logger: init logging config before colorama

Closes #8376

.github/CODEOWNERS

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

.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/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: "3.10"
       - uses: iterative/dvc-bench@main
         with:
             pytest_options: "-k 'test_init or test_help'"
-

.github/workflows/packages.yaml

@@ -6,152 +6,92 @@ on:
   release:
     types: [released, prereleased]
   workflow_dispatch:
+
 jobs:
-  deb:
-    runs-on: ubuntu-18.04
+  binary:
+    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
       uses: actions/setup-python@v4
       with:
         python-version: 3.9
-    - name: Set up Ruby 2.4
-      uses: actions/[email protected]
+        cache: pip
+        cache-dependency-path: |
+          setup.cfg
+          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 -U wheel pip 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
+        cache: pip
+        cache-dependency-path: setup.cfg
+
     - name: Install
-      run: pip install ".[all,tests]"
+      run: |
+        pip install -U wheel pip build setuptools
+        pip install ".[all,tests]"
+
     - name: Build packages
       run: |
         pip install -U pip
         ./scripts/build_package.sh
+
     - name: Publish packages to PyPI
       if: github.event_name == 'release'
-      uses: pypa/gh-action-pypi-publish@master
+      uses: pypa/gh-action-pypi-publish@release/v1
       with:
-        user: __token__
         password: ${{ secrets.PYPI_TOKEN }}

.github/workflows/tests.yaml

@@ -1,25 +1,27 @@
 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
+
 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
@@ -31,7 +33,7 @@ jobs:
       run: |
         pip install --upgrade pip setuptools wheel
         pip install collective.checkdocs==0.2
-        pip install ".[dev]" pre-commit
+        pip install ".[dev]"
     - name: Check README
       run: python setup.py checkdocs
     - uses: pre-commit/[email protected]
@@ -43,6 +45,12 @@ jobs:
       matrix:
         os: [ubuntu-20.04, windows-latest, macos-latest]
         pyv: ["3.8", "3.9", "3.10"]
+        pytest-filter: 
+        - "import or plot or live or experiment"
+        - "not (import or plot or live or experiment)"
+        include:
+        - {os: ubuntu-latest, pyv: "3.11-dev"}
+
     steps:
     - uses: actions/checkout@v3
       with:
@@ -51,18 +59,8 @@ 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: setup.cfg
     - name: install
       if: steps.cache.pip-cache-dir.cache-hit != 'true'
       run: |
@@ -70,10 +68,7 @@ jobs:
         pip install -e ".[dev]"
     - name: run tests
       timeout-minutes: 40
-      run: >-
-        python -m tests -n=auto
-        --cov-report=xml --cov-report=term
-        ${{ env.extra_test_args }}
+      run: pytest -nauto -ra --durations 100 --cov=dvc --cov-report=xml --cov-report=term -k "${{ matrix.pytest-filter }}"
     - name: upload coverage report
       uses: codecov/codecov-action@v3
       with:

.pre-commit-config.yaml

@@ -6,7 +6,7 @@ repos:
       - id: black
         language_version: python3
     repo: https://github.com/ambv/black
-    rev: 22.6.0
+    rev: 22.8.0
   - repo: https://github.com/pre-commit/pre-commit-hooks
     rev: v4.3.0
     hooks:
@@ -16,9 +16,9 @@ repos:
       - id: codespell
         args: 
           - --ignore-words-list
-          - ba,datas,fo,uptodate,cachable
+          - ba,datas,fo,uptodate,cachable,falsy
     repo: https://github.com/codespell-project/codespell
-    rev: v2.1.0
+    rev: v2.2.1
   - hooks:
       - id: isort
         language_version: python3

README.rst

@@ -11,7 +11,7 @@
 • `Contributing`_
 • `Community and Support`_
 
-|CI| |Maintainability| |Coverage| |VS Code| |DOI|
+|CI| |Python Version| |Coverage| |VS Code| |DOI|
 
 |PyPI| |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

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

@@ -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,

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/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

@@ -17,9 +17,13 @@ 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,
             )
         except DvcException:
@@ -34,6 +38,8 @@ 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 +72,19 @@ 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",
@@ -94,13 +108,5 @@ def add_parser(subparsers, parent_parser):
         ),
         metavar="<number>",
     )
-    import_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(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
         )

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,
@@ -289,4 +291,9 @@ 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 +3,37 @@
 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
 
 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 +48,23 @@ 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 +81,13 @@ 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:
@@ -82,7 +108,7 @@ class DataCloud:
         **kwargs,
     ):
         from dvc.exceptions import FileTransferError
-        from dvc_data.transfer import TransferError, transfer
+        from dvc_data.hashfile.transfer import TransferError, transfer
 
         try:
             return transfer(src_odb, dest_odb, objs, **kwargs)
@@ -164,7 +190,7 @@ 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

@@ -18,7 +18,7 @@ SCHEMA: Mapping[str, Any] = {
 }
 
 
-def _get(stage, p, info):
+def _get(stage, p, info, fs_config=None):
     if info and info.get(RepoDependency.PARAM_REPO):
         repo = info.pop(RepoDependency.PARAM_REPO)
         return RepoDependency(repo, stage, p, info)
@@ -27,7 +27,7 @@ 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, fs_config=fs_config)
 
 
 def loadd_from(stage, d_list):
@@ -38,10 +38,18 @@ def loadd_from(stage, d_list):
     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/param.py

@@ -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 {}

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/exceptions.py

@@ -278,6 +278,11 @@ 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

@@ -2,6 +2,9 @@ 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
@@ -10,6 +13,8 @@ from dvc_objects.fs import (  # noqa: F401
     Schemes,
     generic,
     get_fs_cls,
+    known_implementations,
+    localfs,
     registry,
     system,
 )
@@ -19,37 +24,30 @@ 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.azure import AzureFileSystem  # noqa: F401
-from dvc_objects.fs.implementations.gdrive import (  # noqa: F401
-    GDriveFileSystem,
-)
-from dvc_objects.fs.implementations.gs import GSFileSystem  # noqa: F401
-from dvc_objects.fs.implementations.hdfs import HDFSFileSystem  # noqa: F401
-from dvc_objects.fs.implementations.local import localfs  # noqa: F401
-from dvc_objects.fs.implementations.oss import OSSFileSystem  # noqa: F401
-from dvc_objects.fs.implementations.s3 import S3FileSystem  # noqa: F401
-from dvc_objects.fs.implementations.ssh import (  # noqa: F401
-    DEFAULT_PORT as DEFAULT_SSH_PORT,
-)
-from dvc_objects.fs.implementations.ssh import SSHFileSystem  # noqa: F401
-from dvc_objects.fs.implementations.webdav import (  # noqa: F401
-    WebDAVFileSystem,
-    WebDAVSFileSystem,
-)
-from dvc_objects.fs.implementations.webhdfs import (  # noqa: F401
-    WebHDFSFileSystem,
-)
 from dvc_objects.fs.path import Path  # noqa: F401
 
 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 get_fs_config(config, **kwargs):
     name = kwargs.get("name")
     if name:
         try:
@@ -60,10 +58,10 @@ 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:
@@ -89,21 +87,22 @@ 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:
@@ -116,11 +115,8 @@ 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

@@ -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):
@@ -74,6 +77,9 @@ class RichCallback(Callback):
 
     def branch(self, path_1, path_2, kwargs, child: Optional[Callback] = None):
         child = child or RichCallback(
-            progress=self.progress, desc=path_1, bytes=True
+            progress=self.progress,
+            desc=path_1,
+            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/info.py

@@ -14,6 +14,14 @@ 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"]
     )
 
@@ -88,13 +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 in registry:
-        if scheme in [Schemes.LOCAL, Schemes.MEMORY]:
+        if scheme in [Schemes.LOCAL, Schemes.MEMORY, "dvc", "git"]:
+            continue
+
+        try:
+            fs_cls = registry[scheme]
+        except ImportError:
             continue
 
-        fs_cls = registry[scheme]
         if not fs_cls.get_missing_deps():
             dependencies = []
             for requirement in fs_cls.REQUIRES:

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
 
 

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
@@ -17,29 +17,25 @@ 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_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 .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
 
 logger = logging.getLogger(__name__)
@@ -65,10 +61,12 @@ 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 +83,10 @@ 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)
         ret.append(
             _get(
                 stage,
@@ -98,9 +97,10 @@ def loadd_from(stage, d_list):
                 plot=plot,
                 persist=persist,
                 checkpoint=checkpoint,
-                desc=desc,
                 live=live,
                 remote=remote,
+                **annot,
+                files=files,
             )
         )
     return ret
@@ -192,6 +192,7 @@ def load_from_pipeline(stage, data, typ="outs"):
                 Output.PARAM_PERSIST,
                 Output.PARAM_CHECKPOINT,
                 Output.PARAM_REMOTE,
+                *ANNOTATION_FIELDS,
             ],
         )
 
@@ -245,6 +246,7 @@ 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"
@@ -257,7 +259,6 @@ 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"
@@ -289,11 +290,29 @@ class Output:
         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,
     ):
+        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 +344,38 @@ 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.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.hash_info and self.hash_info.isdir:
+            self.meta.isdir = True
+
     def _parse_path(self, fs, fs_path):
         parsed = urlparse(self.def_path)
         if (
@@ -380,6 +416,12 @@ 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 +461,27 @@ 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 +500,17 @@ 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 changed_checksum(self):
         return self.hash_info != self.get_hash()
 
