.hash_info,
-                self.path_info,
+                self.fs_path,
                 self.fs.scheme if self.fs else None,
             )
         )
@@ -62,9 +58,9 @@ class HashFile:
     def check(self, odb: "ObjectDB", check_hash: bool = True):
         if not check_hash:
             assert self.fs
-            if not self.fs.exists(self.path_info):
+            if not self.fs.exists(self.fs_path):
                 raise FileNotFoundError(
-                    errno.ENOENT, os.strerror(errno.ENOENT), self.path_info
+                    errno.ENOENT, os.strerror(errno.ENOENT), self.fs_path
                 )
             else:
                 return None
@@ -74,13 +70,13 @@ class HashFile:
     def _check_hash(self, odb):
         from .stage import get_file_hash
 
-        actual = get_file_hash(
-            self.path_info, self.fs, self.hash_info.name, odb.state
+        _, actual = get_file_hash(
+            self.fs_path, self.fs, self.hash_info.name, odb.state
         )
 
         logger.trace(
             "cache '%s' expected '%s' actual '%s'",
-            self.path_info,
+            self.fs_path,
             self.hash_info,
             actual,
         )

dvc/objects/meta.py

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

dvc/objects/reference.py

@@ -8,7 +8,7 @@ from .errors import ObjectFormatError
 from .file import HashFile
 
 if TYPE_CHECKING:
-    from dvc.fs.base import BaseFileSystem
+    from dvc.fs.base import FileSystem
     from dvc.hash_info import HashInfo
     from dvc.types import AnyPath
 
@@ -25,23 +25,23 @@ class ReferenceHashFile(HashFile):
 
     def __init__(
         self,
-        path_info: "AnyPath",
-        fs: "BaseFileSystem",
+        fs_path: "AnyPath",
+        fs: "FileSystem",
         hash_info: "HashInfo",
         checksum: Optional[str] = None,
         **kwargs,
     ):
-        super().__init__(path_info, fs, hash_info, **kwargs)
-        self.checksum = checksum or fs.checksum(path_info)
+        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.path_info}"
+        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.path_info):
+        if not self.fs.exists(self.fs_path):
             raise FileNotFoundError(
-                errno.ENOENT, os.strerror(errno.ENOENT), self.path_info
+                errno.ENOENT, os.strerror(errno.ENOENT), self.fs_path
             )
         if self.checksum != self._get_checksum():
             raise ObjectFormatError(f"{self} is changed")
@@ -50,23 +50,22 @@ class ReferenceHashFile(HashFile):
 
     def _get_checksum(self) -> str:
         assert self.fs
-        return self.fs.checksum(self.path_info)
+        return self.fs.checksum(self.fs_path)
 
     def to_bytes(self):
         from dvc.fs.repo import RepoFileSystem
-        from dvc.path_info import PathInfo
 
         # 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
-        path_info = self.path_info
+        fs_path = self.fs_path
         if isinstance(self.fs, RepoFileSystem):
-            path_info = PathInfo(path_info).relative_to(self.fs.root_dir)
+            fs_path = self.fs.path.relpath(fs_path, self.fs.root_dir)
 
         dict_ = {
-            self.PARAM_PATH: path_info,
+            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),
@@ -87,7 +86,7 @@ class ReferenceHashFile(HashFile):
             raise ObjectFormatError("ReferenceHashFile is corrupted") from exc
 
         try:
-            path_info = dict_[cls.PARAM_PATH]
+            fs_path = dict_[cls.PARAM_PATH]
             hash_info = dict_[cls.PARAM_HASH]
         except KeyError as exc:
             raise ObjectFormatError("ReferenceHashFile is corrupted") from exc
@@ -98,19 +97,19 @@ class ReferenceHashFile(HashFile):
             config = dict(config_pairs)
             if RepoFileSystem.PARAM_REPO_URL in config:
                 fs = RepoFileSystem(**config)
-                path_info = fs.root_dir / path_info
+                fs_path = fs.path.join(fs.root_dir, fs_path)
             else:
-                fs_cls = get_fs_cls(config, scheme=path_info.scheme)
+                fs_cls = get_fs_cls(config, scheme=scheme)
                 fs = fs_cls(**config)
         return ReferenceHashFile(
-            path_info,
+            fs_path,
             fs,
             hash_info,
             checksum=dict_.get(cls.PARAM_CHECKSUM),
         )
 
     @staticmethod
-    def config_tuple(fs: "BaseFileSystem"):
+    def config_tuple(fs: "FileSystem"):
         return (
             fs.scheme,
             tuple(

dvc/objects/stage.py

@@ -1,10 +1,9 @@
 import hashlib
 import logging
 import os
-import pathlib
 from concurrent.futures import ThreadPoolExecutor
 from functools import partial
-from typing import TYPE_CHECKING, Dict, Tuple
+from typing import TYPE_CHECKING, Dict, Optional, Tuple
 
 from dvc.exceptions import DvcIgnoreInCollectedDirError
 from dvc.hash_info import HashInfo
@@ -14,10 +13,11 @@ from dvc.utils import file_md5
 
 from .db.reference import ReferenceObjectDB
 from .file import HashFile
+from .meta import Meta
 
 if TYPE_CHECKING:
-    from dvc.fs.base import BaseFileSystem
-    from dvc.types import AnyPath, DvcPath
+    from dvc.fs.base import FileSystem
+    from dvc.types import AnyPath
 
     from .db.base import ObjectDB
 
@@ -27,74 +27,92 @@ logger = logging.getLogger(__name__)
 _STAGING_MEMFS_PATH = "dvc-staging"
 
 
-def _upload_file(path_info, fs, odb, upload_odb):
+def _upload_file(from_fs_path, fs, odb, upload_odb):
     from dvc.utils import tmp_fname
     from dvc.utils.stream import HashedStreamReader
 
-    tmp_info = upload_odb.path_info / tmp_fname()
-    with fs.open(path_info, mode="rb", chunk_size=fs.CHUNK_SIZE) as stream:
+    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(path_info)
-        upload_odb.fs.upload(stream, tmp_info, desc=path_info.name, total=size)
+        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)
-    return path_info, odb.get(stream.hash_info)
+    meta = Meta(size=size)
+    return from_fs_path, meta, odb.get(stream.hash_info)
 
 
-def _get_file_hash(path_info, fs, name):
-    info = fs.info(path_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")
-        return HashInfo(name, info[name], size=info["size"])
+        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
 
-    func = getattr(fs, name, None)
-    if func:
-        return func(path_info)
-
-    if name == "md5":
-        return HashInfo(
-            name, file_md5(path_info, fs), size=fs.getsize(path_info)
-        )
-
-    raise NotImplementedError
+    meta = Meta(size=info["size"])
+    hash_info = HashInfo(name, hash_value)
+    return meta, hash_info
 
 
-def get_file_hash(path_info, fs, name, state=None):
+def get_file_hash(fs_path, fs, name, state=None):
     if state:
-        hash_info = state.get(  # pylint: disable=assignment-from-none
-            path_info, fs
+        meta, hash_info = state.get(  # pylint: disable=assignment-from-none
+            fs_path, fs
         )
         if hash_info:
-            return hash_info
+            return meta, hash_info
 
-    hash_info = _get_file_hash(path_info, fs, name)
+    meta, hash_info = _get_file_hash(fs_path, fs, name)
 
     if state:
         assert ".dir" not in hash_info.value
-        state.save(path_info, fs, hash_info)
+        state.save(fs_path, fs, hash_info)
 
-    return hash_info
+    return meta, hash_info
 
 
-def _get_file_obj(
-    path_info, fs, name, odb=None, upload_odb=None, dry_run=False
-):
+def _stage_file(fs_path, fs, name, odb=None, upload_odb=None, dry_run=False):
     state = odb.state if odb else None
-    hash_info = get_file_hash(path_info, fs, name, state=state)
+    meta, hash_info = get_file_hash(fs_path, fs, name, state=state)
     if upload_odb and not dry_run:
         assert odb and name == "md5"
-        return _upload_file(path_info, fs, odb, upload_odb)
+        return _upload_file(fs_path, fs, odb, upload_odb)
 
     if dry_run:
-        obj = HashFile(path_info, fs, hash_info)
+        obj = HashFile(fs_path, fs, hash_info)
     else:
-        odb.add(path_info, fs, hash_info, move=False)
+        odb.add(fs_path, fs, hash_info, hardlink=False)
         obj = odb.get(hash_info)
-    return path_info, obj
+
+    return fs_path, meta, obj
 
 
 def _build_objects(
-    path_info,
+    fs_path,
     fs,
     name,
     dvcignore=None,
@@ -103,9 +121,9 @@ def _build_objects(
     **kwargs,
 ):
     if dvcignore:
-        walk_iterator = dvcignore.walk_files(fs, path_info)
+        walk_iterator = dvcignore.find(fs, fs_path)
     else:
-        walk_iterator = fs.walk_files(path_info)
+        walk_iterator = fs.find(fs_path)
     with Tqdm(
         unit="md5",
         desc="Computing file/dir hashes (only done once)",
@@ -113,7 +131,7 @@ def _build_objects(
     ) as pbar:
         worker = pbar.wrap_fn(
             partial(
-                _get_file_obj,
+                _stage_file,
                 fs=fs,
                 name=name,
                 **kwargs,
@@ -125,17 +143,18 @@ def _build_objects(
             yield from executor.map(worker, walk_iterator)
 
 
-def _iter_objects(path_info, fs, name, **kwargs):
-    yield from _build_objects(path_info, fs, name, **kwargs)
+def _iter_objects(fs_path, fs, name, **kwargs):
+    yield from _build_objects(fs_path, fs, name, **kwargs)
 
 
-def _build_tree(path_info, 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_info, obj in _iter_objects(path_info, fs, name, **kwargs):
-        if DvcIgnore.DVCIGNORE_FILE == file_info.name:
-            raise DvcIgnoreInCollectedDirError(file_info.parent)
+    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"
@@ -145,11 +164,18 @@ def _build_tree(path_info, fs, name, **kwargs):
         #
         # Yes, this is a BUG, as long as we permit "/" in
         # filenames on Windows and "\" on Unix
-        tree.add(file_info.relative_to(path_info).parts, obj)
-    return tree
+
+        key = fs.path.relparts(file_fs_path, fs_path)
+        assert key
+        tree.add(key, meta, obj.hash_info)
+
+        tree_meta.size += meta.size
+        tree_meta.nfiles += 1
+
+    return tree_meta, tree
 
 
-def _get_tree_obj(path_info, fs, fs_info, name, odb=None, **kwargs):
+def _stage_tree(fs_path, fs, fs_info, name, odb=None, **kwargs):
     from .tree import Tree
 
     value = fs_info.get(name)
@@ -157,51 +183,50 @@ def _get_tree_obj(path_info, fs, fs_info, name, odb=None, **kwargs):
         hash_info = HashInfo(name, value)
         try:
             tree = Tree.load(odb, hash_info)
-            # NOTE: loaded entries are naive objects with hash_infos but no
-            # path_info. For staging trees, obj.path_info should be relative
-            # to the staging src `path_info` and src fs
-            for key, entry in tree:
-                entry.fs = fs
-                entry.path_info = path_info.joinpath(*key)
-            return tree
+            return Meta(nfiles=len(tree)), tree
         except FileNotFoundError:
             pass
 
-    tree = _build_tree(path_info, fs, name, odb=odb, **kwargs)
+    meta, tree = _build_tree(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
-            path_info, fs
+        _, hash_info = state.get(  # pylint: disable=assignment-from-none
+            fs_path, fs
         )
     tree.digest(hash_info=hash_info)
-    odb.add(tree.path_info, tree.fs, tree.hash_info, move=True)
+    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.path_info)
+        tree.fs.remove(tree.fs_path)
     tree.fs = raw.fs
-    tree.path_info = raw.path_info
-    return tree
+    tree.fs_path = raw.fs_path
+    return meta, tree
 
 
 _url_cache: Dict[str, str] = {}
 
 
-def _make_staging_url(path_info: "AnyPath"):
-    from dvc.path_info import CloudURLInfo
+def _make_staging_url(
+    fs: "FileSystem", odb: "ObjectDB", fs_path: Optional[str]
+):
     from dvc.scheme import Schemes
 
