Create hook-asyncssh.py

Fixes #7577

.codecov.yml

@@ -1,10 +0,0 @@
-comment: false
-coverage:
-  status:
-    project:
-      default:
-        target: 90
-        threshold: 10%
-    patch:
-      default:
-        threshold: 25%

.coveragerc

@@ -1,16 +0,0 @@
-[run]
-branch = True
-source = dvc
-omit =
-    *__main__.py
-
-[report]
-exclude_lines =
-    if __name__ == .__main__.:
-    if typing.TYPE_CHECKING:
-    if TYPE_CHECKING:
-    pass
-    raise NotImplementedError
-    # pragma: no cover
-    #  pragma: no cover
-    \.\.\.$

.dvcignore

@@ -1,4 +1,3 @@
 /tests
 /dvc
 /.github
-/bin

.flake8

@@ -0,0 +1,12 @@
+[flake8]
+ignore=
+    E203, # Whitespace before ':'
+    E266, # Too many leading '#' for block comment
+    W503, # Line break occurred before a binary operator
+    P1,  # unindexed parameters in the str.format, see:
+    # https://pypi.org/project/flake8-string-format/
+max_line_length = 79
+max-complexity = 15
+select = B,C,E,F,W,T4,B902,T,P
+show_source = true
+count = true

.github/CODEOWNERS

@@ -1 +0,0 @@
-*       @iterative/dvc

.github/ISSUE_TEMPLATE/bug_report.md

@@ -8,7 +8,7 @@ about: Create a bug report to help us improve DVC
 <!--
 ## Issue name
 
-Issue names must follow the pattern `command: description` where the command is the dvc command that you are trying to run. The description should describe the consequence of the bug. 
+Issue names must follow the pattern `command: description` where the command is the dvc command that you are trying to run. The description should describe the consequence of the bug.
 
 Example: `repro: doesn't detect input changes`
 -->

.github/ISSUE_TEMPLATE/feature_request.md

@@ -2,4 +2,3 @@
 name: "\U0001F680 Feature Request"
 about: Suggest an idea for this project
 ---
-

.github/codecov.yml

@@ -0,0 +1,12 @@
+codecov:
+  notify:
+    wait_for_ci: true
+coverage:
+  status:
+    project:
+      default:
+        target: auto
+        threshold: 2%
+    patch: off
+github_checks:
+  annotations: false

.github/mergify.yml

@@ -1,9 +0,0 @@
-pull_request_rules:
-  - name: backport patches to 1.11
-    conditions:
-      - base=main
-      - label=backport-1.11
-    actions:
-      backport:
-        branches:
-          - "1.11"

.github/release-drafter.yml

@@ -1,49 +0,0 @@
-# Config for https://github.com/apps/release-drafter
-name-template: "$RESOLVED_VERSION 🦉 "
-tag-template: "$RESOLVED_VERSION"
-categories:
-  - title: "🚀 New Features and Enhancements "
-    labels:
-      - "feature"
-      - "enhancement"
-      - "ui"
-  - title: "💣 Breaking Changes "
-    labels:
-      - "breaking-change"
-  - title: "🏇 Optimizations "
-    labels:
-      - "optimize"
-      - "performance"
-  - title: "🐛 Bug Fixes "
-    labels:
-      - "bugfix"
-      - "bug"
-  - title: "🔨 Maintenance "
-    labels:
-      - "maintenance"
-      - "refactoring"
-      - "chore"
-      - "build"
-      - "ci"
-change-template: "- $TITLE (#$NUMBER) @$AUTHOR"
-version-resolver:
-  major:
-    labels:
-      - "major"
-  minor:
-    labels:
-      - "minor"
-  patch:
-    labels:
-      - "patch"
-  default: patch
-exclude-labels:
-  - "skip-changelog"
-template: |
-  > Refer to https://dvc.org/doc/install for installation instructions.
-
-  ## Changes
-
-  $CHANGES
-
-  Thanks again to $CONTRIBUTORS for the contributions! 🎉

.github/release.yml

@@ -0,0 +1,16 @@
+changelog:
+  exclude:
+    labels: ["skip-changelog"]
+  categories:
+    - title: 🛠 Breaking Changes
+      labels: ["breaking-change"]
+    - title: 🚀 New Features and Enhancements
+      labels: ["feature", "enhancement", "ui"]
+    - title: ⚡ Optimizations
+      labels: ["optimize", "performance"]
+    - title: 🐛 Bug Fixes
+      labels: ["bugfix", "bug"]
+    - title: 🔨 Maintenance
+      labels: ["maintenance", "refactoring", "chore", "build", "ci"]
+    - title: Other Changes
+      labels: ["*"]

.github/workflows/benchmarks.yaml

@@ -1,23 +1,18 @@
 name: benchmarks
 on: [pull_request, workflow_dispatch]
 
+concurrency:
+  group: ${{ github.workflow }}-${{ github.head_ref || github.run_id }}
+  cancel-in-progress: true
+
 jobs:
   bench:
     name: run benchmarks
-    runs-on: ${{ matrix.os }}
-    strategy:
-      fail-fast: false
-      matrix:
-        os: [ubuntu-18.04]
+    runs-on: ubuntu-latest
     steps:
-      - name: Cancel Previous Runs
-        uses: styfle/[email protected]
-        with:
-          access_token: ${{ github.token }}
       - uses: actions/setup-python@v4
         with:
-            python-version: 3.8
+            python-version: "3.11"
       - uses: iterative/dvc-bench@main
         with:
             pytest_options: "-k 'test_init or test_help'"
-

.github/workflows/codeql.yml

@@ -0,0 +1,39 @@
+name: "CodeQL"
+
+on:
+  push:
+    branches: ["main"]
+  pull_request:
+    branches: ["main"]
+  schedule:
+    - cron: '32 19 * * 2'
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+jobs:
+  analyze:
+    name: Analyze
+    runs-on: ubuntu-latest
+    permissions:
+      actions: read
+      contents: read
+      security-events: write
+
+    strategy:
+      fail-fast: false
+      matrix:
+        language: ['python']
+
+    steps:
+    - name: Checkout repository
+      uses: actions/checkout@v3
+
+    - name: Initialize CodeQL
+      uses: github/codeql-action/init@v2
+      with:
+        languages: ${{ matrix.language }}
+
+    - name: Perform CodeQL Analysis
+      uses: github/codeql-action/analyze@v2

.github/workflows/packages.yaml

@@ -6,152 +6,107 @@ on:
   release:
     types: [released, prereleased]
   workflow_dispatch:
+
+permissions:  # added using https://github.com/step-security/secure-workflows
+  contents: read
+
 jobs:
-  deb:
-    runs-on: ubuntu-18.04
+  binary:
+    permissions:
+      contents: write  # for actions/upload-release-asset to upload release asset
+    strategy:
+      matrix:
+        include:
+         - {pkg: "deb",  os: "ubuntu-18.04", asset: "dvc_${{ github.event.release.tag_name }}_amd64.deb"}
+         - {pkg: "rpm", os: "ubuntu-18.04", asset: "dvc-${{ github.event.release.tag_name }}-1.x86_64.rpm"}
+         - {pkg: "osxpkg", os: "macos-10.15", asset: "dvc-${{ github.event.release.tag_name }}.pkg"}
+         - {pkg: "exe",  os: "windows-2019", asset: "dvc-${{ github.event.release.tag_name }}.exe"}
+
+    name: ${{ matrix.pkg }}
+    runs-on: ${{ matrix.os }}
     steps:
     - uses: actions/checkout@v3
       with:
           fetch-depth: 0
-    - name: Set up Python 3.9
+
+    - name: Set up Python 3.10
       uses: actions/setup-python@v4
       with:
-        python-version: 3.9
-    - name: Set up Ruby 2.4
-      uses: actions/[email protected]
+        python-version: "3.10"
+        cache: pip
+        cache-dependency-path: |
+          pyproject.toml
+          scripts/build-requirements.txt
+
+    - name: Set up Ruby 2.6
+      uses: ruby/setup-ruby@v1
+      if: matrix.pkg != 'exe'
       with:
-        ruby-version: 2.4
-    - name: Install deps
-      run: |
-        pip install .[all]
-        pip install -r scripts/build-requirements.txt
-        gem install --no-document fpm
-    - name: Build deb
-      run: python scripts/build.py deb
-    - name: Publish deb
-      if: github.event_name == 'release'
-      uses: actions/[email protected]
-      env:
-        GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-      with:
-        upload_url: ${{ github.event.release.upload_url }}
-        asset_path: scripts/fpm/dvc_${{ github.event.release.tag_name }}_amd64.deb
-        asset_name: dvc_${{ github.event.release.tag_name }}_amd64.deb
-        asset_content_type: binary/octet-stream
-  rpm:
-    runs-on: ubuntu-18.04
-    steps:
-    - uses: actions/checkout@v3
-      with:
-        fetch-depth: 0
-    - name: Set up Python 3.9
-      uses: actions/setup-python@v4
-      with:
-        python-version: 3.9
-    - name: Set up Ruby 2.4
-      uses: actions/[email protected]
-      with:
-        ruby-version: 2.4
-    - name: Install deps
-      run: |
-        pip install .[all]
-        pip install -r scripts/build-requirements.txt
-        gem install --no-document fpm
-    - name: Build rpm
-      run: python scripts/build.py rpm
-    - name: Publish rpm
-      if: github.event_name == 'release'
-      uses: actions/[email protected]
-      env:
-        GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-      with:
-        upload_url: ${{ github.event.release.upload_url }}
-        asset_path: scripts/fpm/dvc-${{ github.event.release.tag_name }}-1.x86_64.rpm
-        asset_name: dvc-${{ github.event.release.tag_name }}-1.x86_64.rpm
-        asset_content_type: binary/octet-stream
-  osxpkg:
-    runs-on: macos-10.15
-    steps:
-    - uses: actions/checkout@v3
-      with:
-        fetch-depth: 0
-    - name: Set up Python 3.9
-      uses: actions/setup-python@v4
-      with:
-        python-version: 3.9
-    - name: Set up Ruby 2.4
-      uses: actions/[email protected]
-      with:
-        ruby-version: 2.4
-    - name: Install deps
-      run: |
-        # available lxml (webdav dependency) wheels are built with an old
-        # SDK, which breaks osxpkg notarization. Building from scratch
-        # instead.
-        pip install .[all] --no-binary lxml
-        pip install -r scripts/build-requirements.txt
-        gem install --no-document fpm
-    - name: Build
-      run: python scripts/build.py osxpkg
-    - name: Publish
-      if: github.event_name == 'release'
-      uses: actions/[email protected]
-      env:
-        GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-      with:
-        upload_url: ${{ github.event.release.upload_url }}
-        asset_path: scripts/fpm/dvc-${{ github.event.release.tag_name }}.pkg
-        asset_name: dvc-${{ github.event.release.tag_name }}.pkg
-        asset_content_type: binary/octet-stream
-  exe:
-    runs-on: windows-2019
-    steps:
-    - uses: actions/checkout@v3
-      with:
-        fetch-depth: 0
-    - name: Set up Python 3.9
-      uses: actions/setup-python@v4
-      with:
-        python-version: 3.9
+        ruby-version: '2.6'
+
+    - name: Install fpm
+      if: matrix.pkg != 'exe'
+      run: gem install --no-document fpm
+
     - name: Install deps
       run: |
+        pip install --upgrade pip wheel setuptools
         pip install .[all]
         pip install -r scripts/build-requirements.txt
+
     - name: Pull images
-      run: |
-        dvc pull
-    - name: Build
-      run: python scripts\build.py exe
-      shell: cmd
-    - name: Publish
+      run: dvc pull
+
+    - name: Build ${{ matrix.pkg }}
+      run: python scripts/build.py ${{ matrix.pkg }}
+
+    - name: Publish ${{ matrix.pkg }}
       if: github.event_name == 'release'
+      uses: actions/[email protected]
       env:
         GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-      uses: actions/[email protected]
       with:
         upload_url: ${{ github.event.release.upload_url }}
-        asset_path: scripts/innosetup/dvc-${{ github.event.release.tag_name }}.exe
-        asset_name: dvc-${{ github.event.release.tag_name }}.exe
+        asset_path: scripts/${{ matrix.pkg == 'exe' && 'innosetup' || 'fpm' }}/${{ matrix.asset }}
+        asset_name: ${{ matrix.asset }}
         asset_content_type: binary/octet-stream
+
   pip:
-    runs-on: ubuntu-18.04
+    runs-on: ubuntu-20.04
     steps:
     - uses: actions/checkout@v3
       with:
         fetch-depth: 0
+
     - name: Set up Python 3.8
       uses: actions/setup-python@v4
       with:
         python-version: 3.8
+
     - name: Install
-      run: pip install ".[all,tests]"
-    - name: Build packages
+      run: pip install --upgrade pip wheel
+
+    - name: Force version for Test PyPI uploads
+      if: ${{ !startsWith(github.ref, 'refs/tags') }}
       run: |
-        pip install -U pip
-        ./scripts/build_package.sh
+        pip install setuptools_scm
+        echo version=$(python -m setuptools_scm | awk -F+ '{print $1}' | tail -1) >> $GITHUB_ENV
+
+    - name: Build packages
+      run: ./scripts/build_package.sh
+      env:
+        SETUPTOOLS_SCM_PRETEND_VERSION: ${{ env.version }}
+
     - name: Publish packages to PyPI
       if: github.event_name == 'release'
-      uses: pypa/gh-action-pypi-publish@master
+      uses: pypa/gh-action-pypi-publish@release/v1
       with:
-        user: __token__
         password: ${{ secrets.PYPI_TOKEN }}
+
+    - name: Publish to Test PyPI
+      if: ${{ github.event_name == 'release' || (github.event_name == 'push' && github.ref == 'refs/heads/main') }}
+      uses: pypa/gh-action-pypi-publish@release/v1
+      with:
+        password: ${{ secrets.TEST_PYPI_TOKEN }}
+        repository_url: https://test.pypi.org/legacy/
+        skip_existing: true

.github/workflows/tests.yaml

@@ -1,25 +1,30 @@
 name: Tests
 
 on:
-  pull_request: {}
-  push: {}
+  push:
+    branches: [main]
+  pull_request:
   schedule:
     - cron: '5 1 * * *'  # every day at 01:05
+  workflow_dispatch:
 
 env:
   DVC_TEST: "true"
   HOMEBREW_NO_AUTO_UPDATE: 1
   SHELL: /bin/bash
 
+concurrency:
+  group: ${{ github.workflow }}-${{ github.head_ref || github.run_id }}
+  cancel-in-progress: true
+
+permissions:  # added using https://github.com/step-security/secure-workflows
+  contents: read
+
 jobs:
   lint:
     timeout-minutes: 10
     runs-on: ubuntu-latest
     steps:
-    - name: Cancel Previous Runs
-      uses: styfle/[email protected]
-      with:
-        access_token: ${{ github.token }}
     - uses: actions/checkout@v3
       with:
         fetch-depth: 0
@@ -27,13 +32,12 @@ jobs:
       uses: actions/setup-python@v4
       with:
         python-version: 3.8
+        cache: pip
+        cache-dependency-path: pyproject.toml
     - name: Install requirements
       run: |
-        pip install --upgrade pip setuptools wheel
-        pip install collective.checkdocs==0.2
-        pip install ".[dev]" pre-commit
-    - name: Check README
-      run: python setup.py checkdocs
+        pip install --upgrade pip wheel
+        pip install ".[dev]"
     - uses: pre-commit/[email protected]
   tests:
     timeout-minutes: 45
@@ -41,8 +45,17 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        os: [ubuntu-20.04, windows-latest, macos-latest]
-        pyv: ["3.8", "3.9", "3.10"]
+        os: [ubuntu-20.04, windows-latest]
+        pyv: ["3.8", "3.9", "3.10", "3.11"]
+        pytest-filter:
+        - "import or plot or live or experiment"
+        - "not (import or plot or live or experiment)"
+        include:
+        - {os: macos-latest, pyv: "3.8"}
+        - {os: macos-latest, pyv: "3.9"}
+        - {os: macos-latest, pyv: "3.10"}
+        - {os: macos-latest, pyv: "3.11"}
+
     steps:
     - uses: actions/checkout@v3
       with:
@@ -51,39 +64,31 @@ jobs:
       uses: actions/setup-python@v4
       with:
         python-version: ${{ matrix.pyv }}
-    - name: get pip cache dir
-      id: pip-cache-dir
-      run: |
-        echo "::set-output name=dir::$(pip cache dir)"
-    - name: set pip cache
-      id: pip-cache
-      uses: actions/cache@v3
-      with:
-        path: ${{ steps.pip-cache-dir.outputs.dir }}
-        key: ${{ runner.os }}-pip-${{ hashFiles('setup.cfg') }}
-        restore-keys: |
-          ${{ runner.os }}-pip-
+        cache: pip
+        cache-dependency-path: pyproject.toml
     - name: install
-      if: steps.cache.pip-cache-dir.cache-hit != 'true'
       run: |
-        pip install --upgrade pip setuptools wheel
+        pip install --upgrade pip wheel
         pip install -e ".[dev]"
     - name: run tests
-      timeout-minutes: 40
-      run: >-
-        python -m tests -n=auto
-        --cov-report=xml --cov-report=term
-        ${{ env.extra_test_args }}
+      timeout-minutes: 20
+      run: pytest -nauto --durations 100 --cov --cov-report=xml --cov-report=term -k "${{ matrix.pytest-filter }}"
     - name: upload coverage report
       uses: codecov/codecov-action@v3
       with:
         file: ./coverage.xml
         fail_ci_if_error: false
+  check:
+    if: always()
+    needs: [lint, tests]
+    runs-on: ubuntu-latest
+    steps:
+      - uses: re-actors/alls-green@release/v1
+        with:
+          jobs: ${{ toJSON(needs) }}
   notify:
     if: github.ref == 'refs/heads/main' && failure()
-    needs:
-      - lint
-      - tests
+    needs: [lint, tests]
     runs-on: ubuntu-latest
     steps:
     - name: Slack Notification

.pre-commit-config.yaml

@@ -1,39 +1,62 @@
+default_language_version:
+  python: python3
+
 ci:
   skip: [mypy, pylint, dvc-pre-commit]
 
 repos:
-  - hooks:
-      - id: black
-        language_version: python3
-    repo: https://github.com/ambv/black
-    rev: 22.6.0
   - repo: https://github.com/pre-commit/pre-commit-hooks
     rev: v4.3.0
     hooks:
+      - id: check-added-large-files
+      - id: check-case-conflict
+      - id: check-executables-have-shebangs
+      - id: check-json
+      - id: check-merge-conflict
+        exclude: "tests/func/utils/test_strict_yaml.py"
+        args: ['--assume-in-merge']
       - id: check-toml
       - id: check-yaml
-  - hooks:
-      - id: codespell
-        args: 
-          - --ignore-words-list
-          - ba,datas,fo,uptodate,cachable
-    repo: https://github.com/codespell-project/codespell
-    rev: v2.1.0
-  - hooks:
-      - id: isort
-        language_version: python3
-    repo: https://github.com/timothycrosley/isort
+      - id: debug-statements
+        exclude: "dvc/_debug.py"
+      - id: end-of-file-fixer
+      - id: mixed-line-ending
+      - id: sort-simple-yaml
+      - id: trailing-whitespace
+  - repo: https://github.com/lovesegfault/beautysh
+    rev: v6.2.1
+    hooks:
+      - id: beautysh
+        args: ["-i", "2"]
+  - repo: https://github.com/asottile/pyupgrade
+    rev: v3.2.0
+    hooks:
+      - id: pyupgrade
+        args:
+          - --py38-plus
+  - repo: https://github.com/psf/black
+    rev: 22.10.0
+    hooks:
+      - id: black
+  - repo: https://github.com/PyCQA/isort
     rev: 5.10.1
-  - hooks:
+    hooks:
+      - id: isort
+  - repo: https://github.com/codespell-project/codespell
+    rev: v2.2.2
+    hooks:
+      - id: codespell
+        additional_dependencies: ["tomli"]
+  - repo: https://github.com/PyCQA/flake8
+    rev: 5.0.4
+    hooks:
       - id: flake8
-        language_version: python3
         additional_dependencies:
+          - flake8-broken-line
           - flake8-bugbear
           - flake8-comprehensions
           - flake8-debugger
           - flake8-string-format
-    repo: https://gitlab.com/pycqa/flake8
-    rev: 3.9.2
   - repo: local
     hooks:
     - id: mypy
@@ -47,14 +70,6 @@ repos:
       entry: pylint
       language: system
       types: [python]
-  - hooks:
-      - args:
-          - -i
-          - "2"
-        id: beautysh
-        language_version: python3
-    repo: https://github.com/lovesegfault/beautysh
-    rev: v6.2.1
   - hooks:
       - args:
           - git-hook

CODE_OF_CONDUCT.md

@@ -74,4 +74,3 @@ available at https://www.contributor-covenant.org/version/1/4/code-of-conduct.ht
 
 For answers to common questions about this code of conduct, see
 https://www.contributor-covenant.org/faq
-

MANIFEST.in

@@ -1,3 +0,0 @@
-include LICENSE
-recursive-include requirements *.txt
-recursive-include dvc/repo/plots/templates *.json

README.rst

@@ -11,9 +11,9 @@
 • `Contributing`_
 • `Community and Support`_
 
-|CI| |Maintainability| |Coverage| |VS Code| |DOI|
+|CI| |Python Version| |Coverage| |VS Code| |DOI|
 
-|PyPI| |Packages| |Brew| |Conda| |Choco| |Snap|
+|PyPI| |PyPI Downloads| |Packages| |Brew| |Conda| |Choco| |Snap|
 
 |
 
@@ -82,7 +82,7 @@ To share and back up the *data cache*, DVC supports multiple remote storage plat
 
 |Flowchart|
 
-`DVC pipelines <https://dvc.org/doc/start/data-management/pipelines>`_ (computational graphs) connect code and data together.
+`DVC pipelines <https://dvc.org/doc/start/data-management/data-pipelines>`_ (computational graphs) connect code and data together.
 They specify all steps required to produce a model: input dependencies including code, data, commands to run; and output information to be saved.
 
 Last but not least, `DVC Experiment Versioning <https://dvc.org/doc/start/experiments>`_ lets you prepare and run a large number of experiments.
@@ -248,6 +248,10 @@ Barrak, A., Eghan, E.E. and Adams, B. `On the Co-evolution of ML Pipelines and S
    :target: https://codeclimate.com/github/iterative/dvc
    :alt: Code Climate
 
+.. |Python Version| image:: https://img.shields.io/pypi/pyversions/dvc
+   :target: https://pypi.org/project/dvc
+   :alt: Python Version
+
 .. |Coverage| image:: https://codecov.io/gh/iterative/dvc/branch/main/graph/badge.svg
    :target: https://codecov.io/gh/iterative/dvc
    :alt: Codecov
@@ -272,6 +276,10 @@ Barrak, A., Eghan, E.E. and Adams, B. `On the Co-evolution of ML Pipelines and S
    :target: https://pypi.org/project/dvc
    :alt: PyPI
 
+.. |PyPI Downloads| image:: https://img.shields.io/pypi/dm/dvc.svg?color=blue&label=Downloads&logo=pypi&logoColor=gold
+   :target: https://pypi.org/project/dvc
+   :alt: PyPI Downloads
+
 .. |Packages| image:: https://img.shields.io/github/v/release/iterative/dvc?label=deb|pkg|rpm|exe&logo=GitHub
    :target: https://github.com/iterative/dvc/releases/latest
    :alt: deb|pkg|rpm|exe
@@ -288,6 +296,6 @@ Barrak, A., Eghan, E.E. and Adams, B. `On the Co-evolution of ML Pipelines and S
    :target: https://github.com/iterative/dvc/graphs/contributors
    :alt: Contributors
 
-.. |VS Code| image:: https://vsmarketplacebadge.apphb.com/version/Iterative.dvc.svg
+.. |VS Code| image:: https://img.shields.io/visual-studio-marketplace/v/Iterative.dvc?color=blue&label=VSCode&logo=visualstudiocode&logoColor=blue
    :target: https://marketplace.visualstudio.com/items?itemName=Iterative.dvc
    :alt: VS Code Extension

bin/dvc

@@ -1,3 +0,0 @@
-#!/bin/bash
-
-PYTHONPATH=$DVC_HOME python -m dvc $@

bin/dvc.bat

@@ -1,3 +0,0 @@
-@echo OFF
-set PYTHONPATH=%DVC_HOME%
-python -m dvc %*

dvc/__init__.py

@@ -4,6 +4,6 @@ DVC
 Make your data science projects reproducible and shareable.
 """
 import dvc.logger
-from dvc.version import __version__  # noqa: F401
+from dvc.version import __version__, version_tuple  # noqa: F401
 
 dvc.logger.setup()

dvc/_debug.py

@@ -1,16 +1,14 @@
 from contextlib import ExitStack, contextmanager
 from datetime import datetime
-from typing import TYPE_CHECKING, Callable, Union
+from typing import TYPE_CHECKING, Callable, Optional, Union
 
 if TYPE_CHECKING:
     from argparse import Namespace
+    from types import FrameType
 
 
 @contextmanager
-def viztracer_profile(
-    path: Union[Callable[[], str], str],
-    max_stack_depth: int = -1,
-):
+def viztracer_profile(path: Union[Callable[[], str], str], depth: int = -1):
     try:
         import viztracer  # pylint: disable=import-error
     except ImportError:
@@ -18,7 +16,7 @@ def viztracer_profile(
         yield
         return
 
-    tracer = viztracer.VizTracer(max_stack_depth=max_stack_depth)
+    tracer = viztracer.VizTracer(max_stack_depth=depth)
 
     tracer.start()
     yield
@@ -29,7 +27,9 @@ def viztracer_profile(
 
 @contextmanager
 def yappi_profile(
-    path: Union[Callable[[], str], str] = None, wall_clock: bool = True
+    path: Union[Callable[[], str], str] = None,
+    wall_clock: Optional[bool] = True,
+    separate_threads: Optional[bool] = False,
 ):
     try:
         import yappi  # pylint: disable=import-error
@@ -44,14 +44,26 @@ def yappi_profile(
     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")
+    threads = yappi.get_thread_stats()
+    stats = {}
+    if separate_threads:
+        for thread in threads:
+            ctx_id = thread.id
+            stats[ctx_id] = yappi.get_func_stats(ctx_id=ctx_id)
     else:
-        yappi.get_func_stats().print_all()
-        yappi.get_thread_stats().print_all()
+        stats[None] = yappi.get_func_stats()
+
+    fpath = path() if callable(path) else path
+    for ctx_id, st in stats.items():
+        if fpath:
+            out = f"{fpath}-{ctx_id}" if ctx_id is not None else fpath
+            st.save(out, type="callgrind")
+        else:
+            if ctx_id is not None:
+                print(f"\nThread {ctx_id}")
+            st.print_all()  # pylint:disable=no-member
+            if ctx_id is None:
+                threads.print_all()  # pylint:disable=no-member
 
     yappi.clear_stats()
 
@@ -109,6 +121,38 @@ def debug():
         raise  # prevent from jumping ahead
 
 
+def _sigshow(_, frame: Optional["FrameType"]) -> None:
+    from shutil import get_terminal_size
+    from traceback import format_stack
+
+    from dvc.ui import ui
+
+    lines = "\u2015" * get_terminal_size().columns
+    ui.error_write(lines, "\n", *format_stack(frame), lines, sep="")
+
+
+@contextmanager
+def show_stack():
+    r"""Show stack trace on SIGQUIT (Ctrl-\) or SIGINFO (Ctrl-T on macOS)."""
+    import signal
+
+    signal.signal(signal.SIGQUIT, _sigshow)
+    try:
+        signal.signal(signal.SIGINFO, _sigshow)  # only available on macOS
+    except AttributeError:
+        pass
+    yield
+
+
+def _get_path_func(tool: str, ext: str):
+    fmt = f"{tool}.dvc-{{now:%Y%m%d}}_{{now:%H%M%S}}.{ext}"
+
+    def func(now: Optional["datetime"] = None) -> str:
+        return fmt.format(now=now or datetime.now())
+
+    return func
+
+
 @contextmanager
 def debugtools(args: "Namespace" = None, **kwargs):
     kw = vars(args) if args else {}
@@ -121,19 +165,21 @@ 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("show_stack", False):
+            stack.enter_context(show_stack())
         if kw.get("yappi"):
-            output = "callgrind.dvc-{0:%Y%m%d}_{0:%H%M%S}.out"
+            path_func = _get_path_func("callgrind", "out")
             stack.enter_context(
-                yappi_profile(path=lambda: output.format(datetime.now()))
-            )
-        if kw.get("viztracer") or kw.get("viztracer_depth"):
-            output = "viztracer.dvc-{0:%Y%m%d}_{0:%H%M%S}.json"
-            stack.enter_context(
-                viztracer_profile(
-                    path=lambda: output.format(datetime.now()),
-                    max_stack_depth=kw.get("viztracer_depth") or -1,
+                yappi_profile(
+                    path=path_func,
+                    separate_threads=kw.get("yappi_separate_threads"),
                 )
             )
+        if kw.get("viztracer") or kw.get("viztracer_depth"):
+            path_func = _get_path_func("viztracer", "json")
+            depth = kw.get("viztracer_depth") or -1
+            prof = viztracer_profile(path=path_func, depth=depth)
+            stack.enter_context(prof)
         yield
 
 
@@ -146,6 +192,12 @@ def add_debugging_flags(parser):
     parser.add_argument(
         "--yappi", action="store_true", default=False, help=SUPPRESS
     )
+    parser.add_argument(
+        "--yappi-separate-threads",
+        action="store_true",
+        default=False,
+        help=SUPPRESS,
+    )
     parser.add_argument(
         "--viztracer", action="store_true", default=False, help=SUPPRESS
     )
@@ -160,3 +212,10 @@ def add_debugging_flags(parser):
     parser.add_argument(
         "--instrument-open", action="store_true", default=False, help=SUPPRESS
     )
+    parser.add_argument(
+        "--show-stack",
+        "--ss",
+        action="store_true",
+        default=False,
+        help=SUPPRESS,
+    )

dvc/analytics.py

@@ -102,6 +102,8 @@ def _runtime_info():
     """
     Gather information from the environment where DVC runs to fill a report.
     """
+    from iterative_telemetry import find_or_create_user_id
+
     from dvc import __version__
     from dvc.utils import is_binary
 
@@ -110,7 +112,7 @@ def _runtime_info():
         "is_binary": is_binary(),
         "scm_class": _scm_in_use(),
         "system_info": _system_info(),
-        "user_id": _find_or_create_user_id(),
+        "user_id": find_or_create_user_id(),
     }
 
 
@@ -146,45 +148,3 @@ def _system_info():
 