@@ -469,6 +528,11 @@ 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"}
@@ -551,11 +615,12 @@ class Output:
                 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
+            self.files = None
             if not self.IS_DEPENDENCY:
                 logger.debug(
                     "Output '%s' doesn't use cache. Skipping saving.", self
@@ -568,18 +633,19 @@ class Output:
             self.odb,
             self.fs_path,
             self.fs,
-            self.odb.fs.PARAM_CHECKSUM,
+            self.hash_name,
             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 +676,7 @@ 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 +705,7 @@ 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 +719,10 @@ 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(
@@ -668,9 +736,7 @@ class Output:
         if self.IS_DEPENDENCY:
             return ret
 
-        if self.desc:
-            ret[self.PARAM_DESC] = self.desc
-
+        ret.update(self.annot.to_dict())
         if not self.use_cache:
             ret[self.PARAM_CACHE] = self.use_cache
 
@@ -699,6 +765,15 @@ class Output:
         if self.remote:
             ret[self.PARAM_REMOTE] = self.remote
 
+        if 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
 
     def verify_metric(self):
@@ -731,14 +806,22 @@ class Output:
         ) 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
 
@@ -858,6 +941,7 @@ class Output:
         )
 
         self.hash_info = obj.hash_info
+        self.files = None
         return obj
 
     def get_files_number(self, filter_info=None):
@@ -928,6 +1012,7 @@ 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)
             )
@@ -1020,7 +1105,7 @@ class Output:
         other = out.dumpd()
 
         ignored = [
-            self.fs.PARAM_CHECKSUM,
+            self.hash_name,
             Meta.PARAM_SIZE,
             Meta.PARAM_NFILES,
         ]
@@ -1039,9 +1124,9 @@ 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 +1141,11 @@ 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 +1153,7 @@ 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),
@@ -1086,21 +1176,33 @@ class Output:
         return bool(self.plot) or bool(self.live)
 
 
+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_FILES: [DIR_FILES_SCHEMA],
 }

dvc/parsing/context.py

@@ -159,7 +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)

dvc/render/convert.py

@@ -2,7 +2,13 @@ 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
 
@@ -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/vega.py

@@ -127,13 +127,20 @@ class VegaConverter(Converter):
                 ),
             )
         )
+        self.steps.append(
+            (
+                "infer_y",
+                partial(
+                    self._infer_y,
+                ),
+            )
+        )
 
         self.steps.append(
             (
                 "generate_y",
                 partial(
                     self._generate_y_values,
-                    y_values=self.plot_properties.get("y", None),
                 ),
             )
         )
@@ -156,7 +163,7 @@ class VegaConverter(Converter):
             } - {None}
             self.inferred_properties["fields"] = fields
 
-    def _infer_y(self, datapoints: List[Dict]):
+    def _infer_y(self, datapoints: List[Dict], **kwargs):
         if "y" not in self.plot_properties:
             data_fields = list(first(datapoints))
             skip = (
@@ -178,6 +185,7 @@ class VegaConverter(Converter):
                     )
             else:
                 self.inferred_properties["y"] = inferred_y
+        return datapoints
 
     def convert(
         self,
@@ -197,7 +205,6 @@ class VegaConverter(Converter):
         processed = deepcopy(data)
 
         for step_name, step in self.steps:
-
             if step_name not in skip:
                 processed = step(  # type: ignore
                     processed,
@@ -205,25 +212,28 @@ class VegaConverter(Converter):
                     filename=filename,
                 )
 
-        self._infer_y(processed)  # type: ignore
         return processed, {**self.plot_properties, **self.inferred_properties}
 
     def _generate_y_values(  # noqa: C901
         self,
         datapoints: List[Dict],
-        y_values: Optional[Union[str, List, Dict]],
         revision: str,
         filename: str,
         **kwargs,
     ) -> List[Dict]:
 
+        y_values = self.plot_properties.get(
+            "y", None
+        ) or self.inferred_properties.get("y", None)
+
+        assert y_values is not None
+
         result = []
         properties_update = {}
 
         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):
@@ -250,12 +260,6 @@ class VegaConverter(Converter):
 
             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(
@@ -290,13 +294,15 @@ class VegaConverter(Converter):
             ):
                 # if we use the same field from all files,
                 # we dont have to generate it
+                y_field = all_fields.pop()
                 for datapoint in datapoints:
                     result.append(
                         _add_version_info(
-                            datapoint, _version_info(revision, filename)
+                            datapoint,
+                            _version_info(revision, filename, y_field),
                         )
                     )
-                properties_update.update({"y": all_fields.pop()})
+                properties_update.update({"y": y_field})
             else:
                 for def_filename, val in y_values.items():
                     if isinstance(val, str):

dvc/repo/__init__.py

@@ -3,7 +3,7 @@ import os
 from collections import defaultdict
 from contextlib import contextmanager
 from functools import wraps
-from typing import TYPE_CHECKING, Callable, Optional
+from typing import TYPE_CHECKING, Callable, List, Optional
 
 from funcy import cached_property
 
@@ -18,6 +18,7 @@ if TYPE_CHECKING:
     from dvc.fs import FileSystem
     from dvc.repo.scm_context import SCMContext
     from dvc.scm import Base
+    from dvc.stage import Stage
 
 logger = logging.getLogger(__name__)
 
@@ -67,6 +68,7 @@ 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 +81,7 @@ 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 +189,7 @@ 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)
@@ -444,6 +447,44 @@ class Repo:
 
         return used
 
+    def partial_imports(
+        self,
+        targets=None,
+        all_branches=False,
+        all_tags=False,
+        all_commits=False,
+        all_experiments=False,
+        commit_date: Optional[str] = None,
+        recursive=False,
+        revs=None,
+        num=1,
+    ) -> List["Stage"]:
+        """Get the stages related to the given target and collect dependencies
+        which are missing outputs.
+
+        This is useful to retrieve files which have been imported to the repo
+        using --no-download.
+
+        Returns:
+            A list of partially imported stages
+        """
+        from itertools import chain
+
+        partial_imports = chain.from_iterable(
+            self.index.partial_imports(targets, recursive=recursive)
+            for _ in self.brancher(
+                revs=revs,
+                all_branches=all_branches,
+                all_tags=all_tags,
+                all_commits=all_commits,
+                all_experiments=all_experiments,
+                commit_date=commit_date,
+                num=num,
+            )
+        )
+
+        return list(partial_imports)
+
     @property
     def stages(self):  # obsolete, only for backward-compatibility
         return self.index.stages
@@ -487,9 +528,9 @@ 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 +542,13 @@ 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

@@ -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/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/data.py

@@ -1,7 +1,21 @@
 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,
+    cast,
+)
 
+from dvc.fs.git import GitFileSystem
 from dvc.ui import ui
 
 if TYPE_CHECKING:
@@ -10,19 +24,20 @@ if TYPE_CHECKING:
     from dvc.output import Output
     from dvc.repo import Repo
     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(
@@ -62,19 +76,12 @@ 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
 
-    drop_root = False
-    trees = isinstance(old_obj, Tree) or isinstance(new_obj, Tree)
-    if trees:
-        drop_root = not with_dirs
-
-    def path_join(root: str, *paths: str) -> str:
-        if not trees and paths == ROOT:
+    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])
 
@@ -83,38 +90,46 @@ def _granular_diff(
     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):
@@ -141,6 +156,8 @@ 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,
@@ -152,7 +169,7 @@ 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:
@@ -175,8 +192,18 @@ 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
@@ -188,7 +215,9 @@ 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:
@@ -197,15 +226,20 @@ 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
@@ -216,28 +250,36 @@ 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)
@@ -265,3 +307,18 @@ 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

@@ -16,7 +16,7 @@ from .exceptions import (
     InvalidExpRefError,
     MultipleBranchError,
 )
-from .executor.base import BaseExecutor, ExecutorInfo
+from .executor.base import BaseExecutor, ExecutorInfo, TaskStatus
 from .queue.base import BaseStashQueue, QueueEntry
 from .queue.celery import LocalCeleryQueue
 from .queue.tempdir import TempDirQueue
@@ -463,6 +463,8 @@ class Experiments:
             if rev == "workspace":
                 # If we are appending to a checkpoint branch in a workspace
                 # run, show the latest checkpoint as running.
+                if info.status > TaskStatus.RUNNING:
+                    return result
                 last_rev = self.scm.get_ref(EXEC_BRANCH)
                 if last_rev:
                     result[last_rev] = info.asdict()

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,
@@ -35,6 +36,8 @@ from ..refs import (
     EXEC_BASELINE,
     EXEC_BRANCH,
     EXEC_CHECKPOINT,
+    EXEC_HEAD,
+    EXEC_MERGE,
     EXEC_NAMESPACE,
     EXPS_NAMESPACE,
     EXPS_STASH,
@@ -47,10 +50,9 @@ 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 +66,16 @@ 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 +85,16 @@ 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 +151,7 @@ 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 +165,17 @@ 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 +215,7 @@ class BaseExecutor(ABC):
             dvc_dir=self.dvc_dir,
             name=self.name,
             wdir=self.wdir,
+            status=self.status,
             **result_dict,
         )
 