-    url = CloudURLInfo(f"{Schemes.MEMORY}://{_STAGING_MEMFS_PATH}")
-    if path_info:
-        if isinstance(path_info, (str, pathlib.PurePath)):
-            path = os.path.abspath(path_info)
-        else:
-            path = str(path_info)
-        if path not in _url_cache:
-            _url_cache[path] = hashlib.sha256(path.encode("utf-8")).hexdigest()
-        url /= _url_cache[path]
+    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
 
 
@@ -214,18 +239,18 @@ def _get_staging(odb: "ObjectDB") -> "ObjectDB":
     from dvc.fs.memory import MemoryFileSystem
 
     fs = MemoryFileSystem()
-    path_info = _make_staging_url(odb.path_info)
+    fs_path = _make_staging_url(fs, odb, odb.fs_path)
     state = odb.state
-    return ReferenceObjectDB(fs, path_info, state=state)
+    return ReferenceObjectDB(fs, fs_path, state=state)
 
 
-def _load_from_state(odb, staging, path_info, fs, name):
+def _load_from_state(odb, staging, fs_path, fs, name):
     from . import check, load
     from .errors import ObjectFormatError
     from .tree import Tree
 
     state = odb.state
-    hash_info = state.get(path_info, fs)
+    meta, hash_info = state.get(fs_path, fs)
     if hash_info:
         for odb_ in (odb, staging):
             if odb_.exists(hash_info):
@@ -233,16 +258,9 @@ def _load_from_state(odb, staging, path_info, fs, name):
                     obj = load(odb_, hash_info)
                     check(odb_, obj, check_hash=False)
                     if isinstance(obj, Tree):
-                        obj.hash_info.nfiles = len(obj)
-                        for key, entry in obj:
-                            entry.fs = fs
-                            entry.path_info = path_info.joinpath(*key)
-                    else:
-                        obj.fs = fs
-                        obj.path_info = path_info
+                        meta.nfiles = len(obj)
                     assert obj.hash_info.name == name
-                    obj.hash_info.size = hash_info.size
-                    return odb_, obj
+                    return odb_, meta, obj
                 except (ObjectFormatError, FileNotFoundError):
                     pass
     raise FileNotFoundError
@@ -255,10 +273,10 @@ def _stage_external_tree_info(odb, tree, name):
     # that would be universal for all clouds.
     assert odb and name != "md5"
 
-    odb.add(tree.path_info, tree.fs, tree.hash_info)
+    odb.add(tree.fs_path, tree.fs, tree.hash_info)
     raw = odb.get(tree.hash_info)
-    hash_info = get_file_hash(raw.path_info, raw.fs, name, state=odb.state)
-    tree.path_info = raw.path_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
@@ -269,13 +287,13 @@ def _stage_external_tree_info(odb, tree, name):
 
 def stage(
     odb: "ObjectDB",
-    path_info: "DvcPath",
-    fs: "BaseFileSystem",
+    fs_path: "AnyPath",
+    fs: "FileSystem",
     name: str,
     upload: bool = False,
     dry_run: bool = False,
     **kwargs,
-) -> Tuple["ObjectDB", "HashFile"]:
+) -> Tuple["ObjectDB", "Meta", "HashFile"]:
     """Stage (prepare) objects from the given path for addition to an ODB.
 
     Returns at tuple of (staging_odb, object) where addition to the ODB can
@@ -289,19 +307,20 @@ def stage(
     ODB filesystem, and staged objects will reference the uploaded path rather
     than the original source path.
     """
-    assert path_info and path_info.scheme == fs.scheme
+    assert fs_path
+    # assert scheme(fs_path) == fs.scheme
 
-    details = fs.info(path_info)
+    details = fs.info(fs_path)
     staging = _get_staging(odb)
     if odb:
         try:
-            return _load_from_state(odb, staging, path_info, fs, name)
+            return _load_from_state(odb, staging, fs_path, fs, name)
         except FileNotFoundError:
             pass
 
     if details["type"] == "directory":
-        obj = _get_tree_obj(
-            path_info,
+        meta, obj = _stage_tree(
+            fs_path,
             fs,
             details,
             name,
@@ -314,8 +333,8 @@ def stage(
         if name != "md5":
             obj = _stage_external_tree_info(odb, obj, name)
     else:
-        _, obj = _get_file_obj(
-            path_info,
+        _, meta, obj = _stage_file(
+            fs_path,
             fs,
             name,
             odb=staging,
@@ -324,6 +343,6 @@ def stage(
         )
 
     if odb and odb.state and obj.hash_info:
-        odb.state.save(path_info, fs, obj.hash_info)
+        odb.state.save(fs_path, fs, obj.hash_info)
 
-    return staging, obj
+    return staging, meta, obj

dvc/objects/status.py

@@ -56,7 +56,7 @@ def _indexed_dir_hashes(odb, index, dir_objs, name, cache_odb):
                 tree = Tree.load(cache_odb, HashInfo(name, dir_hash))
             except FileNotFoundError:
                 continue
-        file_hashes = [entry.hash_info.value for _, entry in tree]
+        file_hashes = [oid.value for _, _, oid in tree]
         if dir_hash not in index:
             logger.debug(
                 "Indexing new .dir '%s' with '%s' nested files",
@@ -86,7 +86,7 @@ def status(
         exists: objs that exist in odb
         missing: objs that do not exist in ODB
     """
-    logger.debug("Preparing to collect status from '%s'", odb.path_info)
+    logger.debug("Preparing to collect status from '%s'", odb.fs_path)
     if not name:
         name = odb.fs.PARAM_CHECKSUM
 
@@ -102,9 +102,9 @@ def status(
                 tree = None
             else:
                 tree = Tree.load(cache_odb, hash_info)
-                for _, entry in tree:
-                    assert entry.hash_info and entry.hash_info.value
-                    hash_infos[entry.hash_info.value] = entry.hash_info
+                for _, _, oid in tree:
+                    assert oid and oid.value
+                    hash_infos[oid.value] = oid
             if index:
                 dir_objs[hash_info.value] = tree
         hash_infos[hash_info.value] = hash_info
@@ -116,7 +116,7 @@ def status(
     hashes: Set[str] = set(hash_infos.keys())
     exists: Set[str] = set()
 
-    logger.debug("Collecting status from '%s'", odb.path_info)
+    logger.debug("Collecting status from '%s'", odb.fs_path)
     if index and hashes:
         if dir_objs:
             exists = hashes.intersection(
@@ -128,9 +128,7 @@ def status(
             hashes.difference_update(exists)
 
     if hashes:
-        exists.update(
-            odb.hashes_exist(hashes, name=str(odb.path_info), **kwargs)
-        )
+        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)},

dvc/objects/transfer.py

@@ -20,9 +20,9 @@ logger = logging.getLogger(__name__)
 
 def _log_exceptions(func):
     @wraps(func)
-    def wrapper(path_info, *args, **kwargs):
+    def wrapper(fs_path, *args, **kwargs):
         try:
-            func(path_info, *args, **kwargs)
+            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
@@ -31,7 +31,7 @@ def _log_exceptions(func):
             if isinstance(exc, OSError) and exc.errno == errno.EMFILE:
                 raise
 
-            logger.exception("failed to transfer '%s'", path_info)
+            logger.exception("failed to transfer '%s'", fs_path)
             return 1
 
     return wrapper
@@ -74,7 +74,7 @@ def _do_transfer(
         dir_obj = find_tree_by_obj_id([cache_odb, src], dir_hash)
         assert dir_obj
 
-        entry_ids = {entry.hash_info for _, entry in dir_obj}
+        entry_ids = {oid for _, _, oid in dir_obj}
         bound_file_ids = all_file_ids & entry_ids
         all_file_ids -= entry_ids
 
@@ -87,8 +87,8 @@ def _do_transfer(
             )
             logger.error(
                 "failed to upload '%s' to '%s'",
-                src.get(dir_obj.hash_info).path_info,
-                dest.get(dir_obj.hash_info).path_info,
+                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):
@@ -117,7 +117,7 @@ def _do_transfer(
     # index successfully pushed dirs
     if dest_index:
         for dir_obj in succeeded_dir_objs:
-            file_hashes = {entry.hash_info.value for _, entry in dir_obj}
+            file_hashes = {oid.value for _, _, oid in dir_obj}
             logger.debug(
                 "Indexing pushed dir '%s' with '%s' nested files",
                 dir_obj.hash_info,
@@ -133,7 +133,7 @@ def transfer(
     obj_ids: Iterable["HashInfo"],
     jobs: Optional[int] = None,
     verify: bool = False,
-    move: bool = False,
+    hardlink: bool = False,
     **kwargs,
 ) -> int:
     """Transfer (copy) the specified objects from one ODB to another.
@@ -144,8 +144,8 @@ def transfer(
 
     logger.debug(
         "Preparing to transfer data from '%s' to '%s'",
-        src.path_info,
-        dest.path_info,
+        src.fs_path,
+        dest.fs_path,
     )
     if src == dest:
         return 0
@@ -157,7 +157,11 @@ def transfer(
     def func(hash_info: "HashInfo") -> None:
         obj = src.get(hash_info)
         return dest.add(
-            obj.path_info, obj.fs, obj.hash_info, verify=verify, move=move
+            obj.fs_path,
+            obj.fs,
+            obj.hash_info,
+            verify=verify,
+            hardlink=hardlink,
         )
 
     total = len(status.new)

dvc/objects/tree.py

@@ -12,6 +12,8 @@ from .stage import get_file_hash
 if TYPE_CHECKING:
     from dvc.hash_info import HashInfo
 
+    from .meta import Meta
+
 logger = logging.getLogger(__name__)
 
 
@@ -20,7 +22,7 @@ class Tree(HashFile):
 
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
-        self._dict: Dict[Tuple[str], "HashFile"] = {}
+        self._dict: Dict[Tuple[str], Tuple["Meta", "HashInfo"]] = {}
 
     @cached_property
     def trie(self):
@@ -28,38 +30,34 @@ class Tree(HashFile):
 
         return Trie(self._dict)
 
-    def add(self, key, obj):
+    def add(self, key: Tuple[str], meta: "Meta", oid: "HashInfo"):
         self.__dict__.pop("trie", None)
-        self._dict[key] = obj
+        self._dict[key] = (meta, oid)
 
     def digest(self, hash_info: Optional["HashInfo"] = None):
         from dvc.fs.memory import MemoryFileSystem
-        from dvc.path_info import CloudURLInfo
         from dvc.utils import tmp_fname
 
         memfs = MemoryFileSystem()
-        path_info = CloudURLInfo("memory://{}".format(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.fs_path = fs_path
         if hash_info:
             self.hash_info = hash_info
         else:
-            self.hash_info = get_file_hash(path_info, memfs, "md5")
+            _, self.hash_info = get_file_hash(fs_path, memfs, "md5")
             assert self.hash_info.value
             self.hash_info.value += ".dir"
-        try:
-            self.hash_info.size = sum(obj.size for _, obj in self)
-        except TypeError:
-            self.hash_info.size = None
-        self.hash_info.nfiles = len(self)
 
     def __len__(self):
         return len(self._dict)
 
     def __iter__(self):
-        yield from self._dict.items()
+        yield from (
+            (key, value[0], value[1]) for key, value in self._dict.items()
+        )
 
     def as_dict(self):
         return self._dict.copy()
@@ -73,10 +71,10 @@ class Tree(HashFile):
                 {
                     # NOTE: not using hash_info.to_dict() because we don't want
                     # size/nfiles fields at this point.
-                    obj.hash_info.name: obj.hash_info.value,
+                    oid.name: oid.value,
                     self.PARAM_RELPATH: posixpath.sep.join(parts),
                 }
-                for parts, obj in self._dict.items()  # noqa: B301
+                for parts, _, oid in self  # noqa: B301
             ),
             key=itemgetter(self.PARAM_RELPATH),
         )
@@ -94,8 +92,7 @@ class Tree(HashFile):
             relpath = entry.pop(cls.PARAM_RELPATH)
             parts = tuple(relpath.split(posixpath.sep))
             hash_info = HashInfo.from_dict(entry)
-            obj = HashFile(None, None, hash_info)
-            tree.add(parts, obj)
+            tree.add(parts, None, hash_info)
         return tree
 
     @classmethod
@@ -103,7 +100,7 @@ class Tree(HashFile):
         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
@@ -111,15 +108,13 @@ 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
@@ -129,43 +124,42 @@ class Tree(HashFile):
         inside prefix.
 
         The returned tree will contain the original tree's hash_info and
-        path_info.
+        fs_path.
 