     # We don't collect data for any other system.
     raise NotImplementedError
-
-
-def _find_or_create_user_id():
-    """
-    The user's ID is stored on a file under the global config directory.
-
-    The file should contain a JSON with a "user_id" key:
-
-        {"user_id": "16fd2706-8baf-433b-82eb-8c7fada847da"}
-
-    IDs are generated randomly with UUID.
-    """
-    import uuid
-
-    from dvc.config import Config
-    from dvc.lock import Lock, LockError
-    from dvc.utils.fs import makedirs
-
-    config_dir = Config.get_dir("global")
-    fname = os.path.join(config_dir, "user_id")
-    lockfile = os.path.join(config_dir, "user_id.lock")
-
-    # Since the `fname` and `lockfile` are under the global config,
-    # we need to make sure such directory exist already.
-    makedirs(config_dir, exist_ok=True)
-
-    try:
-        with Lock(lockfile):
-            try:
-                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", encoding="utf-8") as fobj:
-                    json.dump({"user_id": user_id}, fobj)
-
-            return user_id
-
-    except LockError:
-        logger.debug("Failed to acquire '%s'", lockfile)

dvc/annotations.py

@@ -0,0 +1,35 @@
+from dataclasses import asdict, dataclass, field, fields
+from typing import Any, ClassVar, Dict, List, Optional
+
+from funcy import compact
+
+
+@dataclass
+class Annotation:
+    PARAM_DESC: ClassVar[str] = "desc"
+    PARAM_TYPE: ClassVar[str] = "type"
+    PARAM_LABELS: ClassVar[str] = "labels"
+    PARAM_META: ClassVar[str] = "meta"
+
+    desc: Optional[str] = None
+    type: Optional[str] = None
+    labels: List[str] = field(default_factory=list)
+    meta: Dict[str, Any] = field(default_factory=dict)
+
+    def update(self, **kwargs) -> "Annotation":
+        for attr, value in kwargs.items():
+            if value and hasattr(self, attr):
+                setattr(self, attr, value)
+        return self
+
+    def to_dict(self) -> Dict[str, str]:
+        return compact(asdict(self))
+
+
+ANNOTATION_FIELDS = [field.name for field in fields(Annotation)]
+ANNOTATION_SCHEMA = {
+    Annotation.PARAM_DESC: str,
+    Annotation.PARAM_TYPE: str,
+    Annotation.PARAM_LABELS: [str],
+    Annotation.PARAM_META: object,
+}

dvc/api/__init__.py

@@ -1,9 +1,15 @@
-from .data import (  # noqa, pylint: disable=redefined-builtin
-    get_url,
-    open,
-    read,
-)
+from dvc.fs.dvc import _DVCFileSystem as DVCFileSystem
+
+from .data import open  # pylint: disable=redefined-builtin
+from .data import get_url, read
 from .experiments import make_checkpoint
 from .params import params_show
 
-__all__ = ["get_url", "make_checkpoint", "open", "params_show", "read"]
+__all__ = [
+    "get_url",
+    "make_checkpoint",
+    "open",
+    "params_show",
+    "read",
+    "DVCFileSystem",
+]

dvc/api/params.py

@@ -37,6 +37,10 @@ def params_show(
             stages to retrieve parameters from.
             Defaults to `None`.
             If `None`, all parameters from all stages will be retrieved.
+            If this method is called from a different location to the one where
+            the `dvc.yaml` is found, the relative path to the `dvc.yaml` must
+            be provided as a prefix with the syntax `{relpath}:{stage}`.
+            For example: `subdir/dvc.yaml:stage-0` or `../dvc.yaml:stage-1`.
         rev (str, optional): Name of the `Git revision`_ to retrieve parameters
             from.
             Defaults to `None`.

dvc/cli/__init__.py

@@ -121,7 +121,7 @@ def _log_exceptions(exc: Exception) -> Optional[int]:
         )
         return None
 
-    from dvc_data.build import IgnoreInCollectedDirError
+    from dvc_data.hashfile.build import IgnoreInCollectedDirError
 
     if isinstance(exc, IgnoreInCollectedDirError):
         logger.exception("")

dvc/cli/actions.py

@@ -0,0 +1,20 @@
+from argparse import _AppendAction
+
+
+class CommaSeparatedArgs(_AppendAction):  # pylint: disable=protected-access
+    def __call__(self, parser, namespace, values, option_string=None):
+        from funcy import ldistinct
+
+        items = getattr(namespace, self.dest) or []
+        items.extend(map(str.strip, values.split(",")))
+        setattr(namespace, self.dest, ldistinct(items))
+
+
+class KeyValueArgs(_AppendAction):
+    def __call__(self, parser, namespace, values, option_string=None):
+        items = getattr(namespace, self.dest) or {}
+        for value in filter(bool, values):
+            key, _, value = value.partition("=")
+            items[key.strip()] = value
+
+        setattr(namespace, self.dest, items)

dvc/cli/parser.py

@@ -2,8 +2,8 @@
 import argparse
 import logging
 import os
-import sys
 
+from dvc import __version__
 from dvc.commands import (
     add,
     cache,
@@ -29,6 +29,7 @@ from dvc.commands import (
     init,
     install,
     ls,
+    ls_url,
     machine,
     metrics,
     move,
@@ -75,6 +76,7 @@ COMMANDS = [
     install,
     root,
     ls,
+    ls_url,
     freeze,
     dag,
     daemon,
@@ -126,17 +128,6 @@ class DvcParser(argparse.ArgumentParser):
         return args
 
 
-class VersionAction(argparse.Action):  # pragma: no cover
-    # pylint: disable=too-few-public-methods
-    """Shows DVC version and exits."""
-
-    def __call__(self, parser, namespace, values, option_string=None):
-        from dvc import __version__
-
-        print(__version__)
-        sys.exit(0)
-
-
 def get_parent_parser():
     """Create instances of a parser containing common arguments shared among
     all the commands.
@@ -184,13 +175,11 @@ def get_main_parser():
         help="Show this help message and exit.",
     )
 
-    # NOTE: On some python versions action='version' prints to stderr
-    # instead of stdout https://bugs.python.org/issue18920
     parser.add_argument(
         "-V",
         "--version",
-        action=VersionAction,
-        nargs=0,
+        action="version",
+        version=__version__,
         help="Show program's version.",
     )
 

dvc/cli/utils.py

@@ -21,10 +21,13 @@ def append_doc_link(help_message, path):
     )
 
 
-def fix_plumbing_subparsers(subparsers):
-    # metavar needs to be explicitly set in order to hide plumbing subcommands
+def hide_subparsers_from_help(subparsers):
+    # metavar needs to be explicitly set in order to hide subcommands
     # from the 'positional arguments' choices list
     # see: https://bugs.python.org/issue22848
+    # Need to set `add_help=False`, but avoid setting `help`
+    # (not even to `argparse.SUPPPRESS`).
+    # NOTE: The argument is the parent subparser, not the subcommand parser.
     cmds = [
         cmd for cmd, parser in subparsers.choices.items() if parser.add_help
     ]

dvc/commands/add.py

@@ -2,12 +2,43 @@ import argparse
 import logging
 
 from dvc.cli import completion
+from dvc.cli.actions import KeyValueArgs
 from dvc.cli.command import CmdBase
 from dvc.cli.utils import append_doc_link
 
 logger = logging.getLogger(__name__)
 
 
+def _add_annotating_args(parser: argparse.ArgumentParser) -> None:
+    parser.add_argument(
+        "--desc",
+        type=str,
+        metavar="<text>",
+        help="User description of the data.",
+    )
+    parser.add_argument(
+        "--meta",
+        metavar="key=value",
+        nargs=1,
+        action=KeyValueArgs,
+        help="Custom metadata to add to the data.",
+    )
+    parser.add_argument(
+        "--label",
+        dest="labels",
+        type=str,
+        action="append",
+        metavar="<str>",
+        help="Label for the data.",
+    )
+    parser.add_argument(
+        "--type",
+        type=str,
+        metavar="<str>",
+        help="Type of the data.",
+    )
+
+
 class CmdAdd(CmdBase):
     def run(self):
         from dvc.exceptions import (
@@ -28,6 +59,9 @@ class CmdAdd(CmdBase):
                 glob=self.args.glob,
                 desc=self.args.desc,
                 out=self.args.out,
+                type=self.args.type,
+                labels=self.args.labels,
+                meta=self.args.meta,
                 remote=self.args.remote,
                 to_remote=self.args.to_remote,
                 jobs=self.args.jobs,
@@ -109,15 +143,8 @@ def add_parser(subparsers, parent_parser):
         ),
         metavar="<number>",
     )
-    parser.add_argument(
-        "--desc",
-        type=str,
-        metavar="<text>",
-        help=(
-            "User description of the data (optional). "
-            "This doesn't affect any DVC operations."
-        ),
-    )
+
+    _add_annotating_args(parser)
     parser.add_argument(
         "targets", nargs="+", help="Input files/directories to add."
     ).complete = completion.FILE

dvc/commands/checkout.py

@@ -110,7 +110,7 @@ def add_parser(subparsers, parent_parser):
         nargs="*",
         help=(
             "Limit command scope to these tracked files/directories, "
-            ".dvc files, or stage names."
+            ".dvc files and stage names."
         ),
     ).complete = completion.DVC_FILE
     checkout_parser.set_defaults(func=CmdCheckout)

dvc/commands/commit.py

@@ -70,7 +70,9 @@ def add_parser(subparsers, parent_parser):
     commit_parser.add_argument(
         "targets",
         nargs="*",
-        help="stages or .dvc files to commit. Optional. "
-        "(Finds all DVC files in the workspace by default.)",
+        help=(
+            "Limit command scope to these tracked files/directories, "
+            ".dvc files and stage names."
+        ),
     ).complete = completion.DVCFILES_AND_STAGE
     commit_parser.set_defaults(func=CmdCommit)

dvc/commands/dag.py

@@ -204,7 +204,7 @@ def add_parser(subparsers, parent_parser):
     dag_parser.add_argument(
         "target",
         nargs="?",
-        help="Stage or output to show pipeline for (optional). "
+        help="Stage name or output to show pipeline for. "
         "Finds all stages in the workspace by default.",
     )
     dag_parser.set_defaults(func=CmdDAG)

dvc/commands/data.py

@@ -1,11 +1,18 @@
 import argparse
 import logging
-from typing import TYPE_CHECKING
+from operator import itemgetter
+from typing import TYPE_CHECKING, Any, Dict, Iterable, Set
 
 from funcy import chunks, compact, log_durations
 
+from dvc.cli import completion
+from dvc.cli.actions import CommaSeparatedArgs
 from dvc.cli.command import CmdBase
-from dvc.cli.utils import append_doc_link, fix_subparsers
+from dvc.cli.utils import (
+    append_doc_link,
+    fix_subparsers,
+    hide_subparsers_from_help,
+)
 from dvc.ui import ui
 from dvc.utils import colorize
 
@@ -31,15 +38,22 @@ class CmdDataStatus(CmdBase):
         "unchanged": "DVC unchanged files",
     }
     HINTS = {
-        "not_in_cache": 'use "dvc pull <file>..." '
-        "to update your local storage",
-        "committed": "git commit the corresponding dvc files "
-        "to update the repo",
-        "uncommitted": 'use "dvc commit <file>..." to track changes',
-        "untracked": 'use "git add <file> ..." or '
-        'dvc add <file>..." to commit to git or to dvc',
-        "git_dirty": "there are {}changes not tracked by dvc, "
-        'use "git status" to see',
+        "not_in_cache": ('use "dvc fetch <file>..." to download files',),
+        "committed": (
+            "git commit the corresponding dvc files to update the repo",
+        ),
+        "uncommitted": (
+            'use "dvc commit <file>..." to track changes',
+            'use "dvc checkout <file>..." to discard changes',
+        ),
+        "untracked": (
+            'use "git add <file> ..." or '
+            '"dvc add <file>..." to commit to git or to dvc',
+        ),
+        "git_dirty": (
+            "there are {}changes not tracked by dvc, "
+            'use "git status" to see',
+        ),
     }
 
     @staticmethod
@@ -76,8 +90,9 @@ class CmdDataStatus(CmdBase):
             color = cls.COLORS.get(stage, None)
 
             ui.write(header)
-            if hint := cls.HINTS.get(stage):
-                ui.write(f"  ({hint})")
+            if hints := cls.HINTS.get(stage):
+                for hint in hints:
+                    ui.write(f"  ({hint})")
 
             if isinstance(stage_status, dict):
                 items = [
@@ -92,9 +107,10 @@ class CmdDataStatus(CmdBase):
                 out = "\n".join(tabs + item for item in chunk)
                 ui.write(colorize(out, color))
 
-        if (hint := cls.HINTS.get("git_dirty")) and git_info.get("is_dirty"):
-            message = hint.format("other " if result else "")
-            ui.write(f"[blue]({message})[/]", styled=True)
+        if (hints := cls.HINTS.get("git_dirty")) and git_info.get("is_dirty"):
+            for hint in hints:
+                message = hint.format("other " if result else "")
+                ui.write(f"[blue]({message})[/]", styled=True)
         return 0
 
     def run(self) -> int:
@@ -102,7 +118,6 @@ class CmdDataStatus(CmdBase):
             status = self.repo.data_status(
                 granular=self.args.granular,
                 untracked_files=self.args.untracked_files,
-                with_dirs=self.args.with_dirs,
             )
 
         if not self.args.unchanged:
@@ -116,6 +131,65 @@ class CmdDataStatus(CmdBase):
         return self._show_status(status)
 
 
+class CmdDataLs(CmdBase):
+    @staticmethod
+    def _show_table(
+        d: Iterable[Dict[str, Any]],
+        filter_types: Set[str],
+        filter_labels: Set[str],
+        markdown: bool = False,
+    ) -> None:
+        from rich.style import Style
+
+        from dvc.compare import TabularData
+
+        td = TabularData(
+            columns=["Path", "Type", "Labels", "Description"], fill_value="-"
+        )
+        for entry in sorted(d, key=itemgetter("path")):
+            typ = entry.get("type", "")
+            desc = entry.get("desc", "-")
+            labels = entry.get("labels", [])
+
+            if filter_types and typ not in filter_types:
+                continue
+            if filter_labels and filter_labels.isdisjoint(labels):
+                continue
+
+            rich_label = ui.rich_text()
+            for index, label in enumerate(labels):
+                if index:
+                    rich_label.append(",")
+                style = Style(bold=label in filter_labels, color="green")
+                rich_label.append(label, style=style)
+
+            if markdown and desc:
+                desc = desc.partition("\n")[0]
+
+            path = ui.rich_text(entry["path"], style="cyan")
+            type_style = Style(bold=typ in filter_types, color="yellow")
+            typ = ui.rich_text(entry.get("type", ""), style=type_style)
+            td.append([path, typ or "-", rich_label or "-", desc])
+
+        td.render(markdown=markdown, rich_table=True)
+
+    def run(self):
+        from dvc.repo.data import ls
+
+        filter_labels = set(self.args.labels)
+        filter_types = set(self.args.type)
+        d = ls(
+            self.repo, targets=self.args.targets, recursive=self.args.recursive
+        )
+        self._show_table(
+            d,
+            filter_labels=filter_labels,
+            filter_types=filter_types,
+            markdown=self.args.markdown,
+        )
+        return 0
+
+
 def add_parser(subparsers, parent_parser):
     data_parser = subparsers.add_parser(
         "data",
@@ -172,10 +246,52 @@ def add_parser(subparsers, parent_parser):
         nargs="?",
         help="Show untracked files.",
     )
-    data_status_parser.add_argument(
-        "--with-dirs",
+    data_status_parser.set_defaults(func=CmdDataStatus)
+
+    DATA_LS_HELP = "List data tracked by DVC with its metadata."
+    data_ls_parser = data_subparsers.add_parser(
+        "ls",
+        aliases=["list"],
+        parents=[parent_parser],
+        description=append_doc_link(DATA_LS_HELP, "data/ls"),
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+        add_help=False,
+    )
+    data_ls_parser.add_argument(
+        "--md",
+        "--show-md",
+        dest="markdown",
         action="store_true",
         default=False,
-        help=argparse.SUPPRESS,
+        help="Show tabulated output in the Markdown format (GFM).",
     )
-    data_status_parser.set_defaults(func=CmdDataStatus)
+    data_ls_parser.add_argument(
+        "--type",
+        action=CommaSeparatedArgs,
+        default=[],
+        help="Comma-separated list of type to filter.",
+    )
+    data_ls_parser.add_argument(
+        "--labels",
+        action=CommaSeparatedArgs,
+        default=[],
+        help="Comma-separated list of labels to filter.",
+    )
+    data_ls_parser.add_argument(
+        "-R",
+        "--recursive",
+        action="store_true",
+        default=False,
+        help="Recursively list from the specified directories.",
+    )
+    data_ls_parser.add_argument(
+        "targets",
+        default=None,
+        nargs="*",
+        help=(
+            "Limit command scope to these tracked files/directories, "
+            ".dvc files, or stage names."
+        ),
+    ).complete = completion.DVCFILES_AND_STAGE
+    data_ls_parser.set_defaults(func=CmdDataLs)
+    hide_subparsers_from_help(data_subparsers)

dvc/commands/data_sync.py

@@ -43,7 +43,8 @@ class CmdDataPull(CmdDataBase):
             )
             self.log_summary(stats)
         except (CheckoutError, DvcException) as exc:
-            self.log_summary(getattr(exc, "stats", {}))
+            if stats := getattr(exc, "stats", {}):
+                self.log_summary(stats)
             logger.exception("failed to pull data from the cloud")
             return 1
 
@@ -119,7 +120,7 @@ def shared_parent_parser():
         nargs="*",
         help=(
             "Limit command scope to these tracked files/directories, "
-            ".dvc files, or stage names."
+            ".dvc files and stage names."
         ),
     ).complete = completion.DVC_FILE
 
@@ -194,7 +195,7 @@ def add_parser(subparsers, _parent_parser):
         "--glob",
         action="store_true",
         default=False,
-        help="Pull cache for targets matching shell-style wildcards.",
+        help=argparse.SUPPRESS,
     )
     pull_parser.set_defaults(func=CmdDataPull)
 

dvc/commands/experiments/__init__.py

@@ -2,8 +2,8 @@ import argparse
 
 from dvc.cli.utils import (
     append_doc_link,
-    fix_plumbing_subparsers,
     fix_subparsers,
+    hide_subparsers_from_help,
 )
 from dvc.commands.experiments import (
     apply,
@@ -59,7 +59,7 @@ def add_parser(subparsers, parent_parser):
     fix_subparsers(experiments_subparsers)
     for cmd in SUB_COMMANDS:
         cmd.add_parser(experiments_subparsers, parent_parser)
-    fix_plumbing_subparsers(experiments_subparsers)
+    hide_subparsers_from_help(experiments_subparsers)
 
 
 def add_rev_selection_flags(

dvc/commands/experiments/diff.py

@@ -27,7 +27,7 @@ class CmdExperimentsDiff(CmdBase):
 
         if self.args.json:
             ui.write_json(diff)
-        else:
+        elif diff:
             from dvc.compare import show_diff
 
             precision = self.args.precision or DEFAULT_PRECISION

dvc/commands/experiments/ls.py

@@ -16,13 +16,15 @@ class CmdExperimentsList(CmdBase):
             num=self.args.num,
             git_remote=self.args.git_remote,
         )
+        tags = self.repo.scm.describe(exps)
+        remained = {baseline for baseline, tag in tags.items() if tag is None}
+        base = "refs/heads"
+        ref_heads = self.repo.scm.describe(remained, base=base)
+
         for baseline in exps:
-            tag = self.repo.scm.describe(baseline)
-            if not tag:
-                branch = self.repo.scm.describe(baseline, base="refs/heads")
-                if branch:
-                    tag = branch.split("/")[-1]
-            name = tag if tag else baseline[:7]
+            name = baseline[:7]
+            if tags[baseline] or ref_heads[baseline]:
+                name = tags[baseline] or ref_heads[baseline][len(base) + 1 :]
             if not name_only:
                 print(f"{name}:")
             for exp_name in exps[baseline]:

dvc/commands/experiments/show.py

@@ -95,11 +95,8 @@ def _collect_rows(
 
         exp = results.get("data", {})
 
-        if exp.get("running"):
-            state = "Running"
-        elif exp.get("queued"):
-            state = "Queued"
-        else:
+        state = exp.get("status")
+        if state == "Success":
             state = fill_value
 
         is_baseline = rev == "baseline"
@@ -476,6 +473,8 @@ class CmdExperimentsShow(CmdBase):
                 all_branches=self.args.all_branches,
                 all_tags=self.args.all_tags,
                 all_commits=self.args.all_commits,
+                hide_queued=self.args.hide_queued,
+                hide_failed=self.args.hide_failed,
                 revs=self.args.rev,
                 num=self.args.num,
                 sha_only=self.args.sha,
@@ -594,6 +593,18 @@ def add_parser(experiments_subparsers, parent_parser):
         default=False,
         help="Always show git commit SHAs instead of branch/tag names.",
     )
+    experiments_show_parser.add_argument(
+        "--hide-failed",
+        action="store_true",
+        default=False,
+        help="Hide failed experiments in the table.",
+    )
+    experiments_show_parser.add_argument(
+        "--hide-queued",
+        action="store_true",
+        default=False,
+        help="Hide queued experiments in the table.",
+    )
     experiments_show_parser.add_argument(
         "--json",
         "--show-json",

dvc/commands/git_hook.py

@@ -74,7 +74,7 @@ class CmdMergeDriver(CmdHookBase):
             our = Dvcfile(dvc, self.args.our, verify=False)
             their = Dvcfile(dvc, self.args.their, verify=False)
 
-            our.merge(ancestor, their)
+            our.merge(ancestor, their, allowed=["add", "remove", "change"])
 
             return 0
         finally:

dvc/commands/imp.py

@@ -19,7 +19,11 @@ class CmdImport(CmdBase):
                 fname=self.args.file,
                 rev=self.args.rev,
                 no_exec=self.args.no_exec,
+                no_download=self.args.no_download,
                 desc=self.args.desc,
+                type=self.args.type,
+                labels=self.args.labels,
+                meta=self.args.meta,
                 jobs=self.args.jobs,
             )
         except DvcException:
@@ -33,6 +37,8 @@ class CmdImport(CmdBase):
 
 
 def add_parser(subparsers, parent_parser):
+    from .add import _add_annotating_args
+
     IMPORT_HELP = (
         "Download file or directory tracked by DVC or by Git "
         "into the workspace, and track it."
@@ -69,20 +75,19 @@ def add_parser(subparsers, parent_parser):
         help="Specify name of the .dvc file this command will generate.",
         metavar="<filename>",
     )
-    import_parser.add_argument(
+    no_download_exec_group = import_parser.add_mutually_exclusive_group()
+    no_download_exec_group.add_argument(
         "--no-exec",
         action="store_true",
         default=False,
-        help="Only create .dvc file without actually downloading it.",
+        help="Only create .dvc file without actually importing target data.",
     )
-    import_parser.add_argument(
-        "--desc",
-        type=str,
-        metavar="<text>",
-        help=(
-            "User description of the data (optional). "
-            "This doesn't affect any DVC operations."
-        ),
+    no_download_exec_group.add_argument(
+        "--no-download",
+        action="store_true",
+        default=False,
+        help="Create .dvc file including target data hash value(s)"
+        " but do not actually download the file(s).",
     )
     import_parser.add_argument(
         "-j",
@@ -94,4 +99,6 @@ def add_parser(subparsers, parent_parser):
         ),
         metavar="<number>",
     )
+
+    _add_annotating_args(import_parser)
     import_parser.set_defaults(func=CmdImport)

dvc/commands/imp_url.py

@@ -17,10 +17,15 @@ class CmdImportUrl(CmdBase):
                 out=self.args.out,
                 fname=self.args.file,
                 no_exec=self.args.no_exec,
+                no_download=self.args.no_download,
                 remote=self.args.remote,
                 to_remote=self.args.to_remote,
                 desc=self.args.desc,
+                type=self.args.type,
+                labels=self.args.labels,
+                meta=self.args.meta,
                 jobs=self.args.jobs,
+                version_aware=self.args.version_aware,
             )
         except DvcException:
             logger.exception(
@@ -34,6 +39,8 @@ class CmdImportUrl(CmdBase):
 
 
 def add_parser(subparsers, parent_parser):
+    from .add import _add_annotating_args
+
     IMPORT_HELP = (
         "Download or copy file from URL and take it under DVC control."
     )
@@ -66,11 +73,19 @@ def add_parser(subparsers, parent_parser):
         help="Specify name of the .dvc file this command will generate.",
         metavar="<filename>",
     ).complete = completion.DIR
-    import_parser.add_argument(
+    no_download_exec_group = import_parser.add_mutually_exclusive_group()
+    no_download_exec_group.add_argument(
         "--no-exec",
         action="store_true",
         default=False,
-        help="Only create .dvc file without actually downloading it.",
+        help="Only create .dvc file without actually importing target data.",
+    )
+    no_download_exec_group.add_argument(
+        "--no-download",
+        action="store_true",
+        default=False,
+        help="Create .dvc file including target data hash value(s)"
+        " but do not actually download the file(s).",
     )
     import_parser.add_argument(
         "--to-remote",
@@ -95,12 +110,13 @@ def add_parser(subparsers, parent_parser):
         metavar="<number>",
     )
     import_parser.add_argument(
-        "--desc",
-        type=str,
-        metavar="<text>",
+        "--version-aware",
+        action="store_true",
+        default=False,
         help=(
-            "User description of the data (optional). "
-            "This doesn't affect any DVC operations."
+            "Import using cloud versioning. Implied if the URL contains a "
+            "version ID."
         ),
     )
+    _add_annotating_args(import_parser)
     import_parser.set_defaults(func=CmdImportUrl)

dvc/commands/ls_url.py

@@ -0,0 +1,43 @@
+import argparse
+import logging
+
+from dvc.cli.command import CmdBaseNoRepo
+from dvc.cli.utils import append_doc_link
+from dvc.ui import ui
+
+from .ls import _prettify
+
+logger = logging.getLogger(__name__)
+
+
+class CmdListUrl(CmdBaseNoRepo):
+    def run(self):
+        from dvc.repo import Repo
+
+        entries = Repo.ls_url(self.args.url, recursive=self.args.recursive)
+        if entries:
+            entries = _prettify(entries, with_color=True)
+            ui.write("\n".join(entries))
+        return 0
+
+
+def add_parser(subparsers, parent_parser):
+    LS_HELP = "List directory contents from URL."
+    lsurl_parser = subparsers.add_parser(
+        "list-url",
+        aliases=["ls-url"],
+        parents=[parent_parser],
+        description=append_doc_link(LS_HELP, "list-url"),
+        help=LS_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    lsurl_parser.add_argument(
+        "url", help="See `dvc import-url -h` for full list of supported URLs."
+    )
+    lsurl_parser.add_argument(
+        "-R",
+        "--recursive",
+        action="store_true",
+        help="Recursively list files.",
+    )
+    lsurl_parser.set_defaults(func=CmdListUrl)

dvc/commands/plots.py

@@ -25,7 +25,7 @@ def _show_json(renderers, split=False):
 
 
 def _adjust_vega_renderers(renderers):
-    from dvc.render import VERSION_FIELD
+    from dvc.render import REVISION_FIELD, VERSION_FIELD
     from dvc_render import VegaRenderer
 
     for r in renderers:
@@ -42,6 +42,7 @@ def _adjust_vega_renderers(renderers):
                         dp["rev"] = "::".join(vi.values())
             else:
                 for dp in r.datapoints:
+                    dp[REVISION_FIELD] = dp[VERSION_FIELD]["revision"]
                     dp.pop(VERSION_FIELD, {})
 
 
@@ -66,18 +67,6 @@ def _data_versions_count(renderer):
     return len(set(x))
 
 
-def _filter_unhandled_renderers(renderers):
-    # filtering out renderers currently unhandled by vscode extension
-    from dvc_render import VegaRenderer
-
-    def _is_json_viable(r):
-        return not (
-            isinstance(r, VegaRenderer) and _data_versions_count(r) > 1
-        )
-
-    return list(filter(_is_json_viable, renderers))
-
-
 class CmdPlots(CmdBase):
     def _func(self, *args, **kwargs):
         raise NotImplementedError
@@ -155,17 +144,16 @@ class CmdPlots(CmdBase):
                 out=renderers_out,
                 templates_dir=self.repo.plots.templates_dir,
             )
+            if self.args.json:
+                _show_json(renderers, self.args.split)
+                return 0
+
+            _adjust_vega_renderers(renderers)
             if self.args.show_vega:
                 renderer = first(filter(lambda r: r.TYPE == "vega", renderers))
                 if renderer:
                     ui.write_json(json.loads(renderer.get_filled_template()))
                 return 0
-            if self.args.json:
-                renderers = _filter_unhandled_renderers(renderers)
-                _show_json(renderers, self.args.split)
-                return 0
-
-            _adjust_vega_renderers(renderers)
 
             output_file: Path = (Path.cwd() / out).resolve() / "index.html"
 
@@ -222,31 +210,22 @@ class CmdPlotsModify(CmdPlots):
 
 
 class CmdPlotsTemplates(CmdBase):
-    TEMPLATES_CHOICES = [
-        "simple",
-        "linear",
-        "confusion",
-        "confusion_normalized",
-        "scatter",
-        "smooth",
-        "bar_horizontal_sorted",
-        "bar_horizontal",
-    ]
-
     def run(self):
-        from dvc_render.vega_templates import dump_templates
+        from dvc.exceptions import InvalidArgumentError
+        from dvc_render.vega_templates import TEMPLATES
 
         try:
-            out = (
-                os.path.join(os.getcwd(), self.args.out)
-                if self.args.out
-                else self.repo.plots.templates_dir
-            )
+            target = self.args.template
+            if target:
+                for template in TEMPLATES:
+                    if target == template.DEFAULT_NAME:
+                        ui.write_json(template.DEFAULT_CONTENT)
+                        return 0
+                raise InvalidArgumentError(f"Unexpected template: {target}.")
 