@@ -212,6 +237,7 @@ 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 +248,6 @@ class BaseExecutor(ABC):
     def from_stash_entry(
         cls: Type[_T],
         repo: "Repo",
-        stash_rev: str,
         entry: "ExpStashEntry",
         **kwargs,
     ) -> _T:
@@ -232,7 +257,6 @@ class BaseExecutor(ABC):
     def _from_stash_entry(
         cls: Type[_T],
         repo: "Repo",
-        stash_rev: str,
         entry: "ExpStashEntry",
         root_dir: str,
         **kwargs,
@@ -241,12 +265,11 @@ 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 +282,12 @@ 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
@@ -324,11 +351,12 @@ 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
 
@@ -411,11 +439,9 @@ 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 +517,6 @@ 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 +573,7 @@ class BaseExecutor(ABC):
     def _repro_dvc(
         cls,
         info: "ExecutorInfo",
+        infofile: str = None,
         log_errors: bool = True,
         **kwargs,
     ):
@@ -557,6 +581,9 @@ 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()
@@ -568,15 +595,28 @@ class BaseExecutor(ABC):
         try:
             logger.debug("Running repro in '%s'", os.getcwd())
             yield dvc
+            info.status = TaskStatus.SUCCESS
+            if infofile is not None:
+                info.dump_json(infofile)
+
         except CheckpointKilledError:
+            info.status = TaskStatus.FAILED
+            if infofile is not None:
+                info.dump_json(infofile)
             raise
         except DvcException:
             if log_errors:
                 logger.exception("")
+            info.status = TaskStatus.FAILED
+            if infofile is not None:
+                info.dump_json(infofile)
             raise
         except Exception:
             if log_errors:
                 logger.exception("unexpected error")
+            info.status = TaskStatus.FAILED
+            if infofile is not None:
+                info.dump_json(infofile)
             raise
         finally:
             dvc.close()
@@ -714,3 +754,23 @@ 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
@@ -46,8 +46,8 @@ class BaseLocalExecutor(BaseExecutor):
     def scm(self):
         return SCM(self.root_dir)
 
-    def cleanup(self):
-        super().cleanup()
+    def cleanup(self, infofile: str):
+        super().cleanup(infofile)
         self.scm.close()
         del self.scm
 
@@ -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/queue/base.py

@@ -1,5 +1,6 @@
 import logging
 import os
+import sys
 from abc import ABC, abstractmethod
 from dataclasses import asdict, dataclass
 from typing import (
@@ -19,6 +20,7 @@ from typing import (
 
 from funcy import cached_property
 
+from dvc.dependency import ParamsDependency
 from dvc.env import DVCLIVE_RESUME
 from dvc.exceptions import DvcException
 from dvc.ui import ui
@@ -31,7 +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
 
@@ -520,14 +522,36 @@ class BaseStashQueue(ABC):
                 provided via `exp run --set-param`.
 
         .. _Hydra Override:
-            https://hydra.cc/docs/next/advanced/override_grammar/basic/
+            https://hydra.cc/docs/advanced/override_grammar/basic/
         """
         logger.debug("Using experiment params '%s'", params)
 
-        from dvc.utils.hydra import apply_overrides
+        try:
+            from dvc.utils.hydra import apply_overrides, compose_and_dump
+        except ValueError:
+            if sys.version_info >= (3, 11):
+                raise DvcException(
+                    "--set-param is not supported in Python >= 3.11"
+                )
+            raise
 
+        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():
-            apply_overrides(path, overrides)
+            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
@@ -537,7 +561,7 @@ class BaseStashQueue(ABC):
 
     @staticmethod
     @scm_locked
-    def setup_executor(
+    def init_executor(
         exp: "Experiments",
         queue_entry: QueueEntry,
         executor_cls: Type[BaseExecutor] = WorkspaceExecutor,
@@ -552,21 +576,23 @@ 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,

dvc/repo/experiments/queue/celery.py

@@ -23,7 +23,12 @@ from dvc.exceptions import DvcException
 from dvc.ui import ui
 
 from ..exceptions import UnresolvedQueueExpNamesError
-from ..executor.base import EXEC_TMP_DIR, ExecutorInfo, ExecutorResult
+from ..executor.base import (
+    EXEC_TMP_DIR,
+    ExecutorInfo,
+    ExecutorResult,
+    TaskStatus,
+)
 from .base import BaseStashQueue, QueueDoneResult, QueueEntry, QueueGetResult
 from .tasks import run_exp
 
@@ -248,7 +253,7 @@ class LocalCeleryQueue(BaseStashQueue):
 
         def _load_collected(rev: str) -> Optional[ExecutorResult]:
             executor_info = _load_info(rev)
-            if executor_info.collected:
+            if executor_info.status > TaskStatus.SUCCESS:
                 return executor_info.result
             raise FileNotFoundError
 

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)()

dvc/repo/experiments/queue/tempdir.py

@@ -11,6 +11,7 @@ from ..executor.base import (
     BaseExecutor,
     ExecutorInfo,
     ExecutorResult,
+    TaskStatus,
 )
 from ..executor.local import TempDirExecutor
 from .base import BaseStashQueue, QueueEntry, QueueGetResult
@@ -60,7 +61,7 @@ class TempDirQueue(WorkspaceQueue):
             stash_entry.name,
             stash_entry.head_rev,
         )
-        executor = self.setup_executor(
+        executor = self.init_executor(
             self.repo.experiments,
             entry,
             TempDirExecutor,
@@ -75,8 +76,9 @@ class TempDirQueue(WorkspaceQueue):
         for stash_rev in self.proc:
             infofile = self.get_infofile_path(stash_rev)
             executor_info = ExecutorInfo.load_json(infofile)
-            if not executor_info.collected and os.path.exists(
-                executor_info.root_dir
+            if (
+                not executor_info.status <= TaskStatus.SUCCESS
+                and os.path.exists(executor_info.root_dir)
             ):
                 yield QueueEntry(
                     self.repo.root_dir,

dvc/repo/experiments/queue/workspace.py

@@ -1,12 +1,10 @@
 import logging
-import os
 from collections import defaultdict
 from typing import TYPE_CHECKING, Collection, Dict, Generator, Optional
 
 from funcy import first
 
 from dvc.exceptions import DvcException
-from dvc.utils.fs import remove
 
 from ..exceptions import ExpQueueEmptyError
 from ..executor.base import BaseExecutor, ExecutorResult
@@ -41,7 +39,7 @@ class WorkspaceQueue(BaseStashQueue):
             stash_entry.name,
             stash_entry.head_rev,
         )
-        executor = self.setup_executor(self.repo.experiments, entry)
+        executor = self.init_executor(self.repo.experiments, entry)
         return QueueGetResult(entry, executor)
 
     def iter_queued(self) -> Generator[QueueEntry, None, None]:
@@ -118,9 +116,7 @@ class WorkspaceQueue(BaseStashQueue):
                 f"Failed to reproduce experiment '{rev[:7]}'"
             ) from exc
         finally:
-            if self._EXEC_NAME == exec_name:
-                remove(os.path.join(self.pid_dir, exec_name))
-            executor.cleanup()
+            executor.cleanup(infofile)
         return results
 
     @staticmethod
@@ -135,6 +131,7 @@ class WorkspaceQueue(BaseStashQueue):
             assert exec_result.exp_hash
             logger.debug("Collected experiment '%s'.", exp_rev[:7])
             results[exp_rev] = exec_result.exp_hash
+
         return results
 
     def get_result(self, entry: QueueEntry) -> Optional[ExecutorResult]:

dvc/repo/experiments/run.py

@@ -1,6 +1,8 @@
 import logging
 from typing import Dict, Iterable, Optional
 
+from dvc.dependency.param import ParamsDependency
+from dvc.exceptions import InvalidArgumentError
 from dvc.repo import locked
 from dvc.ui import ui
 from dvc.utils.cli_parse import to_path_overrides
@@ -26,26 +28,59 @@ def run(
     Returns a dict mapping new experiment SHAs to the results
     of `repro` for that experiment.
     """
+    from dvc.utils.hydra import to_hydra_overrides
+
     if run_all:
         entries = list(repo.experiments.celery_queue.iter_queued())
         return repo.experiments.reproduce_celery(entries, jobs=jobs)
 
     if params:
-        params = to_path_overrides(params)
+        path_overrides = to_path_overrides(params)
+    else:
+        path_overrides = {}
 