         Returns an empty tree if no object exists at the specified prefix.
         """
-        tree = Tree(self.path_info, self.fs, self.hash_info)
+        tree = Tree(self.fs_path, self.fs, self.hash_info)
         try:
-            for key, obj in self.trie.items(prefix):
-                tree.add(key, obj)
+            for key, (meta, oid) in self.trie.items(prefix):
+                tree.add(key, meta, oid)
         except KeyError:
             pass
         return tree
 
-    def get(self, prefix: Tuple[str]) -> Optional[HashFile]:
+    def get(self, odb, prefix: Tuple[str]) -> Optional[HashFile]:
         """Return object at the specified prefix in this tree.
 
         Returns None if no object exists at the specified prefix.
         """
-        obj = self._dict.get(prefix)
-        if obj:
-            return obj
+        _, oid = self._dict.get(prefix) or (None, None)
+        if oid:
+            return odb.get(oid)
 
         tree = Tree(None, None, None)
         depth = len(prefix)
         try:
-            for key, obj in self.trie.items(prefix):
-                tree.add(key[depth:], obj)
+            for key, (meta, entry_oid) in self.trie.items(prefix):
+                tree.add(key[depth:], meta, entry_oid)
         except KeyError:
             return None
         tree.digest()
         return tree
 
 
-def _get_dir_size(odb, tree):
+def du(odb, tree):
     try:
         return sum(
-            odb.fs.getsize(odb.hash_to_path_info(obj.hash_info.value))
-            for _, obj in tree
+            odb.fs.getsize(odb.hash_to_path(oid.value)) for _, _, oid in tree
         )
     except FileNotFoundError:
         return None
@@ -226,11 +220,8 @@ def merge(odb, ancestor_info, our_info, their_info):
     merged_dict = _merge(ancestor.as_dict(), our.as_dict(), their.as_dict())
 
     merged = Tree(None, None, None)
-    for key, hi in merged_dict.items():
-        merged.add(key, hi)
+    for key, (meta, oid) in merged_dict.items():
+        merged.add(key, meta, oid)
     merged.digest()
 
-    odb.add(merged.path_info, merged.fs, merged.hash_info)
-    hash_info = merged.hash_info
-    hash_info.size = _get_dir_size(odb, merged)
-    return hash_info
+    return merged

dvc/output.py

@@ -1,7 +1,6 @@
 import logging
 import os
 from collections import defaultdict
-from copy import copy
 from typing import TYPE_CHECKING, Dict, Optional, Set, Type
 from urllib.parse import urlparse
 
@@ -9,7 +8,6 @@ from funcy import collecting, project
 from voluptuous import And, Any, Coerce, Length, Lower, Required, SetTo
 
 from dvc import objects, prompt
-from dvc.checkout import checkout
 from dvc.exceptions import (
     CheckoutError,
     CollectCacheError,
@@ -17,6 +15,7 @@ from dvc.exceptions import (
     MergeError,
     RemoteCacheRequiredError,
 )
+from dvc.objects.checkout import checkout
 
 from .fs import get_cloud_fs
 from .fs.hdfs import HDFSFileSystem
@@ -26,6 +25,7 @@ from .hash_info import HashInfo
 from .istextfile import istextfile
 from .objects import Tree
 from .objects.errors import ObjectFormatError
+from .objects.meta import Meta
 from .objects.stage import stage as ostage
 from .objects.transfer import transfer as otransfer
 from .scheme import Schemes
@@ -292,7 +292,7 @@ class Output:
     ):
         self.repo = stage.repo if stage else None
 
-        fs_cls, fs_config, path_info = get_cloud_fs(self.repo, url=path)
+        fs_cls, fs_config, fs_path = get_cloud_fs(self.repo, url=path)
         self.fs = fs_cls(**fs_config)
 
         if (
@@ -309,13 +309,13 @@ class Output:
         # here is a list and comments:
         #
         #   .def_path - path from definition in DVC file
-        #   .path_info - PathInfo/URLInfo structured resolved path
         #   .fspath - local only, resolved
         #   .__str__ - for presentation purposes, def_path/relpath
         #
         # By resolved path, which contains actual location,
         # should be absolute and don't contain remote:// refs.
         self.stage = stage
+        self.meta = Meta.from_dict(info)
         self.hash_info = HashInfo.from_dict(info)
         self.use_cache = False if self.IS_DEPENDENCY else cache
         self.metric = False if self.IS_DEPENDENCY else metric
@@ -325,32 +325,32 @@ class Output:
         self.live = live
         self.desc = desc
 
-        self.path_info = self._parse_path(self.fs, path_info)
+        self.fs_path = self._parse_path(self.fs, fs_path)
         if self.use_cache and self.odb is None:
-            raise RemoteCacheRequiredError(self.path_info)
+            raise RemoteCacheRequiredError(self.fs.scheme, self.fs_path)
 
         self.obj = None
         self.isexec = False if self.IS_DEPENDENCY else isexec
 
         self.remote = remote
 
-    def _parse_path(self, fs, path_info):
+    def _parse_path(self, fs, fs_path):
         if fs.scheme != "local":
-            return path_info
+            return fs_path
 
         parsed = urlparse(self.def_path)
         if parsed.scheme != "remote":
             # NOTE: we can path either from command line or .dvc file,
             # so we should expect both posix and windows style paths.
-            # PathInfo accepts both, i.e. / works everywhere, \ only on win.
+            # paths accepts both, i.e. / works everywhere, \ only on win.
             #
             # FIXME: if we have Windows path containing / or posix one with \
             # then we have #2059 bug and can't really handle that.
-            if self.stage and not path_info.is_absolute():
-                path_info = self.stage.wdir / path_info
+            if self.stage and not os.path.isabs(fs_path):
+                fs_path = fs.path.join(self.stage.wdir, fs_path)
 
-        abs_p = os.path.abspath(os.path.normpath(path_info))
-        return fs.PATH_CLS(abs_p)
+        abs_p = os.path.abspath(os.path.normpath(fs_path))
+        return abs_p
 
     def __repr__(self):
         return "{class_name}: '{def_path}'".format(
@@ -370,9 +370,9 @@ class Output:
 
         cur_dir = os.getcwd()
         if path_isin(cur_dir, self.repo.root_dir):
-            return relpath(self.path_info, cur_dir)
+            return relpath(self.fs_path, cur_dir)
 
-        return relpath(self.path_info, self.repo.root_dir)
+        return relpath(self.fs_path, self.repo.root_dir)
 
     @property
     def scheme(self):
@@ -390,7 +390,7 @@ class Output:
             return False
 
         return self.repo and path_isin(
-            os.path.realpath(self.path_info), self.repo.root_dir
+            os.path.realpath(self.fs_path), self.repo.root_dir
         )
 
     @property
@@ -406,7 +406,9 @@ class Output:
 
     @property
     def cache_path(self):
-        return self.odb.hash_to_path_info(self.hash_info.value).url
+        return self.odb.fs.unstrip_protocol(
+            self.odb.hash_to_path(self.hash_info.value)
+        )
 
     def get_hash(self):
         if self.use_cache:
@@ -415,9 +417,9 @@ class Output:
         else:
             odb = self.repo.odb.local
             name = self.fs.PARAM_CHECKSUM
-        _, obj = ostage(
+        _, _, obj = ostage(
             odb,
-            self.path_info,
+            self.fs_path,
             self.fs,
             name,
             dvcignore=self.dvcignore,
@@ -437,10 +439,10 @@ class Output:
 
     @property
     def exists(self):
-        if self._is_path_dvcignore(self.path_info):
+        if self._is_path_dvcignore(self.fs_path):
             return False
 
-        return self.fs.exists(self.path_info)
+        return self.fs.exists(self.fs_path)
 
     def changed_checksum(self):
         return self.hash_info != self.get_hash()
@@ -490,17 +492,17 @@ class Output:
 
     @property
     def is_empty(self):
-        return self.fs.is_empty(self.path_info)
+        return self.fs.is_empty(self.fs_path)
 
     def isdir(self):
-        if self._is_path_dvcignore(self.path_info):
+        if self._is_path_dvcignore(self.fs_path):
             return False
-        return self.fs.isdir(self.path_info)
+        return self.fs.isdir(self.fs_path)
 
     def isfile(self):
-        if self._is_path_dvcignore(self.path_info):
+        if self._is_path_dvcignore(self.fs_path):
             return False
-        return self.fs.isfile(self.path_info)
+        return self.fs.isfile(self.fs_path)
 
     # pylint: disable=no-member
 
@@ -511,13 +513,13 @@ class Output:
         if self.repo.scm.is_tracked(self.fspath):
             raise OutputAlreadyTrackedError(self)
 
-        self.repo.scm.ignore(self.fspath)
+        self.repo.scm_context.ignore(self.fspath)
 
     def ignore_remove(self):
         if not self.use_scm_ignore:
             return
 
-        self.repo.scm.ignore_remove(self.fspath)
+        self.repo.scm_context.ignore_remove(self.fspath)
 
     # pylint: enable=no-member
 
@@ -537,7 +539,15 @@ class Output:
             self.verify_metric()
 
         if not self.use_cache:
-            self.hash_info = self.get_hash()
+            _, self.meta, obj = ostage(
+                self.repo.odb.local,
+                self.fs_path,
+                self.fs,
+                self.fs.PARAM_CHECKSUM,
+                dvcignore=self.dvcignore,
+                dry_run=True,
+            )
+            self.hash_info = obj.hash_info
             if not self.IS_DEPENDENCY:
                 logger.debug(
                     "Output '%s' doesn't use cache. Skipping saving.", self
@@ -546,23 +556,19 @@ class Output:
 
         assert not self.IS_DEPENDENCY
 
-        if not self.changed():
-            logger.debug("Output '%s' didn't change. Skipping saving.", self)
-            return
-
-        _, self.obj = ostage(
+        _, self.meta, self.obj = ostage(
             self.odb,
-            self.path_info,
+            self.fs_path,
             self.fs,
             self.odb.fs.PARAM_CHECKSUM,
             dvcignore=self.dvcignore,
         )
         self.hash_info = self.obj.hash_info
-        self.isexec = self.isfile() and self.fs.isexec(self.path_info)
+        self.isexec = self.isfile() and self.fs.isexec(self.fs_path)
 
     def set_exec(self):
         if self.isfile() and self.isexec:
-            self.odb.set_exec(self.path_info)
+            self.odb.set_exec(self.fs_path)
 
     def commit(self, filter_info=None):
         if not self.exists:
@@ -574,14 +580,14 @@ class Output:
             granular = (
                 self.is_dir_checksum
                 and filter_info
-                and filter_info != self.path_info
+                and filter_info != self.fs_path
             )
             if granular:
                 obj = self._commit_granular_dir(filter_info)
             else:
-                staging, obj = ostage(
+                staging, _, obj = ostage(
                     self.odb,
-                    filter_info or self.path_info,
+                    filter_info or self.fs_path,
                     self.fs,
                     self.odb.fs.PARAM_CHECKSUM,
                     dvcignore=self.dvcignore,
@@ -591,10 +597,10 @@ class Output:
                     self.odb,
                     {obj.hash_info},
                     shallow=False,
-                    move=True,
+                    hardlink=True,
                 )
             checkout(
-                filter_info or self.path_info,
+                filter_info or self.fs_path,
                 self.fs,
                 obj,
                 self.odb,
@@ -605,30 +611,34 @@ class Output:
             self.set_exec()
 
     def _commit_granular_dir(self, filter_info):
-        prefix = filter_info.relative_to(self.path_info).parts
-        staging, save_obj = ostage(
+        prefix = self.fs.path.parts(
+            self.fs.path.relpath(filter_info, self.fs_path)
+        )
+        staging, _, save_obj = ostage(
             self.odb,
-            self.path_info,
+            self.fs_path,
             self.fs,
             self.odb.fs.PARAM_CHECKSUM,
             dvcignore=self.dvcignore,
         )
         save_obj = save_obj.filter(prefix)
-        checkout_obj = save_obj.get(prefix)
+        checkout_obj = save_obj.get(self.odb, prefix)
         otransfer(
             staging,
             self.odb,
-            {save_obj.hash_info} | {entry.hash_info for _, entry in save_obj},
+            {save_obj.hash_info} | {oid for _, _, oid in save_obj},
             shallow=True,
-            move=True,
+            hardlink=True,
         )
         return checkout_obj
 
     def dumpd(self):
-        ret = copy(self.hash_info.to_dict())
+        ret = {**self.hash_info.to_dict(), **self.meta.to_dict()}
 
         if self.is_in_repo:
-            path = self.path_info.relpath(self.stage.wdir).as_posix()
+            path = self.fs.path.as_posix(
+                relpath(self.fs_path, self.stage.wdir)
+            )
         else:
             path = self.def_path
 