-            targets = [self.args.target] if self.args.target else None
-            dump_templates(output=out, targets=targets)
-            templates_path = os.path.relpath(out, os.getcwd())
-            ui.write(f"Templates have been written into '{templates_path}'.")
+            else:
+                for template in TEMPLATES:
+                    ui.write(template.DEFAULT_NAME)
 
             return 0
         except DvcException:
@@ -356,8 +335,7 @@ def add_parser(subparsers, parent_parser):
     plots_modify_parser.set_defaults(func=CmdPlotsModify)
 
     TEMPLATES_HELP = (
-        "Write built-in plots templates to a directory "
-        "(.dvc/plots by default)."
+        "List built-in plots templates or show JSON specification for one."
     )
     plots_templates_parser = plots_subparsers.add_parser(
         "templates",
@@ -367,13 +345,14 @@ def add_parser(subparsers, parent_parser):
         formatter_class=argparse.RawDescriptionHelpFormatter,
     )
     plots_templates_parser.add_argument(
-        "target",
+        "template",
         default=None,
         nargs="?",
-        choices=CmdPlotsTemplates.TEMPLATES_CHOICES,
-        help="Template to write. Writes all templates by default.",
+        help=(
+            "Template for which to show JSON specification. "
+            "List all template names by default."
+        ),
     )
-    _add_output_argument(plots_templates_parser, typ="templates")
     plots_templates_parser.set_defaults(func=CmdPlotsTemplates)
 
 

dvc/commands/repro.py

@@ -55,7 +55,16 @@ def add_arguments(repro_parser):
     repro_parser.add_argument(
         "targets",
         nargs="*",
-        help="Stages to reproduce. 'dvc.yaml' by default.",
+        help=(
+            """\
+Stages to reproduce. 'dvc.yaml' by default.
+The targets can be path to a dvc.yaml file or `.dvc` file,
+or a stage name from dvc.yaml file from
+current working directory. To run a stage from dvc.yaml
+from other directories, the target must be a path followed by colon `:`
+and then the stage name name.
+"""
+        ),
     ).complete = completion.DVCFILES_AND_STAGE
     repro_parser.add_argument(
         "-f",

dvc/commands/stage.py

@@ -76,7 +76,7 @@ class CmdStageList(CmdBase):
         # removing duplicates while maintaining order
         collected = chain.from_iterable(
             self.repo.stage.collect(
-                target=target, recursive=self.args.recursive, accept_group=True
+                target=target, recursive=self.args.recursive
             )
             for target in self.args.targets
         )
@@ -238,26 +238,6 @@ def _add_common_args(parser):
         help="Declare output plot file (do not put into DVC cache).",
         metavar="<path>",
     )
-    parser.add_argument(
-        "--live", help="Declare output as dvclive.", metavar="<path>"
-    )
-    parser.add_argument(
-        "--live-no-cache",
-        help="Declare output as dvclive (do not put into DVC cache).",
-        metavar="<path>",
-    )
-    parser.add_argument(
-        "--live-no-summary",
-        action="store_true",
-        default=False,
-        help=argparse.SUPPRESS,
-    )
-    parser.add_argument(
-        "--live-no-html",
-        action="store_true",
-        default=False,
-        help="Signal dvclive logger to not produce training report.",
-    )
     parser.add_argument(
         "-w",
         "--wdir",

dvc/commands/update.py

@@ -18,6 +18,7 @@ class CmdUpdate(CmdBase):
                 rev=self.args.rev,
                 recursive=self.args.recursive,
                 to_remote=self.args.to_remote,
+                no_download=self.args.no_download,
                 remote=self.args.remote,
                 jobs=self.args.jobs,
             )
@@ -55,6 +56,13 @@ def add_parser(subparsers, parent_parser):
         default=False,
         help="Update all stages in the specified directory.",
     )
+    update_parser.add_argument(
+        "--no-download",
+        action="store_true",
+        default=False,
+        help="Update .dvc file git revision/hash value(s)"
+        " but do not download the file(s).",
+    )
     update_parser.add_argument(
         "--to-remote",
         action="store_true",

dvc/config.py

@@ -176,7 +176,10 @@ class Config(dict):
 
         if fs.exists(filename):
             with fs.open(filename) as fobj:
-                conf_obj = ConfigObj(fobj)
+                try:
+                    conf_obj = ConfigObj(fobj)
+                except UnicodeDecodeError as exc:
+                    raise ConfigError(str(exc)) from exc
         else:
             conf_obj = ConfigObj()
         return _parse_named(_lower_keys(conf_obj.dict()))

dvc/config_schema.py

@@ -78,7 +78,9 @@ REMOTE_COMMON = {
     "url": str,
     "checksum_jobs": All(Coerce(int), Range(1)),
     "jobs": All(Coerce(int), Range(1)),
+    Optional("worktree"): Bool,
     Optional("no_traverse"): Bool,  # obsoleted
+    Optional("version_aware"): Bool,
 }
 LOCAL_COMMON = {
     "type": supported_cache_type,
@@ -168,6 +170,7 @@ SCHEMA = {
                 "gs": {
                     "projectname": str,
                     "credentialpath": str,
+                    "endpointurl": str,
                     Optional("verify", default=False): Bool,
                     **REMOTE_COMMON,
                 },
@@ -177,6 +180,8 @@ SCHEMA = {
                     "user": str,
                     "password": str,
                     "ask_password": Bool,
+                    "passphrase": str,
+                    "ask_passphrase": Bool,
                     "keyfile": str,
                     "timeout": Coerce(int),
                     "gss_auth": Bool,
@@ -288,4 +293,9 @@ SCHEMA = {
         "bool": All(Lower, Choices("store_true", "boolean_optional")),
         "list": All(Lower, Choices("nargs", "append")),
     },
+    "hydra": {
+        Optional("enabled", default=False): Bool,
+        "config_dir": str,
+        "config_name": str,
+    },
 }

dvc/data_cloud.py

@@ -3,16 +3,38 @@
 import logging
 from typing import TYPE_CHECKING, Iterable, Optional
 
-from dvc_data.db import get_index
+from funcy import cached_property
+
+from dvc_data.hashfile.db import get_index
 
 if TYPE_CHECKING:
     from dvc_data.hashfile.db import HashFileDB
     from dvc_data.hashfile.hash_info import HashInfo
-    from dvc_data.status import CompareStatusResult
+    from dvc_data.hashfile.status import CompareStatusResult
+    from dvc_data.hashfile.transfer import TransferResult
 
 logger = logging.getLogger(__name__)
 
 
+class Remote:
+    def __init__(self, path, fs, **config):
+        self.path = path
+        self.fs = fs
+
+        self.worktree = config.pop("worktree", False)
+        self.config = config
+
+    @cached_property
+    def odb(self):
+        from dvc_data.hashfile.db import get_odb
+
+        path = self.path
+        if self.worktree:
+            path = self.fs.path.join(path, ".dvc", "cache")
+
+        return get_odb(self.fs, path, **self.config)
+
+
 class DataCloud:
     """Class that manages dvc remotes.
 
@@ -27,18 +49,23 @@ class DataCloud:
     def __init__(self, repo):
         self.repo = repo
 
-    def get_remote_odb(
+    def get_remote(
         self,
         name: Optional[str] = None,
         command: str = "<command>",
-    ) -> "HashFileDB":
+    ) -> "Remote":
         from dvc.config import NoRemoteError
 
         if not name:
             name = self.repo.config["core"].get("remote")
 
         if name:
-            return self._init_odb(name)
+            from dvc.fs import get_cloud_fs
+
+            cls, config, fs_path = get_cloud_fs(self.repo, name=name)
+            fs = cls(**config)
+            config["tmp_dir"] = self.repo.index_db_dir
+            return Remote(fs_path, fs, **config)
 
         if bool(self.repo.config["remote"]):
             error_msg = (
@@ -55,13 +82,13 @@ class DataCloud:
 
         raise NoRemoteError(error_msg)
 
-    def _init_odb(self, name):
-        from dvc.fs import get_cloud_fs
-        from dvc_data.db import get_odb
-
-        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 get_remote_odb(
+        self,
+        name: Optional[str] = None,
+        command: str = "<command>",
+    ) -> "HashFileDB":
+        remote = self.get_remote(name=name, command=command)
+        return remote.odb
 
     def _log_missing(self, status: "CompareStatusResult"):
         if status.missing:
@@ -80,14 +107,10 @@ class DataCloud:
         dest_odb: "HashFileDB",
         objs: Iterable["HashInfo"],
         **kwargs,
-    ):
-        from dvc.exceptions import FileTransferError
-        from dvc_data.transfer import TransferError, transfer
+    ) -> "TransferResult":
+        from dvc_data.hashfile.transfer import transfer
 
-        try:
-            return transfer(src_odb, dest_odb, objs, **kwargs)
-        except TransferError as exc:
-            raise FileTransferError(exc.fails) from exc
+        return transfer(src_odb, dest_odb, objs, **kwargs)
 
     def push(
         self,
@@ -95,7 +118,7 @@ class DataCloud:
         jobs: Optional[int] = None,
         remote: Optional[str] = None,
         odb: Optional["HashFileDB"] = None,
-    ):
+    ) -> "TransferResult":
         """Push data items in a cloud-agnostic way.
 
         Args:
@@ -122,7 +145,7 @@ class DataCloud:
         jobs: Optional[int] = None,
         remote: Optional[str] = None,
         odb: Optional["HashFileDB"] = None,
-    ):
+    ) -> "TransferResult":
         """Pull data items in a cloud-agnostic way.
 
         Args:
@@ -164,7 +187,7 @@ class DataCloud:
             log_missing: log warning messages if file doesn't exist
                 neither in cache, neither in cloud.
         """
-        from dvc_data.status import compare_status
+        from dvc_data.hashfile.status import compare_status
 
         if not odb:
             odb = self.get_remote_odb(remote, "status")

dvc/dependency/__init__.py

@@ -1,7 +1,7 @@
 from collections import defaultdict
 from typing import Any, Mapping
 
-from dvc.output import ARTIFACT_SCHEMA, Output
+from dvc.output import ARTIFACT_SCHEMA, DIR_FILES_SCHEMA, Output
 
 from .base import Dependency
 from .param import ParamsDependency
@@ -15,10 +15,11 @@ SCHEMA: Mapping[str, Any] = {
     **ARTIFACT_SCHEMA,
     **RepoDependency.REPO_SCHEMA,
     **ParamsDependency.PARAM_SCHEMA,
+    Output.PARAM_FILES: [DIR_FILES_SCHEMA],
 }
 
 
-def _get(stage, p, info):
+def _get(stage, p, info, **kwargs):
     if info and info.get(RepoDependency.PARAM_REPO):
         repo = info.pop(RepoDependency.PARAM_REPO)
         return RepoDependency(repo, stage, p, info)
@@ -27,21 +28,30 @@ def _get(stage, p, info):
         params = info.pop(ParamsDependency.PARAM_PARAMS)
         return ParamsDependency(stage, p, params)
 
-    return Dependency(stage, p, info)
+    return Dependency(stage, p, info, **kwargs)
 
 
 def loadd_from(stage, d_list):
     ret = []
     for d in d_list:
         p = d.pop(Output.PARAM_PATH, None)
-        ret.append(_get(stage, p, d))
+        files = d.pop(Output.PARAM_FILES, None)
+        ret.append(_get(stage, p, d, files=files))
     return ret
 
 
-def loads_from(stage, s_list, erepo=None):
+def loads_from(stage, s_list, erepo=None, fs_config=None):
     assert isinstance(s_list, list)
     info = {RepoDependency.PARAM_REPO: erepo} if erepo else {}
-    return [_get(stage, s, info.copy()) for s in s_list]
+    return [
+        _get(
+            stage,
+            s,
+            info.copy(),
+            fs_config=fs_config,
+        )
+        for s in s_list
+    ]
 
 
 def _merge_params(s_list):

dvc/dependency/base.py

@@ -1,6 +1,7 @@
-from typing import Type
+from typing import Dict, Type
 
 from dvc.exceptions import DvcException
+from dvc.fs import download as fs_download
 from dvc.output import Output
 
 
@@ -30,5 +31,39 @@ class Dependency(Output):
     )  # type: Type[DvcException]
     IsStageFileError = DependencyIsStageFileError  # type: Type[DvcException]
 
+    def workspace_status(self) -> Dict[str, str]:
+        if self.fs.version_aware:
+            old_fs_path = self.fs_path
+            try:
+                self.fs_path = self.fs.path.version_path(self.fs_path, None)
+                if self.changed_meta():
+                    return {str(self): "update available"}
+            finally:
+                self.fs_path = old_fs_path
+        return super().workspace_status()
+
     def update(self, rev=None):
-        pass
+        if self.fs.version_aware:
+            self.fs_path = self.fs.path.version_path(self.fs_path, rev)
+            self.meta = self.get_meta()
+            self.def_path = self.fs.path.version_path(
+                self.def_path, self.meta.version_id
+            )
+            self.fs_path = self.fs.path.version_path(
+                self.fs_path, self.meta.version_id
+            )
+
+    def download(self, to, jobs=None):
+        fs_download(self.fs, self.fs_path, to, jobs=jobs)
+
+    def save(self):
+        super().save()
+        if self.fs.version_aware:
+            self.fs_path = self.fs.path.version_path(
+                self.fs_path, self.meta.version_id
+            )
+
+    def dumpd(self, **kwargs):
+        if self.fs.version_aware:
+            kwargs["with_files"] = True
+        return super().dumpd(**kwargs)

dvc/dependency/param.py

@@ -8,7 +8,7 @@ import dpath.util
 from voluptuous import Any
 
 from dvc.exceptions import DvcException
-from dvc.utils.serialize import LOADERS, ParseError
+from dvc.utils.serialize import ParseError, load_path
 from dvc_data.hashfile.hash_info import HashInfo
 
 from .base import Dependency
@@ -39,14 +39,15 @@ class ParamsDependency(Dependency):
 
     def __init__(self, stage, path, params=None, repo=None):
         self.params = list(params) if params else []
-        info = (
-            {self.PARAM_PARAMS: params} if isinstance(params, dict) else None
-        )
+        hash_info = HashInfo()
+        if isinstance(params, dict):
+            hash_info = HashInfo(self.PARAM_PARAMS, params)
         repo = repo or stage.repo
         path = path or os.path.join(repo.root_dir, self.DEFAULT_PARAMS_FILE)
-        super().__init__(stage, path, info=info, repo=repo)
+        super().__init__(stage, path, repo=repo)
+        self.hash_info = hash_info
 
-    def dumpd(self):
+    def dumpd(self, **kwargs):
         ret = super().dumpd()
         if not self.hash_info:
             ret[self.PARAM_PARAMS] = self.params or {}
@@ -139,12 +140,9 @@ class ParamsDependency(Dependency):
             )
 
     def read_file(self):
-        _, ext = os.path.splitext(self.fs_path)
-        loader = LOADERS[ext]
-
         self.validate_filepath()
         try:
-            return loader(self.fs_path, fs=self.repo.fs)
+            return load_path(self.fs_path, self.repo.fs)
         except ParseError as exc:
             raise BadParamFileError(
                 f"Unable to read parameters from '{self}'"

dvc/dependency/repo.py

@@ -6,6 +6,7 @@ from typing import TYPE_CHECKING, Dict, Optional, Set, Tuple
 from voluptuous import Required
 
 from dvc.prompt import confirm
+from dvc_data.hashfile.meta import Meta
 
 from .base import Dependency
 
@@ -32,6 +33,7 @@ class RepoDependency(Dependency):
     def __init__(self, def_repo, stage, *args, **kwargs):
         self.def_repo = def_repo
         self._objs: Dict[str, "HashFile"] = {}
+        self._meta: Dict[str, Meta] = {}
         super().__init__(stage, *args, **kwargs)
 
     def _parse_path(self, fs, fs_path):
@@ -59,11 +61,11 @@ class RepoDependency(Dependency):
     def save(self):
         pass
 
-    def dumpd(self):
+    def dumpd(self, **kwargs):
         return {self.PARAM_PATH: self.def_path, self.PARAM_REPO: self.def_repo}
 
     def download(self, to, jobs=None):
-        from dvc_data.checkout import checkout
+        from dvc_data.hashfile.checkout import checkout
 
         for odb, objs in self.get_used_objs().items():
             self.repo.cloud.pull(objs, jobs=jobs, odb=odb)
@@ -94,17 +96,17 @@ class RepoDependency(Dependency):
     def get_used_objs(
         self, **kwargs
     ) -> Dict[Optional["ObjectDB"], Set["HashInfo"]]:
-        used, _ = self._get_used_and_obj(**kwargs)
+        used, _, _ = self._get_used_and_obj(**kwargs)
         return used
 
     def _get_used_and_obj(
         self, obj_only=False, **kwargs
-    ) -> Tuple[Dict[Optional["ObjectDB"], Set["HashInfo"]], "HashFile"]:
+    ) -> Tuple[Dict[Optional["ObjectDB"], Set["HashInfo"]], Meta, "HashFile"]:
         from dvc.config import NoRemoteError
         from dvc.exceptions import NoOutputOrStageError, PathMissingError
         from dvc.utils import as_posix
-        from dvc_data.build import build
-        from dvc_data.objects.tree import Tree, TreeError
+        from dvc_data.hashfile.build import build
+        from dvc_data.hashfile.tree import Tree, TreeError
 
         local_odb = self.repo.odb.local
         locked = kwargs.pop("locked", True)
@@ -131,7 +133,7 @@ class RepoDependency(Dependency):
                     pass
 
             try:
-                object_store, _, obj = build(
+                object_store, meta, obj = build(
                     local_odb,
                     as_posix(self.def_path),
                     repo.dvcfs,
@@ -145,16 +147,18 @@ class RepoDependency(Dependency):
             object_store.read_only = True
 
             self._objs[rev] = obj
+            self._meta[rev] = meta
+
             used_obj_ids[object_store].add(obj.hash_info)
             if isinstance(obj, Tree):
                 used_obj_ids[object_store].update(oid for _, _, oid in obj)
-            return used_obj_ids, obj
+            return used_obj_ids, meta, obj
 
     def _check_circular_import(self, odb, obj_ids):
         from dvc.exceptions import CircularImportError
-        from dvc.fs.dvc import DvcFileSystem
-        from dvc_data.db.reference import ReferenceHashFileDB
-        from dvc_data.objects.tree import Tree
+        from dvc.fs.dvc import DVCFileSystem
+        from dvc_data.hashfile.db.reference import ReferenceHashFileDB
+        from dvc_data.hashfile.tree import Tree
 
         if not isinstance(odb, ReferenceHashFileDB):
             return
@@ -169,7 +173,7 @@ class RepoDependency(Dependency):
 
         checked_urls = set()
         for obj in iter_objs():
-            if not isinstance(obj.fs, DvcFileSystem):
+            if not isinstance(obj.fs, DVCFileSystem):
                 continue
             if (
                 obj.fs.repo_url in checked_urls
@@ -190,11 +194,21 @@ class RepoDependency(Dependency):
         rev = self._get_rev(locked=locked)
         if rev in self._objs:
             return self._objs[rev]
-        _, obj = self._get_used_and_obj(
+        _, _, obj = self._get_used_and_obj(
             obj_only=True, filter_info=filter_info, **kwargs
         )
         return obj
 
+    def get_meta(self, filter_info=None, **kwargs):
+        locked = kwargs.get("locked", True)
+        rev = self._get_rev(locked=locked)
+        if rev in self._meta:
+            return self._meta[rev]
+        _, meta, _ = self._get_used_and_obj(
+            obj_only=True, filter_info=filter_info, **kwargs
+        )
+        return meta
+
     def _make_repo(self, locked=True, **kwargs):
         from dvc.external_repo import external_repo
 

dvc/dvcfile.py

@@ -172,7 +172,7 @@ class FileMixin:
     def dump(self, stage, **kwargs):
         raise NotImplementedError
 
-    def merge(self, ancestor, other):
+    def merge(self, ancestor, other, allowed=None):
         raise NotImplementedError
 
 
@@ -198,18 +198,18 @@ class SingleStageFile(FileMixin):
         if self.verify:
             check_dvcfile_path(self.repo, self.path)
         logger.debug("Saving information to '%s'.", relpath(self.path))
-        dump_yaml(self.path, serialize.to_single_stage_file(stage))
+        dump_yaml(self.path, serialize.to_single_stage_file(stage, **kwargs))
         self.repo.scm_context.track_file(self.relpath)
 
     def remove_stage(self, stage):  # pylint: disable=unused-argument
         self.remove()
 
-    def merge(self, ancestor, other):
+    def merge(self, ancestor, other, allowed=None):
         assert isinstance(ancestor, SingleStageFile)
         assert isinstance(other, SingleStageFile)
 
         stage = self.stage
-        stage.merge(ancestor.stage, other.stage)
+        stage.merge(ancestor.stage, other.stage, allowed=allowed)
         self.dump(stage)
 
 
@@ -237,15 +237,17 @@ class PipelineFile(FileMixin):
             self._dump_pipeline_file(stage)
 
         if update_lock:
-            self._dump_lockfile(stage)
+            self._dump_lockfile(stage, **kwargs)
 
-    def _dump_lockfile(self, stage):
-        self._lockfile.dump(stage)
+    def _dump_lockfile(self, stage, **kwargs):
+        self._lockfile.dump(stage, **kwargs)
 
     @staticmethod
-    def _check_if_parametrized(stage):
+    def _check_if_parametrized(stage, action: str = "dump") -> None:
         if stage.raw_data.parametrized:
-            raise ParametrizedDumpError(f"cannot dump a parametrized {stage}")
+            raise ParametrizedDumpError(
+                f"cannot {action} a parametrized {stage}"
+            )
 
     def _dump_pipeline_file(self, stage):
         self._check_if_parametrized(stage)
@@ -291,6 +293,7 @@ class PipelineFile(FileMixin):
         self._lockfile.remove()
 
     def remove_stage(self, stage):
+        self._check_if_parametrized(stage, "remove")
         self._lockfile.remove_stage(stage)
         if not self.exists():
             return
@@ -307,7 +310,7 @@ class PipelineFile(FileMixin):
         else:
             super().remove()
 
-    def merge(self, ancestor, other):
+    def merge(self, ancestor, other, allowed=None):
         raise NotImplementedError
 
 
@@ -363,7 +366,7 @@ class Lockfile(FileMixin):
         return {SCHEMA_KWD: version}
 
     def dump(self, stage, **kwargs):
-        stage_data = serialize.to_lockfile(stage)
+        stage_data = serialize.to_lockfile(stage, **kwargs)
 
         with modify_yaml(self.path, fs=self.repo.fs) as data:
             version = LOCKFILE_VERSION.from_dict(data)
@@ -408,7 +411,7 @@ class Lockfile(FileMixin):
         else:
             self.remove()
 
-    def merge(self, ancestor, other):
+    def merge(self, ancestor, other, allowed=None):
         raise NotImplementedError
 
 

dvc/env.py

@@ -2,9 +2,6 @@ DVC_CHECKPOINT = "DVC_CHECKPOINT"
 DVC_DAEMON = "DVC_DAEMON"
 DVC_PAGER = "DVC_PAGER"
 DVC_ROOT = "DVC_ROOT"
-DVCLIVE_PATH = "DVCLIVE_PATH"
-DVCLIVE_SUMMARY = "DVCLIVE_SUMMARY"
-DVCLIVE_HTML = "DVCLIVE_HTML"
 DVCLIVE_RESUME = "DVCLIVE_RESUME"
 DVC_IGNORE_ISATTY = "DVC_IGNORE_ISATTY"
 DVC_EXP_GIT_REMOTE = "DVC_EXP_GIT_REMOTE"

dvc/exceptions.py

@@ -132,10 +132,17 @@ class NotDvcRepoError(DvcException):
 class CyclicGraphError(DvcException):
     def __init__(self, stages):
         assert isinstance(stages, list)
-        msg = "Pipeline has a cycle involving: {}.".format(
-            ", ".join(s.addressing for s in stages)
+        stage_part = "stage" if len(stages) == 1 else "stages"
+        msg = (
+            "Same item(s) are defined as both a dependency and an output "
+            "in {stage_part}: {stage}."
+        )
+        super().__init__(
+            msg.format(
+                stage_part=stage_part,
+                stage=", ".join(s.addressing for s in stages),
+            )
         )
-        super().__init__(msg)
 
 
 class ConfirmRemoveError(DvcException):
@@ -278,6 +285,11 @@ class PathMissingError(DvcException):
         self.dvc_only = dvc_only
 
 
+class URLMissingError(DvcException):
+    def __init__(self, url):
+        super().__init__(f"The path '{url}' does not exist")
+
+
 class RemoteCacheRequiredError(DvcException):
     def __init__(self, scheme, fs_path):
 

dvc/fs/__init__.py

@@ -1,26 +1,21 @@
 from urllib.parse import urlparse
 
+from dvc_http import HTTPFileSystem, HTTPSFileSystem  # noqa: F401
+
+from dvc.config import ConfigError as RepoConfigError
+from dvc.config_schema import SCHEMA, Invalid
+
 # pylint: disable=unused-import
 from dvc_objects.fs import utils  # noqa: F401
 from dvc_objects.fs import (  # noqa: F401
-    FS_MAP,
-    AzureFileSystem,
-    GDriveFileSystem,
-    GSFileSystem,
-    HDFSFileSystem,
-    HTTPFileSystem,
-    HTTPSFileSystem,
     LocalFileSystem,
     MemoryFileSystem,
-    OSSFileSystem,
-    S3FileSystem,
     Schemes,
-    SSHFileSystem,
-    WebDAVFileSystem,
-    WebDAVSFileSystem,
-    WebHDFSFileSystem,
     generic,
     get_fs_cls,
+    known_implementations,
+    localfs,
+    registry,
     system,
 )
 from dvc_objects.fs.base import AnyFSPath, FileSystem  # noqa: F401
@@ -29,21 +24,47 @@ from dvc_objects.fs.errors import (  # noqa: F401
     ConfigError,
     RemoteMissingDepsError,
 )
-from dvc_objects.fs.implementations.azure import AzureAuthError  # noqa: F401
-from dvc_objects.fs.implementations.local import localfs  # noqa: F401
-from dvc_objects.fs.implementations.ssh import (  # noqa: F401
-    DEFAULT_PORT as DEFAULT_SSH_PORT,
-)
 from dvc_objects.fs.path import Path  # noqa: F401
 
+from .callbacks import Callback
 from .data import DataFileSystem  # noqa: F401
-from .dvc import DvcFileSystem  # noqa: F401
+from .dvc import DVCFileSystem  # noqa: F401
 from .git import GitFileSystem  # noqa: F401
 
+known_implementations.update(
+    {
+        "dvc": {
+            "class": "dvc.fs.dvc.DVCFileSystem",
+            "err": "dvc is supported, but requires 'dvc' to be installed",
+        },
+        "git": {
+            "class": "dvc.fs.git.GitFileSystem",
+            "err": "git is supported, but requires 'dvc' to be installed",
+        },
+    }
+)
+
+
 # pylint: enable=unused-import
 
 
-def get_fs_config(repo, config, **kwargs):
+def download(fs, fs_path, to, jobs=None):
+    with Callback.as_tqdm_callback(
+        desc=f"Downloading {fs.path.name(fs_path)}",
+        unit="files",
+    ) as cb:
+        fs.get(fs_path, to.fs_path, batch_size=jobs, callback=cb)
+
+
+def parse_external_url(url, config=None):
+    remote_config = dict(config) if config else {}
+    remote_config["url"] = url
+    fs_cls, fs_config, fs_path = get_cloud_fs(None, **remote_config)
+    fs = fs_cls(**fs_config)
+    return fs, fs_path
+
+
+def get_fs_config(config, **kwargs):
     name = kwargs.get("name")
     if name:
         try:
@@ -54,10 +75,10 @@ def get_fs_config(repo, config, **kwargs):
             raise RemoteNotFoundError(f"remote '{name}' doesn't exist")
     else:
         remote_conf = kwargs
-    return _resolve_remote_refs(repo, config, remote_conf)
+    return _resolve_remote_refs(config, remote_conf)
 
 
-def _resolve_remote_refs(repo, config, remote_conf):
+def _resolve_remote_refs(config, remote_conf):
     # Support for cross referenced remotes.
     # This will merge the settings, shadowing base ref with remote_conf.
     # For example, having:
@@ -83,21 +104,22 @@ def _resolve_remote_refs(repo, config, remote_conf):
     if parsed.scheme != "remote":
         return remote_conf
 
-    base = get_fs_config(repo, config, name=parsed.netloc)
-    cls, _, _ = get_cloud_fs(repo, **base)
+    base = get_fs_config(config, name=parsed.netloc)
+    cls, _, _ = _get_cloud_fs(config, **base)
     relpath = parsed.path.lstrip("/").replace("/", cls.sep)
     url = cls.sep.join((base["url"], relpath))
     return {**base, **remote_conf, "url": url}
 
 
 def get_cloud_fs(repo, **kwargs):
-    from dvc.config import ConfigError as RepoConfigError
-    from dvc.config_schema import SCHEMA, Invalid
-
     repo_config = repo.config if repo else {}
+    return _get_cloud_fs(repo_config, **kwargs)
+
+
+def _get_cloud_fs(repo_config, **kwargs):
     core_config = repo_config.get("core", {})
 
-    remote_conf = get_fs_config(repo, repo_config, **kwargs)
+    remote_conf = get_fs_config(repo_config, **kwargs)
     try:
         remote_conf = SCHEMA["remote"][str](remote_conf)
     except Invalid as exc:
@@ -110,11 +132,8 @@ def get_cloud_fs(repo, **kwargs):
 
     cls = get_fs_cls(remote_conf)
 
-    if cls == GDriveFileSystem and repo:
-        remote_conf["gdrive_credentials_tmp_dir"] = repo.tmp_dir
-
     url = remote_conf.pop("url")
-    if issubclass(cls, WebDAVFileSystem):
+    if cls.protocol in ["webdav", "webdavs"]:
         # For WebDAVFileSystem, provided url is the base path itself, so it
         # should be treated as being a root path.
         fs_path = cls.root_marker

dvc/fs/callbacks.py

@@ -1,6 +1,6 @@
 # pylint: disable=unused-import
 from contextlib import ExitStack
-from typing import TYPE_CHECKING, Optional
+from typing import TYPE_CHECKING, Any, BinaryIO, Dict, Optional, Union
 
 from funcy import cached_property
 
@@ -24,6 +24,7 @@ class RichCallback(Callback):
         bytes: bool = False,  # pylint: disable=redefined-builtin
         unit: str = None,
         disable: bool = False,
+        transient: bool = True,
     ) -> None:
         self._progress = progress
         self.disable = disable
@@ -35,6 +36,7 @@ class RichCallback(Callback):
             "visible": False,
             "progress_type": None if bytes else "summary",
         }
+        self._transient = transient
         self._stack = ExitStack()
         super().__init__(size=size, value=value)
 
@@ -47,7 +49,7 @@ class RichCallback(Callback):
             return self._progress
 
         progress = RichTransferProgress(
-            transient=True,
+            transient=self._transient,
             disable=self.disable,
             console=ui.error_console,
         )
@@ -61,7 +63,8 @@ class RichCallback(Callback):
         return self
 
     def close(self):
-        self.progress.clear_task(self.task)
+        if self._transient:
+            self.progress.clear_task(self.task)
         self._stack.close()
 
     def call(self, hook_name=None, **kwargs):
@@ -72,8 +75,17 @@ class RichCallback(Callback):
             visible=not self.disable,
         )
 
-    def branch(self, path_1, path_2, kwargs, child: Optional[Callback] = None):
+    def branch(
+        self,
+        path_1: Union[str, BinaryIO],
+        path_2: str,
+        kwargs: Dict[str, Any],
+        child: "Callback" = None,
+    ):
         child = child or RichCallback(
-            progress=self.progress, desc=path_1, bytes=True
+            progress=self.progress,
+            desc=path_1 if isinstance(path_1, str) else path_2,
+            bytes=True,
+            transient=self._transient,
         )
         return super().branch(path_1, path_2, kwargs, child=child)

dvc/fs/dvc.py

@@ -1,17 +1,16 @@
+import errno
 import logging
 import ntpath
 import os
 import posixpath
 import threading
 from contextlib import suppress
-from itertools import takewhile
-from typing import TYPE_CHECKING, Callable, Optional, Tuple, Type, Union
+from typing import TYPE_CHECKING, Any, Callable, Optional, Tuple, Type, Union
 
 from fsspec.spec import AbstractFileSystem
 from funcy import cached_property, wrap_prop, wrap_with
 
 from dvc_objects.fs.base import FileSystem
-from dvc_objects.fs.callbacks import DEFAULT_CALLBACK
 from dvc_objects.fs.path import Path
 
 from .data import DataFileSystem
@@ -22,29 +21,19 @@ if TYPE_CHECKING:
 logger = logging.getLogger(__name__)
 
 RepoFactory = Union[Callable[[str], "Repo"], Type["Repo"]]
+Key = Tuple[str, ...]
 
 
 def as_posix(path: str) -> str:
     return path.replace(ntpath.sep, posixpath.sep)
 
 
-def _wrap_walk(dvc_fs, *args, **kwargs):
-    for root, dnames, fnames in dvc_fs.walk(*args, **kwargs):
-        yield dvc_fs.path.join(dvc_fs.repo.root_dir, root), dnames, fnames
+# NOT the same as dvc.dvcfile.is_dvc_file()!
+def _is_dvc_file(fname):
+    from dvc.dvcfile import is_valid_filename
+    from dvc.ignore import DvcIgnore
 
-
-def _ls(fs, path):
-    dnames = []
-    fnames = []
-
-    for entry in fs.ls(path, detail=True):
-        name = fs.path.name(entry["name"])
-        if entry["type"] == "directory":
-            dnames.append(name)
-        else:
-            fnames.append(name)
-
-    return dnames, fnames
+    return is_valid_filename(fname) or fname == DvcIgnore.DVCIGNORE_FILE
 
 
 def _merge_info(repo, fs_info, dvc_info):
@@ -70,40 +59,68 @@ def _merge_info(repo, fs_info, dvc_info):
     return ret
 
 
-class _DvcFileSystem(AbstractFileSystem):  # pylint:disable=abstract-method
-    """DVC + git-tracked files fs.
+def _get_dvc_path(dvc_fs, subkey):
+    return dvc_fs.path.join(*subkey) if subkey else ""
 
-    Args:
-        repo: DVC or git repo.
-        subrepos: traverse to subrepos (by default, it ignores subrepos)
-        repo_factory: A function to initialize subrepo with, default is Repo.
-        kwargs: Additional keyword arguments passed to the `DataFileSystem()`.
-    """
 
+class _DVCFileSystem(AbstractFileSystem):  # pylint:disable=abstract-method
+    cachable = False
     root_marker = "/"
 
-    PARAM_REPO_URL = "repo_url"
-    PARAM_REPO_ROOT = "repo_root"
-    PARAM_REV = "rev"
-    PARAM_CACHE_DIR = "cache_dir"
-    PARAM_CACHE_TYPES = "cache_types"
-    PARAM_SUBREPOS = "subrepos"
-
     def __init__(
         self,
+        url: Optional[str] = None,
+        rev: Optional[str] = None,
         repo: Optional["Repo"] = None,
-        subrepos=False,
-        repo_factory: RepoFactory = None,
-        **kwargs,
-    ):
-        super().__init__()
+        subrepos: bool = False,
+        repo_factory: Optional[RepoFactory] = None,
+        **repo_kwargs: Any,
+    ) -> None:
+        """DVC + git-tracked files fs.
 