-    if queue:
-        if not kwargs.get("checkpoint_resume", None):
-            kwargs["reset"] = True
+    hydra_enabled = repo.config.get("hydra", {}).get("enabled", False)
+    hydra_output_file = ParamsDependency.DEFAULT_PARAMS_FILE
+    if hydra_enabled and hydra_output_file not in path_overrides:
+        # Force `_update_params` even if `--set-param` was not used
+        path_overrides[hydra_output_file] = []
+
+    hydra_sweep = any(
+        x.is_sweep_override()
+        for param_file in path_overrides
+        for x in to_hydra_overrides(path_overrides[param_file])
+    )
+
+    if hydra_sweep and not queue:
+        raise InvalidArgumentError(
+            "Sweep overrides can't be used without `--queue`"
+        )
+
+    if not queue:
+        return repo.experiments.reproduce_one(
+            targets=targets, params=path_overrides, tmp_dir=tmp_dir, **kwargs
+        )
+
+    if hydra_sweep:
+        from dvc.utils.hydra import get_hydra_sweeps
+
+        sweeps = get_hydra_sweeps(path_overrides)
+    else:
+        sweeps = [path_overrides]
+
+    if not kwargs.get("checkpoint_resume", None):
+        kwargs["reset"] = True
+
+    for sweep_overrides in sweeps:
         queue_entry = repo.experiments.queue_one(
             repo.experiments.celery_queue,
             targets=targets,
-            params=params,
+            params=sweep_overrides,
             **kwargs,
         )
+        if sweep_overrides:
+            ui.write(f"Queueing with overrides '{sweep_overrides}'.")
         name = queue_entry.name or queue_entry.stash_rev[:7]
         ui.write(f"Queued experiment '{name}' for future execution.")
-        return {}
 
-    return repo.experiments.reproduce_one(
-        targets=targets, params=params, tmp_dir=tmp_dir, **kwargs
-    )
+    return {}

dvc/repo/experiments/show.py

@@ -142,6 +142,9 @@ def show(
     fetch_running: bool = True,
 ):
 
+    if repo.scm.no_commits:
+        return {}
+
     if onerror is None:
         onerror = onerror_collect
 

dvc/repo/fetch.py

@@ -1,6 +1,8 @@
 import logging
+from contextlib import suppress
 from typing import TYPE_CHECKING, Optional
 
+from dvc.config import NoRemoteError
 from dvc.exceptions import DownloadError, FileTransferError
 from dvc.fs import Schemes
 
@@ -12,6 +14,20 @@ if TYPE_CHECKING:
 logger = logging.getLogger(__name__)
 
 
+def _fetch_worktree(repo, remote):
+    from dvc_data.index import save
+
+    index = repo.index.data["repo"]
+    for key, entry in index.iteritems():
+        entry.fs = remote.fs
+        entry.path = remote.fs.path.join(
+            remote.path,
+            *key,
+        )
+    save(index)
+    return len(index)
+
+
 @locked
 def fetch(
     self,
@@ -42,6 +58,11 @@ def fetch(
     if isinstance(targets, str):
         targets = [targets]
 
+    with suppress(NoRemoteError):
+        _remote = self.cloud.get_remote(name=remote)
+        if _remote.worktree:
+            return _fetch_worktree(self, _remote)
+
     used = self.used_objs(
         targets,
         all_branches=all_branches,
@@ -93,6 +114,18 @@ def fetch(
             downloaded += d
             failed += f
 
+    d, f = _fetch_partial_imports(
+        self,
+        targets,
+        all_branches=all_branches,
+        all_tags=all_tags,
+        all_commits=all_commits,
+        recursive=recursive,
+        revs=revs,
+    )
+    downloaded += d
+    failed += f
+
     if failed:
         raise DownloadError(failed)
 
@@ -107,3 +140,39 @@ def _fetch(repo, obj_ids, **kwargs):
     except FileTransferError as exc:
         failed += exc.amount
     return downloaded, failed
+
+
+def _fetch_partial_imports(repo, targets, **kwargs):
+    from dvc.stage.exceptions import DataSourceChanged
+
+    downloaded = 0
+    failed = 0
+    for stage in repo.partial_imports(targets, **kwargs):
+        dep = stage.deps[0]
+        out = stage.outs[0]
+        try:
+            if dep.changed_checksum():
+                raise DataSourceChanged(f"{stage} ({dep})")
+
+            logger.info("Importing '%s' -> '%s'", dep, out)
+            if stage.is_repo_import:
+                obj = dep.get_obj()
+                meta = dep.get_meta()
+                out.hash_info = obj.hash_info
+                out.meta = meta
+            else:
+                out.transfer(
+                    dep.def_path, jobs=kwargs.get("jobs"), odb=repo.odb.local
+                )
+        except DataSourceChanged as exc:
+            logger.warning(f"{exc}")
+            failed += 1
+            continue
+        if not any(
+            kwargs.get(kw, None)
+            for kw in ("all_branches", "all_tags", "all_commits", "revs")
+        ):
+            stage.dump()
+
+        downloaded += 1
+    return downloaded, failed

dvc/repo/gc.py

@@ -60,8 +60,8 @@ def gc(
     from contextlib import ExitStack
 
     from dvc.repo import Repo
-    from dvc_data.db import get_index
-    from dvc_data.gc import gc as ogc
+    from dvc_data.hashfile.db import get_index
+    from dvc_data.hashfile.gc import gc as ogc
 
     if not repos:
         repos = []

dvc/repo/imp_url.py

@@ -17,11 +17,16 @@ def imp_url(
     fname=None,
     erepo=None,
     frozen=True,
+    no_download=False,
     no_exec=False,
     remote=None,
     to_remote=False,
     desc=None,
+    type=None,  # pylint: disable=redefined-builtin
+    labels=None,
+    meta=None,
     jobs=None,
+    fs_config=None,
 ):
     from dvc.dvcfile import Dvcfile
     from dvc.stage import Stage, create_stage, restore_fields
@@ -31,9 +36,9 @@ def imp_url(
         self, out, always_local=to_remote and not out
     )
 
-    if to_remote and no_exec:
+    if to_remote and (no_exec or no_download):
         raise InvalidArgumentError(
-            "--no-exec can't be combined with --to-remote"
+            "--no-exec/--no-download cannot be combined with --to-remote"
         )
 
     if not to_remote and remote:
@@ -57,12 +62,12 @@ def imp_url(
         deps=[url],
         outs=[out],
         erepo=erepo,
+        fs_config=fs_config,
     )
     restore_fields(stage)
 
-    if desc:
-        stage.outs[0].desc = desc
-
+    out_obj = stage.outs[0]
+    out_obj.annot.update(desc=desc, type=type, labels=labels, meta=meta)
     dvcfile = Dvcfile(self, stage.path)
     dvcfile.remove()
 
@@ -80,7 +85,7 @@ def imp_url(
         stage.save_deps()
         stage.md5 = stage.compute_md5()
     else:
-        stage.run(jobs=jobs)
+        stage.run(jobs=jobs, no_download=no_download)
 
     stage.frozen = frozen
 

dvc/repo/index.py

@@ -183,15 +183,7 @@ class Index:
             if not out.use_cache:
                 continue
 
-            if out.is_in_repo:
-                workspace = "repo"
-                key = self.repo.fs.path.relparts(
-                    out.fs_path, self.repo.root_dir
-                )
-            else:
-                workspace = out.fs.protocol
-                no_drive = out.fs.path.flavour.splitdrive(out.fs_path)[1]
-                key = out.fs.path.parts(no_drive)[1:]
+            workspace, key = out.index_key
 
             try:
                 remote = self.repo.cloud.get_remote_odb(out.remote)
@@ -200,11 +192,15 @@ class Index:
 
             data_index = by_workspace[workspace]
 
+            if out.files:
+                out.obj = out.get_obj()
+
             data_index[key] = DataIndexEntry(
                 meta=out.meta,
                 obj=out.obj,
                 hash_info=out.hash_info,
                 odb=out.odb,
+                cache=out.odb,
                 remote=remote,
             )
 
@@ -246,6 +242,26 @@ class Index:
                 used[odb].update(objs)
         return used
 
+    def partial_imports(
+        self,
+        targets: "TargetType" = None,
+        recursive: bool = False,
+    ) -> List["Stage"]:
+        from itertools import chain
+
+        from dvc.utils.collections import ensure_list
+
+        collect_targets: Sequence[Optional[str]] = (None,)
+        if targets:
+            collect_targets = ensure_list(targets)
+
+        pairs = chain.from_iterable(
+            self.stage_collector.collect_granular(target, recursive=recursive)
+            for target in collect_targets
+        )
+
+        return [stage for stage, _ in pairs if stage.is_partial_import]
+
     # Following methods help us treat the collection as a set-like structure
     # and provides faux-immutability.
     # These methods do not preserve stages order.

dvc/repo/ls.py

@@ -1,11 +1,10 @@
 import os
-from itertools import chain
 from typing import TYPE_CHECKING, Optional
 
 from dvc.exceptions import PathMissingError
 
 if TYPE_CHECKING:
-    from dvc.fs.dvc import DvcFileSystem
+    from dvc.fs.dvc import DVCFileSystem
 
     from . import Repo
 
@@ -56,7 +55,7 @@ def ls(
 def _ls(
     repo: "Repo", path: str, recursive: bool = None, dvc_only: bool = False
 ):
-    fs: "DvcFileSystem" = repo.dvcfs
+    fs: "DVCFileSystem" = repo.dvcfs
     fs_path = fs.from_os_path(path)
 
     try:
@@ -66,15 +65,16 @@ def _ls(
 
     infos = {}
     for root, dirs, files in fs.walk(
-        fs_path, dvcfiles=True, dvc_only=dvc_only
+        fs_path, dvcfiles=True, dvc_only=dvc_only, detail=True
     ):
-        entries = chain(files, dirs) if not recursive else files
+        if not recursive:
+            files.update(dirs)
 