@@ -679,22 +689,21 @@ class Output:
         if not self.metric or self.plot:
             return
 
-        path = os.fspath(self.path_info)
-        if not os.path.exists(path):
+        if not os.path.exists(self.fs_path):
             return
 
         name = "metrics" if self.metric else "plot"
-        if os.path.isdir(path):
+        if os.path.isdir(self.fs_path):
             msg = "directory '%s' cannot be used as %s."
-            logger.debug(msg, str(self.path_info), name)
+            logger.debug(msg, str(self), name)
             return
 
-        if not istextfile(path, self.fs):
+        if not istextfile(self.fs_path, self.fs):
             msg = "binary file '{}' cannot be used as {}."
-            raise DvcException(msg.format(self.path_info, name))
+            raise DvcException(msg.format(self.fs_path, name))
 
     def download(self, to, jobs=None):
-        self.fs.download(self.path_info, to.path_info, jobs=jobs)
+        self.fs.download(self.fs_path, to.fs_path, jobs=jobs)
 
     def get_obj(self, filter_info=None, **kwargs):
         if self.obj:
@@ -707,9 +716,10 @@ class Output:
         else:
             return None
 
-        if filter_info and filter_info != self.path_info:
-            prefix = filter_info.relative_to(self.path_info).parts
-            obj = obj.get(prefix)
+        fs_path = self.fs.path
+        if filter_info and filter_info != self.fs_path:
+            prefix = fs_path.relparts(filter_info, self.fs_path)
+            obj = obj.get(self.odb, prefix)
 
         return obj
 
@@ -726,12 +736,12 @@ class Output:
         if not self.use_cache:
             if progress_callback:
                 progress_callback(
-                    str(self.path_info), self.get_files_number(filter_info)
+                    self.fs_path, self.get_files_number(filter_info)
                 )
             return None
 
         obj = self.get_obj(filter_info=filter_info)
-        if not obj and (filter_info and filter_info != self.path_info):
+        if not obj and (filter_info and filter_info != self.fs_path):
             # backward compatibility
             return None
 
@@ -744,7 +754,7 @@ class Output:
 
         try:
             modified = checkout(
-                filter_info or self.path_info,
+                filter_info or self.fs_path,
                 self.fs,
                 obj,
                 self.odb,
@@ -762,7 +772,7 @@ class Output:
         return added, False if added else modified
 
     def remove(self, ignore_remove=False):
-        self.fs.remove(self.path_info)
+        self.fs.remove(self.fs_path)
         if self.scheme != Schemes.LOCAL:
             return
 
@@ -772,16 +782,16 @@ class Output:
     def move(self, out):
         # pylint: disable=no-member
         if self.scheme == "local" and self.use_scm_ignore:
-            self.repo.scm.ignore_remove(self.fspath)
+            self.repo.scm_context.ignore_remove(self.fspath)
 
-        self.fs.move(self.path_info, out.path_info)
+        self.fs.move(self.fs_path, out.fs_path)
         self.def_path = out.def_path
-        self.path_info = out.path_info
+        self.fs_path = out.fs_path
         self.save()
         self.commit()
 
         if self.scheme == "local" and self.use_scm_ignore:
-            self.repo.scm.ignore(self.fspath)
+            self.repo.scm_context.ignore(self.fspath)
 
     def transfer(
         self, source, odb=None, jobs=None, update=False, no_progress_bar=False
@@ -803,7 +813,7 @@ class Output:
 
         upload = not (update and from_fs.isdir(from_info))
         jobs = jobs or min((from_fs.jobs, odb.fs.jobs))
-        staging, obj = ostage(
+        staging, self.meta, obj = ostage(
             odb,
             from_info,
             from_fs,
@@ -817,7 +827,7 @@ class Output:
             odb,
             {obj.hash_info},
             jobs=jobs,
-            move=upload,
+            hardlink=False,
             shallow=False,
         )
 
@@ -831,15 +841,15 @@ class Output:
         if not self.hash_info.isdir:
             return 1
 
-        if not filter_info or filter_info == self.path_info:
-            return self.hash_info.nfiles or 0
+        if not filter_info or filter_info == self.fs_path:
+            return self.meta.nfiles or 0
 
         obj = self.get_obj(filter_info=filter_info)
         return len(obj) if obj else 0
 
     def unprotect(self):
         if self.exists:
-            self.odb.unprotect(self.path_info)
+            self.odb.unprotect(self.fs_path)
 
     def get_dir_cache(self, **kwargs):
         if not self.is_dir_checksum:
@@ -881,7 +891,7 @@ class Output:
                 "Cache for files inside will be lost. "
                 "Would you like to continue? Use '-f' to force."
             )
-            if not force and not prompt.confirm(msg.format(self.path_info)):
+            if not force and not prompt.confirm(msg.format(self.fs_path)):
                 raise CollectCacheError(
                     "unable to fully collect used cache"
                     " without cache for directory '{}'".format(self)
@@ -889,8 +899,10 @@ class Output:
             return None
 
         obj = self.get_obj()
-        if filter_info and filter_info != self.path_info:
-            prefix = filter_info.relative_to(self.path_info).parts
+        if filter_info and filter_info != self.fs_path:
+            prefix = self.fs.path.parts(
+                self.fs.path.relpath(filter_info, self.fs_path)
+            )
             obj = obj.filter(prefix)
         return obj
 
@@ -946,9 +958,9 @@ class Output:
         obj.hash_info.obj_name = name
         oids = {obj.hash_info}
         if isinstance(obj, Tree):
-            for key, entry_obj in obj:
-                entry_obj.hash_info.obj_name = self.fs.sep.join([name, *key])
-                oids.add(entry_obj.hash_info)
+            for key, _, oid in obj:
+                oid.obj_name = self.fs.sep.join([name, *key])
+                oids.add(oid)
         return oids
 
     def get_used_external(
@@ -981,8 +993,8 @@ class Output:
 
         ignored = [
             self.fs.PARAM_CHECKSUM,
-            HashInfo.PARAM_SIZE,
-            HashInfo.PARAM_NFILES,
+            Meta.PARAM_SIZE,
+            Meta.PARAM_NFILES,
         ]
 
         for opt in ignored:
@@ -1000,7 +1012,7 @@ class Output:
             )
 
     def merge(self, ancestor, other):
-        from dvc.objects.tree import merge
+        from dvc.objects.tree import du, merge
 
         assert other
 
@@ -1013,13 +1025,20 @@ class Output:
         self._check_can_merge(self)
         self._check_can_merge(other)
 
-        self.hash_info = merge(
+        merged = merge(
             self.odb, ancestor_info, self.hash_info, other.hash_info
         )
+        self.odb.add(merged.fs_path, merged.fs, merged.hash_info)
+
+        self.hash_info = merged.hash_info
+        self.meta = Meta(
+            size=du(self.odb, merged),
+            nfiles=len(merged),
+        )
 
     @property
     def fspath(self):
-        return self.path_info.fspath
+        return self.fs_path
 
     @property
     def is_decorated(self) -> bool:
@@ -1040,8 +1059,8 @@ ARTIFACT_SCHEMA = {
     Output.PARAM_PLOT: bool,
     Output.PARAM_PERSIST: bool,
     Output.PARAM_CHECKPOINT: bool,
-    HashInfo.PARAM_SIZE: int,
-    HashInfo.PARAM_NFILES: int,
+    Meta.PARAM_SIZE: int,
+    Meta.PARAM_NFILES: int,
     Output.PARAM_ISEXEC: bool,
 }
 

dvc/parsing/__init__.py

@@ -17,7 +17,6 @@ from funcy import cached_property, collecting, first, isa, join, reraise
 
 from dvc.exceptions import DvcException
 from dvc.parsing.interpolate import ParseError
-from dvc.path_info import PathInfo
 from dvc.utils import relpath
 
 from .context import (
@@ -134,10 +133,10 @@ def make_definition(
 
 
 class DataResolver:
-    def __init__(self, repo: "Repo", wdir: PathInfo, d: dict):
+    def __init__(self, repo: "Repo", wdir: str, d: dict):
         self.fs = fs = repo.fs
         self.wdir = wdir
-        self.relpath = relpath(self.wdir / "dvc.yaml")
+        self.relpath = relpath(fs.path.join(self.wdir, "dvc.yaml"))
 
         vars_ = d.get(VARS_KWD, [])
         check_interpolations(vars_, VARS_KWD, self.relpath)
@@ -232,7 +231,7 @@ class EntryDefinition:
 
     def _resolve_wdir(
         self, context: Context, name: str, wdir: str = None
-    ) -> PathInfo:
+    ) -> str:
         if not wdir:
             return self.wdir
 
@@ -240,7 +239,7 @@ class EntryDefinition:
             wdir = to_str(context.resolve_str(wdir))
         except (ContextError, ParseError) as exc:
             format_and_raise(exc, f"'{self.where}.{name}.wdir'", self.relpath)
-        return self.wdir / wdir
+        return self.resolver.fs.path.join(self.wdir, wdir)
 
     def resolve(self, **kwargs):
         try:

dvc/parsing/context.py

@@ -19,7 +19,6 @@ from dvc.parsing.interpolate import (
     recurse,
     str_interpolate,
 )
-from dvc.path_info import PathInfo
 from dvc.utils import relpath
 
 logger = logging.getLogger(__name__)
@@ -356,7 +355,7 @@ class Context(CtxDict):
 
     @classmethod
     def load_from(
-        cls, fs, path: PathInfo, select_keys: List[str] = None
+        cls, fs, path: str, select_keys: List[str] = None
     ) -> "Context":
         from dvc.utils.serialize import LOADERS
 
@@ -396,18 +395,17 @@ class Context(CtxDict):
             raise ReservedKeyError(matches)
         return super().merge_update(other, overwrite=overwrite)
 
-    def merge_from(self, fs, item: str, wdir: PathInfo, overwrite=False):
+    def merge_from(self, fs, item: str, wdir: str, overwrite=False):
         path, _, keys_str = item.partition(":")
         select_keys = lfilter(bool, keys_str.split(",")) if keys_str else None
 
-        abspath = os.path.abspath(wdir / path)
-        path_info = PathInfo(abspath)
+        abspath = os.path.abspath(fs.path.join(wdir, path))
         if abspath in self.imports:
             if not select_keys and self.imports[abspath] is None:
                 return  # allow specifying complete filepath multiple times
             self.check_loaded(abspath, item, select_keys)
 