+        Args:
+            path (str, optional): URL or path to a DVC/Git repository.
+                Defaults to a DVC repository in the current working directory.
+                Both HTTP and SSH protocols are supported for remote Git repos
+                (e.g. [user@]server:project.git).
+            rev (str, optional): Any Git revision such as a branch or tag name,
+                a commit hash or a dvc experiment name.
+                Defaults to the default branch in case of remote repositories.
+                In case of a local repository, if rev is unspecified, it will
+                default to the working directory.
+                If the repo is not a Git repo, this option is ignored.
+            repo (:obj:`Repo`, optional): `Repo` instance.
+            subrepos (bool): traverse to subrepos.
+                By default, it ignores subrepos.
+            repo_factory (callable): A function to initialize subrepo with.
+                The default is `Repo`.
+
+        Examples:
+            - Opening a filesystem from repo in current working directory
+
+            >>> fs = DVCFileSystem()
+
+            - Opening a filesystem from local repository
+
+            >>> fs = DVCFileSystem("path/to/local/repository")
+
+            - Opening a remote repository
+
+            >>> fs = DVCFileSystem(
+            ...    "https://github.com/iterative/example-get-started",
+            ...    rev="main",
+            ... )
+        """
         from pygtrie import Trie
 
+        super().__init__()
         if repo is None:
-            repo, repo_factory = self._repo_from_fs_config(
-                subrepos=subrepos, **kwargs
+            repo = self._make_repo(
+                url=url, rev=rev, subrepos=subrepos, **repo_kwargs
             )
+            assert repo
+            # pylint: disable=protected-access
+            repo_factory = repo._fs_conf["repo_factory"]
 
         if not repo_factory:
             from dvc.repo import Repo
@@ -137,78 +154,35 @@ class _DvcFileSystem(AbstractFileSystem):  # pylint:disable=abstract-method
         if hasattr(repo, "dvc_dir"):
             self._datafss[key] = DataFileSystem(index=repo.index.data["repo"])
 
-    def _get_key(self, path):
+    def _get_key(self, path) -> Key:
         parts = self.repo.fs.path.relparts(path, self.repo.root_dir)
         if parts == (".",):
             parts = ()
         return parts
 
+    def _get_key_from_relative(self, path) -> Key:
+        parts = self.path.relparts(path, self.root_marker)
+        if parts and parts[0] == os.curdir:
+            parts = parts[1:]
+        return parts
+
+    def _from_key(self, parts: Key) -> str:
+        return self.repo.fs.path.join(self.repo.root_dir, *parts)
+
     @property
     def repo_url(self):
         if self.repo is None:
             return None
         return self.repo.url
 
-    @property
-    def config(self):
-        return {
-            self.PARAM_REPO_URL: self.repo_url,
-            self.PARAM_REPO_ROOT: self.repo.root_dir,
-            self.PARAM_REV: getattr(self.repo.fs, "rev", None),
-            self.PARAM_CACHE_DIR: os.path.abspath(self.repo.odb.local.path),
-            self.PARAM_CACHE_TYPES: self.repo.odb.local.cache_types,
-            self.PARAM_SUBREPOS: self._traverse_subrepos,
-        }
-
     @classmethod
-    def _repo_from_fs_config(
-        cls, **config
-    ) -> Tuple["Repo", Optional["RepoFactory"]]:
-        from dvc.external_repo import erepo_factory, external_repo
+    def _make_repo(cls, **kwargs) -> "Repo":
         from dvc.repo import Repo
 
-        url = config.get(cls.PARAM_REPO_URL)
-        root = config.get(cls.PARAM_REPO_ROOT)
-        assert url or root
+        with Repo.open(uninitialized=True, **kwargs) as repo:
+            return repo
 
-        def _open(*args, **kwargs):
-            # NOTE: if original repo was an erepo (and has a URL),
-            # we cannot use Repo.open() since it will skip erepo
-            # cache/remote setup for local URLs
-            if url is None:
-                return Repo.open(*args, **kwargs)
-            return external_repo(*args, **kwargs)
-
-        cache_dir = config.get(cls.PARAM_CACHE_DIR)
-        cache_config = (
-            {}
-            if not cache_dir
-            else {
-                "cache": {
-                    "dir": cache_dir,
-                    "type": config.get(cls.PARAM_CACHE_TYPES),
-                }
-            }
-        )
-        repo_kwargs: dict = {
-            "rev": config.get(cls.PARAM_REV),
-            "subrepos": config.get(cls.PARAM_SUBREPOS, False),
-            "uninitialized": True,
-        }
-        factory: Optional["RepoFactory"] = None
-        if url is None:
-            repo_kwargs["config"] = cache_config
-        else:
-            repo_kwargs["cache_dir"] = cache_dir
-            factory = erepo_factory(url, root, cache_config)
-
-        with _open(
-            url if url else root,
-            **repo_kwargs,
-        ) as repo:
-            return repo, factory
-
-    def _get_repo(self, path: str) -> "Repo":
+    def _get_repo(self, key: Key) -> "Repo":
         """Returns repo that the path falls in, using prefix.
 
         If the path is already tracked/collected, it just returns the repo.
@@ -216,33 +190,21 @@ class _DvcFileSystem(AbstractFileSystem):  # pylint:disable=abstract-method
         Otherwise, it collects the repos that might be in the path's parents
         and then returns the appropriate one.
         """
-        if not self.repo.fs.path.isin_or_eq(path, self.repo.root_dir):
-            # outside of repo
-            return self.repo
-
-        key = self._get_key(path)
         repo = self._subrepos_trie.get(key)
         if repo:
             return repo
 
         prefix_key, repo = self._subrepos_trie.longest_prefix(key)
-        prefix = self.repo.fs.path.join(
-            self.repo.root_dir,
-            *prefix_key,  # pylint: disable=not-an-iterable
-        )
-
-        parents = (parent for parent in self.repo.fs.path.parents(path))
-        dirs = [path] + list(takewhile(lambda p: p != prefix, parents))
-        dirs.reverse()
-        self._update(dirs, starting_repo=repo)
+        dir_keys = (key[:i] for i in range(len(prefix_key) + 1, len(key) + 1))
+        self._update(dir_keys, starting_repo=repo)
         return self._subrepos_trie.get(key) or self.repo
 
     @wrap_with(threading.Lock())
-    def _update(self, dirs, starting_repo):
+    def _update(self, dir_keys, starting_repo):
         """Checks for subrepo in directories and updates them."""
         repo = starting_repo
-        for d in dirs:
-            key = self._get_key(d)
+        for key in dir_keys:
+            d = self._from_key(key)
             if self._is_dvc_repo(d):
                 repo = self.repo_factory(
                     d,
@@ -265,79 +227,68 @@ class _DvcFileSystem(AbstractFileSystem):  # pylint:disable=abstract-method
         repo_path = self.repo.fs.path.join(dir_path, Repo.DVC_DIR)
         return self.repo.fs.isdir(repo_path)
 
-    def _get_fs_pair(
-        self, path
-    ) -> Tuple[
-        Optional["Repo"],
-        Optional[FileSystem],
-        Optional[str],
-        Optional[DataFileSystem],
-        Optional[str],
-    ]:
+    def _get_subrepo_info(
+        self, key: Key
+    ) -> Tuple["Repo", Optional[DataFileSystem], Key]:
         """
-        Returns a pair of fss based on repo the path falls in, using prefix.
+        Returns information about the subrepo the key is part of.
         """
-        parts = self.path.relparts(path, self.root_marker)
-        if parts and parts[0] == os.curdir:
-            parts = parts[1:]
-
-        fs_path = self.repo.fs.path.join(self.repo.root_dir, *parts)
-        repo = self._get_repo(fs_path)
-        fs = repo.fs
-
-        repo_parts = fs.path.relparts(repo.root_dir, self.repo.root_dir)
-        if repo_parts[0] == os.curdir:
-            repo_parts = repo_parts[1:]
-
-        dvc_parts = parts[len(repo_parts) :]
-        if dvc_parts and dvc_parts[0] == os.curdir:
-            dvc_parts = dvc_parts[1:]
-
-        key = self._get_key(repo.root_dir)
-        dvc_fs = self._datafss.get(key)
-        if dvc_fs:
-            dvc_path = dvc_fs.path.join(*dvc_parts) if dvc_parts else ""
+        repo = self._get_repo(key)
+        repo_key: Key
+        if repo is self.repo:
+            repo_key = ()
+            subkey = key
         else:
-            dvc_path = None
+            repo_key = self._get_key(repo.root_dir)
+            subkey = key[len(repo_key) :]
 
-        return repo, fs, fs_path, dvc_fs, dvc_path
+        dvc_fs = self._datafss.get(repo_key)
+        return repo, dvc_fs, subkey
 
-    def open(
-        self, path, mode="r", encoding="utf-8", **kwargs
+    def _open(
+        self, path, mode="rb", **kwargs
     ):  # pylint: disable=arguments-renamed, arguments-differ
-        if "b" in mode:
-            encoding = None
+        if mode != "rb":
+            raise OSError(errno.EROFS, os.strerror(errno.EROFS))
 
-        _, fs, fs_path, dvc_fs, dvc_path = self._get_fs_pair(path)
+        key = self._get_key_from_relative(path)
+        fs_path = self._from_key(key)
         try:
-            return fs.open(fs_path, mode=mode, encoding=encoding)
+            return self.repo.fs.open(fs_path, mode=mode)
         except FileNotFoundError:
+            _, dvc_fs, subkey = self._get_subrepo_info(key)
             if not dvc_fs:
                 raise
 
-        return dvc_fs.open(dvc_path, mode=mode, encoding=encoding, **kwargs)
+        dvc_path = _get_dvc_path(dvc_fs, subkey)
+        return dvc_fs.open(dvc_path, mode=mode)
 
-    def isdvc(self, path, **kwargs):
-        _, _, _, dvc_fs, dvc_path = self._get_fs_pair(path)
+    def isdvc(self, path, **kwargs) -> bool:
+        """Is this entry dvc-tracked?"""
+        key = self._get_key_from_relative(path)
+        _, dvc_fs, subkey = self._get_subrepo_info(key)
+        dvc_path = _get_dvc_path(dvc_fs, subkey)
         return dvc_fs is not None and dvc_fs.isdvc(dvc_path, **kwargs)
 
     def ls(  # pylint: disable=arguments-differ
         self, path, detail=True, dvc_only=False, **kwargs
     ):
-        repo, fs, fs_path, dvc_fs, dvc_path = self._get_fs_pair(path)
-
-        dvcignore = repo.dvcignore
-        ignore_subrepos = kwargs.get("ignore_subrepos", True)
+        key = self._get_key_from_relative(path)
+        repo, dvc_fs, subkey = self._get_subrepo_info(key)
 
         names = set()
         if dvc_fs:
             with suppress(FileNotFoundError):
+                dvc_path = _get_dvc_path(dvc_fs, subkey)
                 for entry in dvc_fs.ls(dvc_path, detail=False):
                     names.add(dvc_fs.path.name(entry))
 
-        if not dvc_only and fs:
+        ignore_subrepos = kwargs.get("ignore_subrepos", True)
+        if not dvc_only:
+            fs = self.repo.fs
+            fs_path = self._from_key(key)
             try:
-                for entry in dvcignore.ls(
+                for entry in repo.dvcignore.ls(
                     fs, fs_path, detail=False, ignore_subrepos=ignore_subrepos
                 ):
                     names.add(fs.path.name(entry))
@@ -345,26 +296,21 @@ class _DvcFileSystem(AbstractFileSystem):  # pylint:disable=abstract-method
                 pass
 
         dvcfiles = kwargs.get("dvcfiles", False)
-
-        def _func(fname):
-            from dvc.dvcfile import is_valid_filename
-            from dvc.ignore import DvcIgnore
-
-            if dvcfiles:
-                return True
-
-            return not (
-                is_valid_filename(fname) or fname == DvcIgnore.DVCIGNORE_FILE
-            )
-
-        names = filter(_func, names)
+        if not dvcfiles:
+            names = (name for name in names if not _is_dvc_file(name))
 
         infos = []
         paths = []
         for name in names:
             entry_path = self.path.join(path, name)
+            entry_key = key + (name,)
             try:
-                info = self.info(entry_path, ignore_subrepos=ignore_subrepos)
+                info = self._info(
+                    entry_key,
+                    entry_path,
+                    ignore_subrepos=ignore_subrepos,
+                    check_ignored=False,
+                )
             except FileNotFoundError:
                 continue
             infos.append(info)
@@ -375,48 +321,39 @@ class _DvcFileSystem(AbstractFileSystem):  # pylint:disable=abstract-method
 
         return infos
 
-    def get_file(  # pylint: disable=arguments-differ
-        self, rpath, lpath, callback=DEFAULT_CALLBACK, **kwargs
-    ):
-        _, fs, fs_path, dvc_fs, dvc_path = self._get_fs_pair(rpath)
-
-        if fs:
-            try:
-                fs.get_file(fs_path, lpath, callback=callback, **kwargs)
-                return
-            except FileNotFoundError:
-                if not dvc_fs:
-                    raise
-        dvc_fs.get_file(dvc_path, lpath, callback=callback, **kwargs)
-
     def info(self, path, **kwargs):
-        repo, fs, fs_path, dvc_fs, dvc_path = self._get_fs_pair(path)
-
-        dvcignore = repo.dvcignore
+        key = self._get_key_from_relative(path)
         ignore_subrepos = kwargs.get("ignore_subrepos", True)
+        return self._info(key, path, ignore_subrepos=ignore_subrepos)
+
+    def _info(self, key, path, ignore_subrepos=True, check_ignored=True):
+        repo, dvc_fs, subkey = self._get_subrepo_info(key)
 
         dvc_info = None
         if dvc_fs:
             try:
-                dvc_info = dvc_fs.info(dvc_path)
+                dvc_info = dvc_fs.fs.index.info(subkey)
+                dvc_path = _get_dvc_path(dvc_fs, subkey)
+                dvc_info["name"] = dvc_path
             except FileNotFoundError:
                 pass
 
         fs_info = None
-        if fs:
-            try:
-                fs_info = fs.info(fs_path)
-                if dvcignore.is_ignored(
-                    fs, fs_path, ignore_subrepos=ignore_subrepos
-                ):
-                    fs_info = None
-            except (FileNotFoundError, NotADirectoryError):
-                if not dvc_info:
-                    raise
+        fs = self.repo.fs
+        fs_path = self._from_key(key)
+        try:
+            fs_info = fs.info(fs_path)
+            if check_ignored and repo.dvcignore.is_ignored(
+                fs, fs_path, ignore_subrepos=ignore_subrepos
+            ):
+                fs_info = None
+        except (FileNotFoundError, NotADirectoryError):
+            if not dvc_info:
+                raise
 
         # NOTE: if some parent in fs_path turns out to be a file, it means
         # that the whole repofs branch doesn't exist.
-        if fs and not fs_info and dvc_info:
+        if dvc_info and not fs_info:
             for parent in fs.path.parents(fs_path):
                 try:
                     if fs.info(parent)["type"] != "directory":
@@ -432,16 +369,8 @@ class _DvcFileSystem(AbstractFileSystem):  # pylint:disable=abstract-method
         info["name"] = path
         return info
 
-    def checksum(self, path):
-        _, fs, fs_path, dvc_fs, dvc_path = self._get_fs_pair(path)
 
-        try:
-            return fs.checksum(fs_path)
-        except FileNotFoundError:
-            return dvc_fs.checksum(dvc_path)
-
-
-class DvcFileSystem(FileSystem):
+class DVCFileSystem(FileSystem):
     protocol = "local"
     PARAM_CHECKSUM = "md5"
 
@@ -451,7 +380,7 @@ class DvcFileSystem(FileSystem):
     @wrap_prop(threading.Lock())
     @cached_property
     def fs(self):
-        return _DvcFileSystem(**self.fs_args)
+        return _DVCFileSystem(**self.fs_args)
 
     def isdvc(self, path, **kwargs):
         return self.fs.isdvc(path, **kwargs)
@@ -468,10 +397,6 @@ class DvcFileSystem(FileSystem):
     def repo_url(self):
         return self.fs.repo_url
 
-    @property
-    def config(self):
-        return self.fs.config
-
     def from_os_path(self, path):
         if os.path.isabs(path):
             path = os.path.relpath(path, self.repo.root_dir)

dvc/fs/git.py

@@ -15,6 +15,7 @@ class GitFileSystem(FileSystem):  # pylint:disable=abstract-method
     """Proxies the repo file access methods to Git objects"""
 
     protocol = "local"
+    PARAM_CHECKSUM = "md5"
 
     def __init__(
         self,

dvc/ignore.py

@@ -300,12 +300,25 @@ class DvcIgnoreFilter:
         return [fs_dict[name] for name in chain(dirs, nondirs)]
 
     def walk(self, fs: FileSystem, path: AnyFSPath, **kwargs):
+        detail = kwargs.get("detail", False)
         ignore_subrepos = kwargs.pop("ignore_subrepos", True)
         if fs.protocol == Schemes.LOCAL:
             for root, dirs, files in fs.walk(path, **kwargs):
-                dirs[:], files[:] = self(
-                    root, dirs, files, ignore_subrepos=ignore_subrepos
-                )
+                if detail:
+                    all_dnames = set(dirs.keys())
+                    all_fnames = set(files.keys())
+                    dnames, fnames = self(
+                        root,
+                        all_dnames,
+                        all_fnames,
+                        ignore_subrepos=ignore_subrepos,
+                    )
+                    list(map(dirs.pop, all_dnames - set(dnames)))
+                    list(map(files.pop, all_fnames - set(fnames)))
+                else:
+                    dirs[:], files[:] = self(
+                        root, dirs, files, ignore_subrepos=ignore_subrepos
+                    )
                 yield root, dirs, files
         else:
             yield from fs.walk(path, **kwargs)

dvc/info.py

@@ -8,12 +8,20 @@ import psutil
 
 from dvc import __version__
 from dvc.exceptions import NotDvcRepoError
-from dvc.fs import FS_MAP, generic, get_fs_cls, get_fs_config
+from dvc.fs import Schemes, generic, get_fs_cls, get_fs_config, registry
 from dvc.repo import Repo
 from dvc.scm import SCMError
 from dvc.utils import error_link
 from dvc.utils.pkg import PKG
 
+SUBPROJECTS = (
+    "dvc_data",
+    "dvc_objects",
+    "dvc_render",
+    "dvc_task",
+    "dvclive",
+    "scmrepo",
+)
 package = "" if PKG is None else f"({PKG})"
 
 
@@ -23,6 +31,7 @@ def get_dvc_info():
         "---------------------------------",
         f"Platform: Python {platform.python_version()} on "
         f"{platform.platform()}",
+        f"Subprojects:{_get_subprojects()}",
         f"Supports:{_get_supported_remotes()}",
     ]
 
@@ -67,7 +76,7 @@ def _get_caches(cache):
 
 def _get_remotes(config):
     schemes = (
-        get_fs_cls(get_fs_config(None, config, name=remote)).protocol
+        get_fs_cls(get_fs_config(config, name=remote)).protocol
         for remote in config["remote"]
     )
 
@@ -88,9 +97,29 @@ def _get_linktype_support_info(repo):
     return ", ".join(links)
 
 
+def _get_subprojects():
+    subprojects = []
+    for subproject in SUBPROJECTS:
+        try:
+            version = importlib_metadata.version(subproject)
+            subprojects.append(f"{subproject} = {version}")
+        except ImportError:
+            pass
+
+    return "\n\t" + "\n\t".join(subprojects)
+
+
 def _get_supported_remotes():
     supported_remotes = []
-    for scheme, fs_cls in FS_MAP.items():
+    for scheme in registry:
+        if scheme in [Schemes.LOCAL, Schemes.MEMORY, "dvc", "git"]:
+            continue
+
+        try:
+            fs_cls = registry[scheme]
+        except ImportError:
+            continue
+
         if not fs_cls.get_missing_deps():
             dependencies = []
             for requirement in fs_cls.REQUIRES:

dvc/lock.py

@@ -70,6 +70,8 @@ class LockNoop(LockBase):
         self._lock = True
 
     def unlock(self):
+        if not self.is_locked:
+            raise DvcException("Unlock called on an unlocked lock")
         self._lock = False
 
     @property
@@ -93,6 +95,7 @@ class Lock(LockBase):
         super().__init__(lockfile)
         self._friendly = friendly
         self._lock = None
+        self._lock_failed = False
 
     @property
     def files(self):
@@ -100,6 +103,7 @@ class Lock(LockBase):
 
     def _do_lock(self):
         try:
+            self._lock_failed = False
             with Tqdm(
                 bar_format="{desc}",
                 disable=not self._friendly,
@@ -111,6 +115,7 @@ class Lock(LockBase):
             ):
                 self._lock = zc.lockfile.LockFile(self._lockfile)
         except zc.lockfile.LockError:
+            self._lock_failed = True
             raise LockError(FAILED_TO_LOCK_MESSAGE)
 
     def lock(self):
@@ -120,6 +125,12 @@ class Lock(LockBase):
         lock_retry()
 
     def unlock(self):
+        if self._lock_failed:
+            assert self._lock is None
+            return
+
+        if not self.is_locked:
+            raise DvcException("Unlock called on an unlocked lock")
         self._lock.close()
         self._lock = None
 

dvc/logger.py

@@ -213,8 +213,6 @@ def set_loggers_level(level: int = logging.INFO) -> None:
 
 
 def setup(level: int = logging.INFO) -> None:
-    colorama.init()
-
     if level >= logging.DEBUG:
         # Unclosed session errors for asyncio/aiohttp are only available
         # on the tracing mode for extensive debug purposes. They are really
@@ -297,3 +295,8 @@ def setup(level: int = logging.INFO) -> None:
             "disable_existing_loggers": False,
         }
     )
+
+    # NOTE: colorama init needs to go after logging config to avoid potential
+    # conflicts with libs that modify logging (like tensorflow/abseil)
+    # https://github.com/iterative/dvc/issues/8387
+    colorama.init()

dvc/machine/backend/base.py

@@ -3,7 +3,8 @@ from contextlib import contextmanager
 from typing import TYPE_CHECKING, Iterator, Optional
 
 if TYPE_CHECKING:
-    from dvc.fs import SSHFileSystem
+    from dvc_ssh import SSHFileSystem
+
     from dvc.types import StrPath
 
 
@@ -25,7 +26,7 @@ class BaseMachineBackend(ABC):
     ) -> Iterator[dict]:
         """Iterate over status of all instances of the specified machine."""
 
-    def close(self):
+    def close(self):  # noqa: B027
         pass
 
     @abstractmethod

dvc/machine/backend/terraform.py

@@ -3,8 +3,9 @@ from contextlib import contextmanager
 from functools import partial, partialmethod
 from typing import TYPE_CHECKING, Iterator, Optional
 
+from dvc_ssh import DEFAULT_PORT, SSHFileSystem
+
 from dvc.exceptions import DvcException
-from dvc.fs import DEFAULT_SSH_PORT, SSHFileSystem
 from dvc.utils.fs import makedirs
 
 from .base import BaseMachineBackend
@@ -68,7 +69,7 @@ class TerraformBackend(BaseMachineBackend):
             resource = tpi.default_resource(name)
         return {
             "host": resource["instance_ip"],
-            "port": DEFAULT_SSH_PORT,
+            "port": DEFAULT_PORT,
             "username": "ubuntu",
             "fs_factory": partial(_sshfs, dict(resource)),
         }

dvc/odbmgr.py

@@ -1,5 +1,5 @@
 from dvc.fs import GitFileSystem, Schemes
-from dvc_data.db import get_odb
+from dvc_data.hashfile.db import get_odb
 
 
 def _get_odb(repo, settings, fs=None):

dvc/output.py

@@ -1,10 +1,10 @@
 import logging
 import os
 from collections import defaultdict
-from typing import TYPE_CHECKING, Dict, Optional, Set, Type
+from typing import TYPE_CHECKING, Dict, List, Optional, Set, Tuple, Type
 from urllib.parse import urlparse
 
-from funcy import collecting, project
+from funcy import cached_property, collecting, project
 from voluptuous import And, Any, Coerce, Length, Lower, Required, SetTo
 
 from dvc import prompt
@@ -17,31 +17,31 @@ from dvc.exceptions import (
     MergeError,
     RemoteCacheRequiredError,
 )
-from dvc_data import Tree
-from dvc_data import check as ocheck
-from dvc_data import load as oload
-from dvc_data.build import build
-from dvc_data.checkout import checkout
+from dvc_data.hashfile import Tree
+from dvc_data.hashfile import check as ocheck
+from dvc_data.hashfile import load as oload
+from dvc_data.hashfile.build import build
+from dvc_data.hashfile.checkout import checkout
 from dvc_data.hashfile.hash_info import HashInfo
 from dvc_data.hashfile.istextfile import istextfile
 from dvc_data.hashfile.meta import Meta
-from dvc_data.transfer import transfer as otransfer
+from dvc_data.hashfile.transfer import transfer as otransfer
+from dvc_data.index import DataIndexEntry
 from dvc_objects.errors import ObjectFormatError
 
-from .fs import (
-    HDFSFileSystem,
-    LocalFileSystem,
-    RemoteMissingDepsError,
-    S3FileSystem,
-    Schemes,
-    get_cloud_fs,
-)
+from .annotations import ANNOTATION_FIELDS, ANNOTATION_SCHEMA, Annotation
+from .fs import LocalFileSystem, RemoteMissingDepsError, Schemes, get_cloud_fs
+from .fs.callbacks import DEFAULT_CALLBACK
 from .utils import relpath
 from .utils.fs import path_isin
 
 if TYPE_CHECKING:
+    from dvc_data.hashfile.obj import HashFile
+    from dvc_data.index import DataIndexKey
     from dvc_objects.db import ObjectDB
 
+    from .fs.callbacks import Callback
+
 logger = logging.getLogger(__name__)
 
 
@@ -65,10 +65,12 @@ CASE_SENSITIVE_CHECKSUM_SCHEMA = Any(
 #
 # so when a few types of outputs share the same name, we only need
 # specify it once.
+HDFS_PARAM_CHECKSUM = "checksum"
+S3_PARAM_CHECKSUM = "etag"
 CHECKSUMS_SCHEMA = {
     LocalFileSystem.PARAM_CHECKSUM: CHECKSUM_SCHEMA,
-    HDFSFileSystem.PARAM_CHECKSUM: CHECKSUM_SCHEMA,
-    S3FileSystem.PARAM_CHECKSUM: CASE_SENSITIVE_CHECKSUM_SCHEMA,
+    HDFS_PARAM_CHECKSUM: CHECKSUM_SCHEMA,
+    S3_PARAM_CHECKSUM: CASE_SENSITIVE_CHECKSUM_SCHEMA,
 }
 
 
@@ -85,9 +87,10 @@ def loadd_from(stage, d_list):
         plot = d.pop(Output.PARAM_PLOT, False)
         persist = d.pop(Output.PARAM_PERSIST, False)
         checkpoint = d.pop(Output.PARAM_CHECKPOINT, False)
-        desc = d.pop(Output.PARAM_DESC, False)
-        live = d.pop(Output.PARAM_LIVE, False)
         remote = d.pop(Output.PARAM_REMOTE, None)
+        annot = {field: d.pop(field, None) for field in ANNOTATION_FIELDS}
+        files = d.pop(Output.PARAM_FILES, None)
+        push = d.pop(Output.PARAM_PUSH, True)
         ret.append(
             _get(
                 stage,
@@ -98,9 +101,10 @@ def loadd_from(stage, d_list):
                 plot=plot,
                 persist=persist,
                 checkpoint=checkpoint,
-                desc=desc,
-                live=live,
                 remote=remote,
+                **annot,
+                files=files,
+                push=push,
             )
         )
     return ret
@@ -114,8 +118,8 @@ def loads_from(
     plot=False,
     persist=False,
     checkpoint=False,
-    live=False,
     remote=None,
+    push=True,
 ):
     return [
         _get(
@@ -127,8 +131,8 @@ def loads_from(
             plot=plot,
             persist=persist,
             checkpoint=checkpoint,
-            live=live,
             remote=remote,
+            push=push,
         )
         for s in s_list
     ]
@@ -162,29 +166,21 @@ def load_from_pipeline(stage, data, typ="outs"):
         stage.PARAM_OUTS,
         stage.PARAM_METRICS,
         stage.PARAM_PLOTS,
-        stage.PARAM_LIVE,
     ):
         raise ValueError(f"'{typ}' key is not allowed for pipeline files.")
 
     metric = typ == stage.PARAM_METRICS
     plot = typ == stage.PARAM_PLOTS
-    live = typ == stage.PARAM_LIVE
-    if live:
-        # `live` is single object
-        data = [data]
 
     d = _merge_data(data)
 
     for path, flags in d.items():
-        plt_d, live_d = {}, {}
+        plt_d = {}
         if plot:
             from dvc.schema import PLOT_PROPS
 
             plt_d, flags = _split_dict(flags, keys=PLOT_PROPS.keys())
-        if live:
-            from dvc.schema import LIVE_PROPS
 
-            live_d, flags = _split_dict(flags, keys=LIVE_PROPS.keys())
         extra = project(
             flags,
             [
@@ -192,6 +188,8 @@ def load_from_pipeline(stage, data, typ="outs"):
                 Output.PARAM_PERSIST,
                 Output.PARAM_CHECKPOINT,
                 Output.PARAM_REMOTE,
+                Output.PARAM_PUSH,
+                *ANNOTATION_FIELDS,
             ],
         )
 
@@ -201,7 +199,6 @@ def load_from_pipeline(stage, data, typ="outs"):
             info={},
             plot=plt_d or plot,
             metric=metric,
-            live=live_d or live,
             **extra,
         )
 
@@ -245,6 +242,7 @@ class Output:
     PARAM_PATH = "path"
     PARAM_CACHE = "cache"
     PARAM_CHECKPOINT = "checkpoint"
+    PARAM_FILES = "files"
     PARAM_METRIC = "metric"
     PARAM_METRIC_TYPE = "type"
     PARAM_METRIC_XPATH = "xpath"
@@ -257,11 +255,8 @@ class Output:
     PARAM_PLOT_TITLE = "title"
     PARAM_PLOT_HEADER = "header"
     PARAM_PERSIST = "persist"
-    PARAM_DESC = "desc"
-    PARAM_LIVE = "live"
-    PARAM_LIVE_SUMMARY = "summary"
-    PARAM_LIVE_HTML = "html"
     PARAM_REMOTE = "remote"
+    PARAM_PUSH = "push"
 