-        for entry in entries:
-            entry_fs_path = fs.path.join(root, entry)
+        for name, entry in files.items():
+            entry_fs_path = fs.path.join(root, name)
             relparts = fs.path.relparts(entry_fs_path, fs_path)
             name = os.path.join(*relparts)
-            infos[name] = fs.info(entry_fs_path)
+            infos[name] = entry
 
         if not recursive:
             break
@@ -85,11 +85,10 @@ def _ls(
     ret = {}
     for name, info in infos.items():
         dvc_info = info.get("dvc_info", {})
-        if fs.isdvc(info["name"], recursive=True) or not dvc_only:
-            ret[name] = {
-                "isout": dvc_info.get("isout", False),
-                "isdir": info["type"] == "directory",
-                "isexec": info.get("isexec", False),
-            }
+        ret[name] = {
+            "isout": dvc_info.get("isout", False),
+            "isdir": info["type"] == "directory",
+            "isexec": info.get("isexec", False),
+        }
 
     return ret

dvc/repo/ls_url.py

@@ -0,0 +1,37 @@
+from dvc.exceptions import URLMissingError
+from dvc.fs import get_cloud_fs
+
+
+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 ls_url(url, *, config=None, recursive=False):
+    fs, fs_path = parse_external_url(url, config=config)
+    try:
+        info = fs.info(fs_path)
+    except FileNotFoundError as exc:
+        raise URLMissingError(url) from exc
+    if info["type"] != "directory":
+        return [{"path": info["name"], "isdir": False}]
+
+    ret = []
+    for _, dirs, files in fs.walk(fs_path, detail=True):
+        if not recursive:
+            files.update(dirs)
+
+        for info in files.values():
+            ls_info = {
+                "path": fs.path.relpath(info["name"], fs_path),
+                "isdir": info["type"] == "directory",
+            }
+            ret.append(ls_info)
+
+        if not recursive:
+            break
+
+    return ret

dvc/repo/metrics/show.py

@@ -4,7 +4,7 @@ from typing import List
 
 from scmrepo.exceptions import SCMError
 
-from dvc.fs.dvc import DvcFileSystem
+from dvc.fs.dvc import DVCFileSystem
 from dvc.output import Output
 from dvc.repo import locked
 from dvc.repo.collect import StrPaths, collect
@@ -77,7 +77,7 @@ def _read_metric(path, fs, rev, **kwargs):
 
 
 def _read_metrics(repo, metrics, rev, onerror=None):
-    fs = DvcFileSystem(repo=repo)
+    fs = DVCFileSystem(repo=repo)
 
     relpath = ""
     if repo.root_dir != repo.fs.path.getcwd():
@@ -85,12 +85,14 @@ def _read_metrics(repo, metrics, rev, onerror=None):
 
     res = {}
     for metric in metrics:
+        rel_metric_path = os.path.join(relpath, *fs.path.parts(metric))
         if not fs.isfile(metric):
-            continue
+            if fs.isfile(rel_metric_path):
+                metric = rel_metric_path
+            else:
+                continue
 
-        res[os.path.join(relpath, *fs.path.parts(metric))] = _read_metric(
-            metric, fs, rev, onerror=onerror
-        )
+        res[rel_metric_path] = _read_metric(metric, fs, rev, onerror=onerror)
 
     return res
 

dvc/repo/params/show.py

@@ -56,7 +56,9 @@ def _collect_configs(
             default_params
         ):
             fs_paths.append(default_params)
-
+    if targets and (deps or stages) and not params:
+        # A target has been provided but it is not used in the stages
+        fs_paths = []
     return params, fs_paths
 
 

dvc/repo/plots/__init__.py

@@ -100,7 +100,7 @@ class Plots:
                             "filename":
                             {
                                 "data_source": callable loading the data,
-                                "props": propreties for the file if it is
+                                "props": properties for the file if it is
                                          plots type output
                             }
                         }
@@ -152,9 +152,9 @@ class Plots:
         props: Optional[Dict] = None,
         onerror: Optional[Callable] = None,
     ):
-        from dvc.fs.dvc import DvcFileSystem
+        from dvc.fs.dvc import DVCFileSystem
 
-        fs = DvcFileSystem(repo=self.repo)
+        fs = DVCFileSystem(repo=self.repo)
 
         props = props or {}
 
@@ -386,30 +386,56 @@ def _collect_output_plots(
     return result
 
 
-def _adjust_definitions_to_cwd(fs, config_relpath, plots_definitions):
-    result = defaultdict(dict)
+def _id_is_path(plot_props=None):
+    if not plot_props:
+        return True
 
-    config_dirname = fs.path.dirname(config_relpath)
+    y_def = plot_props.get("y")
+    return not isinstance(y_def, dict)
 
-    for plot_id, plot_def in plots_definitions.items():
 
-        y_def = plot_def.get("y", None) if plot_def else None
-        if y_def is None or not isinstance(y_def, dict):
-            # plot_id is filename
-            new_plot_id = _normpath(fs.path.join(config_dirname, plot_id))
-            result[new_plot_id] = plot_def or {}
+def _adjust_sources(fs, plot_props, config_dir):
+    new_plot_props = deepcopy(plot_props)
+    old_y = new_plot_props.pop("y", {})
+    new_y = {}
+    for filepath, val in old_y.items():
+        new_y[_normpath(fs.path.join(config_dir, filepath))] = val
+    new_plot_props["y"] = new_y
+    return new_plot_props
+
+
+def _resolve_definitions(
+    fs, targets, props, config_path, definitions, onerror=None
+):
+    config_dir = fs.path.dirname(config_path)
+    result: Dict[str, Dict] = {}
+    for plot_id, plot_props in definitions.items():
+        if plot_props is None:
+            plot_props = {}
+        if _id_is_path(plot_props):
+            data_path = _normpath(fs.path.join(config_dir, plot_id))
+            if _matches(targets, config_path, plot_id):
+                unpacked = unpack_if_dir(
+                    fs,
+                    data_path,
+                    props={**plot_props, **props},
+                    onerror=onerror,
+                )
+                dpath.util.merge(
+                    result,
+                    unpacked,
+                )
         else:
-            new_plot_def = deepcopy(plot_def)
-            old_y = new_plot_def.pop("y")
-            new_y = {}
-            for filepath, val in old_y.items():
-                new_y[_normpath(fs.path.join(config_dirname, filepath))] = val
-            new_plot_def["y"] = new_y
-            result[plot_id] = new_plot_def
-    return dict(result)
+            if _matches(targets, config_path, plot_id):
+                adjusted_props = _adjust_sources(fs, plot_props, config_dir)
+                dpath.util.merge(
+                    result, {"data": {plot_id: {**adjusted_props, **props}}}
+                )
+
+    return result
 
 
-def _collect_pipeline_files(repo, targets: List[str], props):
+def _collect_pipeline_files(repo, targets: List[str], props, onerror=None):
     from dvc.dvcfile import PipelineFile
 
     result: Dict[str, Dict] = {}
@@ -417,21 +443,19 @@ def _collect_pipeline_files(repo, targets: List[str], props):
     for dvcfile in dvcfiles:
         if isinstance(dvcfile, PipelineFile):
             dvcfile_path = _relpath(repo.dvcfs, dvcfile.path)
-            dvcfile_defs = _adjust_definitions_to_cwd(
-                repo.fs, dvcfile_path, dvcfile.load().get("plots", {})
+            dvcfile_defs = dvcfile.load().get("plots", {})
+            resolved = _resolve_definitions(
+                repo.dvcfs,
+                targets,
+                props,
+                dvcfile_path,
+                dvcfile_defs,
+                onerror=onerror,
+            )
+            dpath.util.merge(
+                result,
+                {dvcfile_path: resolved},
             )
-            for plot_id, plot_props in dvcfile_defs.items():
-                if plot_props is None:
-                    plot_props = {}
-                if _matches(targets, dvcfile_path, plot_id):
-                    dpath.util.merge(
-                        result,
-                        {
-                            dvcfile_path: {
-                                "data": {plot_id: {**plot_props, **props}}
-                            }
-                        },
-                    )
     return result
 
 
@@ -448,12 +472,15 @@ def _collect_definitions(
     result: Dict = defaultdict(dict)
     props = props or {}
 
-    from dvc.fs.dvc import DvcFileSystem
+    from dvc.fs.dvc import DVCFileSystem
 