-        ctx = Context.load_from(fs, path_info, select_keys)
+        ctx = Context.load_from(fs, abspath, select_keys)
 
         try:
             self.merge_update(ctx, overwrite=overwrite)
@@ -439,12 +437,12 @@ class Context(CtxDict):
         self,
         fs,
         vars_: List,
-        wdir: PathInfo,
+        wdir: str,
         stage_name: str = None,
         default: str = None,
     ):
         if default:
-            to_import = wdir / default
+            to_import = fs.path.join(wdir, default)
             if fs.exists(to_import):
                 self.merge_from(fs, default, wdir)
             else:

dvc/proc/exceptions.py

@@ -0,0 +1,25 @@
+from dvc.exceptions import DvcException
+
+
+class ProcessNotTerminatedError(DvcException):
+    def __init__(self, name):
+        super().__init__(f"Managed process '{name}' has not been terminated.")
+
+
+class ProcessNotFoundError(DvcException):
+    def __init__(self, name):
+        super().__init__(f"Managed process '{name}' does not exist.")
+
+
+class TimeoutExpired(DvcException):
+    def __init__(self, cmd, timeout):
+        super().__init__(
+            f"'{cmd}' did not complete before timeout '{timeout}'"
+        )
+        self.cmd = cmd
+        self.timeout = timeout
+
+
+class UnsupportedSignalError(DvcException):
+    def __init__(self, sig):
+        super().__init__(f"Unsupported signal: {sig}")

dvc/proc/manager.py

@@ -0,0 +1,154 @@
+"""Serverless process manager."""
+
+import json
+import logging
+import os
+import signal
+import sys
+from typing import Generator, List, Optional, Tuple, Union
+
+from funcy.flow import reraise
+from shortuuid import uuid
+
+from .exceptions import ProcessNotTerminatedError, UnsupportedSignalError
+from .process import ManagedProcess, ProcessInfo
+
+logger = logging.getLogger(__name__)
+
+
+class ProcessManager:
+    """Manager for controlling background ManagedProcess(es).
+
+    Spawned process entries are kept in the manager directory until they
+    are explicitly removed (with remove() or cleanup()) so that return
+    value and log information can be accessed after a process has completed.
+    """
+
+    def __init__(self, wdir: Optional[str] = None):
+        self.wdir = wdir or "."
+
+    def __iter__(self) -> Generator[str, None, None]:
+        if not os.path.exists(self.wdir):
+            return
+        for name in os.listdir(self.wdir):
+            yield name
+
+    def __getitem__(self, key: str) -> "ProcessInfo":
+        info_path = os.path.join(self.wdir, key, f"{key}.json")
+        try:
+            with open(info_path, encoding="utf-8") as fobj:
+                return ProcessInfo.from_dict(json.load(fobj))
+        except FileNotFoundError:
+            raise KeyError
+
+    @reraise(FileNotFoundError, KeyError)
+    def __setitem__(self, key: str, value: "ProcessInfo"):
+        info_path = os.path.join(self.wdir, key, f"{key}.json")
+        with open(info_path, "w", encoding="utf-8") as fobj:
+            return json.dump(value.asdict(), fobj)
+
+    def __delitem__(self, key: str) -> None:
+        from dvc.utils.fs import remove
+
+        path = os.path.join(self.wdir, key)
+        if os.path.exists(path):
+            remove(path)
+
+    def get(self, key: str, default=None):
+        try:
+            return self[key]
+        except KeyError:
+            return default
+
+    def processes(self) -> Generator[Tuple[str, "ProcessInfo"], None, None]:
+        for name in self:
+            try:
+                yield name, self[name]
+            except KeyError:
+                continue
+
+    def spawn(self, args: Union[str, List[str]], name: Optional[str] = None):
+        """Run the given command in the background."""
+        name = name or uuid()
+        pid = ManagedProcess.spawn(
+            args,
+            wdir=os.path.join(self.wdir, name),
+            name=name,
+        )
+        logger.debug(
+            "Spawned managed process '%s' (PID: '%d')",
+            name,
+            pid,
+        )
+
+    def send_signal(self, name: str, sig: int):
+        """Send `signal` to the specified named process."""
+        process_info = self[name]
+        if sys.platform == "win32":
+            if sig not in (
+                signal.SIGTERM,
+                signal.CTRL_C_EVENT,
+                signal.CTRL_BREAK_EVENT,
+            ):
+                raise UnsupportedSignalError(sig)
+
+        def handle_closed_process():
+            logging.warning(
+                f"Process {name} had already aborted unexpectedly."
+            )
+            process_info.returncode = -1
+            self[name] = process_info
+
+        if process_info.returncode is None:
+            try:
+                os.kill(process_info.pid, sig)
+            except ProcessLookupError:
+                handle_closed_process()
+                raise
+            except OSError as exc:
+                if sys.platform == "win32":
+                    if exc.winerror == 87:
+                        handle_closed_process()
+                        raise ProcessLookupError from exc
+                raise
+
+    def terminate(self, name: str):
+        """Terminate the specified named process."""
+        self.send_signal(name, signal.SIGTERM)
+
+    def kill(self, name: str):
+        """Kill the specified named process."""
+        if sys.platform == "win32":
+            self.send_signal(name, signal.SIGTERM)
+        else:
+            self.send_signal(name, signal.SIGKILL)
+
+    def remove(self, name: str, force: bool = False):
+        """Remove the specified named process from this manager.
+
+        If the specified process is still running, it will be forcefully killed
+        if `force` is True`, otherwise an exception will be raised.
+
+        Raises:
+            ProcessNotTerminatedError if the specified process is still
+            running and was not forcefully killed.
+        """
+        try:
+            process_info = self[name]
+        except KeyError:
+            return
+        if process_info.returncode is None and not force:
+            raise ProcessNotTerminatedError(name)
+        try:
+            self.kill(name)
+        except ProcessLookupError:
+            pass
+        del self[name]
+
+    def cleanup(self, force: bool = False):
+        """Remove stale (terminated) processes from this manager."""
+        for name in self:
+            try:
+                self.remove(name, force)
+            except ProcessNotTerminatedError:
+                continue

dvc/proc/process.py

@@ -0,0 +1,180 @@
+import json
+import logging
+import os
+import shlex
+import subprocess
+from contextlib import AbstractContextManager
+from dataclasses import asdict, dataclass
+from typing import List, Optional, TextIO, Union
+
+from funcy import cached_property
+from shortuuid import uuid
+
+from dvc.utils.fs import makedirs
+
+from .exceptions import TimeoutExpired
+
+logger = logging.getLogger(__name__)
+
+
+@dataclass
+class ProcessInfo:
+    pid: int
+    stdin: Optional[str]
+    stdout: Optional[str]
+    stderr: Optional[str]
+    returncode: Optional[int]
+
+    @classmethod
+    def from_dict(cls, d):
+        return cls(**d)
+
+    def asdict(self):
+        return asdict(self)
+
+
+class ManagedProcess(AbstractContextManager):
+    """Run the specified command with redirected output.
+
+    stdout and stderr will both be redirected to <name>.out.
+    Interactive processes (requiring stdin input) are currently unsupported.
+
+    Parameters:
+        args: Command to be run.
+        wdir: If specified, redirected output files will be placed in `wdir`.
+        name: Name to use for this process, if not specified a UUID will be
+            generated instead.
+    """
+
+    def __init__(
+        self,
+        args: Union[str, List[str]],
+        wdir: Optional[str] = None,
+        name: Optional[str] = None,
+    ):
+        self.args: List[str] = (
+            shlex.split(args, posix=os.name == "posix")
+            if isinstance(args, str)
+            else list(args)
+        )
+        self.wdir = wdir
+        self.name = name or uuid()
+        self.returncode: Optional[int] = None
+        self._stdout: Optional[TextIO] = None
+        self._stderr: Optional[TextIO] = None
+
+        self._run()
+
+    def __exit__(self, exc_type, exc_value, traceback):
+        self.wait()
+
+    def _close_fds(self):
+        if self._stdout:
+            self._stdout.close()
+            self._stdout = None
+        if self._stderr:
+            self._stderr.close()
+            self._stderr = None
+
+    def _make_path(self, path: str) -> str:
+        return os.path.join(self.wdir, path) if self.wdir else path
+
+    @cached_property
+    def stdout_path(self) -> str:
+        return self._make_path(f"{self.name}.out")
+
+    @cached_property
+    def info_path(self) -> str:
+        return self._make_path(f"{self.name}.json")
+
+    @cached_property
+    def pidfile_path(self) -> str:
+        return self._make_path(f"{self.name}.pid")
+
+    @property
+    def info(self) -> "ProcessInfo":
+        return ProcessInfo(
+            pid=self.pid,
+            stdin=None,
+            stdout=self.stdout_path,
+            stderr=None,
+            returncode=self.returncode,
+        )
+
+    def _make_wdir(self):
+        if self.wdir:
+            makedirs(self.wdir, exist_ok=True)
+
+    def _dump(self):
+        self._make_wdir()
+        with open(self.info_path, "w", encoding="utf-8") as fobj:
+            json.dump(self.info.asdict(), fobj)
+        with open(self.pidfile_path, "w", encoding="utf-8") as fobj:
+            fobj.write(str(self.pid))
+
+    def _run(self):
+        self._make_wdir()
+        logger.debug(
+            "Appending output to '%s'",
+            self.stdout_path,
+        )
+        self._stdout = open(self.stdout_path, "ab")
+        try:
+            self._proc = subprocess.Popen(
+                self.args,
+                stdin=subprocess.DEVNULL,
+                stdout=self._stdout,
+                stderr=subprocess.STDOUT,
+                close_fds=True,
+                shell=False,
+            )
+            self.pid: int = self._proc.pid
+            self._dump()
+        except Exception:
+            if self._proc is not None:
+                self._proc.kill()
+            self._close_fds()
+            raise
+
+    def wait(self, timeout: Optional[int] = None) -> Optional[int]:
+        """Block until a process started with `run` has completed.
+
+        Raises:
+            TimeoutExpired if `timeout` was set and the process
+            did not terminate after `timeout` seconds.
+        """
+        if self.returncode is not None:
+            return self.returncode
+        try:
+            self._proc.wait(timeout=timeout)
+        except subprocess.TimeoutExpired as exc:
+            raise TimeoutExpired(exc.cmd, exc.timeout) from exc
+        self.returncode = self._proc.returncode
+        self._close_fds()
+        self._dump()
+        return self.returncode
+
+    @classmethod
+    def spawn(cls, *args, **kwargs) -> Optional[int]:
+        """Spawn a ManagedProcess command in the background.
+
+        Returns: The spawned process PID.
+        """
+        import multiprocessing as mp
+
+        proc = mp.Process(
+            target=cls._spawn,
+            args=args,
+            kwargs=kwargs,
+            daemon=True,
+        )
+        proc.start()
+        # Do not terminate the child daemon when the main process exits
+        # pylint: disable=protected-access
+        mp.process._children.discard(proc)  # type: ignore[attr-defined]
+        return proc.pid
+
+    @classmethod
+    def _spawn(cls, *args, **kwargs):
+        with cls(*args, **kwargs):
+            pass

dvc/progress.py

@@ -136,8 +136,8 @@ class Tqdm(tqdm):
 
         return wrapped
 
-    def as_callback(self, fs, path_info):
-        return FsspecCallback(fs, path_info, self)
+    def as_callback(self):
+        return FsspecCallback(self)
 
     def close(self):
         self.postfix["info"] = ""
@@ -167,21 +167,15 @@ class Tqdm(tqdm):
 
 
 class FsspecCallback(fsspec.Callback):
-    def __init__(self, fs, path_info, progress_bar):
-        self.fs = fs
-        self.path_info = path_info
+    def __init__(self, progress_bar):
         self.progress_bar = progress_bar
         super().__init__()
 
     def set_size(self, size):
-        """``set_size()`` is an API that might be called with ``None`` if the
-        information is not already present on the caller. In that case, we'll
-        retrieve the size from an ``info()`` call."""
-        if size is None:
-            size = self.fs.info(self.path_info)["size"]
-        self.progress_bar.total = size
-        self.progress_bar.refresh()
-        super().set_size(size)
+        if size is not None:
+            self.progress_bar.total = size
+            self.progress_bar.refresh()
+            super().set_size(size)
 
     def relative_update(self, inc=1):
         self.progress_bar.update(inc)

dvc/render/__init__.py

@@ -1,33 +1,4 @@
-import logging
-from typing import TYPE_CHECKING, Dict
+from .image import ImageRenderer
+from .vega import VegaRenderer
 
-if TYPE_CHECKING:
-    from dvc.types import StrPath
-
-logger = logging.getLogger(__name__)
-
-
-class Renderer:
-    def __init__(self, data: Dict):
-        self.data = data
-
-        from dvc.render.utils import get_files
-
-        files = get_files(self.data)
-
-        # we assume comparison of same file between revisions for now
-        assert len(files) == 1
-        self.filename = files.pop()
-
-    def _convert(self, path: "StrPath"):
-        raise NotImplementedError
-
-    @property
-    def DIV(self):
-        raise NotImplementedError
-
-    def generate_html(self, path: "StrPath"):
-        """this method might edit content of path"""
-        partial = self._convert(path)
-        div_id = f"plot_{self.filename.replace('.', '_').replace('/', '_')}"
-        return self.DIV.format(id=div_id, partial=partial)
+RENDERERS = [ImageRenderer, VegaRenderer]

dvc/render/base.py

@@ -0,0 +1,61 @@
+import abc
+from typing import TYPE_CHECKING, Dict
+
+from dvc.exceptions import DvcException
+
+if TYPE_CHECKING:
+    from dvc.types import StrPath
+
+
+REVISION_FIELD = "rev"
+INDEX_FIELD = "step"
+
+
+class BadTemplateError(DvcException):
+    pass
+
+
+class Renderer(abc.ABC):
+    def __init__(self, data: Dict, templates=None):
+        self.data = data
+        self.templates = templates
+
+        from dvc.render.utils import get_files
+
+        files = get_files(self.data)
+
+        # we assume comparison of same file between revisions for now
+        assert len(files) == 1
+        self.filename = files.pop()
+
+    def _convert(self, path: "StrPath"):
+        raise NotImplementedError
+
+    @property
+    @abc.abstractmethod
+    def DIV(self):
+        raise NotImplementedError
+
+    @property
+    @abc.abstractmethod
+    def SCRIPTS(self):
+        raise NotImplementedError
+
+    @abc.abstractmethod
+    def as_json(self):
+        raise NotImplementedError
+
+    @staticmethod
+    def _remove_special_chars(string: str):
+        return string.translate(
+            {ord(c): "_" for c in r"!@#$%^&*()[]{};,<>?\/:.|`~=_+"}
+        )
+
+    def generate_html(self, path: "StrPath"):
+        """this method might edit content of path"""
+        partial = self._convert(path)
+
+        div_id = self._remove_special_chars(self.filename)
+        div_id = f"plot_{div_id}"
+
+        return self.DIV.format(id=div_id, partial=partial)

dvc/render/html.py

@@ -5,7 +5,7 @@ from typing import TYPE_CHECKING, Dict, List, Optional
 from dvc.exceptions import DvcException
 
 if TYPE_CHECKING:
-    from dvc.render import Renderer
+    from dvc.render.base import Renderer
     from dvc.types import StrPath
 