     METRIC_SCHEMA = Any(
         None,
@@ -287,13 +282,31 @@ class Output:
         plot=False,
         persist=False,
         checkpoint=False,
-        live=False,
         desc=None,
+        type=None,  # pylint: disable=redefined-builtin
+        labels=None,
+        meta=None,
         remote=None,
         repo=None,
+        fs_config=None,
+        files: List[Dict[str, Any]] = None,
+        push: bool = True,
     ):
+        self.annot = Annotation(
+            desc=desc, type=type, labels=labels or [], meta=meta or {}
+        )
         self.repo = stage.repo if not repo and stage else repo
-        fs_cls, fs_config, fs_path = get_cloud_fs(self.repo, url=path)
+        meta = Meta.from_dict(info or {})
+        # NOTE: when version_aware is not passed into get_cloud_fs, it will be
+        # set based on whether or not path is versioned
+        fs_kwargs = {"version_aware": True} if meta.version_id else {}
+
+        if fs_config is not None:
+            fs_kwargs.update(**fs_config)
+
+        fs_cls, fs_config, fs_path = get_cloud_fs(
+            self.repo, url=path, **fs_kwargs
+        )
         self.fs = fs_cls(**fs_config)
 
         if (
@@ -325,21 +338,42 @@ class Output:
         # By resolved path, which contains actual location,
         # should be absolute and don't contain remote:// refs.
         self.stage = stage
-        self.meta = Meta.from_dict(info)
-        self.hash_info = HashInfo.from_dict(info)
+        self.meta = meta
+        self.files = files
         self.use_cache = False if self.IS_DEPENDENCY else cache
         self.metric = False if self.IS_DEPENDENCY else metric
         self.plot = False if self.IS_DEPENDENCY else plot
         self.persist = persist
         self.checkpoint = checkpoint
-        self.live = live
-        self.desc = desc
+        self.can_push = push
 
         self.fs_path = self._parse_path(self.fs, fs_path)
-        self.obj = None
+        self.obj: Optional["HashFile"] = None
 
         self.remote = remote
 
+        if self.fs.version_aware:
+            _, version_id = self.fs.path.coalesce_version(
+                self.def_path, self.meta.version_id
+            )
+            self.meta.version_id = version_id
+
+        if self.is_in_repo:
+            self.hash_name = "md5"
+        else:
+            self.hash_name = self.fs.PARAM_CHECKSUM
+
+        self.hash_info = HashInfo(
+            name=self.hash_name,
+            value=getattr(self.meta, self.hash_name, None),
+        )
+        if self.meta.nfiles or self.hash_info and self.hash_info.isdir:
+            self.meta.isdir = True
+            if not self.hash_info and self.hash_name != "md5":
+                md5 = getattr(self.meta, "md5", None)
+                if md5:
+                    self.hash_info = HashInfo("md5", md5)
+
     def _parse_path(self, fs, fs_path):
         parsed = urlparse(self.def_path)
         if (
@@ -380,6 +414,12 @@ class Output:
 
         return self.fs.path.relpath(self.fs_path, self.repo.root_dir)
 
+    def clear(self):
+        self.hash_info = HashInfo.from_dict({})
+        self.meta = Meta.from_dict({})
+        self.obj = None
+        self.files = None
+
     @property
     def protocol(self):
         return self.fs.protocol
@@ -419,21 +459,27 @@ class Output:
         )
 
     def get_hash(self):
+        _, hash_info = self._get_hash_meta()
+        return hash_info
+
+    def _get_hash_meta(self):
         if self.use_cache:
             odb = self.odb
-            name = self.odb.fs.PARAM_CHECKSUM
         else:
             odb = self.repo.odb.local
-            name = self.fs.PARAM_CHECKSUM
-        _, _, obj = build(
+        _, meta, obj = build(
             odb,
             self.fs_path,
             self.fs,
-            name,
+            self.hash_name,
             ignore=self.dvcignore,
             dry_run=not self.use_cache,
         )
-        return obj.hash_info
+        return meta, obj.hash_info
+
+    def get_meta(self) -> Meta:
+        meta, _ = self._get_hash_meta()
+        return meta
 
     @property
     def is_dir_checksum(self):
@@ -452,6 +498,47 @@ class Output:
 
         return self.fs.exists(self.fs_path)
 
+    @cached_property
+    def index_key(self) -> Tuple[str, "DataIndexKey"]:
+        if self.is_in_repo:
+            workspace = "repo"
+            key = self.repo.fs.path.relparts(self.fs_path, self.repo.root_dir)
+        else:
+            workspace = self.fs.protocol
+            no_drive = self.fs.path.flavour.splitdrive(self.fs_path)[1]
+            key = self.fs.path.parts(no_drive)[1:]
+        return workspace, key
+
+    def get_entry(self) -> "DataIndexEntry":
+        from dvc.config import NoRemoteError
+
+        try:
+            remote = self.repo.cloud.get_remote_odb(self.remote)
+        except NoRemoteError:
+            remote = None
+
+        if self.files and not self.obj:
+            self.obj = self.get_obj()
+
+        entry = DataIndexEntry(
+            meta=self.meta,
+            obj=self.obj,
+            hash_info=self.hash_info,
+            odb=self.odb,
+            cache=self.odb,
+            remote=remote,
+        )
+        if self.stage.is_import and not self.stage.is_repo_import:
+            dep = self.stage.deps[0]
+            entry.fs = dep.fs
+            entry.path = dep.fs_path
+            entry.meta = dep.meta
+            if not entry.obj:
+                if not dep.obj and dep.files:
+                    dep.obj = dep.get_obj()
+                entry.obj = dep.obj
+        return entry
+
     def changed_checksum(self):
         return self.hash_info != self.get_hash()
 
@@ -469,6 +556,11 @@ class Output:
         except (FileNotFoundError, ObjectFormatError):
             return True
 
+    def changed_meta(self) -> bool:
+        if self.fs.version_aware and self.meta.version_id:
+            return self.meta.version_id != self.get_meta().version_id
+        return False
+
     def workspace_status(self):
         if not self.exists:
             return {str(self): "deleted"}
@@ -546,40 +638,38 @@ class Output:
         if self.metric:
             self.verify_metric()
 
-        if not self.use_cache:
-            _, self.meta, obj = build(
+        if self.use_cache:
+            _, self.meta, self.obj = build(
+                self.odb,
+                self.fs_path,
+                self.fs,
+                self.hash_name,
+                ignore=self.dvcignore,
+            )
+        else:
+            _, self.meta, self.obj = build(
                 self.repo.odb.local,
                 self.fs_path,
                 self.fs,
-                self.fs.PARAM_CHECKSUM,
+                self.hash_name,
                 ignore=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
                 )
-            return
 
-        assert not self.IS_DEPENDENCY
-
-        _, self.meta, self.obj = build(
-            self.odb,
-            self.fs_path,
-            self.fs,
-            self.odb.fs.PARAM_CHECKSUM,
-            ignore=self.dvcignore,
-        )
         self.hash_info = self.obj.hash_info
+        self.files = None
 
     def set_exec(self):
         if self.isfile() and self.meta.isexec:
             self.odb.set_exec(self.fs_path)
 
     def _checkout(self, *args, **kwargs):
-        from dvc_data.checkout import CheckoutError as _CheckoutError
-        from dvc_data.checkout import LinkError, PromptError
+        from dvc_data.hashfile.checkout import CheckoutError as _CheckoutError
+        from dvc_data.hashfile.checkout import LinkError, PromptError
 
         kwargs.setdefault("ignore", self.dvcignore)
         try:
@@ -610,7 +700,7 @@ class Output:
                     self.odb,
                     filter_info or self.fs_path,
                     self.fs,
-                    self.odb.fs.PARAM_CHECKSUM,
+                    self.hash_name,
                     ignore=self.dvcignore,
                 )
                 otransfer(
@@ -639,7 +729,7 @@ class Output:
             self.odb,
             self.fs_path,
             self.fs,
-            self.odb.fs.PARAM_CHECKSUM,
+            self.hash_name,
             ignore=self.dvcignore,
         )
         save_obj = save_obj.filter(prefix)
@@ -653,8 +743,10 @@ class Output:
         )
         return checkout_obj
 
-    def dumpd(self):
-        ret = {**self.hash_info.to_dict(), **self.meta.to_dict()}
+    def dumpd(self, **kwargs):
+        meta = self.meta.to_dict()
+        meta.pop("isdir", None)
+        ret = {**self.hash_info.to_dict(), **meta}
 
         if self.is_in_repo:
             path = self.fs.path.as_posix(
@@ -665,39 +757,46 @@ class Output:
 
         ret[self.PARAM_PATH] = path
 
-        if self.IS_DEPENDENCY:
-            return ret
+        if not self.IS_DEPENDENCY:
+            ret.update(self.annot.to_dict())
+            if not self.use_cache:
+                ret[self.PARAM_CACHE] = self.use_cache
 
-        if self.desc:
-            ret[self.PARAM_DESC] = self.desc
+            if isinstance(self.metric, dict):
+                if (
+                    self.PARAM_METRIC_XPATH in self.metric
+                    and not self.metric[self.PARAM_METRIC_XPATH]
+                ):
+                    del self.metric[self.PARAM_METRIC_XPATH]
 
-        if not self.use_cache:
-            ret[self.PARAM_CACHE] = self.use_cache
+            if self.metric:
+                ret[self.PARAM_METRIC] = self.metric
 
-        if isinstance(self.metric, dict):
-            if (
-                self.PARAM_METRIC_XPATH in self.metric
-                and not self.metric[self.PARAM_METRIC_XPATH]
-            ):
-                del self.metric[self.PARAM_METRIC_XPATH]
+            if self.plot:
+                ret[self.PARAM_PLOT] = self.plot
 
-        if self.metric:
-            ret[self.PARAM_METRIC] = self.metric
+            if self.persist:
+                ret[self.PARAM_PERSIST] = self.persist
 
-        if self.plot:
-            ret[self.PARAM_PLOT] = self.plot
+            if self.checkpoint:
+                ret[self.PARAM_CHECKPOINT] = self.checkpoint
 
-        if self.persist:
-            ret[self.PARAM_PERSIST] = self.persist
+            if self.remote:
+                ret[self.PARAM_REMOTE] = self.remote
 
-        if self.checkpoint:
-            ret[self.PARAM_CHECKPOINT] = self.checkpoint
+            if not self.can_push:
+                ret[self.PARAM_PUSH] = self.can_push
 
-        if self.live:
-            ret[self.PARAM_LIVE] = self.live
-
-        if self.remote:
-            ret[self.PARAM_REMOTE] = self.remote
+        if (
+            (not self.IS_DEPENDENCY or self.stage.is_import)
+            and self.hash_info.isdir
+            and (kwargs.get("with_files") or self.files is not None)
+        ):
+            if self.obj:
+                obj = self.obj
+            else:
+                obj = self.get_obj()
+            ret[self.PARAM_FILES] = obj.as_list(with_meta=True)
 
         return ret
 
@@ -722,26 +821,26 @@ class Output:
                 f"binary file '{self.fs_path}' cannot be used as metrics."
             )
 
-    def download(self, to, jobs=None):
-        from dvc.fs.callbacks import Callback
-
-        with Callback.as_tqdm_callback(
-            desc=f"Downloading {self.fs.path.name(self.fs_path)}",
-            unit="files",
-        ) as cb:
-            self.fs.get(self.fs_path, to.fs_path, batch_size=jobs, callback=cb)
-
-    def get_obj(self, filter_info=None, **kwargs):
+    def get_obj(
+        self, filter_info: Optional[str] = None, **kwargs
+    ) -> Optional["HashFile"]:
+        obj: Optional["HashFile"] = None
         if self.obj:
             obj = self.obj
         elif self.hash_info:
-            try:
-                obj = oload(self.odb, self.hash_info)
-            except FileNotFoundError:
-                return None
+            if self.files:
+                tree = Tree.from_list(self.files, hash_name=self.hash_name)
+                tree.digest()
+                obj = tree
+            else:
+                try:
+                    obj = oload(self.odb, self.hash_info)
+                except (FileNotFoundError, ObjectFormatError):
+                    return None
         else:
             return None
 
+        assert obj
         fs_path = self.fs.path
         if filter_info and filter_info != self.fs_path:
             prefix = fs_path.relparts(filter_info, self.fs_path)
@@ -752,7 +851,7 @@ class Output:
     def checkout(
         self,
         force=False,
-        progress_callback=None,
+        progress_callback: "Callback" = DEFAULT_CALLBACK,
         relink=False,
         filter_info=None,
         allow_missing=False,
@@ -760,9 +859,9 @@ class Output:
         **kwargs,
     ):
         if not self.use_cache:
-            if progress_callback:
-                progress_callback(
-                    self.fs_path, self.get_files_number(filter_info)
+            if progress_callback != DEFAULT_CALLBACK:
+                progress_callback.relative_update(
+                    self.get_files_number(filter_info)
                 )
             return None
 
@@ -858,6 +957,7 @@ class Output:
         )
 
         self.hash_info = obj.hash_info
+        self.files = None
         return obj
 
     def get_files_number(self, filter_info=None):
@@ -928,6 +1028,7 @@ class Output:
 
         obj = self.get_obj()
         if filter_info and filter_info != self.fs_path:
+            assert obj
             prefix = self.fs.path.parts(
                 self.fs.path.relpath(filter_info, self.fs_path)
             )
@@ -942,9 +1043,15 @@ class Output:
         if not self.use_cache:
             return {}
 
+        push: bool = kwargs.pop("push", False)
         if self.stage.is_repo_import:
+            if push:
+                return {}
             return self.get_used_external(**kwargs)
 
+        if push and not self.can_push:
+            return {}
+
         if not self.hash_info:
             msg = (
                 "Output '{}'({}) is missing version info. "
@@ -1020,7 +1127,7 @@ class Output:
         other = out.dumpd()
 
         ignored = [
-            self.fs.PARAM_CHECKSUM,
+            self.hash_name,
             Meta.PARAM_SIZE,
             Meta.PARAM_NFILES,
         ]
@@ -1039,9 +1146,9 @@ class Output:
                 "unable to auto-merge outputs that are not directories"
             )
 
-    def merge(self, ancestor, other):
-        from dvc_data.objects.tree import MergeError as TreeMergeError
-        from dvc_data.objects.tree import du, merge
+    def merge(self, ancestor, other, allowed=None):
+        from dvc_data.hashfile.tree import MergeError as TreeMergeError
+        from dvc_data.hashfile.tree import du, merge
 
         assert other
 
@@ -1056,7 +1163,11 @@ class Output:
 
         try:
             merged = merge(
-                self.odb, ancestor_info, self.hash_info, other.hash_info
+                self.odb,
+                ancestor_info,
+                self.hash_info,
+                other.hash_info,
+                allowed=allowed,
             )
         except TreeMergeError as exc:
             raise MergeError(str(exc)) from exc
@@ -1064,6 +1175,7 @@ class Output:
         self.odb.add(merged.path, merged.fs, merged.oid)
 
         self.hash_info = merged.hash_info
+        self.files = None
         self.meta = Meta(
             size=du(self.odb, merged),
             nfiles=len(merged),
@@ -1079,28 +1191,69 @@ class Output:
 
     @property
     def is_metric(self) -> bool:
-        return bool(self.metric) or bool(self.live)
+        return bool(self.metric)
 
     @property
     def is_plot(self) -> bool:
-        return bool(self.plot) or bool(self.live)
+        return bool(self.plot)
 
+    def restore_fields(self, other: "Output"):
+        """Restore attributes that need to be preserved when serialized."""
+        self.annot = other.annot
+        self.remote = other.remote
+        self.can_push = other.can_push
+
+    def merge_version_meta(self, other: "Output"):
+        """Merge version meta for files which are unchanged from other."""
+        if not self.hash_info:
+            return
+        if self.hash_info.isdir:
+            return self._merge_dir_version_meta(other)
+        if self.hash_info != other.hash_info:
+            return
+        self.meta = other.meta
+
+    def _merge_dir_version_meta(self, other: "Output"):
+        from dvc_data.hashfile.tree import update_meta
+
+        if not self.obj or not other.hash_info.isdir:
+            return
+        other_obj = other.obj if other.obj is not None else other.get_obj()
+        assert isinstance(self.obj, Tree) and isinstance(other_obj, Tree)
+        updated = update_meta(self.obj, other_obj)
+        assert updated.hash_info == self.obj.hash_info
+        self.obj = updated
+        self.files = updated.as_list(with_meta=True)
+
+
+META_SCHEMA = {
+    Meta.PARAM_SIZE: int,
+    Meta.PARAM_NFILES: int,
+    Meta.PARAM_ISEXEC: bool,
+    Meta.PARAM_VERSION_ID: str,
+}
 
 ARTIFACT_SCHEMA = {
     **CHECKSUMS_SCHEMA,
+    **META_SCHEMA,
     Required(Output.PARAM_PATH): str,
     Output.PARAM_PLOT: bool,
     Output.PARAM_PERSIST: bool,
     Output.PARAM_CHECKPOINT: bool,
-    Meta.PARAM_SIZE: int,
-    Meta.PARAM_NFILES: int,
-    Meta.PARAM_ISEXEC: bool,
+}
+
+DIR_FILES_SCHEMA: Dict[str, Any] = {
+    **CHECKSUMS_SCHEMA,
+    **META_SCHEMA,
+    Required(Tree.PARAM_RELPATH): str,
 }
 
 SCHEMA = {
     **ARTIFACT_SCHEMA,
+    **ANNOTATION_SCHEMA,
     Output.PARAM_CACHE: bool,
     Output.PARAM_METRIC: Output.METRIC_SCHEMA,
-    Output.PARAM_DESC: str,
     Output.PARAM_REMOTE: str,
+    Output.PARAM_PUSH: bool,
+    Output.PARAM_FILES: [DIR_FILES_SCHEMA],
 }

dvc/parsing/__init__.py

@@ -36,7 +36,7 @@ from .interpolate import (
 )
 
 if TYPE_CHECKING:
-    from typing_extensions import NoReturn
+    from typing import NoReturn
 
     from dvc.repo import Repo
 

dvc/parsing/context.py

@@ -1,5 +1,4 @@
 import logging
-import os
 from abc import ABC, abstractmethod
 from collections import defaultdict
 from collections.abc import Mapping, MutableMapping, MutableSequence, Sequence
@@ -160,7 +159,7 @@ class Value(Node):
 PRIMITIVES = (int, float, str, bytes, bool)
 
 
-class Container(Node, ABC):
+class Container(Node, ABC):  # noqa: B024
     meta: Meta
     data: Union[list, dict]
     _key_transform = staticmethod(identity)
@@ -357,17 +356,14 @@ class Context(CtxDict):
     def load_from(
         cls, fs, path: str, select_keys: List[str] = None
     ) -> "Context":
-        from dvc.utils.serialize import LOADERS
+        from dvc.utils.serialize import load_path
 
         if not fs.exists(path):
             raise ParamsLoadError(f"'{path}' does not exist")
         if fs.isdir(path):
             raise ParamsLoadError(f"'{path}' is a directory")
 
-        _, ext = os.path.splitext(path)
-        loader = LOADERS[ext]
-
-        data = loader(path, fs=fs)
+        data = load_path(path, fs)
         if not isinstance(data, Mapping):
             typ = type(data).__name__
             raise ParamsLoadError(

dvc/parsing/interpolate.py

@@ -1,3 +1,4 @@
+import os
 import re
 import typing
 from collections.abc import Iterable, Mapping
@@ -9,10 +10,9 @@ from dvc.exceptions import DvcException
 from dvc.utils.flatten import flatten
 
 if typing.TYPE_CHECKING:
-    from typing import List, Match
+    from typing import List, Match, NoReturn
 
     from pyparsing import ParseException
-    from typing_extensions import NoReturn
 
     from .context import Context
 
@@ -71,6 +71,16 @@ def embrace(s: str):
     return BRACE_OPEN + s + BRACE_CLOSE
 
 
+def escape_str(value):
+    if os.name == "nt":
+        from subprocess import list2cmdline
+
+        return list2cmdline([value])
+    from shlex import quote
+
+    return quote(value)
+
+
 @singledispatch
 def to_str(obj) -> str:
     return str(obj)
@@ -98,12 +108,12 @@ def _(obj: dict):
                     result += f"--no-{k} "
 
         elif isinstance(v, str):
-            result += f"--{k} '{v}' "
+            result += f"--{k} {escape_str(v)} "
 
         elif isinstance(v, Iterable):
             for n, i in enumerate(v):
                 if isinstance(i, str):
-                    i = f"'{i}'"
+                    i = escape_str(i)
                 elif isinstance(i, Iterable):
                     raise ParseError(
                         f"Cannot interpolate nested iterable in '{k}'"

dvc/render/convert.py

@@ -2,20 +2,26 @@ import json
 from collections import defaultdict
 from typing import Dict, List, Union
 
-from dvc.render import REVISION_FIELD, REVISIONS_KEY, SRC_FIELD, TYPE_KEY
+from dvc.render import (
+    REVISION_FIELD,
+    REVISIONS_KEY,
+    SRC_FIELD,
+    TYPE_KEY,
+    VERSION_FIELD,
+)
 from dvc.render.converter.image import ImageConverter
 from dvc.render.converter.vega import VegaConverter
 
 
 def _get_converter(
-    renderer_class, props
+    renderer_class, renderer_id, props, data
 ) -> Union[VegaConverter, ImageConverter]:
     from dvc_render import ImageRenderer, VegaRenderer
 
     if renderer_class.TYPE == VegaRenderer.TYPE:
-        return VegaConverter(props)
+        return VegaConverter(renderer_id, data, props)
     if renderer_class.TYPE == ImageRenderer.TYPE:
-        return ImageConverter(props)
+        return ImageConverter(renderer_id, data, props)
 
     raise ValueError(f"Invalid renderer class {renderer_class}")
 
@@ -23,7 +29,7 @@ def _get_converter(
 def _group_by_rev(datapoints):
     grouped = defaultdict(list)
     for datapoint in datapoints:
-        rev = datapoint.pop(REVISION_FIELD)
+        rev = datapoint.get(VERSION_FIELD, {}).get("revision")
         grouped[rev].append(datapoint)
     return dict(grouped)
 

dvc/render/converter/__init__.py

@@ -1,9 +1,19 @@
-from typing import Dict, Optional
+from typing import Any, Dict, List, Optional, Tuple
 
 
 class Converter:
-    def __init__(self, plot_properties: Optional[Dict] = None):
-        self.plot_properties = plot_properties or {}
+    def __init__(
+        self,
+        plot_id: str,
+        data: Dict[str, Any] = None,
+        properties: Optional[Dict] = None,
+    ):
+        self.plot_id = plot_id
+        self.properties = properties or {}
+        self.data = data or {}
 
-    def convert(self, data, revision: str, filename: str, **kwargs):
+    def convert(self) -> Tuple[List[Tuple[str, str, Any]], Dict]:
+        raise NotImplementedError
+
+    def flat_datapoints(self, revision: str) -> Tuple[List[Dict], Dict]:
         raise NotImplementedError

dvc/render/converter/image.py

@@ -1,6 +1,6 @@
 import base64
 import os
-from typing import TYPE_CHECKING, Dict, List, Tuple
+from typing import TYPE_CHECKING, Any, Dict, List, Tuple
 
 from dvc.render import FILENAME_FIELD, REVISION_FIELD, SRC_FIELD
 
@@ -34,25 +34,33 @@ class ImageConverter(Converter):
         base64_str = base64.b64encode(image_data).decode()
         return f"data:image;base64,{base64_str}"
 
-    def convert(
-        self, data, revision: str, filename: str, **kwargs
-    ) -> Tuple[List[Dict], Dict]:
+    def convert(self) -> Tuple[List[Tuple[str, str, Any]], Dict]:
+        datas = []
+        for filename, image_data in self.data.items():
+            datas.append((filename, "", image_data))
+        return datas, self.properties
+
+    def flat_datapoints(self, revision: str) -> Tuple[List[Dict], Dict]:
         """
         Convert the DVC Plots content to DVC Render datapoints.
         Return both generated datapoints and updated properties.
         """
-        path = self.plot_properties.get("out")
-        if path:
-            if not os.path.isdir(path):
-                os.makedirs(path, exist_ok=True)
-            src = self._write_image(
-                os.path.abspath(path), revision, filename, data
-            )
-        else:
-            src = self._encode_image(data)
-        datapoint = {
-            REVISION_FIELD: revision,
-            FILENAME_FIELD: filename,
-            SRC_FIELD: src,
-        }
-        return [datapoint], self.plot_properties
+        path = self.properties.get("out")
+        datapoints = []
+        datas, properties = self.convert()
+        for filename, _, image_data in datas:
+            if path:
+                if not os.path.isdir(path):
+                    os.makedirs(path, exist_ok=True)
+                src = self._write_image(
+                    os.path.abspath(path), revision, filename, image_data
+                )
+            else:
+                src = self._encode_image(image_data)
+            datapoint = {
+                REVISION_FIELD: revision,
+                FILENAME_FIELD: filename,
+                SRC_FIELD: src,
+            }
+            datapoints.append(datapoint)
+        return datapoints, properties

dvc/render/converter/vega.py

@@ -1,90 +1,101 @@
 from copy import deepcopy
-from functools import partial
-from typing import Dict, Iterable, List, Optional, Set, Union
+from typing import Any, Dict, Iterable, List, Tuple, Union
 
-from funcy import first, project
+from funcy import first, last
 
 from dvc.exceptions import DvcException
-from dvc.render import (
-    FILENAME_FIELD,
-    INDEX_FIELD,
-    REVISION_FIELD,
-    VERSION_FIELD,
-)
+from dvc.render import FILENAME_FIELD, INDEX_FIELD, VERSION_FIELD
 
 from . import Converter
 
 
-class FieldsNotFoundError(DvcException):
-    def __init__(self, expected_fields, found_fields):
-        expected_str = ", ".join(expected_fields)
+class FieldNotFoundError(DvcException):
+    def __init__(self, expected_field, found_fields):
         found_str = ", ".join(found_fields)
         super().__init__(
-            f"Could not find all provided fields ('{expected_str}') "
+            f"Could not find provided field ('{expected_field}') "
             f"in data fields ('{found_str}')."
         )
 
 
-class PlotDataStructureError(DvcException):
-    def __init__(self):
-        super().__init__(
-            "Plot data extraction failed. Please see "
-            "https://man.dvc.org/plots for supported data formats."
-        )
+def _lists(blob: Union[Dict, List]) -> Iterable[List]:
+    if isinstance(blob, list):
+        yield blob
+    else:
+        for _, value in blob.items():
+            if isinstance(value, dict):
+                yield from _lists(value)
+            elif isinstance(value, list):
+                yield value
 
 
-def _filter_fields(
-    datapoints: List[Dict], fields: Set, **kwargs
-) -> List[Dict]:
-    if not fields:
-        return datapoints
-    assert isinstance(fields, set)
-
-    new_data = []
-    for data_point in datapoints:
-        keys = set(data_point.keys())
-        if not fields <= keys:
-            raise FieldsNotFoundError(fields, keys)
-
-        new_data.append(project(data_point, fields))
-
-    return new_data
+def _file_field(*args):
+    for axis_def in args:
+        if axis_def is not None:
+            for file, val in axis_def.items():
+                if isinstance(val, str):
+                    yield file, val
+                elif isinstance(val, list):
+                    for field in val:
+                        yield file, field
 
 
-def _lists(dictionary: Dict):
-    for _, value in dictionary.items():
-        if isinstance(value, dict):
-            yield from _lists(value)
-        elif isinstance(value, list):
-            yield value
+def _find(
+    filename: str,
+    field: str,
+    data_series: List[Tuple[str, str, Any]],
+):
+    for data_file, data_field, data in data_series:
+        if data_file == filename and data_field == field:
+            return data_file, data_field, data
+    return None
 
 
-def _find_first_list(
-    data: Union[Dict, List], fields: Set, **kwargs
-) -> List[Dict]:
-    fields = fields or set()
-
-    if not isinstance(data, dict):
-        return data
-
-    for lst in _lists(data):
-        if (
-            all(isinstance(dp, dict) for dp in lst)
-            # if fields is empty, it will match any set
-            and set(first(lst).keys()) & fields == fields
-        ):
-            return lst
-
-    raise PlotDataStructureError()
+def _verify_field(file2datapoints: Dict[str, List], filename: str, field: str):
+    if filename in file2datapoints:
+        datapoint = first(file2datapoints[filename])
+        if field not in datapoint:
+            raise FieldNotFoundError(field, datapoint.keys())
+    return
 
 
-def _append_index(datapoints: List[Dict], **kwargs) -> List[Dict]:
-    if INDEX_FIELD in first(datapoints).keys():
-        return datapoints
+def _get_x(properties: Dict, file2datapoints: Dict[str, List[Dict]]):
+    x = properties.get("x", None)
+    if x is not None and isinstance(x, dict):
+        filename, field = first(_file_field(x))
+        _verify_field(file2datapoints, filename, field)
+        return filename, field
+    return None
 
-    for index, data_point in enumerate(datapoints):
-        data_point[INDEX_FIELD] = index
-    return datapoints
+
+def _get_ys(properties, file2datapoints: Dict[str, List[Dict]]):
+    y = properties.get("y", None)
+    if y is not None:
+        for filename, field in _file_field(y):
+            _verify_field(file2datapoints, filename, field)
+            yield filename, field
+
+
+def _is_datapoints(lst: List[Dict]):
+    """
+    check if dict keys match, datapoints with different keys mgiht lead
+    to unexpected behavior
+    """
+
+    return all(isinstance(item, dict) for item in lst) and set(
+        first(lst).keys()
+    ) == {key for keys in lst for key in keys}
+
+
+def get_datapoints(file_content: Dict):
+    result: List[Dict[str, Any]] = []
+    for lst in _lists(file_content):
+        if _is_datapoints(lst):
+            for index, datapoint in enumerate(lst):
+                if len(result) <= index:
+                    result.append({})
+                result[index].update(datapoint)
+    return result
 
 
 class VegaConverter(Converter):
@@ -95,222 +106,215 @@ class VegaConverter(Converter):
     ('x', 'y') it will attempt to fill in the blanks.
     """
 
-    def __init__(self, plot_properties: Optional[Dict] = None):
-        super().__init__(plot_properties)
+    def __init__(
+        self, plot_id: str, data: Dict = None, properties: Dict = None
+    ):
+        super().__init__(plot_id, data, properties)
+        self.plot_id = plot_id
         self.inferred_properties: Dict = {}
 
-        self.steps = []
+        # TODO we should be handling that in `convert`,
+        #      to avoid stateful `self.inferred_properties`
+        self._infer_x_y()
 
-        self._infer_x()
-        self._infer_fields()
+    def _infer_y_from_data(self):
+        if self.plot_id in self.data:
+            for lst in _lists(self.data[self.plot_id]):
+                if all(isinstance(item, dict) for item in lst):
+                    datapoint = first(lst)
+                    field = last(datapoint.keys())
+                    self.inferred_properties["y"] = {self.plot_id: field}
+                    break
 
-        self.steps.append(
-            (
-                "find_data",
-                partial(
-                    _find_first_list,
-                    fields=self.inferred_properties.get("fields", set())
-                    - {INDEX_FIELD},
-                ),
-            )
-        )
+    def _infer_x_y(self):
+        def _infer_files(from_name, to_name):
+            from_value = self.properties.get(from_name, None)
+            to_value = self.properties.get(to_name, None)
 