-    fs = DvcFileSystem(repo=repo)
+    fs = DVCFileSystem(repo=repo)
 
     if not config_files:
-        dpath.util.merge(result, _collect_pipeline_files(repo, targets, props))
+        dpath.util.merge(
+            result,
+            _collect_pipeline_files(repo, targets, props, onerror=onerror),
+        )
 
     if targets or (not targets and not config_files):
         dpath.util.merge(
@@ -463,12 +490,20 @@ def _collect_definitions(
 
     if config_files:
         for path in config_files:
-            definitions = parse(fs, path)
-            definitions = _adjust_definitions_to_cwd(
-                repo.fs, path, definitions
-            )
+            definitions = parse(fs, path).get("data", {})
             if definitions:
-                dpath.util.merge(result, {path: definitions})
+                resolved = _resolve_definitions(
+                    repo.dvcfs,
+                    targets,
+                    props,
+                    path,
+                    definitions,
+                    onerror=onerror,
+                )
+                dpath.util.merge(
+                    result,
+                    {path: resolved},
+                )
 
     for target in targets:
         if not result or fs.exists(target):

dvc/repo/plots/diff.py

@@ -12,6 +12,8 @@ def _revisions(repo, revs, experiment):
 
 
 def diff(repo, *args, revs=None, experiment=False, **kwargs):
+    if repo.scm.no_commits:
+        return {}
     return repo.plots.show(
         *args, revs=_revisions(repo, revs, experiment), **kwargs
     )

dvc/repo/push.py

@@ -9,6 +9,49 @@ if TYPE_CHECKING:
     from dvc_objects.db.base import ObjectDB
 
 
+def _push_worktree(repo, remote):
+    from dvc_data.hashfile.tree import tree_from_index
+    from dvc_data.index import checkout
+
+    index = repo.index.data["repo"]
+    checkout(index, remote.path, remote.fs)
+
+    for stage in repo.index.stages:
+        for out in stage.outs:
+            if not out.use_cache:
+                continue
+
+            if not out.is_in_repo:
+                continue
+
+            workspace, key = out.index_key
+            index = repo.index.data[workspace]
+            entry = index[key]
+            if out.isdir():
+                old_tree = out.get_obj()
+                entry.hash_info = old_tree.hash_info
+                entry.meta = out.meta
+                for subkey, entry in index.iteritems(key):
+                    if entry.meta.isdir:
+                        continue
+                    fs_path = repo.fs.path.join(repo.root_dir, *subkey)
+                    _, hash_info = old_tree.get(
+                        repo.fs.path.relparts(fs_path, out.fs_path)
+                    )
+                    entry.hash_info = hash_info
+                tree_meta, new_tree = tree_from_index(index, key)
+                new_tree.digest()
+                out.obj = new_tree
+                out.hash_info = new_tree.hash_info
+                out.meta = tree_meta
+            else:
+                out.hash_info = entry.hash_info
+                out.meta = entry.meta
+        stage.dvcfile.dump(stage, with_files=True, update_pipeline=False)
+
+    return len(index)
+
+
 @locked
 def push(
     self,
@@ -26,6 +69,10 @@ def push(
     odb: Optional["ObjectDB"] = None,
     include_imports=False,
 ):
+    _remote = self.cloud.get_remote(name=remote)
+    if _remote.worktree:
+        return _push_worktree(self, _remote)
+
     used_run_cache = (
         self.stage_cache.push(remote, odb=odb) if run_cache else []
     )

dvc/repo/remove.py

@@ -1,6 +1,13 @@
 import logging
 import typing
 
+from dvc.dvcfile import DVC_FILE_SUFFIX
+from dvc.stage.exceptions import (
+    StageFileDoesNotExistError,
+    StageFileIsNotDvcFileError,
+    StageNotFound,
+)
+
 from . import locked
 
 if typing.TYPE_CHECKING:
@@ -11,7 +18,16 @@ logger = logging.getLogger(__name__)
 
 @locked
 def remove(self: "Repo", target: str, outs: bool = False):
-    stages = self.stage.from_target(target)
+    try:
+        stages = self.stage.from_target(target, accept_group=False)
+    except (StageNotFound, StageFileDoesNotExistError) as e:
+        # If the user specified a tracked file as a target instead of a stage,
+        # e.g. `data.csv` instead of `data.csv.dvc`,
+        # give a more helpful error message.
+        if self.fs.exists(target + DVC_FILE_SUFFIX):
+            raise StageFileIsNotDvcFileError(target) from e
+        else:
+            raise
 
     for stage in stages:
         stage.remove(remove_outs=outs, force=outs)

dvc/repo/reproduce.py

@@ -102,7 +102,6 @@ def reproduce(
     from .graph import get_pipeline, get_pipelines
 
     glob = kwargs.pop("glob", False)
-    accept_group = not glob
 
     if isinstance(targets, str):
         targets = [targets]
@@ -137,7 +136,6 @@ def reproduce(
                 self.stage.collect(
                     target,
                     recursive=recursive,
-                    accept_group=accept_group,
                     glob=glob,
                 )
             )

dvc/repo/stage.py

@@ -82,7 +82,6 @@ def _collect_specific_target(
     target: str,
     with_deps: bool,
     recursive: bool,
-    accept_group: bool,
 ) -> Tuple[StageIter, "OptStr", "OptStr"]:
     from dvc.dvcfile import is_valid_filename
 
@@ -98,13 +97,11 @@ def _collect_specific_target(
         msg = "Checking if stage '%s' is in '%s'"
         logger.debug(msg, target, PIPELINE_FILE)
         if not (recursive and loader.fs.isdir(target)):
-            stages = _maybe_collect_from_dvc_yaml(
-                loader, target, with_deps, accept_group=accept_group
-            )
+            stages = _maybe_collect_from_dvc_yaml(loader, target, with_deps)
             if stages:
                 return stages, file, name
     elif not with_deps and is_valid_filename(file):
-        stages = loader.load_all(file, name, accept_group=accept_group)
+        stages = loader.load_all(file, name)
         return stages, file, name
     return [], file, name
 
@@ -209,7 +206,7 @@ class StageLoad:
         return stage
 
     def from_target(
-        self, target: str, accept_group: bool = False, glob: bool = False
+        self, target: str, accept_group: bool = True, glob: bool = False
     ) -> StageList:
         """
         Returns a list of stage from the provided target.
@@ -228,10 +225,9 @@ class StageLoad:
         path, name = parse_target(target)
         return self.load_one(path=path, name=name)
 
-    @staticmethod
-    def _get_filepath(path: str = None, name: str = None) -> str:
+    def _get_filepath(self, path: str = None, name: str = None) -> str:
         if path:
-            return path
+            return self.repo.fs.path.realpath(path)
 
         path = PIPELINE_FILE
         logger.debug("Assuming '%s' to be a stage inside '%s'", name, path)
@@ -250,15 +246,12 @@ class StageLoad:
         self,
         stages: "StageLoader",
         name: str = None,
-        accept_group: bool = False,
+        accept_group: bool = True,
         glob: bool = False,
     ) -> Iterable[str]:
 
-        assert not (accept_group and glob)
-
         if not name:
             return stages.keys()
-
         if accept_group and stages.is_foreach_generated(name):
             return self._get_group_keys(stages, name)
         if glob:
@@ -269,7 +262,7 @@ class StageLoad:
         self,
         path: str = None,
         name: str = None,
-        accept_group: bool = False,
+        accept_group: bool = True,
         glob: bool = False,
     ) -> StageList:
         """Load a list of stages from a file.
@@ -338,7 +331,6 @@ class StageLoad:
         with_deps: bool = False,
         recursive: bool = False,
         graph: "DiGraph" = None,
-        accept_group: bool = False,
         glob: bool = False,
     ) -> StageIter:
         """Collect list of stages from the provided target.
@@ -347,15 +339,14 @@ class StageLoad:
             target: if not provided, all of the stages in the graph are
                 returned.
                 Target can be:
-                - a stage name in the `dvc.yaml` file.
+                - a foreach group name or a stage name in the `dvc.yaml` file.
+                - a generated stage name from a foreach group.
                 - a path to `dvc.yaml` or `.dvc` file.
                 - in case of a stage to a dvc.yaml file in a different
                   directory than current working directory, it can be a path
                   to dvc.yaml file, followed by a colon `:`, followed by stage
                   name (eg: `../dvc.yaml:build`).
                 - in case of `recursive`, it can be a path to a directory.
-                - in case of `accept_group`, it can be a group name of
-                    `foreach` generated stage.
                 - in case of `glob`, it can be a wildcard pattern to match
                   stages. Example: `build*` for stages in `dvc.yaml` file, or
                   `../dvc.yaml:build*` for stages in dvc.yaml in a different
@@ -367,8 +358,6 @@ class StageLoad:
             recursive: if true and if `target` is a directory, all of the
                 stages inside that directory is returned.
             graph: graph to use. Defaults to `repo.graph`.
-            accept_group: if true, all of the `foreach` generated stages of
-                the specified target is returned.
             glob: Use `target` as a pattern to match stages in a file.
         """
         if not target:
@@ -380,7 +369,7 @@ class StageLoad:
             path = self.fs.path.abspath(target)
             return collect_inside_path(path, graph or self.graph)
 
-        stages = self.from_target(target, accept_group=accept_group, glob=glob)
+        stages = self.from_target(target, glob=glob)
         if not with_deps:
             return stages
 
@@ -392,14 +381,14 @@ class StageLoad:
         with_deps: bool = False,
         recursive: bool = False,
         graph: "DiGraph" = None,
-        accept_group: bool = False,
     ) -> List[StageInfo]:
         """Collects a list of (stage, filter_info) from the given target.
 