 PAGE_HTML = """<!DOCTYPE html>
@@ -13,9 +13,7 @@ PAGE_HTML = """<!DOCTYPE html>
 <head>
     {refresh_tag}
     <title>DVC Plot</title>
-    <script src="https://cdn.jsdelivr.net/npm/[email protected]"></script>
-    <script src="https://cdn.jsdelivr.net/npm/[email protected]"></script>
-    <script src="https://cdn.jsdelivr.net/npm/[email protected]"></script>
+    {scripts}
 </head>
 <body>
     {plot_divs}
@@ -29,8 +27,9 @@ class MissingPlaceholderError(DvcException):
 
 
 class HTML:
-    PLACEHOLDER = "plot_divs"
-    PLACEHOLDER_FORMAT_STR = f"{{{PLACEHOLDER}}}"
+    SCRIPTS_PLACEHOLDER = "scripts"
+    PLOTS_PLACEHOLDER = "plot_divs"
+    PLOTS_PLACEHOLDER_FORMAT_STR = f"{{{PLOTS_PLACEHOLDER}}}"
     REFRESH_PLACEHOLDER = "refresh_tag"
     REFRESH_TAG = '<meta http-equiv="refresh" content="{}">'
 
@@ -40,11 +39,12 @@ class HTML:
         refresh_seconds: Optional[int] = None,
     ):
         template = template or PAGE_HTML
-        if self.PLACEHOLDER_FORMAT_STR not in template:
-            raise MissingPlaceholderError(self.PLACEHOLDER_FORMAT_STR)
+        if self.PLOTS_PLACEHOLDER_FORMAT_STR not in template:
+            raise MissingPlaceholderError(self.PLOTS_PLACEHOLDER_FORMAT_STR)
 
         self.template = template
         self.elements: List[str] = []
+        self.scripts: str = ""
         self.refresh_tag = ""
         if refresh_seconds is not None:
             self.refresh_tag = self.REFRESH_TAG.format(refresh_seconds)
@@ -65,13 +65,19 @@ class HTML:
         self.elements.append(tabulate.tabulate(rows, header, tablefmt="html"))
         return self
 
+    def with_scripts(self, scripts: str) -> "HTML":
+        if scripts not in self.scripts:
+            self.scripts += f"\n{scripts}"
+        return self
+
     def with_element(self, html: str) -> "HTML":
         self.elements.append(html)
         return self
 
     def embed(self) -> str:
         kwargs = {
-            self.PLACEHOLDER: "\n".join(self.elements),
+            self.SCRIPTS_PLACEHOLDER: self.scripts,
+            self.PLOTS_PLACEHOLDER: "\n".join(self.elements),
             self.REFRESH_PLACEHOLDER: self.refresh_tag,
         }
         return self.template.format(**kwargs)
@@ -89,7 +95,7 @@ def write(
 
     page_html = None
     if template_path:
-        with open(template_path) as fobj:
+        with open(template_path, encoding="utf-8") as fobj:
             page_html = fobj.read()
 
     document = HTML(page_html, refresh_seconds=refresh_seconds)
@@ -98,10 +104,11 @@ def write(
         document.with_element("<br>")
 
     for renderer in renderers:
+        document.with_scripts(renderer.SCRIPTS)
         document.with_element(renderer.generate_html(path))
 
     index = Path(os.path.join(path, "index.html"))
 
-    with open(index, "w") as fd:
+    with open(index, "w", encoding="utf-8") as fd:
         fd.write(document.embed())
     return index

dvc/render/image.py

@@ -1,7 +1,7 @@
 import os
 from typing import TYPE_CHECKING
 
-from dvc.render import Renderer
+from dvc.render.base import Renderer
 from dvc.render.utils import get_files
 from dvc.utils import relpath
 
@@ -13,10 +13,13 @@ class ImageRenderer(Renderer):
     DIV = """
         <div
             id="{id}"
-            style="border: 1px solid;">
+            style="border:1px solid black;text-align:center;
+            white-space: nowrap;overflow-y:hidden;">
             {partial}
         </div>"""
 
+    SCRIPTS = ""
+
     def _write_image(
         self,
         page_dir_path: "StrPath",
@@ -30,15 +33,16 @@ class ImageRenderer(Renderer):
         img_path = os.path.join(
             static, f"{revision}_{filename.replace(os.sep, '_')}"
         )
-        rel_img_path = relpath(img_path, page_dir_path)
         with open(img_path, "wb") as fd:
             fd.write(image_data)
         return """
-        <div>
+        <div
+            style="border:1px dotted black;margin:2px;display: inline-block;
+            overflow:hidden;margin-left:8px;">
             <p>{title}</p>
             <img src="{src}">
         </div>""".format(
-            title=revision, src=rel_img_path
+            title=revision, src=(relpath(img_path, page_dir_path))
         )
 
     def _convert(self, path: "StrPath"):
@@ -57,6 +61,9 @@ class ImageRenderer(Renderer):
             return "\n".join(div_content)
         return ""
 
+    def as_json(self):
+        raise NotImplementedError
+
     @staticmethod
     def matches(data):
         files = get_files(data)

dvc/render/utils.py

@@ -1,7 +1,7 @@
 import os.path
 from typing import Dict, List
 
-import dpath
+import dpath.util
 
 
 def get_files(data: Dict) -> List:
@@ -26,20 +26,18 @@ def find_vega(repo, plots_data, target):
     from dvc.render.vega import VegaRenderer
 
     if found and VegaRenderer.matches(found):
-        return VegaRenderer(found, repo.plots.templates).get_vega()
+        return VegaRenderer(found, repo.plots.templates).as_json()
     return ""
 
 
 def match_renderers(plots_data, templates):
-    from dvc.render.image import ImageRenderer
-    from dvc.render.vega import VegaRenderer
+    from dvc.render import RENDERERS
 
     renderers = []
     for g in group_by_filename(plots_data):
-        if VegaRenderer.matches(g):
-            renderers.append(VegaRenderer(g, templates))
-        if ImageRenderer.matches(g):
-            renderers.append(ImageRenderer(g))
+        for renderer_class in RENDERERS:
+            if renderer_class.matches(g):
+                renderers.append(renderer_class(g, templates))
     return renderers
 
 

dvc/render/vega.py

@@ -5,30 +5,23 @@ from typing import TYPE_CHECKING, Dict, List, Optional, Union
 from funcy import first
 
 from dvc.exceptions import DvcException
-from dvc.render import Renderer
+from dvc.render.base import (
+    INDEX_FIELD,
+    REVISION_FIELD,
+    BadTemplateError,
+    Renderer,
+)
 from dvc.render.utils import get_files
 
 if TYPE_CHECKING:
-    from dvc.repo.plots.template import PlotTemplates
     from dvc.types import StrPath
 
-REVISION_FIELD = "rev"
-INDEX_FIELD = "step"
-
-
-class PlotMetricTypeError(DvcException):
-    def __init__(self, file):
-        super().__init__(
-            "'{}' - file type error\n"
-            "Only JSON, YAML, CSV and TSV formats are supported.".format(file)
-        )
-
 
 class PlotDataStructureError(DvcException):
     def __init__(self):
         super().__init__(
             "Plot data extraction failed. Please see "
-            "https://man.dvc.org/plot for supported data formats."
+            "https://man.dvc.org/plots for supported data formats."
         )
 
 
@@ -108,9 +101,11 @@ class VegaRenderer(Renderer):
     </div>
     """
 
-    def __init__(self, data: Dict, templates: "PlotTemplates"):
-        super().__init__(data)
-        self.templates = templates
+    SCRIPTS = """
+    <script src="https://cdn.jsdelivr.net/npm/[email protected]"></script>
+    <script src="https://cdn.jsdelivr.net/npm/[email protected]"></script>
+    <script src="https://cdn.jsdelivr.net/npm/[email protected]"></script>
+    """
 
     def _squash_props(self) -> Dict:
         resolved: Dict[str, str] = {}
@@ -144,10 +139,39 @@ class VegaRenderer(Renderer):
                     datapoints.extend(tmp)
         return datapoints
 
-    def get_vega(self) -> Optional[str]:
+    def fill_template(self, template, datapoints, props=None):
+        props = props or {}
+
+        content = deepcopy(template.content)
+        if template.anchor_str("data") not in template.content:
+            anchor = template.anchor("data")
+            raise BadTemplateError(
+                f"Template '{template.name}' is not using '{anchor}' anchor"
+            )
+
+        if props.get("x"):
+            template.check_field_exists(datapoints, props.get("x"))
+        if props.get("y"):
+            template.check_field_exists(datapoints, props.get("y"))
+
+        content = template.fill_anchor(content, "data", datapoints)
+
+        props.setdefault("title", "")
+        props.setdefault("x_label", props.get("x"))
+        props.setdefault("y_label", props.get("y"))
+
+        names = ["title", "x", "y", "x_label", "y_label"]
+        for name in names:
+            value = props.get(name)
+            if value is not None:
+                content = template.fill_anchor(content, name, value)
+
+        return content
+
+    def as_json(self) -> Optional[str]:
         props = self._squash_props()
 
-        template = self.templates.load(props.get("template") or "default")
+        template = self.templates.load(props.get("template", None))
 
         if not props.get("x") and template.has_anchor("x"):
             props["append_index"] = True
@@ -162,11 +186,14 @@ class VegaRenderer(Renderer):
                 props["y"] = first(
                     f for f in reversed(fields) if f not in skip
                 )
-            return template.render(datapoints, props=props)
+            filled_template = self.fill_template(template, datapoints, props)
+
+            return filled_template
+
         return None
 
     def _convert(self, path: "StrPath"):
-        return self.get_vega()
+        return self.as_json()
 
     @staticmethod
     def matches(data):

dvc/repo/__init__.py

@@ -11,16 +11,15 @@ from dvc.exceptions import FileMissingError
 from dvc.exceptions import IsADirectoryError as DvcIsADirectoryError
 from dvc.exceptions import NotDvcRepoError, OutputNotFoundError
 from dvc.ignore import DvcIgnoreFilter
-from dvc.path_info import PathInfo
 from dvc.utils import env2bool
 from dvc.utils.fs import path_isin
 
 if TYPE_CHECKING:
-    from dvc.fs.base import BaseFileSystem
+    from dvc.fs.base import FileSystem
     from dvc.objects.file import HashFile
+    from dvc.repo.scm_context import SCMContext
     from dvc.scm import Base
 
-
 logger = logging.getLogger(__name__)
 
 
@@ -91,15 +90,18 @@ class Repo:
     ):
         assert bool(scm) == bool(rev)
 
-        from dvc.scm import SCM
-        from dvc.scm.base import SCMError
-        from dvc.scm.git import Git
+        from dvc.fs.git import GitFileSystem
+        from dvc.scm import SCM, Base, Git, SCMError
         from dvc.utils.fs import makedirs
 
         dvc_dir = None
         tmp_dir = None
         try:
-            fs = scm.get_fs(rev) if isinstance(scm, Git) and rev else None
+            fs = (
+                GitFileSystem(scm=scm, rev=rev)
+                if isinstance(scm, Git) and rev
+                else None
+            )
             root_dir = self.find_root(root_dir, fs)
             dvc_dir = os.path.join(root_dir, self.DVC_DIR)
             tmp_dir = os.path.join(dvc_dir, "tmp")
@@ -113,13 +115,39 @@ class Repo:
             except SCMError:
                 scm = SCM(os.curdir, no_scm=True)
 