-        if not self.plot_properties.get("x", None):
-            self.steps.append(("append_index", partial(_append_index)))
-
-        self.steps.append(
-            (
-                "filter_fields",
-                partial(
-                    _filter_fields,
-                    fields=self.inferred_properties.get("fields", set()),
-                ),
-            )
-        )
-
-        self.steps.append(
-            (
-                "generate_y",
-                partial(
-                    self._generate_y_values,
-                    y_values=self.plot_properties.get("y", None),
-                ),
-            )
-        )
-
-    def _infer_x(self):
-        if not self.plot_properties.get("x", None):
-            self.inferred_properties["x"] = INDEX_FIELD
-
-    def skip_step(self, name: str):
-        self.steps = [(_name, fn) for _name, fn in self.steps if _name != name]
-
-    def _infer_fields(self):
-        fields = self.plot_properties.get("fields", set())
-        if fields:
-            fields = {
-                *fields,
-                self.plot_properties.get("x", None),
-                self.plot_properties.get("y", None),
-                self.inferred_properties.get("x", None),
-            } - {None}
-            self.inferred_properties["fields"] = fields
-
-    def _infer_y(self, datapoints: List[Dict]):
-        if "y" not in self.plot_properties:
-            data_fields = list(first(datapoints))
-            skip = (
-                REVISION_FIELD,
-                self.plot_properties.get("x", None)
-                or self.inferred_properties.get("x"),
-                FILENAME_FIELD,
-                VERSION_FIELD,
-            )
-            inferred_y = first(
-                f for f in reversed(data_fields) if f not in skip
-            )
-            if "y" in self.inferred_properties:
-                previous_y = self.inferred_properties["y"]
-                if previous_y != inferred_y:
-                    raise DvcException(
-                        f"Inferred y ('{inferred_y}' value does not match"
-                        f"previously matched one ('f{previous_y}')."
-                    )
+            if isinstance(to_value, str):
+                self.inferred_properties[to_name] = {}
+                if isinstance(from_value, dict):
+                    for file in from_value.keys():
+                        self.inferred_properties[to_name][file] = to_value
+                else:
+                    self.inferred_properties[to_name][self.plot_id] = to_value
             else:
-                self.inferred_properties["y"] = inferred_y
+                self.inferred_properties[to_name] = to_value
+
+        _infer_files("y", "x")
+        _infer_files("x", "y")
+
+        if self.inferred_properties.get("y", None) is None:
+            self._infer_y_from_data()
+
+    def _find_datapoints(self):
+        result = {}
+        for file, content in self.data.items():
+            result[file] = get_datapoints(content)
+
+        return result
+
+    @staticmethod
+    def infer_y_label(properties):
+        y_label = properties.get("y_label", None)
+        if y_label is None:
+            y = properties.get("y", None)
+            if isinstance(y, dict):
+                fields = {field for _, field in _file_field(y)}
+                if len(fields) == 1:
+                    y_label = first(fields)
+                else:
+                    y_label = "y"
+            elif isinstance(y, list):
+                y_label = "y"
+            elif isinstance(y, str):
+                y_label = y
+
+        return y_label
+
+    @staticmethod
+    def infer_x_label(properties):
+        x_label = properties.get("x_label", None)
+        if x_label is None:
+            x = properties.get("x", None)
+            if isinstance(x, dict):
+                fields = {field for _, field in _file_field(x)}
+                if len(fields) == 1:
+                    x_label = first(fields)
+                else:
+                    x_label = "x"
+            else:
+                x_label = INDEX_FIELD
+        return x_label
+
+    def flat_datapoints(self, revision):
+
+        file2datapoints, properties = self.convert()
+
+        props_update = {}
+
+        x = _get_x(properties, file2datapoints)
+
+        # assign "step" if no x provided
+        if not x:
+            x_file, x_field = (
+                None,
+                INDEX_FIELD,
+            )
+        else:
+            x_file, x_field = x
+        props_update["x"] = x_field
+
+        ys = list(_get_ys(properties, file2datapoints))
+
+        all_datapoints = []
+        all_y_fields = {y_field for _, y_field in ys}
+
+        # override to unified y field name if there are different y fields
+        if len(all_y_fields) > 1:
+            props_update["y"] = "dvc_inferred_y_value"
+        else:
+            props_update["y"] = first(all_y_fields)
+
+        for y_file, y_field in ys:
+            datapoints = deepcopy(file2datapoints.get(y_file, []))
+
+            if props_update.get("y", None) == "dvc_inferred_y_value":
+                _update_from_field(
+                    datapoints,
+                    field="dvc_inferred_y_value",
+                    source_field=y_field,
+                )
+
+            if x_field == INDEX_FIELD and x_file is None:
+                _update_from_index(datapoints, INDEX_FIELD)
+            else:
+                x_datapoints = file2datapoints.get(x_file, [])
+                try:
+                    _update_from_field(
+                        datapoints,
+                        field=x_field,
+                        source_datapoints=x_datapoints,
+                    )
+                except IndexError:
+                    raise DvcException(
+                        f"Cannot join '{x_field}' from '{x_file}' and "
+                        "'{y_field}' from '{y_file}'. "
+                        "They have to have same length."
+                    )
+
+            _update_all(
+                datapoints,
+                update_dict={
+                    VERSION_FIELD: {
+                        "revision": revision,
+                        FILENAME_FIELD: y_file,
+                        "field": y_field,
+                    }
+                },
+            )
+
+            all_datapoints.extend(datapoints)
+
+        if not all_datapoints:
+            return [], {}
+
+        properties = {**properties, **props_update}
+
+        return all_datapoints, properties
 
     def convert(
         self,
-        data,
-        revision: str,
-        filename: str,
-        skip: List = None,
-        **kwargs,
     ):
         """
         Convert the data. Fill necessary fields ('x', 'y') and return both
-        generated datapoints and updated properties.
+        generated datapoints and updated properties. If `x` is not provided,
+        leave it as None, fronteds should handle it.
+
+        NOTE: Studio uses this method.
+              The only thing studio FE handles is filling `x` and `y`.
+              `x/y_label` should be filled here.
+
+              Datapoints are not stripped according to config, because users
+              might be utilizing other fields in their custom plots.
         """
-        if not skip:
-            skip = []
+        datapoints = self._find_datapoints()
+        properties = {**self.properties, **self.inferred_properties}
 
-        processed = deepcopy(data)
+        properties["y_label"] = self.infer_y_label(properties)
+        properties["x_label"] = self.infer_x_label(properties)
 
-        for step_name, step in self.steps:
+        return datapoints, properties
 
-            if step_name not in skip:
-                processed = step(  # type: ignore
-                    processed,
-                    revision=revision,
-                    filename=filename,
-                )
 
-        self._infer_y(processed)  # type: ignore
-        return processed, {**self.plot_properties, **self.inferred_properties}
+def _update_from_field(
+    target_datapoints: List[Dict],
+    field: str,
+    source_datapoints: List[Dict] = None,
+    source_field: str = None,
+):
+    if source_datapoints is None:
+        source_datapoints = target_datapoints
+    if source_field is None:
+        source_field = field
 
-    def _generate_y_values(  # noqa: C901
-        self,
-        datapoints: List[Dict],
-        y_values: Optional[Union[str, List, Dict]],
-        revision: str,
-        filename: str,
-        **kwargs,
-    ) -> List[Dict]:
+    if len(source_datapoints) != len(target_datapoints):
+        raise IndexError(
+            "Source and target datapoints must have the same length"
+        )
 
-        result = []
-        properties_update = {}
+    for index, datapoint in enumerate(target_datapoints):
+        source_datapoint = source_datapoints[index]
+        if source_field in source_datapoint:
+            datapoint[field] = source_datapoint[source_field]
 
-        def _add_version_info(datapoint, version_info):
-            tmp = datapoint.copy()
-            tmp[VERSION_FIELD] = version_info
-            tmp[REVISION_FIELD] = version_info["revision"]
-            return tmp
 
-        def _version_info(revision, filename=None, field=None):
-            res = {"revision": revision}
-            if filename is not None:
-                res["filename"] = filename
-            if field is not None:
-                res["field"] = field
-            return res
+def _update_from_index(datapoints: List[Dict], new_field: str):
+    for index, datapoint in enumerate(datapoints):
+        datapoint[new_field] = index
 
-        def _generate_y(datapoint, field):
-            tmp = datapoint.copy()
-            tmp["dvc_inferred_y_value"] = datapoint[field]
-            tmp = _add_version_info(
-                tmp, _version_info(revision, filename, field)
-            )
-            if (
-                "y_label" not in properties_update
-                and "y_label" not in self.plot_properties
-            ):
-                properties_update["y_label"] = "y"
 
-            properties_update["y"] = "dvc_inferred_y_value"
+def _update_all(datapoints: List[Dict], update_dict: Dict):
 
-            return tmp
-
-        if not y_values:
-            for dp in datapoints:
-                result.append(
-                    _add_version_info(dp, _version_info(revision, filename))
-                )
-
-        if isinstance(y_values, str):
-            for datapoint in datapoints:
-                result.append(
-                    _add_version_info(
-                        datapoint, _version_info(revision, filename, y_values)
-                    )
-                )
-
-        if isinstance(y_values, list):
-            for datapoint in datapoints:
-                for y_val in y_values:
-                    if y_val in datapoint:
-                        result.append(_generate_y(datapoint, y_val))
-
-        if isinstance(y_values, dict):
-
-            def _to_set(values: Iterable):
-                result = set()
-                for val in values:
-                    if isinstance(val, list):
-                        for elem in val:
-                            result.add(elem)
-                    else:
-                        result.add(val)
-
-                return result
-
-            all_fields = _to_set(y_values.values())
-            if (
-                all([isinstance(field, str) for field in all_fields])
-                and len(all_fields) == 1
-            ):
-                # if we use the same field from all files,
-                # we dont have to generate it
-                for datapoint in datapoints:
-                    result.append(
-                        _add_version_info(
-                            datapoint, _version_info(revision, filename)
-                        )
-                    )
-                properties_update.update({"y": all_fields.pop()})
-            else:
-                for def_filename, val in y_values.items():
-                    if isinstance(val, str):
-                        fields = [val]
-                    if isinstance(val, list):
-                        fields = val
-                    for datapoint in datapoints:
-                        for field in fields:
-                            if field in datapoint and def_filename in filename:
-                                result.append(_generate_y(datapoint, field))
-
-        self.inferred_properties = {
-            **self.inferred_properties,
-            **properties_update,
-        }
-
-        return result
+    for datapoint in datapoints:
+        datapoint.update(update_dict)

dvc/render/match.py

@@ -42,7 +42,7 @@ class PlotsData:
         return dict(groups)
 
     def get_definition_data(self, target_files, rev):
-        result = []
+        result = {}
         for file in target_files:
             file_content = (
                 self.data.get(rev, {})
@@ -52,7 +52,7 @@ class PlotsData:
                 .get("data", {})
             )
             if file_content:
-                result.append((file, file_content))
+                result[file] = file_content
 
         return result
 
@@ -91,15 +91,14 @@ def match_defs_renderers(
                 renderer_cls = VegaRenderer
                 renderer_id = plot_id
 
-            converter = _get_converter(renderer_cls, props)
+            converter = _get_converter(
+                renderer_cls, inner_id, props, definitions_data
+            )
 
-            for filename, plot_data in definitions_data:
-                dps, final_props = converter.convert(
-                    revision=rev,
-                    filename=filename,
-                    data=plot_data,
-                )
-                plot_datapoints.extend(dps)
+            dps, rev_props = converter.flat_datapoints(rev)
+            if not final_props and rev_props:
+                final_props = rev_props
+            plot_datapoints.extend(dps)
 
         if "title" not in final_props:
             final_props["title"] = renderer_id

dvc/repo/__init__.py

@@ -67,6 +67,7 @@ class Repo:
     from dvc.repo.imp_url import imp_url  # type: ignore[misc]
     from dvc.repo.install import install  # type: ignore[misc]
     from dvc.repo.ls import ls as _ls  # type: ignore[misc]
+    from dvc.repo.ls_url import ls_url as _ls_url  # type: ignore[misc]
     from dvc.repo.move import move  # type: ignore[misc]
     from dvc.repo.pull import pull  # type: ignore[misc]
     from dvc.repo.push import push  # type: ignore[misc]
@@ -79,6 +80,7 @@ class Repo:
     from .data import status as data_status  # type: ignore[misc]
 
     ls = staticmethod(_ls)
+    ls_url = staticmethod(_ls_url)
     get = staticmethod(_get)
     get_url = staticmethod(_get_url)
 
@@ -186,7 +188,7 @@ class Repo:
         self.config = Config(self.dvc_dir, fs=self.fs, config=config)
         self._uninitialized = uninitialized
 
-        # used by DvcFileSystem to determine if it should traverse subrepos
+        # used by DVCFileSystem to determine if it should traverse subrepos
         self.subrepos = subrepos
 
         self.cloud = DataCloud(self)
@@ -267,7 +269,7 @@ class Repo:
         except SCMError:
             if self._uninitialized:
                 # might not be a git/dvc repo at all
-                # used in `params/metrics/plots/live` targets
+                # used in `params/metrics/plots` targets
                 return SCM(self.root_dir, no_scm=True)
             raise
 
@@ -399,6 +401,7 @@ class Repo:
         used_run_cache=None,
         revs=None,
         num=1,
+        push: bool = False,
     ):
         """Get the stages related to the given target and collect
         the `info` of its outputs.
@@ -433,6 +436,7 @@ class Repo:
                 jobs=jobs,
                 recursive=recursive,
                 with_deps=with_deps,
+                push=push,
             ).items():
                 used[odb].update(objs)
 
@@ -487,9 +491,9 @@ class Repo:
 
     @cached_property
     def dvcfs(self):
-        from dvc.fs.dvc import DvcFileSystem
+        from dvc.fs.dvc import DVCFileSystem
 
-        return DvcFileSystem(
+        return DVCFileSystem(
             repo=self, subrepos=self.subrepos, **self._fs_conf
         )
 
@@ -501,13 +505,13 @@ class Repo:
     def open_by_relpath(self, path, remote=None, mode="r", encoding=None):
         """Opens a specified resource as a file descriptor"""
         from dvc.fs.data import DataFileSystem
-        from dvc.fs.dvc import DvcFileSystem
+        from dvc.fs.dvc import DVCFileSystem
 
         if os.path.isabs(path):
             fs = DataFileSystem(index=self.index.data["local"])
             fs_path = path
         else:
-            fs = DvcFileSystem(repo=self, subrepos=True)
+            fs = DVCFileSystem(repo=self, subrepos=True)
             fs_path = fs.from_os_path(path)
 
         try:

dvc/repo/add.py

@@ -187,7 +187,7 @@ def add(  # noqa: C901
                 stage.transfer(source, to_remote=to_remote, odb=odb, **kwargs)
             else:
                 try:
-                    stage.save()
+                    stage.save(merge_versioned=True)
                     if not no_commit:
                         stage.commit()
                 except CacheLinkError:
@@ -264,6 +264,7 @@ def create_stages(
             outs=[out],
             external=external,
         )
-        if kwargs.get("desc"):
-            stage.outs[0].desc = kwargs["desc"]
+
+        out_obj = stage.outs[0]
+        out_obj.annot.update(**kwargs)
         yield stage

dvc/repo/brancher.py

@@ -77,6 +77,7 @@ def brancher(  # noqa: E302
         from dvc.fs import GitFileSystem
 
         for sha, names in found_revs.items():
+            self.__dict__.pop("index", None)
             self.fs = GitFileSystem(scm=scm, rev=sha)
             self.root_dir = self.fs.path.join("/", *repo_root_parts)
 

dvc/repo/checkout.py

@@ -7,7 +7,6 @@ from dvc.exceptions import (
     CheckoutErrorSuggestGit,
     NoOutputOrStageError,
 )
-from dvc.progress import Tqdm
 from dvc.utils import relpath
 
 from . import locked
@@ -80,6 +79,7 @@ def checkout(
     allow_missing=False,
     **kwargs,
 ):
+    from dvc.fs.callbacks import Callback
 
     stats = {"added": [], "deleted": [], "modified": [], "failed": []}
     if not targets:
@@ -91,13 +91,16 @@ def checkout(
 
     pairs = _collect_pairs(self, targets, with_deps, recursive)
     total = get_all_files_numbers(pairs)
-    with Tqdm(
-        total=total, unit="file", desc="Checkout", disable=total == 0
-    ) as pbar:
+    with Callback.as_tqdm_callback(
+        unit="file",
+        desc="Checkout",
+        disable=total == 0,
+    ) as cb:
+        cb.set_size(total)
         for stage, filter_info in pairs:
             result = stage.checkout(
                 force=force,
-                progress_callback=pbar.update_msg,
+                progress_callback=cb,
                 relink=relink,
                 filter_info=filter_info,
                 allow_missing=allow_missing,

dvc/repo/collect.py

@@ -30,9 +30,9 @@ def _collect_paths(
     recursive: bool = False,
     rev: str = None,
 ) -> StrPaths:
-    from dvc.fs.dvc import DvcFileSystem
+    from dvc.fs.dvc import DVCFileSystem
 
-    fs = DvcFileSystem(repo=repo)
+    fs = DVCFileSystem(repo=repo)
     fs_paths = [fs.from_os_path(target) for target in targets]
 
     target_paths: StrPaths = []

dvc/repo/commit.py

@@ -56,10 +56,13 @@ def commit(
     ]
     for stage_info in stages_info:
         stage = stage_info.stage
-        changes = stage.changed_entries()
-        if any(changes):
-            prompt_to_commit(stage, changes, force=force)
+        if force:
             stage.save(allow_missing=allow_missing)
+        else:
+            changes = stage.changed_entries()
+            if any(changes):
+                prompt_to_commit(stage, changes, force=force)
+                stage.save(allow_missing=allow_missing)
         stage.commit(
             filter_info=stage_info.filter_info, allow_missing=allow_missing
         )

dvc/repo/data.py

@@ -1,28 +1,43 @@
 import os
+import posixpath
 from collections import defaultdict
-from typing import TYPE_CHECKING, Any, Dict, List, Optional, TypedDict, cast
+from itertools import chain
+from typing import (
+    TYPE_CHECKING,
+    Any,
+    Dict,
+    Iterable,
+    Iterator,
+    List,
+    Optional,
+    Tuple,
+    TypedDict,
+    Union,
+    cast,
+)
 
+from dvc.fs.git import GitFileSystem
 from dvc.ui import ui
 
 if TYPE_CHECKING:
-    from scmrepo.base import Base
-
     from dvc.output import Output
     from dvc.repo import Repo
+    from dvc.scm import Git, NoSCM
     from dvc_data.hashfile.db import HashFileDB
+    from dvc_data.hashfile.hash_info import HashInfo
     from dvc_data.hashfile.obj import HashFile
 
 
-def _in_cache(obj: Optional["HashFile"], cache: "HashFileDB") -> bool:
+def posixpath_to_os_path(path: str) -> str:
+    return path.replace(posixpath.sep, os.path.sep)
+
+
+def _in_cache(obj: "HashInfo", cache: "HashFileDB") -> bool:
     from dvc_objects.errors import ObjectFormatError
 
-    if not obj:
-        return False
-    if not obj.hash_info.value:
-        return False
-
+    assert obj.value
     try:
-        cache.check(obj.hash_info.value)
+        cache.check(obj.value)
         return True
     except (FileNotFoundError, ObjectFormatError):
         return False
@@ -30,31 +45,30 @@ def _in_cache(obj: Optional["HashFile"], cache: "HashFileDB") -> bool:
 
 def _shallow_diff(
     root: str,
-    old_obj: Optional["HashFile"],
-    new_obj: Optional["HashFile"],
+    old: Optional["HashInfo"],
+    new: Optional["HashInfo"],
     cache: "HashFileDB",
 ) -> Dict[str, List[str]]:
-    # TODO: add support for shallow diff in dvc-data
-    # TODO: we may want to recursively do in_cache check
     d = {}
 
-    from dvc_data.objects.tree import Tree
+    old_root = new_root = root
+    if old and old.isdir:
+        old_root = os.path.sep.join([root, ""])
+    if new and new.isdir:
+        new_root = os.path.sep.join([root, ""])
 
-    if isinstance(new_obj, Tree):
-        root = os.path.sep.join([root, ""])
+    if old and not _in_cache(old, cache):
+        d["not_in_cache"] = [old_root]
 
-    if not _in_cache(old_obj, cache):
-        d["not_in_cache"] = [root]
-
-    if old_obj is None and new_obj is None:
+    if not old and not new:
         return d
-    if old_obj is None:
-        return {"added": [root], **d}
-    if new_obj is None:
-        return {"deleted": [root], **d}
-    if old_obj.hash_info != new_obj.hash_info:
-        return {"modified": [root], **d}
-    return {"unchanged": [root], **d}
+    if not new:
+        return {"deleted": [old_root], **d}
+    if not old:
+        return {"added": [new_root], **d}
+    if old != new:
+        return {"modified": [new_root], **d}
+    return {"unchanged": [new_root], **d}
 
 
 def _granular_diff(
@@ -62,55 +76,60 @@ def _granular_diff(
     old_obj: Optional["HashFile"],
     new_obj: Optional["HashFile"],
     cache: "HashFileDB",
-    with_dirs: bool = False,
 ) -> Dict[str, List[str]]:
-    from dvc_data.diff import ROOT
-    from dvc_data.diff import diff as odiff
-    from dvc_data.objects.tree import Tree
+    from dvc_data.hashfile.diff import ROOT
+    from dvc_data.hashfile.diff import diff as odiff
 
-    def path_join(root: str, *paths: str) -> str:
-        if not isinstance(new_obj, Tree):
+    def path_join(root: str, *paths: str, isdir: bool = False) -> str:
+        if not isdir and paths == ROOT:
             return root
         return os.path.sep.join([root, *paths])
 
     diff_data = odiff(old_obj, new_obj, cache)
-    drop_root = not with_dirs and isinstance(new_obj, Tree)
 
     output: Dict[str, List[str]] = defaultdict(list)
     for state in ("added", "deleted", "modified", "unchanged"):
         items = getattr(diff_data, state)
-        output[state].extend(
-            path_join(root, *item.new.key)
-            for item in items
-            if not (drop_root and item.new.key == ROOT)
-        )
-        # TODO: PERF: diff is checking not_in_cache for each even if we only
-        # need it for the index.
-        # BUG: not_in_cache file also shows up as modified in staged and
-        # unstaged. We currently don't know if it is really modified.
-        output["not_in_cache"].extend(
-            path_join(root, *item.new.key)
-            for item in items
-            if not item.old.in_cache
-            and not (drop_root and item.new.key == ROOT)
-            and state != "added"
-        )
-    return output
+        for item in items:  # pylint: disable=not-an-iterable
+            entry = item.old if state == "deleted" else item.new
+            isdir = entry.oid.isdir if entry.oid else False
+
+            path = path_join(root, *entry.key, isdir=isdir)
+            output[state].append(path)
+            if not item.old.in_cache and state != "added":
+                output["not_in_cache"].append(path)
+    return dict(output)
+
+
+def _get_obj_items(root: str, obj: Optional["HashFile"]) -> List[str]:
+    if not obj:
+        return []
+
+    from dvc_data.hashfile.tree import Tree
+
+    sep = os.path.sep
+    if isinstance(obj, Tree):
+        return [sep.join([root, *key]) for key, _, _ in obj]
+    return [root]
 
 
 def _diff(
     root: str,
+    old_oid: Optional["HashInfo"],
     old_obj: Optional["HashFile"],
+    new_oid: Optional["HashInfo"],
     new_obj: Optional["HashFile"],
-    cache: "HashFileDB",
+    odb: "HashFileDB",
     granular: bool = False,
-    with_dirs: bool = False,
 ) -> Dict[str, List[str]]:
-    if granular:
-        return _granular_diff(
-            root, old_obj, new_obj, cache, with_dirs=with_dirs
-        )
-    return _shallow_diff(root, old_obj, new_obj, cache)
+    if not granular:
+        return _shallow_diff(root, old_oid, new_oid, odb)
+    if (old_oid and not old_obj) or (new_oid and not new_obj):
+        # we don't have enough information to give full details
+        unknown = _get_obj_items(root, new_obj)
+        shallow_diff = _shallow_diff(root, old_oid, new_oid, odb)
+        return {**shallow_diff, "unknown": unknown}
+    return _granular_diff(root, old_obj, new_obj, odb)
 
 
 class GitInfo(TypedDict, total=False):
@@ -121,7 +140,9 @@ class GitInfo(TypedDict, total=False):
     is_dirty: bool
 
 
-def _git_info(scm: "Base", untracked_files: str = "all") -> GitInfo:
+def _git_info(
+    scm: Union["Git", "NoSCM"], untracked_files: str = "all"
+) -> GitInfo:
     from scmrepo.exceptions import SCMError
 
     from dvc.scm import NoSCM
@@ -137,6 +158,8 @@ def _git_info(scm: "Base", untracked_files: str = "all") -> GitInfo:
         empty_repo = False
 
     staged, unstaged, untracked = scm.status(untracked_files=untracked_files)
+    if os.name == "nt":
+        untracked = [posixpath_to_os_path(path) for path in untracked]
     # NOTE: order is important here.
     return GitInfo(
         staged=staged,
@@ -148,7 +171,7 @@ def _git_info(scm: "Base", untracked_files: str = "all") -> GitInfo:
 
 
 def _diff_index_to_wtree(repo: "Repo", **kwargs: Any) -> Dict[str, List[str]]:
-    from dvc_data.build import build
+    from dvc_data.hashfile.build import build
 
     unstaged_diff = defaultdict(list)
     for out in repo.index.outs:
@@ -171,8 +194,18 @@ def _diff_index_to_wtree(repo: "Repo", **kwargs: Any) -> Dict[str, List[str]]:
         cache = repo.odb.repo
         root = str(out)
         old = out.get_obj()
+
         with ui.status(f"Calculating diff for {root} between index/workspace"):
-            d = _diff(root, old, new, cache, **kwargs)
+            d = _diff(
+                root,
+                out.hash_info,
+                old,
+                new.hash_info if new else None,
+                new,
+                cache,
+                **kwargs,
+            )
+
         for state, items in d.items():
             if not items:
                 continue
@@ -184,7 +217,9 @@ def _diff_head_to_index(
     repo: "Repo", head: str = "HEAD", **kwargs: Any
 ) -> Dict[str, List[str]]:
     # we need to store objects from index and the HEAD to diff later
-    objs: Dict[str, Dict[str, "HashFile"]] = defaultdict(dict)
+    objs: Dict[str, Dict[str, Tuple["HashFile", "HashInfo"]]]
+    objs = defaultdict(dict)
+
     staged_diff = defaultdict(list)
     for rev in repo.brancher(revs=[head]):
         for out in repo.index.outs:
@@ -193,15 +228,20 @@ def _diff_head_to_index(
                 continue
 
             root = str(out)
+            if isinstance(out.fs, GitFileSystem):
+                root = posixpath_to_os_path(root)
             typ = "index" if rev == "workspace" else head
-            objs[root][typ] = out.get_obj()
+            objs[root][typ] = (out.get_obj(), out.hash_info)
 
     cache = repo.odb.repo
     for root, obj_d in objs.items():
-        old = obj_d.get(head, None)
-        new = obj_d.get("index", None)
+        old_obj, old_oid = obj_d.get(head, (None, None))
+        new_obj, new_oid = obj_d.get("index", (None, None))
         with ui.status(f"Calculating diff for {root} between head/index"):
-            d = _diff(root, old, new, cache, **kwargs)
+            d = _diff(
+                root, old_oid, old_obj, new_oid, new_obj, cache, **kwargs
+            )
+
         for state, items in d.items():
             if not items:
                 continue
@@ -212,28 +252,36 @@ def _diff_head_to_index(
 
 class Status(TypedDict):
     not_in_cache: List[str]
-    committed: Dict[str, Any]
-    uncommitted: Dict[str, Any]
+    committed: Dict[str, List[str]]
+    uncommitted: Dict[str, List[str]]
     untracked: List[str]
     unchanged: List[str]
     git: GitInfo
 
 
-def _transform_git_paths_to_dvc(repo: "Repo", files: List[str]):
+def _transform_git_paths_to_dvc(
+    repo: "Repo", files: Iterable[str]
+) -> List[str]:
     """Transform files rel. to Git root to DVC root, and drop outside files."""
     rel = repo.fs.path.relpath(repo.root_dir, repo.scm.root_dir).rstrip("/")
-    if rel in (os.curdir, ""):
-        return files
 
-    prefix = rel + os.sep
-    length = len(prefix)
-    return [file[length:] for file in files if file.startswith(prefix)]
+    # if we have repo root in a different location than scm's root,
+    # i.e. subdir repo, all git_paths need to be transformed rel. to the DVC
+    # repo root and anything outside need to be filtered out.
+    if rel not in (os.curdir, ""):
+        prefix = rel + os.sep
+        length = len(prefix)
+        files = (file[length:] for file in files if file.startswith(prefix))
+
+    start = repo.fs.path.relpath(repo.fs.path.getcwd(), repo.root_dir)
+    if start in (os.curdir, ""):
+        return list(files)
+    # we need to convert repo relative paths to curdir relative.
+    return [repo.fs.path.relpath(file, start) for file in files]
 
 
 def status(repo: "Repo", untracked_files: str = "no", **kwargs: Any) -> Status:
-    from scmrepo.exceptions import SCMError
-
-    from dvc.scm import NoSCMError
+    from dvc.scm import NoSCMError, SCMError
 
     head = kwargs.pop("head", "HEAD")
     uncommitted_diff = _diff_index_to_wtree(repo, **kwargs)
@@ -261,3 +309,18 @@ def status(repo: "Repo", untracked_files: str = "no", **kwargs: Any) -> Status:
         unchanged=list(unchanged),
         git=git_info,
     )
+
+
+def ls(
+    repo: "Repo",
+    targets: List[Optional[str]] = None,
+    recursive: bool = False,
+) -> Iterator[Dict[str, Any]]:
+    targets = targets or [None]
+    pairs = chain.from_iterable(
+        repo.stage.collect_granular(target, recursive=recursive)
+        for target in targets
+    )
+    for stage, filter_info in pairs:
+        for out in stage.filter_outs(filter_info):
+            yield {"path": str(out), **out.annot.to_dict()}

dvc/repo/diff.py

@@ -23,9 +23,9 @@ def diff(self, a_rev="HEAD", b_rev=None, targets=None):
     if self.scm.no_commits:
         return {}
 
-    from dvc.fs.dvc import DvcFileSystem
+    from dvc.fs.dvc import DVCFileSystem
 
-    dvcfs = DvcFileSystem(repo=self)
+    dvcfs = DVCFileSystem(repo=self)
     targets = ensure_list(targets)
     targets = [dvcfs.from_os_path(target) for target in targets]
 
@@ -115,7 +115,7 @@ def _paths_checksums(repo, targets):
 
 def _output_paths(repo, targets):
     from dvc.fs import LocalFileSystem
-    from dvc_data.build import build
+    from dvc_data.hashfile.build import build
 
     on_working_fs = isinstance(repo.fs, LocalFileSystem)
 

dvc/repo/experiments/__init__.py

@@ -2,9 +2,9 @@ import logging
 import os
 import re
 import time
-from typing import Any, Dict, Iterable, Optional
+from typing import Dict, Iterable, Optional
 
-from funcy import cached_property, first
+from funcy import cached_property, chain, first
 
 from dvc.exceptions import DvcException
 from dvc.ui import ui
@@ -16,7 +16,7 @@ from .exceptions import (
     InvalidExpRefError,
     MultipleBranchError,
 )
-from .executor.base import BaseExecutor, ExecutorInfo
+from .executor.base import BaseExecutor
 from .queue.base import BaseStashQueue, QueueEntry
 from .queue.celery import LocalCeleryQueue
 from .queue.tempdir import TempDirQueue
@@ -25,7 +25,6 @@ from .refs import (
     CELERY_FAILED_STASH,
     CELERY_STASH,
     EXEC_APPLY,
-    EXEC_BRANCH,
     EXEC_CHECKPOINT,
     EXEC_NAMESPACE,
     EXPS_NAMESPACE,
@@ -172,7 +171,18 @@ class Experiments:
     ) -> Dict[str, str]:
         results: Dict[str, str] = {}
         if entries is None:
-            entries = list(self.celery_queue.iter_queued())
+            entries = list(
+                chain(
+                    self.celery_queue.iter_active(),
+                    self.celery_queue.iter_queued(),
+                )
+            )
+
+        logger.debug(
+            "reproduce all these entries '%s'",
+            entries,
+        )
+
         if not entries:
             return results
 
@@ -190,7 +200,10 @@ class Experiments:
                     time.sleep(1)
                 self.celery_queue.follow(entry)
                 # wait for task collection to complete
-                result = self.celery_queue.get_result(entry)
+                try:
+                    result = self.celery_queue.get_result(entry)
+                except FileNotFoundError:
+                    result = None
                 if result is None or result.exp_hash is None:
                     name = entry.name or entry.stash_rev[:7]
                     failed.append(name)
@@ -211,10 +224,11 @@ class Experiments:
 
     def _log_reproduced(self, revs: Iterable[str], tmp_dir: bool = False):
         names = []
+        rev_names = self.get_exact_name(revs)
         for rev in revs:
-            name = self.get_exact_name(rev)
+            name = rev_names[rev]
             names.append(name if name else rev[:7])
-        ui.write("\nReproduced experiment(s): {}".format(", ".join(names)))
+        ui.write("\nRan experiment(s): {}".format(", ".join(names)))
         if tmp_dir:
             ui.write(
                 "To apply the results of an experiment to your workspace "
@@ -411,86 +425,41 @@ class Experiments:
             raise MultipleBranchError(rev, ref_infos)
         return str(ref_infos[0])
 
-    def get_exact_name(self, rev: str):
+    def get_exact_name(self, revs: Iterable[str]) -> Dict[str, Optional[str]]:
         """Returns preferred name for the specified revision.
 