         Priority is in the order of following in case of ambiguity:
         - .dvc file or .yaml file
         - dir if recursive and directory exists
-        - stage_name
+        - foreach_group_name or stage_name
+        - generated stage name from a foreach group
         - output file
 
         Args:
@@ -418,7 +407,7 @@ class StageLoad:
         target = as_posix(target)
 
         stages, file, _ = _collect_specific_target(
-            self, target, with_deps, recursive, accept_group
+            self, target, with_deps, recursive
         )
         if not stages:
             if not (recursive and self.fs.isdir(target)):
@@ -440,7 +429,6 @@ class StageLoad:
                     with_deps,
                     recursive,
                     graph,
-                    accept_group=accept_group,
                 )
             except StageFileDoesNotExistError as exc:
                 # collect() might try to use `target` as a stage name

dvc/repo/update.py

@@ -10,6 +10,7 @@ def update(
     rev=None,
     recursive=False,
     to_remote=False,
+    no_download=False,
     remote=None,
     jobs=None,
 ):
@@ -21,6 +22,11 @@ def update(
     if isinstance(targets, str):
         targets = [targets]
 
+    if to_remote and no_download:
+        raise InvalidArgumentError(
+            "--to-remote can't be used with --no-download"
+        )
+
     if not to_remote and remote:
         raise InvalidArgumentError(
             "--remote can't be used without --to-remote"
@@ -31,7 +37,13 @@ def update(
         stages.update(self.stage.collect(target, recursive=recursive))
 
     for stage in stages:
-        stage.update(rev, to_remote=to_remote, remote=remote, jobs=jobs)
+        stage.update(
+            rev,
+            to_remote=to_remote,
+            remote=remote,
+            no_download=no_download,
+            jobs=jobs,
+        )
         dvcfile = Dvcfile(self, stage.path)
         dvcfile.dump(stage)
 

dvc/schema.py

@@ -3,11 +3,11 @@ from collections.abc import Mapping
 from voluptuous import Any, Optional, Required, Schema
 
 from dvc import dependency, output
-from dvc.output import CHECKSUMS_SCHEMA, Output
+from dvc.annotations import ANNOTATION_SCHEMA
+from dvc.output import CHECKSUMS_SCHEMA, DIR_FILES_SCHEMA, META_SCHEMA, Output
 from dvc.parsing import DO_KWD, FOREACH_KWD, VARS_KWD
 from dvc.parsing.versions import SCHEMA_KWD, lockfile_version_schema
 from dvc.stage.params import StageParams
-from dvc_data.hashfile.meta import Meta
 
 STAGES = "stages"
 SINGLE_STAGE_SCHEMA = {
@@ -25,10 +25,9 @@ SINGLE_STAGE_SCHEMA = {
 
 DATA_SCHEMA = {
     **CHECKSUMS_SCHEMA,
+    **META_SCHEMA,
     Required("path"): str,
-    Meta.PARAM_SIZE: int,
-    Meta.PARAM_NFILES: int,
-    Meta.PARAM_ISEXEC: bool,
+    Output.PARAM_FILES: [DIR_FILES_SCHEMA],
 }
 LOCK_FILE_STAGE_SCHEMA = {
     Required(StageParams.PARAM_CMD): Any(str, list),
@@ -47,16 +46,15 @@ LOCKFILE_V2_SCHEMA = {
 
 OUT_PSTAGE_DETAILED_SCHEMA = {
     str: {
+        **ANNOTATION_SCHEMA,  # type: ignore
         Output.PARAM_CACHE: bool,
         Output.PARAM_PERSIST: bool,
         Output.PARAM_CHECKPOINT: bool,
-        Output.PARAM_DESC: str,
         Output.PARAM_REMOTE: str,
     }
 }
 
 PLOTS = "plots"
-PLOTS_SCHEMA = dict
 PLOT_PROPS = {
     Output.PARAM_PLOT_TEMPLATE: str,
     Output.PARAM_PLOT_X: str,
@@ -112,6 +110,15 @@ def either_or(primary, fallback, fallback_includes=None):
     return validator
 
 
+PLOT_DEFINITION = {
+    Output.PARAM_PLOT_X: str,
+    Output.PARAM_PLOT_Y: Any(str, [str], {str: Any(str, [str])}),
+    Output.PARAM_PLOT_X_LABEL: str,
+    Output.PARAM_PLOT_Y_LABEL: str,
+    Output.PARAM_PLOT_TITLE: str,
+    Output.PARAM_PLOT_TEMPLATE: str,
+}
+PLOTS_SCHEMA = {str: Any(PLOT_DEFINITION, None)}
 FOREACH_IN = {
     Required(FOREACH_KWD): Any(dict, list, str),
     Required(DO_KWD): STAGE_DEFINITION,

dvc/scm.py

@@ -160,7 +160,7 @@ def resolve_rev(scm: "Git", rev: str) -> str:
     except InternalRevError as exc:
         # `scm` will only resolve git branch and tag names,
         # if rev is not a sha it may be an abbreviated experiment name
-        if not rev.startswith("refs/"):
+        if not (rev == "HEAD" or rev.startswith("refs/")):
             from dvc.repo.experiments.utils import (
                 AmbiguousExpRefInfo,
                 resolve_name,

dvc/stage/__init__.py

@@ -78,6 +78,7 @@ def create_stage(cls, repo, path, external=False, **kwargs):
 
     wdir = os.path.abspath(kwargs.get("wdir", None) or os.curdir)
     path = os.path.abspath(path)
+
     check_dvcfile_path(repo, path)
     check_stage_path(repo, wdir, is_wdir=kwargs.get("wdir"))
     check_stage_path(repo, os.path.dirname(path))
@@ -87,7 +88,7 @@ def create_stage(cls, repo, path, external=False, **kwargs):
     if not external:
         check_no_externals(stage)
     fill_stage_dependencies(
-        stage, **project(kwargs, ["deps", "erepo", "params"])
+        stage, **project(kwargs, ["deps", "erepo", "params", "fs_config"])
     )
     check_circular_dependency(stage)
     check_duplicated_arguments(stage)
@@ -109,16 +110,13 @@ def restore_fields(stage):
     # will be used to restore comments later
     # noqa, pylint: disable=protected-access
     stage._stage_text = old._stage_text
-
     stage.meta = old.meta
     stage.desc = old.desc
 
-    old_fields = {out.def_path: (out.desc, out.remote) for out in old.outs}
-
+    old_fields = {out.def_path: (out.annot, out.remote) for out in old.outs}
     for out in stage.outs:
-        out_fields = old_fields.get(out.def_path, None)
-        if out_fields:
-            out.desc, out.remote = out_fields
+        if out_fields := old_fields.get(out.def_path, None):
+            out.annot, out.remote = out_fields
 
 
 class Stage(params.StageParams):
@@ -242,6 +240,14 @@ class Stage(params.StageParams):
         """Whether the DVC file was created with `dvc import`."""
         return not self.cmd and len(self.deps) == 1 and len(self.outs) == 1
 
+    @property
+    def is_partial_import(self) -> bool:
+        """
+        Whether the DVC file was created using `dvc import --no-download`
+        or `dvc import-url --no-download`.
+        """
+        return self.is_import and (not self.outs[0].hash_info)
+
     @property
     def is_repo_import(self):
         if not self.is_import:
@@ -433,18 +439,30 @@ class Stage(params.StageParams):
 
         return self
 
-    def update(self, rev=None, to_remote=False, remote=None, jobs=None):
+    def update(
+        self,
+        rev=None,
+        to_remote=False,
+        remote=None,
+        no_download=None,
+        jobs=None,
+    ):
         if not (self.is_repo_import or self.is_import):
             raise StageUpdateError(self.relpath)
         update_import(
-            self, rev=rev, to_remote=to_remote, remote=remote, jobs=jobs
+            self,
+            rev=rev,
+            to_remote=to_remote,
+            remote=remote,
+            no_download=no_download,
+            jobs=jobs,
         )
 
     def reload(self):
         return self.dvcfile.stage
 
-    def dumpd(self):
-        return get_dump(self)
+    def dumpd(self, **kwargs):
+        return get_dump(self, **kwargs)
 
     def compute_md5(self):
         # `dvc add`ed files don't need stage md5
@@ -457,6 +475,7 @@ class Stage(params.StageParams):
 
     def save(self, allow_missing=False):
         self.save_deps(allow_missing=allow_missing)
+
         self.save_outs(allow_missing=allow_missing)
         self.md5 = self.compute_md5()
 
@@ -525,14 +544,16 @@ class Stage(params.StageParams):
         no_commit=False,
         force=False,
         allow_missing=False,
+        no_download=False,
         **kwargs,
     ):
         if (self.cmd or self.is_import) and not self.frozen and not dry:
             self.remove_outs(ignore_remove=False, force=False)
 
-        if not self.frozen and self.is_import:
-            jobs = kwargs.get("jobs", None)
-            self._sync_import(dry, force, jobs)
+        if (not self.frozen and self.is_import) or self.is_partial_import:
+            self._sync_import(
+                dry, force, kwargs.get("jobs", None), no_download
+            )
         elif not self.frozen and self.cmd:
             self._run_stage(dry, force, **kwargs)
         else:
@@ -544,9 +565,11 @@ class Stage(params.StageParams):
                 self._check_missing_outputs()
 
         if not dry:
-            if kwargs.get("checkpoint_func", None):
+            if kwargs.get("checkpoint_func", None) or no_download:
                 allow_missing = True
             self.save(allow_missing=allow_missing)
+            if no_download:
+                self.ignore_outs()
             if not no_commit:
                 self.commit(allow_missing=allow_missing)
 