-            from dvc.scm import Base
-
             assert isinstance(scm, Base)
             root_dir = scm.root_dir
 
         return root_dir, dvc_dir, tmp_dir
 
+    def _get_database_dir(self, db_name):
+        # NOTE: by default, store SQLite-based remote indexes and state's
+        # `links` and `md5s` caches in the repository itself to avoid any
+        # possible state corruption in 'shared cache dir' scenario, but allow
+        # user to override this through config when, say, the repository is
+        # located on a mounted volume — see
+        # https://github.com/iterative/dvc/issues/4420
+        base_db_dir = self.config.get(db_name, {}).get("dir", None)
+        if not base_db_dir:
+            return self.tmp_dir
+
+        import hashlib
+
+        from dvc.utils.fs import makedirs
+
+        root_dir_hash = hashlib.sha224(
+            self.root_dir.encode("utf-8")
+        ).hexdigest()
+
+        db_dir = os.path.join(
+            base_db_dir,
+            self.DVC_DIR,
+            f"{os.path.basename(self.root_dir)}-{root_dir_hash[0:7]}",
+        )
+
+        makedirs(db_dir, exist_ok=True)
+        return db_dir
+
     def __init__(
         self,
         root_dir=None,
@@ -133,16 +161,16 @@ class Repo:
     ):
         from dvc.config import Config
         from dvc.data_cloud import DataCloud
+        from dvc.fs.git import GitFileSystem
         from dvc.fs.local import LocalFileSystem
         from dvc.lock import LockNoop, make_lock
-        from dvc.machine import MachineManager
         from dvc.objects.db import ODBManager
         from dvc.repo.live import Live
         from dvc.repo.metrics import Metrics
         from dvc.repo.params import Params
         from dvc.repo.plots import Plots
         from dvc.repo.stage import StageLoad
-        from dvc.scm import SCM
+        from dvc.scm import SCM, Git
         from dvc.stage.cache import StageCache
         from dvc.state import State, StateNoop
 
@@ -157,7 +185,8 @@ class Repo:
         )
 
         if scm:
-            self._fs = scm.get_fs(rev)
+            assert isinstance(scm, Git)
+            self._fs = GitFileSystem(scm=scm, rev=rev)
         else:
             self._fs = LocalFileSystem(url=self.root_dir)
 
@@ -183,10 +212,8 @@ class Repo:
                 friendly=True,
             )
 
-            # NOTE: storing state and link_state in the repository itself to
-            # avoid any possible state corruption in 'shared cache dir'
-            # scenario.
-            self.state = State(self.root_dir, self.tmp_dir, self.dvcignore)
+            state_db_dir = self._get_database_dir("state")
+            self.state = State(self.root_dir, state_db_dir, self.dvcignore)
             self.odb = ODBManager(self)
 
             self.stage_cache = StageCache(self)
@@ -198,14 +225,6 @@ class Repo:
         self.params = Params(self)
         self.live = Live(self)
 
-        if self.tmp_dir and (
-            self.config["feature"].get("machine", False)
-            or env2bool("DVC_TEST")
-        ):
-            self.machine = MachineManager(self)
-        else:
-            self.machine = None
-
         self.stage_collection_error_handler: Optional[
             Callable[[str, Exception], None]
         ] = None
@@ -237,8 +256,7 @@ class Repo:
 
     @cached_property
     def scm(self):
-        from dvc.scm import SCM
-        from dvc.scm.base import SCMError
+        from dvc.scm import SCM, SCMError
 
         if self._scm:
             return self._scm
@@ -253,6 +271,12 @@ class Repo:
                 return SCM(self.root_dir, no_scm=True)
             raise
 
+    @cached_property
+    def scm_context(self) -> "SCMContext":
+        from dvc.repo.scm_context import SCMContext
+
+        return SCMContext(self.scm, self.config)
+
     @cached_property
     def dvcignore(self) -> DvcIgnoreFilter:
 
@@ -263,7 +287,10 @@ class Repo:
 
         assert self.scm
         if isinstance(self.fs, LocalFileSystem):
-            return self.scm.get_rev()
+            from dvc.scm import map_scm_exception
+
+            with map_scm_exception():
+                return self.scm.get_rev()
         return self.fs.rev
 
     @cached_property
@@ -272,12 +299,23 @@ class Repo:
 
         return Experiments(self)
 
+    @cached_property
+    def machine(self):
+        from dvc.machine import MachineManager
+
+        if self.tmp_dir and (
+            self.config["feature"].get("machine", False)
+            or env2bool("DVC_TEST")
+        ):
+            return MachineManager(self)
+        return None
+
     @property
-    def fs(self) -> "BaseFileSystem":
+    def fs(self) -> "FileSystem":
         return self._fs
 
     @fs.setter
-    def fs(self, fs: "BaseFileSystem"):
+    def fs(self, fs: "FileSystem"):
         self._fs = fs
         # Our graph cache is no longer valid, as it was based on the previous
         # fs.
@@ -326,7 +364,7 @@ class Repo:
         )
 
     def unprotect(self, target):
-        return self.odb.local.unprotect(PathInfo(target))
+        return self.odb.local.unprotect(target)
 
     def _ignore(self):
         flist = [self.config.files["local"], self.tmp_dir]
@@ -334,7 +372,8 @@ class Repo:
         if path_isin(self.odb.local.cache_dir, self.root_dir):
             flist += [self.odb.local.cache_dir]
 
-        self.scm.ignore_list(flist)
+        for file in flist:
+            self.scm_context.ignore(file)
 
     def brancher(self, *args, **kwargs):
         from dvc.repo.brancher import brancher
@@ -418,14 +457,18 @@ class Repo:
         outs = outs or self.index.outs_graph
 
         abs_path = os.path.abspath(path)
-        path_info = PathInfo(abs_path)
-        match = path_info.__eq__ if strict else path_info.isin_or_eq
+        fs_path = abs_path
 
         def func(out):
-            if out.scheme == "local" and match(out.path_info):
+            def eq(one, two):
+                return one == two
+
+            match = eq if strict else out.fs.path.isin_or_eq
+
+            if out.scheme == "local" and match(fs_path, out.fs_path):
                 return True
 
-            if recursive and out.path_info.isin(path_info):
+            if recursive and out.fs.path.isin(out.fs_path, fs_path):
                 return True
 
             return False
@@ -452,13 +495,17 @@ class Repo:
 
         return RepoFileSystem(self, subrepos=self.subrepos, **self._fs_conf)
 
+    @cached_property
+    def index_db_dir(self):
+        return self._get_database_dir("index")
+
     @contextmanager
     def open_by_relpath(self, path, remote=None, mode="r", encoding=None):
         """Opens a specified resource as a file descriptor"""
         from dvc.fs.repo import RepoFileSystem
 
         fs = RepoFileSystem(self, subrepos=True)
-        path = PathInfo(self.root_dir) / path
+        path = self.fs.path.join(self.root_dir, path)
         try:
             with fs.open(
                 path, mode=mode, encoding=encoding, remote=remote

dvc/repo/add.py

@@ -87,8 +87,8 @@ def translate_graph_error(stages: Stages) -> Iterator[None]:
         yield
     except OverlappingOutputPathsError as exc:
         msg = OVERLAPPING_OUTPUT_FMT.format(
-            out=exc.overlapping_out.path_info,
-            parent=exc.parent.path_info,
+            out=exc.overlapping_out.fs_path,
+            parent=exc.parent.fs_path,
             parent_stage=exc.parent.stage.addressing,
         )
         raise OverlappingOutputPathsError(exc.parent, exc.overlapping_out, msg)
@@ -228,7 +228,7 @@ def _find_all_targets(
     from dvc.dvcfile import is_dvc_file
 
     if os.path.isdir(target) and recursive:
-        files = map(os.fspath, repo.dvcignore.walk_files(repo.fs, target))
+        files = repo.dvcignore.find(repo.fs, target)
         yield from (
             path
             for path in files

dvc/repo/brancher.py

@@ -1,3 +1,5 @@
+from functools import partial
+
 from funcy import group_by
 
 
@@ -58,8 +60,12 @@ def brancher(  # noqa: E302
 
     try:
         if revs:
-            for sha, names in group_by(scm.resolve_rev, revs).items():
-                self.fs = scm.get_fs(sha)
+            from dvc.fs.git import GitFileSystem
+            from dvc.scm import resolve_rev
+
+            rev_resolver = partial(resolve_rev, scm)
+            for sha, names in group_by(rev_resolver, revs).items():
+                self.fs = GitFileSystem(scm=scm, rev=sha)
                 # ignore revs that don't contain repo root
                 # (i.e. revs from before a subdir=True repo was init'ed)
                 if self.fs.exists(self.root_dir):

dvc/repo/collect.py

@@ -2,8 +2,7 @@ import logging
 import os
 from typing import TYPE_CHECKING, Callable, Iterable, List, Tuple
 
-from dvc.path_info import PathInfo
-from dvc.types import DvcPath
+from dvc.types import AnyPath
 
 if TYPE_CHECKING:
     from dvc.output import Output
@@ -14,7 +13,8 @@ logger = logging.getLogger(__name__)
 
 FilterFn = Callable[["Output"], bool]
 Outputs = List["Output"]
-DvcPaths = List[DvcPath]
+AnyPaths = List[AnyPath]
+StrPaths = List[str]
 
 
 def _collect_outs(
@@ -32,39 +32,40 @@ def _collect_paths(
     rev: str = None,
 ):
     from dvc.fs.repo import RepoFileSystem
+    from dvc.utils import relpath
 
-    path_infos = [PathInfo(os.path.abspath(target)) for target in targets]
+    fs_paths = [os.path.abspath(target) for target in targets]
     fs = RepoFileSystem(repo)
 
-    target_infos = []
-    for path_info in path_infos:
+    target_paths = []
+    for fs_path in fs_paths:
 
-        if recursive and fs.isdir(path_info):
-            target_infos.extend(repo.dvcignore.walk_files(fs, path_info))
+        if recursive and fs.isdir(fs_path):
+            target_paths.extend(repo.dvcignore.find(fs, fs_path))
 
-        if not fs.exists(path_info):
+        if not fs.exists(fs_path):
+            rel = relpath(fs_path)
             if rev == "workspace" or rev == "":
-                logger.warning(
-                    "'%s' was not found in current workspace.", path_info
-                )
+                logger.warning("'%s' was not found in current workspace.", rel)
             else:
-                logger.warning("'%s' was not found at: '%s'.", path_info, rev)
-        target_infos.append(path_info)
-    return target_infos
+                logger.warning("'%s' was not found at: '%s'.", rel, rev)
+        target_paths.append(fs_path)
+    return target_paths
 
 
 def _filter_duplicates(
-    outs: Outputs, path_infos: DvcPaths
-) -> Tuple[Outputs, DvcPaths]:
+    outs: Outputs, fs_paths: StrPaths
+) -> Tuple[Outputs, StrPaths]:
     res_outs: Outputs = []
-    res_infos = path_infos
+    fs_res_paths = fs_paths
 
     for out in outs:
-        if out.path_info in path_infos:
+        if out.fs_path in fs_paths:
             res_outs.append(out)
-            res_infos.remove(out.path_info)
+            # MUTATING THE SAME LIST!!
+            fs_res_paths.remove(out.fs_path)
 
-    return res_outs, res_infos
+    return res_outs, fs_res_paths
 
 
 def collect(
@@ -74,15 +75,15 @@ def collect(
     output_filter: FilterFn = None,
     rev: str = None,
     recursive: bool = False,
-) -> Tuple[Outputs, DvcPaths]:
+) -> Tuple[Outputs, StrPaths]:
     assert targets or output_filter
 
     outs: Outputs = _collect_outs(repo, output_filter=output_filter, deps=deps)
 
     if not targets:
-        path_infos: DvcPaths = []
-        return outs, path_infos
+        fs_paths: StrPaths = []
+        return outs, fs_paths
 
-    target_infos = _collect_paths(repo, targets, recursive=recursive, rev=rev)
+    target_paths = _collect_paths(repo, targets, recursive=recursive, rev=rev)
 