         Prefers tags, branches (heads), experiments in that orer.
         """
+        result: Dict[str, Optional[str]] = {}
         exclude = f"{EXEC_NAMESPACE}/*"
-        ref = self.scm.describe(rev, base=EXPS_NAMESPACE, exclude=exclude)
-        if ref:
-            try:
-                name = ExpRefInfo.from_ref(ref).name
-                if name:
-                    return name
-            except InvalidExpRefError:
-                pass
-        if rev in self.stash_revs:
-            return self.stash_revs[rev].name
-        return None
-
-    def get_running_exps(self, fetch_refs: bool = True) -> Dict[str, Any]:
-        """Return info for running experiments."""
-        result = {}
-        infofile = self.workspace_queue.get_infofile_path("workspace")
-        result.update(
-            self._fetch_running_exp("workspace", infofile, fetch_refs)
+        ref_dict = self.scm.describe(
+            revs, base=EXPS_NAMESPACE, exclude=exclude
         )
-        for queue in (self.tempdir_queue, self.celery_queue):
-            for entry in queue.iter_active():
-                infofile = queue.get_infofile_path(entry.stash_rev)
-                result.update(
-                    self._fetch_running_exp(
-                        entry.stash_rev, infofile, fetch_refs
-                    )
-                )
+        for rev in revs:
+            name: Optional[str] = None
+            ref = ref_dict[rev]
+            if ref:
+                try:
+                    name = ExpRefInfo.from_ref(ref).name
+                except InvalidExpRefError:
+                    pass
+            if not name:
+                if rev in self.stash_revs:
+                    name = self.stash_revs[rev].name
+                elif rev in self.celery_queue.failed_stash.stash_revs:
+                    name = self.celery_queue.failed_stash.stash_revs[rev].name
+            result[rev] = name
         return result
 
-    def _fetch_running_exp(
-        self, rev: str, infofile: str, fetch_refs: bool
-    ) -> Dict[str, Any]:
-        from dvc.scm import InvalidRemoteSCMRepo
-        from dvc.utils.serialize import load_json
-
-        from .executor.local import TempDirExecutor
-
-        result: Dict[str, Any] = {}
-        try:
-            info = ExecutorInfo.from_dict(load_json(infofile))
-        except OSError:
-            return result
-        if info.result is None:
-            if rev == "workspace":
-                # If we are appending to a checkpoint branch in a workspace
-                # run, show the latest checkpoint as running.
-                last_rev = self.scm.get_ref(EXEC_BRANCH)
-                if last_rev:
-                    result[last_rev] = info.asdict()
-                else:
-                    result[rev] = info.asdict()
-            else:
-                result[rev] = info.asdict()
-                if info.git_url and fetch_refs:
-
-                    def on_diverged(_ref: str, _checkpoint: bool):
-                        return False
-
-                    executor = TempDirExecutor.from_info(info)
-                    try:
-                        for ref in executor.fetch_exps(
-                            self.scm,
-                            on_diverged=on_diverged,
-                        ):
-                            logger.debug(
-                                "Updated running experiment '%s'.", ref
-                            )
-                            last_rev = self.scm.get_ref(ref)
-                            result[rev]["last"] = last_rev
-                            if last_rev:
-                                result[last_rev] = info.asdict()
-                    except InvalidRemoteSCMRepo:
-                        # ignore stale info files
-                        del result[rev]
+    def get_running_exps(self, fetch_refs: bool = True) -> Dict[str, Dict]:
+        """Return info for running experiments."""
+        result = {}
+        for queue in (
+            self.workspace_queue,
+            self.tempdir_queue,
+            self.celery_queue,
+        ):
+            result.update(queue.get_running_exps(fetch_refs))
         return result
 
     def apply(self, *args, **kwargs):

dvc/repo/experiments/diff.py

@@ -7,7 +7,7 @@ logger = logging.getLogger(__name__)
 
 
 def diff(repo, *args, a_rev=None, b_rev=None, param_deps=False, **kwargs):
-    from dvc.repo.experiments.show import _collect_experiment_commit
+    from dvc.repo.experiments.show import collect_experiment_commit
     from dvc.scm import resolve_rev
 
     if repo.scm.no_commits:
@@ -15,15 +15,15 @@ def diff(repo, *args, a_rev=None, b_rev=None, param_deps=False, **kwargs):
 
     if a_rev:
         rev = resolve_rev(repo.scm, a_rev)
-        old = _collect_experiment_commit(repo, rev, param_deps=param_deps)
+        old = collect_experiment_commit(repo, rev, param_deps=param_deps)
     else:
-        old = _collect_experiment_commit(repo, "HEAD", param_deps=param_deps)
+        old = collect_experiment_commit(repo, "HEAD", param_deps=param_deps)
 
     if b_rev:
         rev = resolve_rev(repo.scm, b_rev)
-        new = _collect_experiment_commit(repo, rev, param_deps=param_deps)
+        new = collect_experiment_commit(repo, rev, param_deps=param_deps)
     else:
-        new = _collect_experiment_commit(
+        new = collect_experiment_commit(
             repo, "workspace", param_deps=param_deps
         )
 

dvc/repo/experiments/executor/base.py

@@ -4,6 +4,7 @@ import pickle
 from abc import ABC, abstractmethod
 from contextlib import contextmanager
 from dataclasses import asdict, dataclass
+from enum import IntEnum
 from functools import partial
 from typing import (
     TYPE_CHECKING,
@@ -19,6 +20,8 @@ from typing import (
     Union,
 )
 
+from scmrepo.exceptions import SCMError
+
 from dvc.env import DVC_EXP_AUTO_PUSH, DVC_EXP_GIT_REMOTE
 from dvc.exceptions import DvcException
 from dvc.stage.serialize import to_lockfile
@@ -35,6 +38,8 @@ from ..refs import (
     EXEC_BASELINE,
     EXEC_BRANCH,
     EXEC_CHECKPOINT,
+    EXEC_HEAD,
+    EXEC_MERGE,
     EXEC_NAMESPACE,
     EXPS_NAMESPACE,
     EXPS_STASH,
@@ -47,10 +52,9 @@ if TYPE_CHECKING:
     from scmrepo.git import Git
 
     from dvc.repo import Repo
+    from dvc.repo.experiments.stash import ExpStashEntry
     from dvc.stage import PipelineStage
 
-    from ..base import ExpStashEntry
-
 logger = logging.getLogger(__name__)
 
 
@@ -64,6 +68,16 @@ class ExecutorResult(NamedTuple):
     force: bool
 
 
+class TaskStatus(IntEnum):
+    PENDING = 0
+    PREPARING = 1
+    RUNNING = 2
+    SUCCESS = 3
+    FAILED = 4
+    CANCELED = 5
+    FINISHED = 6
+
+
 @dataclass
 class ExecutorInfo:
     git_url: str
@@ -73,13 +87,16 @@ class ExecutorInfo:
     dvc_dir: str
     name: Optional[str] = None
     wdir: Optional[str] = None
-    collected: bool = False
     result_hash: Optional[str] = None
     result_ref: Optional[str] = None
     result_force: bool = False
+    status: TaskStatus = TaskStatus.PENDING
 
     @classmethod
     def from_dict(cls, d):
+        if "collected" in d:
+            if d.pop("collected"):
+                d["status"] = TaskStatus.FINISHED
         return cls(**d)
 
     def asdict(self):
@@ -136,6 +153,7 @@ class BaseExecutor(ABC):
         root_dir: str,
         dvc_dir: str,
         baseline_rev: str,
+        status: TaskStatus,
         wdir: Optional[str] = None,
         name: Optional[str] = None,
         location: Optional[str] = None,
@@ -149,9 +167,17 @@ class BaseExecutor(ABC):
         self.baseline_rev = baseline_rev
         self.location: str = location or self.DEFAULT_LOCATION
         self.result = result
+        self.status = status
 
     @abstractmethod
-    def init_git(self, scm: "Git", branch: Optional[str] = None):
+    def init_git(
+        self,
+        scm: "Git",
+        stash_rev: str,
+        entry: "ExpStashEntry",
+        infofile: Optional[str],
+        branch: Optional[str] = None,
+    ):
         """Init git repo and populate it using exp refs from the specified
         SCM instance.
         """
@@ -191,6 +217,7 @@ class BaseExecutor(ABC):
             dvc_dir=self.dvc_dir,
             name=self.name,
             wdir=self.wdir,
+            status=self.status,
             **result_dict,
         )
 
@@ -212,6 +239,7 @@ class BaseExecutor(ABC):
             root_dir=info.root_dir,
             dvc_dir=info.dvc_dir,
             baseline_rev=info.baseline_rev,
+            status=info.status,
             name=info.name,
             wdir=info.wdir,
             result=result,
@@ -222,7 +250,6 @@ class BaseExecutor(ABC):
     def from_stash_entry(
         cls: Type[_T],
         repo: "Repo",
-        stash_rev: str,
         entry: "ExpStashEntry",
         **kwargs,
     ) -> _T:
@@ -232,7 +259,6 @@ class BaseExecutor(ABC):
     def _from_stash_entry(
         cls: Type[_T],
         repo: "Repo",
-        stash_rev: str,
         entry: "ExpStashEntry",
         root_dir: str,
         **kwargs,
@@ -241,12 +267,11 @@ class BaseExecutor(ABC):
             root_dir=root_dir,
             dvc_dir=relpath(repo.dvc_dir, repo.scm.root_dir),
             baseline_rev=entry.baseline_rev,
+            status=TaskStatus.PREPARING,
             name=entry.name,
             wdir=relpath(os.getcwd(), repo.scm.root_dir),
             **kwargs,
         )
-        executor.init_git(repo.scm, branch=entry.branch)
-        executor.init_cache(repo, stash_rev)
         return executor
 
     @staticmethod
@@ -259,8 +284,12 @@ class BaseExecutor(ABC):
                 exp_data.update(to_lockfile(stage))
         return dict_sha256(exp_data)
 
-    def cleanup(self):
-        pass
+    def cleanup(self, infofile: str):
+        if infofile is not None:
+            info = ExecutorInfo.load_json(infofile)
+            if info.status < TaskStatus.FAILED:
+                info.status = TaskStatus.FINISHED
+            info.dump_json(infofile)
 
     # TODO: come up with better way to stash repro arguments
     @staticmethod
@@ -324,30 +353,35 @@ class BaseExecutor(ABC):
                 logger.debug("Replacing existing experiment '%s'", orig_ref)
                 return True
 
+            if on_diverged:
+                return on_diverged(orig_ref, has_checkpoint)
+
             self._raise_ref_conflict(
                 dest_scm, orig_ref, new_rev, has_checkpoint
             )
-            if on_diverged:
-                on_diverged(orig_ref, has_checkpoint)
             logger.debug("Reproduced existing experiment '%s'", orig_ref)
             return False
 
         # fetch experiments
-        dest_scm.fetch_refspecs(
-            self.git_url,
-            [f"{ref}:{ref}" for ref in refs],
-            on_diverged=on_diverged_ref,
-            force=force,
-            **kwargs,
-        )
-        # update last run checkpoint (if it exists)
-        if has_checkpoint:
+        try:
             dest_scm.fetch_refspecs(
                 self.git_url,
-                [f"{EXEC_CHECKPOINT}:{EXEC_CHECKPOINT}"],
-                force=True,
+                [f"{ref}:{ref}" for ref in refs],
+                on_diverged=on_diverged_ref,
+                force=force,
                 **kwargs,
             )
+            # update last run checkpoint (if it exists)
+            if has_checkpoint:
+                dest_scm.fetch_refspecs(
+                    self.git_url,
+                    [f"{EXEC_CHECKPOINT}:{EXEC_CHECKPOINT}"],
+                    force=True,
+                    **kwargs,
+                )
+        except SCMError:
+            pass
+
         return refs
 
     @classmethod
@@ -411,11 +445,9 @@ class BaseExecutor(ABC):
         exp_ref: Optional["ExpRefInfo"] = None
         repro_force: bool = False
 
-        if infofile is not None:
-            info.dump_json(infofile)
-
         with cls._repro_dvc(
             info,
+            infofile,
             log_errors=log_errors,
             **kwargs,
         ) as dvc:
@@ -491,9 +523,6 @@ class BaseExecutor(ABC):
             info.result_ref = ref
             info.result_force = repro_force
 
-        if infofile is not None:
-            info.dump_json(infofile)
-
         # ideally we would return stages here like a normal repro() call, but
         # stages is not currently picklable and cannot be returned across
         # multiprocessing calls
@@ -550,6 +579,7 @@ class BaseExecutor(ABC):
     def _repro_dvc(
         cls,
         info: "ExecutorInfo",
+        infofile: str = None,
         log_errors: bool = True,
         **kwargs,
     ):
@@ -557,6 +587,9 @@ class BaseExecutor(ABC):
         from dvc.stage.monitor import CheckpointKilledError
 
         dvc = Repo(os.path.join(info.root_dir, info.dvc_dir))
+        info.status = TaskStatus.RUNNING
+        if infofile is not None:
+            info.dump_json(infofile)
         if cls.QUIET:
             dvc.scm_context.quiet = cls.QUIET
         old_cwd = os.getcwd()
@@ -568,17 +601,23 @@ class BaseExecutor(ABC):
         try:
             logger.debug("Running repro in '%s'", os.getcwd())
             yield dvc
+            info.status = TaskStatus.SUCCESS
         except CheckpointKilledError:
+            info.status = TaskStatus.FAILED
             raise
         except DvcException:
             if log_errors:
                 logger.exception("")
+            info.status = TaskStatus.FAILED
             raise
         except Exception:
             if log_errors:
                 logger.exception("unexpected error")
+            info.status = TaskStatus.FAILED
             raise
         finally:
+            if infofile is not None:
+                info.dump_json(infofile)
             dvc.close()
             os.chdir(old_cwd)
 
@@ -714,3 +753,23 @@ class BaseExecutor(ABC):
         disable_other_loggers()
         if level is not None:
             dvc_logger.setLevel(level)
+
+    @contextmanager
+    def set_exec_refs(
+        self, scm: "Git", stash_rev: str, entry: "ExpStashEntry"
+    ):
+        try:
+            # Executor will be initialized with an empty git repo that
+            # we populate by pushing:
+            #   EXEC_HEAD - the base commit for this experiment
+            #   EXEC_MERGE - the unmerged changes (from our stash)
+            #       to be reproduced
+            #   EXEC_BASELINE - the baseline commit for this experiment
+            scm.set_ref(EXEC_HEAD, entry.head_rev)
+            scm.set_ref(EXEC_MERGE, stash_rev)
+            scm.set_ref(EXEC_BASELINE, entry.baseline_rev)
+            yield
+        finally:
+            for ref in (EXEC_HEAD, EXEC_MERGE, EXEC_BASELINE):
+                if scm.get_ref(ref):
+                    scm.remove_ref(ref)

dvc/repo/experiments/executor/local.py

@@ -19,7 +19,7 @@ from ..refs import (
     EXEC_MERGE,
     EXEC_NAMESPACE,
 )
-from .base import EXEC_TMP_DIR, BaseExecutor
+from .base import EXEC_TMP_DIR, BaseExecutor, TaskStatus
 
 if TYPE_CHECKING:
     from scmrepo.git import Git
@@ -46,10 +46,10 @@ class BaseLocalExecutor(BaseExecutor):
     def scm(self):
         return SCM(self.root_dir)
 
-    def cleanup(self):
-        super().cleanup()
+    def cleanup(self, infofile: str):
         self.scm.close()
         del self.scm
+        super().cleanup(infofile)
 
     def collect_cache(
         self, repo: "Repo", exp_ref: "ExpRefInfo", run_cache: bool = True
@@ -67,15 +67,28 @@ class TempDirExecutor(BaseLocalExecutor):
     QUIET = True
     DEFAULT_LOCATION = "tempdir"
 
-    def init_git(self, scm: "Git", branch: Optional[str] = None):
+    def init_git(
+        self,
+        scm: "Git",
+        stash_rev: str,
+        entry: "ExpStashEntry",
+        infofile: Optional[str],
+        branch: Optional[str] = None,
+    ):
         from dulwich.repo import Repo as DulwichRepo
 
         from ..utils import push_refspec
 
         DulwichRepo.init(os.fspath(self.root_dir))
 
-        refspec = f"{EXEC_NAMESPACE}/"
-        push_refspec(scm, self.git_url, refspec, refspec)
+        self.status = TaskStatus.PREPARING
+        if infofile:
+            self.info.dump_json(infofile)
+
+        with self.set_exec_refs(scm, stash_rev, entry):
+            refspec = f"{EXEC_NAMESPACE}/"
+            push_refspec(scm, self.git_url, refspec, refspec)
+
         if branch:
             push_refspec(scm, self.git_url, branch, branch)
             self.scm.set_ref(EXEC_BRANCH, branch, symbolic=True)
@@ -84,17 +97,12 @@ class TempDirExecutor(BaseLocalExecutor):
 
         if self.scm.get_ref(EXEC_CHECKPOINT):
             self.scm.remove_ref(EXEC_CHECKPOINT)
-
         # checkout EXEC_HEAD and apply EXEC_MERGE on top of it without
         # committing
         head = EXEC_BRANCH if branch else EXEC_HEAD
         self.scm.checkout(head, detach=True)
         merge_rev = self.scm.get_ref(EXEC_MERGE)
 
-        for ref in (EXEC_HEAD, EXEC_MERGE, EXEC_BASELINE):
-            if scm.get_ref(ref):
-                scm.remove_ref(ref)
-
         try:
             self.scm.merge(merge_rev, squash=True, commit=False)
         except _SCMError as exc:
@@ -114,8 +122,8 @@ class TempDirExecutor(BaseLocalExecutor):
         """Initialize DVC cache."""
         self._config(repo.odb.repo.path)
 
-    def cleanup(self):
-        super().cleanup()
+    def cleanup(self, infofile: str):
+        super().cleanup(infofile)
         logger.debug("Removing tmpdir '%s'", self.root_dir)
         remove(self.root_dir)
 
@@ -123,7 +131,6 @@ class TempDirExecutor(BaseLocalExecutor):
     def from_stash_entry(
         cls,
         repo: "Repo",
-        stash_rev: str,
         entry: "ExpStashEntry",
         wdir: Optional[str] = None,
         **kwargs,
@@ -132,9 +139,7 @@ class TempDirExecutor(BaseLocalExecutor):
         makedirs(parent_dir, exist_ok=True)
         tmp_dir = mkdtemp(dir=parent_dir)
         try:
-            executor = cls._from_stash_entry(
-                repo, stash_rev, entry, tmp_dir, **kwargs
-            )
+            executor = cls._from_stash_entry(repo, entry, tmp_dir, **kwargs)
             logger.debug("Init temp dir executor in '%s'", tmp_dir)
             return executor
         except Exception:
@@ -152,18 +157,31 @@ class WorkspaceExecutor(BaseLocalExecutor):
     def from_stash_entry(
         cls,
         repo: "Repo",
-        stash_rev: str,
         entry: "ExpStashEntry",
         **kwargs,
     ):
         root_dir = repo.scm.root_dir
-        executor = cls._from_stash_entry(
-            repo, stash_rev, entry, root_dir, **kwargs
+        executor: "WorkspaceExecutor" = cls._from_stash_entry(
+            repo, entry, root_dir, **kwargs
         )
         logger.debug("Init workspace executor in '%s'", root_dir)
         return executor
 
-    def init_git(self, scm: "Git", branch: Optional[str] = None):
+    def init_git(
+        self,
+        scm: "Git",
+        stash_rev: str,
+        entry: "ExpStashEntry",
+        infofile: Optional[str],
+        branch: Optional[str] = None,
+    ):
+        self.status = TaskStatus.PREPARING
+        if infofile:
+            self.info.dump_json(infofile)
+
+        scm.set_ref(EXEC_HEAD, entry.head_rev)
+        scm.set_ref(EXEC_MERGE, stash_rev)
+        scm.set_ref(EXEC_BASELINE, entry.baseline_rev)
         self._detach_stack.enter_context(
             self.scm.detach_head(
                 self.scm.get_ref(EXEC_HEAD),
@@ -184,7 +202,9 @@ class WorkspaceExecutor(BaseLocalExecutor):
     def init_cache(self, repo: "Repo", rev: str, run_cache: bool = True):
         pass
 
-    def cleanup(self):
+    def cleanup(self, infofile: str):
+        super().cleanup(infofile)
+        remove(os.path.dirname(infofile))
         with self._detach_stack:
             self.scm.remove_ref(EXEC_BASELINE)
             self.scm.remove_ref(EXEC_MERGE)
@@ -193,4 +213,3 @@ class WorkspaceExecutor(BaseLocalExecutor):
             checkpoint = self.scm.get_ref(EXEC_CHECKPOINT)
             if checkpoint and checkpoint != self._orig_checkpoint:
                 self.scm.set_ref(EXEC_APPLY, checkpoint)
-        super().cleanup()

dvc/repo/experiments/executor/manager/base.py

@@ -1,269 +0,0 @@
-import logging
-import os
-from abc import ABC
-from collections import defaultdict, deque
-from collections.abc import Mapping
-from typing import TYPE_CHECKING, Deque, Dict, Generator, Optional, Tuple, Type
-
-from dvc.proc.manager import ProcessManager
-
-from ...exceptions import CheckpointExistsError, ExperimentExistsError
-from ...refs import EXEC_BASELINE, EXEC_HEAD, EXEC_MERGE, ExpRefInfo
-from ...stash import ExpStashEntry
-from ..base import EXEC_PID_DIR, BaseExecutor
-from ..local import TempDirExecutor, WorkspaceExecutor
-
-if TYPE_CHECKING:
-    from scmrepo.git import Git
-
-    from dvc.repo import Repo
-
-logger = logging.getLogger(__name__)
-
-
-class BaseExecutorManager(ABC, Mapping):
-    """Manages executors for a collection of experiments to be run."""
-
-    EXECUTOR_CLS: Type = BaseExecutor
-
-    def __init__(
-        self,
-        scm: "Git",
-        wdir: str,
-        **kwargs,
-    ):
-        from dvc.utils.fs import makedirs
-
-        self.scm = scm
-        makedirs(wdir, exist_ok=True)
-        self.wdir = wdir
-        self.proc = ProcessManager(self.pid_dir)
-        self._attached: Dict[str, "BaseExecutor"] = {}
-        self._detached: Dict[str, "BaseExecutor"] = dict(self._load_infos())
-        self._queue: Deque[Tuple[str, "BaseExecutor"]] = deque()
-
-    def __getitem__(self, key: str) -> "BaseExecutor":
-        try:
-            return self._attached[key]
-        except KeyError:
-            pass
-        return self._detached[key]
-
-    def __iter__(self):
-        yield from self._attached
-        yield from self._detached
-
-    def __len__(self):
-        return len(self._attached) + len(self._detached)
-
-    @property
-    def pid_dir(self) -> str:
-        return os.path.join(self.wdir, EXEC_PID_DIR)
-
-    def enqueue(self, rev: str, executor: "BaseExecutor"):
-        assert rev not in self
-        self._queue.append((rev, executor))
-
-    def _load_infos(self) -> Generator[Tuple[str, "BaseExecutor"], None, None]:
-        import json
-        from urllib.parse import urlparse
-
-        from ..base import ExecutorInfo
-        from ..ssh import SSHExecutor
-
-        def make_executor(info: "ExecutorInfo"):
-            if info.git_url:
-                scheme = urlparse(info.git_url).scheme
-                if scheme == "file":
-                    cls: Type = TempDirExecutor
-                elif scheme == "ssh":
-                    cls = SSHExecutor
-                else:
-                    raise NotImplementedError
-            else:
-                cls = WorkspaceExecutor
-            return cls.from_info(info)
-
-        for name in self.proc:
-            infofile = self.get_infofile_path(name)
-            try:
-                with open(infofile, encoding="utf-8") as fobj:
-                    info = ExecutorInfo.from_dict(json.load(fobj))
-                yield name, make_executor(info)
-            except OSError:
-                continue
-
-    def get_infofile_path(self, name: str) -> str:
-        return os.path.join(
-            self.pid_dir,
-            name,
-            f"{name}{BaseExecutor.INFOFILE_EXT}",
-        )
-
-    @classmethod
-    def from_stash_entries(
-        cls,
-        scm: "Git",
-        wdir: str,
-        repo: "Repo",
-        to_run: Dict[str, ExpStashEntry],
-        **kwargs,
-    ):
-        manager = cls(scm, wdir)
-        manager._enqueue_stash_entries(scm, repo, to_run, **kwargs)
-        return manager
-
-    def _enqueue_stash_entries(
-        self,
-        scm: "Git",
-        repo: "Repo",
-        to_run: Dict[str, ExpStashEntry],
-        **kwargs,
-    ):
-        try:
-            for stash_rev, entry in to_run.items():
-                scm.set_ref(EXEC_HEAD, entry.head_rev)
-                scm.set_ref(EXEC_MERGE, stash_rev)
-                scm.set_ref(EXEC_BASELINE, entry.baseline_rev)
-
-                # Executor will be initialized with an empty git repo that
-                # we populate by pushing:
-                #   EXEC_HEAD - the base commit for this experiment
-                #   EXEC_MERGE - the unmerged changes (from our stash)
-                #       to be reproduced
-                #   EXEC_BASELINE - the baseline commit for this experiment
-                executor = self.EXECUTOR_CLS.from_stash_entry(
-                    repo,
-                    stash_rev,
-                    entry,
-                    **kwargs,
-                )
-                self.enqueue(stash_rev, executor)
-        finally:
-            for ref in (EXEC_HEAD, EXEC_MERGE, EXEC_BASELINE):
-                scm.remove_ref(ref)
-
-    def exec_queue(
-        self, repo: "Repo", jobs: Optional[int] = 1, detach: bool = False
-    ):
-        """Run dvc repro for queued executors in parallel."""
-        if detach:
-            raise NotImplementedError
-            # TODO use ProcessManager.spawn() to support detached runs
-        return self._exec_attached(repo, jobs=jobs)
-
-    def _exec_attached(self, repo: "Repo", jobs: Optional[int] = 1):
-        import signal
-        from concurrent.futures import (
-            CancelledError,
-            ProcessPoolExecutor,
-            wait,
-        )
-        from multiprocessing import Manager
-
-        from dvc.stage.monitor import CheckpointKilledError
-
-        result: Dict[str, Dict[str, str]] = defaultdict(dict)
-
-        manager = Manager()
-        pid_q = manager.Queue()
-
-        with ProcessPoolExecutor(max_workers=jobs) as workers:
-            futures = {}
-            while self._queue:
-                rev, executor = self._queue.popleft()
-                infofile = self.get_infofile_path(rev)
-                future = workers.submit(
-                    executor.reproduce,
-                    info=executor.info,
-                    rev=rev,
-                    queue=pid_q,
-                    infofile=infofile,
-                    log_level=logger.getEffectiveLevel(),
-                )
-                futures[future] = (rev, executor)
-                self._attached[rev] = executor
-
-            try:
-                wait(futures)
-            except KeyboardInterrupt:
-                # forward SIGINT to any running executor processes and
-                # cancel any remaining futures
-                workers.shutdown(wait=False)
-                pids = {}
-                for future, (rev, _) in futures.items():
-                    if future.running():
-                        # if future has already been started by the scheduler
-                        # we still have to wait until it tells us its PID
-                        while rev not in pids:
-                            rev, pid = pid_q.get()
-                            pids[rev] = pid
-                        os.kill(pids[rev], signal.SIGINT)
-                    elif not future.done():
-                        future.cancel()
-
-            for future, (rev, executor) in futures.items():
-                rev, executor = futures[future]
-
-                try:
-                    exc = future.exception()
-                    if exc is None:
-                        exec_result = future.result()
-                        result[rev].update(
-                            self._collect_executor(repo, executor, exec_result)
-                        )
-                    elif not isinstance(exc, CheckpointKilledError):
-                        logger.error(
-                            "Failed to reproduce experiment '%s'", rev[:7]
-                        )
-                except CancelledError:
-                    logger.error(
-                        "Cancelled before attempting to reproduce experiment "
-                        "'%s'",
-                        rev[:7],
-                    )
-                finally:
-                    self.cleanup_executor(rev, executor)
-
-        return result
-
-    def _collect_executor(self, repo, executor, exec_result) -> Dict[str, str]:
-        # NOTE: GitPython Repo instances cannot be re-used
-        # after process has received SIGINT or SIGTERM, so we
-        # need this hack to re-instantiate git instances after
-        # checkpoint runs. See:
-        # https://github.com/gitpython-developers/GitPython/issues/427
-        # del self.repo.scm
-
-        results = {}
-
-        def on_diverged(ref: str, checkpoint: bool):
-            ref_info = ExpRefInfo.from_ref(ref)
-            if checkpoint:
-                raise CheckpointExistsError(ref_info.name)
-            raise ExperimentExistsError(ref_info.name)
-
-        for ref in executor.fetch_exps(
-            self.scm,
-            force=exec_result.force,
-            on_diverged=on_diverged,
-        ):
-            exp_rev = self.scm.get_ref(ref)
-            if exp_rev:
-                logger.debug("Collected experiment '%s'.", exp_rev[:7])
-                results[exp_rev] = exec_result.exp_hash
-
-        if exec_result.ref_info is not None:
-            executor.collect_cache(repo, exec_result.ref_info)
-
-        return results
-
-    def cleanup_executor(self, rev: str, executor: "BaseExecutor"):
-        from dvc.utils.fs import remove
-
-        executor.cleanup()
-        try:
-            self.proc.remove(rev)
-        except KeyError:
-            pass
-        remove(os.path.join(self.pid_dir, rev))

dvc/repo/experiments/executor/manager/local.py

@@ -1,107 +0,0 @@
-import logging
-from collections import defaultdict
-from typing import TYPE_CHECKING, Dict, Optional
-
-from ...refs import EXEC_BASELINE, EXEC_BRANCH, EXEC_HEAD, EXEC_MERGE
-from ...stash import ExpStashEntry
-from ..local import TempDirExecutor, WorkspaceExecutor
-from .base import BaseExecutorManager
-
-if TYPE_CHECKING:
-    from scmrepo.git import Git
-
-    from dvc.repo import Repo
-
-logger = logging.getLogger(__name__)
-
-
-class TempDirExecutorManager(BaseExecutorManager):
-    EXECUTOR_CLS = TempDirExecutor
-
-
-class WorkspaceExecutorManager(BaseExecutorManager):
-    EXECUTOR_CLS = WorkspaceExecutor
-
-    @classmethod
-    def from_stash_entries(
-        cls,
-        scm: "Git",
-        wdir: str,
-        repo: "Repo",
-        to_run: Dict[str, ExpStashEntry],
-        **kwargs,
-    ):
-        manager = cls(scm, wdir)
-        try:
-            assert len(to_run) == 1
-            for stash_rev, entry in to_run.items():
-                scm.set_ref(EXEC_HEAD, entry.head_rev)
-                scm.set_ref(EXEC_MERGE, stash_rev)
-                scm.set_ref(EXEC_BASELINE, entry.baseline_rev)
-
-                executor = cls.EXECUTOR_CLS.from_stash_entry(
-                    repo,
-                    stash_rev,
-                    entry,
-                    **kwargs,
-                )
-                manager.enqueue(stash_rev, executor)
-        finally:
-            for ref in (EXEC_MERGE,):
-                scm.remove_ref(ref)
-        return manager
-
-    def _collect_executor(self, repo, executor, exec_result) -> Dict[str, str]:
-        results = {}
-        exp_rev = self.scm.get_ref(EXEC_BRANCH)
-        if exp_rev:
-            logger.debug("Collected experiment '%s'.", exp_rev[:7])
-            results[exp_rev] = exec_result.exp_hash
-        return results
-
-    def exec_queue(
-        self, repo: "Repo", jobs: Optional[int] = 1, detach: bool = False
-    ):
-        """Run a single WorkspaceExecutor.
-
-        Workspace execution is done within the main DVC process
-        (rather than in multiprocessing context)
-        """
-        from dvc.exceptions import DvcException
-        from dvc.stage.monitor import CheckpointKilledError
-
-        assert len(self._queue) == 1
-        assert not detach
-        result: Dict[str, Dict[str, str]] = defaultdict(dict)
-        rev, executor = self._queue.popleft()
-
-        exec_name = "workspace"
-        infofile = self.get_infofile_path(exec_name)
-        try:
-            exec_result = executor.reproduce(
-                info=executor.info,
-                rev=rev,
-                infofile=infofile,
-                log_level=logger.getEffectiveLevel(),
-                log_errors=not isinstance(executor, WorkspaceExecutor),
-            )
-            if not exec_result.exp_hash:
-                raise DvcException(
-                    f"Failed to reproduce experiment '{rev[:7]}'"
-                )
-            if exec_result.ref_info:
-                result[rev].update(
-                    self._collect_executor(repo, executor, exec_result)
-                )
-        except CheckpointKilledError:
-            # Checkpoint errors have already been logged
-            return {}
-        except DvcException:
-            raise
-        except Exception as exc:
-            raise DvcException(
-                f"Failed to reproduce experiment '{rev[:7]}'"
-            ) from exc
-        finally:
-            self.cleanup_executor(exec_name, executor)
-        return result

dvc/repo/experiments/executor/manager/ssh.py

@@ -1,109 +0,0 @@
-import logging
-import posixpath
-from collections import defaultdict
-from typing import TYPE_CHECKING, Callable, Dict, Generator, Optional, Tuple
-
-from ...stash import ExpStashEntry
-from ..base import BaseExecutor
-from ..ssh import SSHExecutor, _sshfs
-from .base import BaseExecutorManager
-
-if TYPE_CHECKING:
-    from scmrepo.git import Git
-
-    from dvc.repo import Repo
-
-logger = logging.getLogger(__name__)
-
-
-class SSHExecutorManager(BaseExecutorManager):
-    EXECUTOR_CLS = SSHExecutor
-
-    def __init__(
-        self,
-        scm: "Git",
-        wdir: str,
-        host: Optional[str] = None,
-        port: Optional[int] = None,
-        username: Optional[str] = None,
-        fs_factory: Optional[Callable] = None,
-        **kwargs,
-    ):
-        assert host
-        super().__init__(scm, wdir, **kwargs)
-        self.host = host
-        self.port = port
-        self.username = username
-        self._fs_factory = fs_factory
-
-    def _load_infos(self) -> Generator[Tuple[str, "BaseExecutor"], None, None]:
-        # TODO: load existing infos using sshfs
-        yield from []
-
-    @classmethod
-    def from_stash_entries(
-        cls,
-        scm: "Git",
-        wdir: str,
-        repo: "Repo",
-        to_run: Dict[str, ExpStashEntry],
-        **kwargs,
-    ):
-        machine_name: Optional[str] = kwargs.get("machine_name", None)
-        manager = cls(
-            scm, wdir, **repo.machine.get_executor_kwargs(machine_name)
-        )
-        manager._enqueue_stash_entries(scm, repo, to_run, **kwargs)
-        return manager
-
-    def sshfs(self):
-        return _sshfs(self._fs_factory, host=self.host, port=self.port)
-
-    def get_infofile_path(self, name: str) -> str:
-        return f"{name}{BaseExecutor.INFOFILE_EXT}"
-
-    def _exec_attached(self, repo: "Repo", jobs: Optional[int] = 1):
-        from dvc.exceptions import DvcException
-        from dvc.stage.monitor import CheckpointKilledError
-
-        assert len(self._queue) == 1
-        result: Dict[str, Dict[str, str]] = defaultdict(dict)
-        rev, executor = self._queue.popleft()
-        info = executor.info
-        infofile = posixpath.join(
-            info.root_dir,
-            info.dvc_dir,
-            "tmp",
-            self.get_infofile_path(rev),
-        )
-        try:
-            exec_result = executor.reproduce(
-                info=executor.info,
-                rev=rev,
-                infofile=infofile,
-                log_level=logger.getEffectiveLevel(),
-                fs_factory=self._fs_factory,
-            )
-            if not exec_result.exp_hash:
-                raise DvcException(
-                    f"Failed to reproduce experiment '{rev[:7]}'"
-                )
-            if exec_result.ref_info:
-                result[rev].update(
-                    self._collect_executor(repo, executor, exec_result)
-                )
-        except CheckpointKilledError:
-            # Checkpoint errors have already been logged
-            return {}
-        except DvcException:
-            raise
-        except Exception as exc:
-            raise DvcException(
-                f"Failed to reproduce experiment '{rev[:7]}'"
-            ) from exc
-        finally:
-            self.cleanup_executor(rev, executor)
-        return result
-
-    def cleanup_executor(self, rev: str, executor: "BaseExecutor"):
-        executor.cleanup()

dvc/repo/experiments/executor/ssh.py

@@ -5,10 +5,9 @@ import sys
 from contextlib import contextmanager
 from typing import TYPE_CHECKING, Callable, Iterable, Optional
 
+from dvc_ssh import SSHFileSystem
 from funcy import first
 
-from dvc.fs import SSHFileSystem
-
 from ..refs import (
     EXEC_BRANCH,
     EXEC_CHECKPOINT,
@@ -16,7 +15,7 @@ from ..refs import (
     EXEC_MERGE,
     EXEC_NAMESPACE,
 )
-from .base import BaseExecutor, ExecutorInfo, ExecutorResult
+from .base import BaseExecutor, ExecutorInfo, ExecutorResult, TaskStatus
 
 if TYPE_CHECKING:
     from queue import Queue
@@ -24,8 +23,8 @@ if TYPE_CHECKING:
     from scmrepo.git import Git
 
     from dvc.repo import Repo
-
-    from ..base import ExpRefInfo, ExpStashEntry
+    from dvc.repo.experiments.refs import ExpRefInfo
+    from dvc.repo.experiments.stash import ExpStashEntry
 
 logger = logging.getLogger(__name__)
 
@@ -76,14 +75,12 @@ class SSHExecutor(BaseExecutor):
     def from_stash_entry(
         cls,
         repo: "Repo",
-        stash_rev: str,
         entry: "ExpStashEntry",
         **kwargs,
     ):
         machine_name: Optional[str] = kwargs.pop("machine_name", None)
         executor = cls._from_stash_entry(
             repo,
-            stash_rev,
             entry,
             cls.gen_dirname(entry.name),
             location=machine_name,
@@ -120,9 +117,20 @@ class SSHExecutor(BaseExecutor):
         }
         return kwargs
 
-    def init_git(self, scm: "Git", branch: Optional[str] = None):
+    def init_git(
+        self,
+        scm: "Git",
+        stash_rev: str,
+        entry: "ExpStashEntry",
+        infofile: Optional[str],
+        branch: Optional[str] = None,
+    ):
         from ..utils import push_refspec
 
+        self.status = TaskStatus.PREPARING
+        if infofile:
+            self.info.dump_json(infofile)
+
         with self.sshfs() as fs:
             fs.makedirs(self.root_dir)
             self._ssh_cmd(fs, "git init .")
@@ -138,8 +146,11 @@ class SSHExecutor(BaseExecutor):
             # TODO: support multiple client key retries in git backends
             # (see https://github.com/iterative/dvc/issues/6508)
             kwargs = self._git_client_args(fs)
-            refspec = f"{EXEC_NAMESPACE}/"
-            push_refspec(scm, self.git_url, refspec, refspec, **kwargs)
+
+            with self.set_exec_refs(scm, stash_rev, entry):
+                refspec = f"{EXEC_NAMESPACE}/"
+                push_refspec(scm, self.git_url, refspec, refspec, **kwargs)
+
             if branch:
                 push_refspec(scm, self.git_url, branch, branch, **kwargs)
                 self._ssh_cmd(fs, f"git symbolic-ref {EXEC_BRANCH} {branch}")

dvc/repo/experiments/init.py

@@ -248,11 +248,6 @@ def init(
         except MissingParamsFile:
             pass
 
-    models = context.get("models")
-    live_path = context.pop("live", None)
-    live_metrics = f"{live_path}.json" if live_path else None
-    live_plots = os.path.join(live_path, "scalars") if live_path else None
-
     if type == "checkpoint":
         outs_key = "checkpoints"
         metrics_key = "metrics_persist_no_cache"
@@ -262,6 +257,13 @@ def init(
         metrics_key = "metrics_no_cache"
         plots_key = "plots_no_cache"
 
+    models = [context.get("models")]
+    metrics = [context.get("metrics")]
+    plots = [context.get("plots")]
+    if live_path := context.pop("live", None):
+        metrics.append(os.path.join(live_path, "metrics.json"))
+        plots.append(os.path.join(live_path, "plots"))
+
     stage = repo.stage.create(
         name=name,
         cmd=context["cmd"],
@@ -269,9 +271,9 @@ def init(
         params=[{params: None}] if params else None,
         force=force,
         **{
-            outs_key: compact([models]),
-            metrics_key: compact([context.get("metrics"), live_metrics]),
-            plots_key: compact([context.get("plots"), live_plots]),
+            outs_key: compact(models),
+            metrics_key: compact(metrics),
+            plots_key: compact(plots),
         },
     )
 

dvc/repo/experiments/queue/base.py

@@ -17,11 +17,12 @@ from typing import (
     Union,
 )
 
-from funcy import cached_property
+from funcy import cached_property, retry
 
-from dvc.dependency.param import MissingParamsError
+from dvc.dependency import ParamsDependency
 from dvc.env import DVCLIVE_RESUME
 from dvc.exceptions import DvcException
+from dvc.lock import LockError
 from dvc.ui import ui
 
 from ..exceptions import CheckpointExistsError, ExperimentExistsError
@@ -32,7 +33,7 @@ from ..executor.base import (
     ExecutorResult,
 )
 from ..executor.local import WorkspaceExecutor
-from ..refs import EXEC_BASELINE, EXEC_HEAD, EXEC_MERGE, ExpRefInfo
+from ..refs import ExpRefInfo
 from ..stash import ExpStash, ExpStashEntry
 from ..utils import exp_refs_by_rev, scm_locked
 
@@ -185,9 +186,12 @@ class BaseStashQueue(ABC):
 
         def _format_entry(
             entry: QueueEntry,
+            exp_result: Optional[ExecutorResult] = None,
             status: str = "Unknown",
         ) -> Dict[str, Any]:
             name = entry.name
+            if not name and exp_result and exp_result.ref_info:
+                name = exp_result.ref_info.name
             # NOTE: We fallback to Unknown status for experiments
             # generated in prior (incompatible) DVC versions
             return {
@@ -210,8 +214,8 @@ class BaseStashQueue(ABC):
             for queue_entry, _ in self.iter_failed()
         )
         result.extend(
-            _format_entry(queue_entry, status="Success")
-            for queue_entry, _ in self.iter_success()
+            _format_entry(queue_entry, exp_result=exp_result, status="Success")
+            for queue_entry, exp_result in self.iter_success()
         )
         return result
 
@@ -264,6 +268,7 @@ class BaseStashQueue(ABC):
                 finish any active experiments before shutting down.
         """
 
+    @abstractmethod
     def logs(
         self,
         rev: str,
@@ -283,7 +288,7 @@ class BaseStashQueue(ABC):
     def _stash_exp(
         self,
         *args,
-        params: Optional[dict] = None,
+        params: Optional[Dict[str, List[str]]] = None,
         resume_rev: Optional[str] = None,
         baseline_rev: Optional[str] = None,
         branch: Optional[str] = None,
@@ -292,10 +297,9 @@ class BaseStashQueue(ABC):
     ) -> QueueEntry:
         """Stash changes from the workspace as an experiment.
 
-        Arguments:
-            params: Optional dictionary of parameter values to be used.
-                Values take priority over any parameters specified in the
-                user's workspace.
+        Args:
+            params: Dict mapping paths to `Hydra Override`_ patterns,
+                provided via `exp run --set-param`.
             resume_rev: Optional checkpoint resume rev.
             baseline_rev: Optional baseline rev for this experiment, defaults
                 to the current SCM rev.
@@ -305,6 +309,9 @@ class BaseStashQueue(ABC):
             name: Optional experiment name. If specified this will be used as
                 the human-readable name in the experiment branch ref. Has no
                 effect of branch is specified.
+
+        .. _Hydra Override:
+            https://hydra.cc/docs/next/advanced/override_grammar/basic/
         """
         with self.scm.detach_head(client="dvc") as orig_head:
             stash_head = orig_head
@@ -508,22 +515,37 @@ class BaseStashQueue(ABC):
             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 NewParamsFound, merge_params
-        from dvc.utils.serialize import MODIFIERS
+    def _update_params(self, params: Dict[str, List[str]]):
+        """Update param files with the provided `Hydra Override`_ patterns.
+
+        Args:
+            params: Dict mapping paths to `Hydra Override`_ patterns,
+                provided via `exp run --set-param`.
+
+        .. _Hydra Override:
+            https://hydra.cc/docs/advanced/override_grammar/basic/
+        """
+        from dvc.utils.hydra import apply_overrides, compose_and_dump
 
         logger.debug("Using experiment params '%s'", params)
 
-        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:
-                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)
+        hydra_config = self.repo.config.get("hydra", {})
+        hydra_enabled = hydra_config.get("enabled", False)
+        hydra_output_file = ParamsDependency.DEFAULT_PARAMS_FILE
+        for path, overrides in params.items():
+            if hydra_enabled and path == hydra_output_file:
+                config_dir = os.path.join(
+                    self.repo.root_dir, hydra_config.get("config_dir", "conf")
+                )
+                config_name = hydra_config.get("config_name", "config")
+                compose_and_dump(
+                    path,
+                    config_dir,
+                    config_name,
+                    overrides,
+                )
+            else:
+                apply_overrides(path, overrides)
 
         # Force params file changes to be staged in git
         # Otherwise in certain situations the changes to params file may be
@@ -532,8 +554,9 @@ class BaseStashQueue(ABC):
         self.scm.add(list(params.keys()))
 
     @staticmethod
+    @retry(180, errors=LockError, timeout=1)
     @scm_locked
-    def setup_executor(
+    def init_executor(
         exp: "Experiments",
         queue_entry: QueueEntry,
         executor_cls: Type[BaseExecutor] = WorkspaceExecutor,
@@ -548,21 +571,23 @@ class BaseStashQueue(ABC):
         )
         if stash_entry.stash_index is not None:
             stash.drop(stash_entry.stash_index)
-
-        scm.set_ref(EXEC_HEAD, stash_entry.head_rev)
-        scm.set_ref(EXEC_MERGE, stash_rev)
-        scm.set_ref(EXEC_BASELINE, stash_entry.baseline_rev)
-
-        # Executor will be initialized with an empty git repo that
-        # we populate by pushing:
-        #   EXEC_HEAD - the base commit for this experiment
-        #   EXEC_MERGE - the unmerged changes (from our stash)
-        #       to be reproduced
-        #   EXEC_BASELINE - the baseline commit for this experiment
-        return executor_cls.from_stash_entry(
-            exp.repo, stash_rev, stash_entry, **kwargs
+        executor = executor_cls.from_stash_entry(
+            exp.repo, stash_entry, **kwargs
         )
 
+        infofile = exp.celery_queue.get_infofile_path(stash_rev)
+        executor.init_git(
+            exp.repo.scm,
+            stash_rev,
+            stash_entry,
+            infofile,
+            branch=stash_entry.branch,
+        )
+
+        executor.init_cache(exp.repo, stash_rev)
+
+        return executor
+
     def get_infofile_path(self, name: str) -> str:
         return os.path.join(
             self.pid_dir,
@@ -571,8 +596,9 @@ class BaseStashQueue(ABC):
         )
 
     @staticmethod
+    @retry(180, errors=LockError, timeout=1)
     @scm_locked
-    def collect_executor(
+    def collect_git(
         exp: "Experiments",
         executor: BaseExecutor,
         exec_result: ExecutorResult,
@@ -596,6 +622,17 @@ class BaseStashQueue(ABC):
                 logger.debug("Collected experiment '%s'.", exp_rev[:7])
                 results[exp_rev] = exec_result.exp_hash
 
+        return results
+
+    @classmethod
+    def collect_executor(
+        cls,
+        exp: "Experiments",
+        executor: BaseExecutor,
+        exec_result: ExecutorResult,
+    ) -> Dict[str, str]:
+        results = cls.collect_git(exp, executor, exec_result)
+
         if exec_result.ref_info is not None:
             executor.collect_cache(exp.repo, exec_result.ref_info)
 
@@ -662,3 +699,11 @@ class BaseStashQueue(ABC):
                 entry.stash_rev,
                 message=f"commit: {msg}",
             )
+
+    @abstractmethod
+    def get_running_exps(self, fetch_refs: bool = True) -> Dict[str, Dict]:
+        """Get the execution info of the currently running experiments
+
+        Args:
+            fetch_ref (bool): fetch completed checkpoints or not.
+        """

dvc/repo/experiments/queue/celery.py

@@ -14,7 +14,8 @@ from typing import (
     Set,
 )
 
-from funcy import cached_property
+from celery.result import AsyncResult
+from funcy import cached_property, first
 from kombu.message import Message
 
 from dvc.daemon import daemonize
@@ -23,9 +24,10 @@ from dvc.ui import ui
 
 from ..exceptions import UnresolvedQueueExpNamesError
 from ..executor.base import EXEC_TMP_DIR, ExecutorInfo, ExecutorResult
-from ..stash import ExpStashEntry
 from .base import BaseStashQueue, QueueDoneResult, QueueEntry, QueueGetResult
+from .exceptions import CannotKillTasksError
 from .tasks import run_exp
+from .utils import fetch_running_exp_from_temp_dir
 
 if TYPE_CHECKING:
     from dvc_task.app import FSApp
@@ -41,7 +43,7 @@ class _MessageEntry(NamedTuple):
 
 
 class _TaskEntry(NamedTuple):
-    task_id: str
+    async_result: AsyncResult
     entry: QueueEntry
 
 
@@ -181,6 +183,7 @@ class LocalCeleryQueue(BaseStashQueue):
                 continue
             args, kwargs, _embed = msg.decode()
             entry_dict = kwargs.get("entry_dict", args[0])
+            logger.debug("Found queued task %s", entry_dict["stash_rev"])
             yield _MessageEntry(msg, QueueEntry.from_dict(entry_dict))
 
     def _iter_processed(self) -> Generator[_MessageEntry, None, None]:
@@ -192,30 +195,39 @@ class LocalCeleryQueue(BaseStashQueue):
             yield _MessageEntry(msg, QueueEntry.from_dict(entry_dict))
 
     def _iter_active_tasks(self) -> Generator[_TaskEntry, None, None]:
-        from celery.result import AsyncResult
 
         for msg, entry in self._iter_processed():
             task_id = msg.headers["id"]
             result: AsyncResult = AsyncResult(task_id)
             if not result.ready():
-                yield _TaskEntry(task_id, entry)
+                logger.debug("Found active task %s", entry.stash_rev)
+                yield _TaskEntry(result, entry)
 
     def _iter_done_tasks(self) -> Generator[_TaskEntry, None, None]:
-        from celery.result import AsyncResult
 
         for msg, entry in self._iter_processed():
             task_id = msg.headers["id"]
             result: AsyncResult = AsyncResult(task_id)
             if result.ready():
-                yield _TaskEntry(task_id, entry)
+                logger.debug("Found done task %s", entry.stash_rev)
+                yield _TaskEntry(result, entry)
 
     def iter_active(self) -> Generator[QueueEntry, None, None]:
         for _, entry in self._iter_active_tasks():
             yield entry
 
     def iter_done(self) -> Generator[QueueDoneResult, None, None]:
-        for _, entry in self._iter_done_tasks():
-            yield QueueDoneResult(entry, self.get_result(entry))
+        for result, entry in self._iter_done_tasks():
+            try:
+                exp_result = self.get_result(entry)
+            except FileNotFoundError:
+                if result.status == "SUCCESS":
+                    raise DvcException(
+                        f"Invalid experiment '{entry.stash_rev[:7]}'."
+                    )
+                elif result.status == "FAILURE":
+                    exp_result = None
+            yield QueueDoneResult(entry, exp_result)
 
     def iter_success(self) -> Generator[QueueDoneResult, None, None]:
         for queue_entry, exp_result in self.iter_done():
@@ -223,84 +235,122 @@ class LocalCeleryQueue(BaseStashQueue):
                 yield QueueDoneResult(queue_entry, exp_result)
 
     def iter_failed(self) -> Generator[QueueDoneResult, None, None]:
-        failed_revs: Dict[str, ExpStashEntry] = (
-            dict(self.failed_stash.stash_revs)
-            if self.failed_stash is not None
-            else {}
-        )
-
         for queue_entry, exp_result in self.iter_done():
-            if exp_result is None and queue_entry.stash_rev in failed_revs:
+            if exp_result is None:
                 yield QueueDoneResult(queue_entry, exp_result)
 
     def reproduce(self) -> Mapping[str, Mapping[str, str]]:
         raise NotImplementedError
 
+    def _load_info(self, rev: str) -> ExecutorInfo:
+        infofile = self.get_infofile_path(rev)
+        return ExecutorInfo.load_json(infofile)
+
+    def _get_done_result(
+        self, entry: QueueEntry, timeout: Optional[float] = None
+    ) -> Optional["ExecutorResult"]:
+        from celery.exceptions import TimeoutError as _CeleryTimeout
+
+        for msg, processed_entry in self._iter_processed():
+            if entry.stash_rev == processed_entry.stash_rev:
+                task_id = msg.headers["id"]
+                result: AsyncResult = AsyncResult(task_id)
+                if not result.ready():
+                    logger.debug(
+                        "Waiting for exp task '%s' to complete", result.id
+                    )
+                    try:
+                        result.get(timeout=timeout)
+                    except _CeleryTimeout as exc:
+                        raise DvcException(
+                            "Timed out waiting for exp to finish."
+                        ) from exc
+                executor_info = self._load_info(entry.stash_rev)
+                return executor_info.result
+        raise FileNotFoundError
+
     def get_result(
         self, entry: QueueEntry, timeout: Optional[float] = None
-    ) -> Optional[ExecutorResult]:
-        from celery.exceptions import TimeoutError as _CeleryTimeout
-        from celery.result import AsyncResult
-
-        def _load_info(rev: str) -> ExecutorInfo:
-            infofile = self.get_infofile_path(rev)
-            return ExecutorInfo.load_json(infofile)
-
-        def _load_collected(rev: str) -> Optional[ExecutorResult]:
-            executor_info = _load_info(rev)
-            if executor_info.collected:
-                return executor_info.result
-            raise FileNotFoundError
+    ) -> Optional["ExecutorResult"]:
 
         try:
-            return _load_collected(entry.stash_rev)
+            return self._get_done_result(entry, timeout)
         except FileNotFoundError:
-            # Infofile will not be created until execution begins
             pass
 
         for queue_entry in self.iter_queued():
             if entry.stash_rev == queue_entry.stash_rev:
                 raise DvcException("Experiment has not been started.")
-        for task_id, active_entry in self._iter_active_tasks():
-            if entry.stash_rev == active_entry.stash_rev:
-                logger.debug("Waiting for exp task '%s' to complete", task_id)
-                try:
-                    result: AsyncResult = AsyncResult(task_id)
-                    result.get(timeout=timeout)
-                except _CeleryTimeout as exc:
-                    raise DvcException(
-                        "Timed out waiting for exp to finish."
-                    ) from exc
-                executor_info = _load_info(entry.stash_rev)
-                return executor_info.result
 
         # NOTE: It's possible for an exp to complete while iterating through
         # other queued and active tasks, in which case the exp will get moved
         # out of the active task list, and needs to be loaded here.
-        try:
-            return _load_collected(entry.stash_rev)
-        except FileNotFoundError:
-            pass
-        raise DvcException(f"Invalid experiment '{entry.stash_rev[:7]}'.")
+        return self._get_done_result(entry, timeout)
+
+    def _get_running_task_ids(self) -> Set[str]:
+        running_task_ids: Set[str] = set()
+        active_workers = self.worker_status()
+        for _, tasks in active_workers.items():
+            task = first(tasks)
+            if task:
+                running_task_ids.add(task["id"])
+        return running_task_ids
+
+    def _try_to_kill_tasks(
+        self, to_kill: Dict[QueueEntry, str]
+    ) -> Dict[QueueEntry, str]:
+        fail_to_kill_entries: Dict[QueueEntry, str] = {}
+        for queue_entry, rev in to_kill.items():
+            try:
+                self.proc.kill(queue_entry.stash_rev)
+                logger.debug(f"Task {rev} had been killed.")
+            except ProcessLookupError:
+                fail_to_kill_entries[queue_entry] = rev
+        return fail_to_kill_entries
+
+    def _mark_inactive_tasks_failure(self, remained_entries):
+        remained_revs: List[str] = []
+        running_ids = self._get_running_task_ids()
+        logger.debug(f"Current running tasks ids: {running_ids}.")
+        for msg, entry in self._iter_processed():
+            if entry not in remained_entries:
+                continue
+            task_id = msg.headers["id"]
+            if task_id in running_ids:
+                remained_revs.append(remained_entries[entry])
+            else:
+                result: AsyncResult = AsyncResult(task_id)
+                if not result.ready():
+                    logger.debug(
+                        f"Task id {task_id} rev {remained_entries[entry]} "
+                        "marked as failure."
+                    )
+                    self.celery.backend.mark_as_failure(task_id, None)
+
+        if remained_revs:
+            raise CannotKillTasksError(remained_revs)
 
     def kill(self, revs: Collection[str]) -> None:
-        to_kill: Set[QueueEntry] = set()
         name_dict: Dict[
             str, Optional[QueueEntry]
         ] = self.match_queue_entry_by_name(set(revs), self.iter_active())
 
-        missing_rev: List[str] = []
+        to_kill: Dict[QueueEntry, str] = {}
+        missing_revs: List[str] = []
         for rev, queue_entry in name_dict.items():
             if queue_entry is None:
-                missing_rev.append(rev)
+                missing_revs.append(rev)
             else:
-                to_kill.add(queue_entry)
+                to_kill[queue_entry] = rev
+        if missing_revs:
+            raise UnresolvedQueueExpNamesError(missing_revs)
 
-        if missing_rev:
-            raise UnresolvedQueueExpNamesError(missing_rev)
+        fail_to_kill_entries: Dict[QueueEntry, str] = self._try_to_kill_tasks(
+            to_kill
+        )
 
-        for queue_entry in to_kill:
-            self.proc.kill(queue_entry.stash_rev)
+        if fail_to_kill_entries:
+            self._mark_inactive_tasks_failure(fail_to_kill_entries)
 
     def shutdown(self, kill: bool = False):
         self.celery.control.shutdown()
@@ -356,10 +406,11 @@ class LocalCeleryQueue(BaseStashQueue):
         ) as fobj:
             ui.write(fobj.read())
 
-    def worker_status(self) -> Dict:
+    def worker_status(self) -> Dict[str, List[Dict]]:
         """Return the current active celery worker"""
         status = self.celery.control.inspect().active() or {}
         logger.debug(f"Worker status: {status}")
+
         return status
 
     def clear(self, *args, **kwargs):
@@ -371,3 +422,18 @@ class LocalCeleryQueue(BaseStashQueue):
         from .remove import celery_remove
 
         return celery_remove(self, *args, **kwargs)
+
+    def get_running_exps(self, fetch_refs: bool = True) -> Dict[str, Dict]:
+        """Get the execution info of the currently running experiments
+
+        Args:
+            fetch_ref (bool): fetch completed checkpoints or not.
+        """
+        result: Dict[str, Dict] = {}
+        for entry in self.iter_active():
+            result.update(
+                fetch_running_exp_from_temp_dir(
+                    self, entry.stash_rev, fetch_refs
+                )
+            )
+        return result

dvc/repo/experiments/queue/exceptions.py

@@ -0,0 +1,12 @@
+from typing import Collection
+
+from dvc.exceptions import DvcException
+
+
+class CannotKillTasksError(DvcException):
+    def __init__(self, revs: Collection[str]):
+        rev_str = ",".join(revs)
+        super().__init__(
+            f"Task '{rev_str}' is initializing, please wait a few seconds "
+            "until the experiments start running to retry the kill operation."
+        )

dvc/repo/experiments/queue/tasks.py

@@ -3,8 +3,6 @@ from typing import Any, Dict
 from celery import shared_task
 from celery.utils.log import get_task_logger
 
-from dvc.utils.fs import remove
-
 from ..executor.base import ExecutorInfo
 from ..executor.local import TempDirExecutor
 from .base import BaseStashQueue, QueueEntry
@@ -13,7 +11,7 @@ logger = get_task_logger(__name__)
 
 
 @shared_task
-def setup_exp(entry_dict: Dict[str, Any]) -> str:
+def setup_exp(entry_dict: Dict[str, Any]) -> TempDirExecutor:
     """Setup an experiment.
 
     Arguments:
@@ -28,7 +26,7 @@ def setup_exp(entry_dict: Dict[str, Any]) -> str:
     repo = Repo(entry.dvc_root)
     # TODO: split executor.init_cache into separate subtask - we can release
     # exp.scm_lock before DVC push
-    executor = BaseStashQueue.setup_executor(
+    executor = BaseStashQueue.init_executor(
         repo.experiments,
         entry,
         TempDirExecutor,
@@ -36,7 +34,7 @@ def setup_exp(entry_dict: Dict[str, Any]) -> str:
     )
     infofile = repo.experiments.celery_queue.get_infofile_path(entry.stash_rev)
     executor.info.dump_json(infofile)
-    return executor.root_dir
+    return executor
 
 
 @shared_task
@@ -79,25 +77,14 @@ def collect_exp(
 
 
 @shared_task
-def cleanup_exp(  # pylint: disable=unused-argument
-    tmp_dir: str, entry_dict: Dict[str, Any]
-) -> None:
+def cleanup_exp(executor: TempDirExecutor, infofile: str) -> None:
     """Cleanup after an experiment.
 
     Arguments:
         tmp_dir: Temp directory to be removed.
         entry_dict: Serialized QueueEntry for this experiment.
     """
-    remove(tmp_dir)
-
-
-def _set_collected(infofile: str):
-    try:
-        executor_info = ExecutorInfo.load_json(infofile)
-        executor_info.collected = True
-        executor_info.dump_json(infofile)
-    except FileNotFoundError:
-        pass
+    executor.cleanup(infofile)
 
 
 @shared_task
@@ -115,11 +102,10 @@ def run_exp(entry_dict: Dict[str, Any]) -> None:
     repo = Repo(entry.dvc_root)
     queue = repo.experiments.celery_queue
     infofile = queue.get_infofile_path(entry.stash_rev)
-    root_dir = setup_exp.s(entry_dict)()
+    executor = setup_exp.s(entry_dict)()
     try:
         cmd = ["dvc", "exp", "exec-run", "--infofile", infofile]
         proc_dict = queue.proc.run_signature(cmd, name=entry.stash_rev)()
         collect_exp.s(proc_dict, entry_dict)()
     finally:
-        cleanup_exp.s(root_dir, entry_dict)()
-        _set_collected(infofile)
+        cleanup_exp.s(executor, infofile)()