-    return _filter_duplicates(outs, target_infos)
+    return _filter_duplicates(outs, target_paths)

dvc/repo/diff.py

@@ -37,14 +37,14 @@ def diff(self, a_rev="HEAD", b_rev=None, targets=None):
             # workspace paths/checksums if b_rev was None
             continue
 
-        targets_path_infos = None
+        targets_paths = None
         if targets is not None:
-            # convert targets to path_infos, and capture any missing targets
-            targets_path_infos, missing_targets[rev] = _targets_to_path_infos(
+            # convert targets to paths, and capture any missing targets
+            targets_paths, missing_targets[rev] = _targets_to_paths(
                 repo_fs, targets
             )
 
-        results[rev] = _paths_checksums(self, targets_path_infos)
+        results[rev] = _paths_checksums(self, targets_paths)
 
     if targets is not None:
         # check for overlapping missing targets between a_rev and b_rev
@@ -133,13 +133,14 @@ def _output_paths(repo, targets):
     for output in repo.index.outs:
         if _exists(output):
             yield_output = targets is None or any(
-                output.path_info.isin_or_eq(target) for target in targets
+                output.fs.path.isin_or_eq(output.fs_path, target)
+                for target in targets
             )
 
             if on_working_fs:
-                _, obj = ostage(
+                _, _, obj = ostage(
                     repo.odb.local,
-                    output.path_info,
+                    output.fs_path,
                     repo.odb.local.fs,
                     "md5",
                     dry_run=True,
@@ -158,19 +159,24 @@ def _output_paths(repo, targets):
 
             if output.is_dir_checksum and (
                 yield_output
-                or any(target.isin(output.path_info) for target in targets)
+                or any(
+                    output.fs.path.isin(target, output.fs_path)
+                    for target in targets
+                )
             ):
-                yield from _dir_output_paths(output.path_info, obj, targets)
+                yield from _dir_output_paths(
+                    output.fs, output.fs_path, obj, targets
+                )
 
 
-def _dir_output_paths(path_info, obj, targets=None):
-    for key, entry_obj in obj:
-        fname = path_info.joinpath(*key)
+def _dir_output_paths(fs, fs_path, obj, targets=None):
+    for key, _, oid in obj:
+        fname = fs.path.join(fs_path, *key)
         if targets is None or any(
-            fname.isin_or_eq(target) for target in targets
+            fs.path.isin_or_eq(fname, target) for target in targets
         ):
             # pylint: disable=no-member
-            yield str(fname), entry_obj.hash_info.value
+            yield fs.path.join(fs.path.name(fs_path), *key), oid.value
 
 
 def _filter_missing(repo_fs, paths):
@@ -185,17 +191,17 @@ def _filter_missing(repo_fs, paths):
             pass
 
 
-def _targets_to_path_infos(repo_fs, targets):
-    path_infos = []
+def _targets_to_paths(repo_fs, targets):
+    paths = []
     missing = []
 
     for target in targets:
         if repo_fs.exists(target):
-            path_infos.append(repo_fs.metadata(target).path_info)
+            paths.append(repo_fs.metadata(target).fs_path)
         else:
             missing.append(target)
 
-    return path_infos, missing
+    return paths, missing
 
 
 def _calculate_renamed(new, old, added, deleted):

dvc/repo/experiments/__init__.py

@@ -10,10 +10,9 @@ from typing import Dict, Iterable, Mapping, Optional
 
 from funcy import cached_property, first
 
+from dvc.dependency.param import MissingParamsError
 from dvc.env import DVCLIVE_RESUME
 from dvc.exceptions import DvcException
-from dvc.path_info import PathInfo
-from dvc.stage.monitor import CheckpointKilledError
 from dvc.utils import relpath
 
 from .base import (
@@ -42,7 +41,9 @@ def scm_locked(f):
     # different sequences of git operations at once
     @wraps(f)
     def wrapper(exp, *args, **kwargs):
-        with exp.scm_lock:
+        from dvc.scm import map_scm_exception
+
+        with map_scm_exception(), exp.scm_lock:
             return f(exp, *args, **kwargs)
 
     return wrapper
@@ -89,7 +90,7 @@ class Experiments:
 
     def __init__(self, repo):
         from dvc.lock import make_lock
-        from dvc.scm.base import NoSCMError
+        from dvc.scm import NoSCMError
 
         if repo.config["core"].get("no_scm", False):
             raise NoSCMError
@@ -98,6 +99,7 @@ class Experiments:
         self.scm_lock = make_lock(
             os.path.join(self.repo.tmp_dir, "exp_scm_lock"),
             tmp_dir=self.repo.tmp_dir,
+            hardlink_lock=repo.config["core"].get("hardlink_lock", False),
         )
 
     @property
@@ -116,7 +118,7 @@ class Experiments:
 
     @cached_property
     def stash(self):
-        from dvc.scm.git import Stash
+        from scmrepo.git import Stash
 
         return Stash(self.scm, EXPS_STASH)
 
@@ -162,7 +164,7 @@ class Experiments:
                 the human-readable name in the experiment branch ref. Has no
                 effect of branch is specified.
         """
-        with self.scm.detach_head() as orig_head:
+        with self.scm.detach_head(client="dvc") as orig_head:
             stash_head = orig_head
             if baseline_rev is None:
                 baseline_rev = orig_head
@@ -350,19 +352,32 @@ class Experiments:
         )
         self.scm.add(self.args_file)
 
+    def _format_new_params_msg(self, new_params, config_path):
+        """Format an error message for when new parameters are identified"""
+        new_param_count = len(new_params)
+        pluralise = "s are" if new_param_count > 1 else " is"
+        param_list = ", ".join(new_params)
+        return (
+            f"{new_param_count} parameter{pluralise} missing "
+            f"from '{config_path}': {param_list}"
+        )
+
     def _update_params(self, params: dict):
         """Update experiment params files with the specified values."""
-        from dvc.utils.collections import merge_params
+        from dvc.utils.collections import NewParamsFound, merge_params
         from dvc.utils.serialize import MODIFIERS
 
         logger.debug("Using experiment params '%s'", params)
 
-        for params_fname in params:
-            path = PathInfo(params_fname)
-            suffix = path.suffix.lower()
+        for path in params:
+            suffix = self.repo.fs.path.suffix(path).lower()
             modify_data = MODIFIERS[suffix]
             with modify_data(path, fs=self.repo.fs) as data:
-                merge_params(data, params[params_fname])
+                try:
+                    merge_params(data, params[path], allow_new=False)
+                except NewParamsFound as e:
+                    msg = self._format_new_params_msg(e.new_params, path)
+                    raise MissingParamsError(msg)
 
         # Force params file changes to be staged in git
         # Otherwise in certain situations the changes to params file may be
@@ -395,7 +410,9 @@ class Experiments:
                 self.scm.reset()
 
         if checkpoint_resume:
-            resume_rev = self.scm.resolve_rev(checkpoint_resume)
+            from dvc.scm import resolve_rev
+
+            resume_rev = resolve_rev(self.scm, checkpoint_resume)
             try:
                 self.check_baseline(resume_rev)
                 checkpoint_resume = resume_rev
@@ -458,6 +475,17 @@ class Experiments:
             "\tdvc exp branch <exp> <branch>\n"
         )
 
+    def _validate_new_ref(self, exp_ref: ExpRefInfo):
+        from .utils import check_ref_format
+
+        if not exp_ref.name:
+            return
+
+        check_ref_format(self.scm, exp_ref)
+
+        if self.scm.get_ref(str(exp_ref)):
+            raise ExperimentExistsError(exp_ref.name)
+
     @scm_locked
     def new(self, *args, checkpoint_resume: Optional[str] = None, **kwargs):
         """Create a new experiment.
@@ -470,6 +498,16 @@ class Experiments:
                 *args, resume_rev=checkpoint_resume, **kwargs
             )
 
+        name = kwargs.get("name", None)
+        baseline_sha = kwargs.get("baseline_rev") or self.repo.scm.get_rev()
+        exp_ref = ExpRefInfo(baseline_sha=baseline_sha, name=name)
+
+        try:
+            self._validate_new_ref(exp_ref)
+        except ExperimentExistsError as err:
+            if not (kwargs.get("force", False) or kwargs.get("reset", False)):
+                raise err
+
         return self._stash_exp(*args, **kwargs)
 
     def _resume_checkpoint(
@@ -638,8 +676,8 @@ class Experiments:
                 name=item.name,
                 branch=item.branch,
                 tmp_dir=base_tmp_dir,
-                cache_dir=self.repo.odb.local.cache_dir,
             )
+            executor.init_cache(self.repo, stash_rev)
             executors[stash_rev] = executor
 
         for ref in (EXEC_HEAD, EXEC_MERGE, EXEC_BASELINE):
@@ -657,6 +695,8 @@ class Experiments:
             dict mapping stash revs to the successfully executed experiments
             for each stash rev.
         """
+        from dvc.stage.monitor import CheckpointKilledError
+
         result: Dict[str, Dict[str, str]] = defaultdict(dict)
 
         manager = Manager()
@@ -759,6 +799,7 @@ class Experiments:
     @unlocked_repo
     def _workspace_repro(self) -> Mapping[str, str]:
         """Run the most recently stashed experiment in the workspace."""
+        from dvc.stage.monitor import CheckpointKilledError
         from dvc.utils.fs import makedirs
 
         from .executor.base import BaseExecutor
@@ -771,7 +812,7 @@ class Experiments:
         # `checkout --force` here will not lose any data (popping stash commit
         # will result in conflict between workspace params and stashed CLI
         # params, but we always want the stashed version).
-        with self.scm.detach_head(entry.rev, force=True):
+        with self.scm.detach_head(entry.rev, force=True, client="dvc"):
             rev = self.stash.pop()
             self.scm.set_ref(EXEC_BASELINE, entry.baseline_rev)
             if entry.branch:
@@ -844,7 +885,9 @@ class Experiments:
         return self._get_baseline(rev)
 
     def _get_baseline(self, rev):
-        rev = self.scm.resolve_rev(rev)
+        from dvc.scm import resolve_rev
+
+        rev = resolve_rev(self.scm, rev)
 
         if rev in self.stash_revs:
             entry = self.stash_revs.get(rev)
@@ -893,7 +936,7 @@ class Experiments:
         from .executor.base import BaseExecutor, ExecutorInfo
 
         result = {}
-        for pidfile in self.repo.fs.walk_files(
+        for pidfile in self.repo.fs.find(
             os.path.join(self.repo.tmp_dir, self.EXEC_PID_DIR)
         ):
             rev, _ = os.path.splitext(os.path.basename(pidfile))

dvc/repo/experiments/apply.py

@@ -3,7 +3,6 @@ import os
 
 from dvc.repo import locked
 from dvc.repo.scm_context import scm_context
-from dvc.scm.base import RevError
 from dvc.utils.fs import remove
 
 from .base import (
@@ -20,13 +19,15 @@ logger = logging.getLogger(__name__)
 @locked
 @scm_context
 def apply(repo, rev, force=True, **kwargs):
+    from scmrepo.exceptions import MergeConflictError
+
     from dvc.repo.checkout import checkout as dvc_checkout
-    from dvc.scm.base import MergeConflictError, SCMError
+    from dvc.scm import RevError, SCMError, resolve_rev
 
     exps = repo.experiments
 
     try:
-        exp_rev = repo.scm.resolve_rev(rev)
+        exp_rev = resolve_rev(repo.scm, rev)
         exps.check_baseline(exp_rev)
     except (RevError, BaselineMismatchError) as exc:
         raise InvalidExpRevError(rev) from exc
@@ -45,7 +46,12 @@ def apply(repo, rev, force=True, **kwargs):
     else:
         workspace = None
 
-    repo.scm.merge(exp_rev, commit=False)
+    from scmrepo.exceptions import SCMError as _SCMError
+
+    try:
+        repo.scm.merge(exp_rev, commit=False)
+    except _SCMError as exc:
+        raise SCMError(str(exc))
 
     if workspace:
         try:
@@ -60,7 +66,7 @@ def apply(repo, rev, force=True, **kwargs):
                 repo.scm.reset(hard=True)
                 repo.scm.stash.pop()
                 raise ApplyConflictError(rev) from exc
-        except SCMError as exc:
+        except _SCMError as exc:
             raise ApplyConflictError(rev) from exc
         repo.scm.stash.drop()
     repo.scm.reset()