@@ -555,8 +578,8 @@ class Stage(params.StageParams):
         return run_stage(self, dry, force, **kwargs)
 
     @rwlocked(read=["deps"], write=["outs"])
-    def _sync_import(self, dry, force, jobs):
-        sync_import(self, dry, force, jobs)
+    def _sync_import(self, dry, force, jobs, no_download):
+        sync_import(self, dry, force, jobs, no_download)
 
     @rwlocked(read=["outs"])
     def _check_missing_outputs(self):
@@ -571,6 +594,8 @@ class Stage(params.StageParams):
     @rwlocked(write=["outs"])
     def checkout(self, allow_missing=False, **kwargs):
         stats = defaultdict(list)
+        if self.is_partial_import:
+            return {}
         for out in self.filter_outs(kwargs.get("filter_info")):
             key, outs = self._checkout(
                 out,
@@ -658,6 +683,9 @@ class Stage(params.StageParams):
         self, *args, **kwargs
     ) -> Dict[Optional["ObjectDB"], Set["HashInfo"]]:
         """Return set of object IDs used by this stage."""
+        if self.is_partial_import and not self.is_repo_import:
+            return {}
+
         used_objs = defaultdict(set)
         for out in self.filter_outs(kwargs.get("filter_info")):
             for odb, objs in out.get_used_objs(*args, **kwargs).items():
@@ -680,7 +708,7 @@ class Stage(params.StageParams):
                 "unable to auto-merge DVC files with deleted outputs"
             )
 
-    def merge(self, ancestor, other):
+    def merge(self, ancestor, other, allowed=None):
         assert other
 
         if not other.outs:
@@ -700,7 +728,7 @@ class Stage(params.StageParams):
         self._check_can_merge(self, ancestor_out)
         self._check_can_merge(other, ancestor_out)
 
-        self.outs[0].merge(ancestor_out, other.outs[0])
+        self.outs[0].merge(ancestor_out, other.outs[0], allowed=allowed)
 
     def dump(self, **kwargs):
         self.dvcfile.dump(self, **kwargs)
@@ -742,5 +770,5 @@ class PipelineStage(Stage):
     def _changed_stage_entry(self):
         return f"'cmd' of {self} has changed."
 
-    def merge(self, ancestor, other):
+    def merge(self, ancestor, other, allowed=None):
         raise NotImplementedError

dvc/stage/cache.py

@@ -9,7 +9,7 @@ from funcy import cached_property, first
 from dvc import fs
 from dvc.exceptions import DvcException
 from dvc.utils import dict_sha256, relpath
-from dvc_data.transfer import _log_exceptions
+from dvc_data.hashfile.transfer import _log_exceptions
 
 if TYPE_CHECKING:
     from dvc_objects.db import ObjectDB

dvc/stage/exceptions.py

@@ -31,7 +31,7 @@ class StageFileIsNotDvcFileError(DvcException):
 
         sname = fname + DVC_FILE_SUFFIX
         if is_dvc_file(sname):
-            msg += f" Do you mean '{sname}'?"
+            msg += f". Do you mean '{sname}'?"
 
         super().__init__(msg)
 
@@ -80,6 +80,11 @@ class MissingDataSource(DvcException):
         super().__init__(msg)
 
 
+class DataSourceChanged(DvcException):
+    def __init__(self, path: str):
+        super().__init__(f"data source changed: {path}")
+
+
 class StageNotFound(DvcException, KeyError):
     def __init__(self, file, name):
         self.file = file.relpath

dvc/stage/imports.py

@@ -17,20 +17,36 @@ def _update_import_on_remote(stage, remote, jobs):
     stage.outs[0].transfer(url, odb=odb, jobs=jobs, update=True)
 
 
-def update_import(stage, rev=None, to_remote=False, remote=None, jobs=None):
+def update_import(
+    stage, rev=None, to_remote=False, remote=None, no_download=None, jobs=None
+):
     stage.deps[0].update(rev=rev)
+    outs = stage.outs
+    deps = stage.deps
+
     frozen = stage.frozen
     stage.frozen = False
+
+    if stage.outs:
+        stage.outs[0].clear()
     try:
         if to_remote:
             _update_import_on_remote(stage, remote, jobs)
         else:
-            stage.reproduce(jobs=jobs)
+            stage.reproduce(no_download=no_download, jobs=jobs)
     finally:
+        if deps == stage.deps:
+            stage.outs = outs
         stage.frozen = frozen
 
 
-def sync_import(stage, dry=False, force=False, jobs=None):
+def sync_import(
+    stage,
+    dry=False,
+    force=False,
+    jobs=None,
+    no_download=False,
+):
     """Synchronize import's outs to the workspace."""
     logger.info("Importing '%s' -> '%s'", stage.deps[0], stage.outs[0])
     if dry:
@@ -40,4 +56,11 @@ def sync_import(stage, dry=False, force=False, jobs=None):
         stage.outs[0].checkout()
     else:
         stage.save_deps()
-        stage.deps[0].download(stage.outs[0], jobs=jobs)
+        if no_download:
+            if stage.is_repo_import:
+                stage.deps[0].update()
+        else:
+            stage.deps[0].download(
+                stage.outs[0],
+                jobs=jobs,
+            )

dvc/stage/loader.py

@@ -69,7 +69,11 @@ class StageLoader(Mapping):
             info = info.copy()
             info.pop("path", None)
             item.meta = Meta.from_dict(info)
-            item.hash_info = HashInfo.from_dict(info)
+            hash_value = getattr(item.meta, item.hash_name, None)
+            item.hash_info = HashInfo(item.hash_name, hash_value)
+            if item.hash_info and item.hash_info.isdir:
+                item.meta.isdir = True
+            item.files = get_in(checksums, [key, path, item.PARAM_FILES])
 
     @classmethod
     def load_stage(cls, dvcfile, name, stage_data, lock_data=None):

dvc/stage/serialize.py

@@ -6,7 +6,7 @@ from typing import TYPE_CHECKING, List, no_type_check
 from funcy import post_processing
 
 from dvc.dependency import ParamsDependency
-from dvc.output import Output
+from dvc.output import Annotation, Output
 from dvc.utils.collections import apply_diff
 from dvc.utils.serialize import parse_yaml_for_update
 
@@ -27,7 +27,7 @@ PARAM_METRIC = Output.PARAM_METRIC
 PARAM_PLOT = Output.PARAM_PLOT
 PARAM_PERSIST = Output.PARAM_PERSIST
 PARAM_CHECKPOINT = Output.PARAM_CHECKPOINT
-PARAM_DESC = Output.PARAM_DESC
+PARAM_DESC = Annotation.PARAM_DESC
 PARAM_REMOTE = Output.PARAM_REMOTE
 
 DEFAULT_PARAMS_FILE = ParamsDependency.DEFAULT_PARAMS_FILE
@@ -38,8 +38,9 @@ sort_by_path = partial(sorted, key=attrgetter("def_path"))
 
 @post_processing(OrderedDict)
 def _get_flags(out):
-    if out.desc:
-        yield PARAM_DESC, out.desc
+    annot = out.annot.to_dict()
+    yield from annot.items()
+
     if not out.use_cache:
         yield PARAM_CACHE, False
     if out.checkpoint:
@@ -151,16 +152,25 @@ def to_pipeline_file(stage: "PipelineStage"):
     }
 
 
-def to_single_stage_lockfile(stage: "Stage") -> dict:
+def to_single_stage_lockfile(stage: "Stage", **kwargs) -> dict:
     assert stage.cmd
 
     def _dumpd(item):
+        meta_d = item.meta.to_dict()
+        meta_d.pop("isdir", None)
         ret = [
             (item.PARAM_PATH, item.def_path),
             *item.hash_info.to_dict().items(),
-            *item.meta.to_dict().items(),
+            *meta_d.items(),
         ]
 
+        if item.hash_info.isdir and kwargs.get("with_files"):
+            if item.obj:
+                obj = item.obj
+            else:
+                obj = item.get_obj()
+            ret.append((item.PARAM_FILES, obj.as_list(with_meta=True)))
+
         return OrderedDict(ret)
 
     res = OrderedDict([("cmd", stage.cmd)])
@@ -180,13 +190,13 @@ def to_single_stage_lockfile(stage: "Stage") -> dict:
     return res
 
 
-def to_lockfile(stage: "PipelineStage") -> dict:
+def to_lockfile(stage: "PipelineStage", **kwargs) -> dict:
     assert stage.name
-    return {stage.name: to_single_stage_lockfile(stage)}
+    return {stage.name: to_single_stage_lockfile(stage, **kwargs)}
 
 
-def to_single_stage_file(stage: "Stage"):
-    state = stage.dumpd()
+def to_single_stage_file(stage: "Stage", **kwargs):
+    state = stage.dumpd(**kwargs)
 
     # When we load a stage we parse yaml with a fast parser, which strips
     # off all the comments and formatting. To retain those on update we do

dvc/stage/utils.py

@@ -4,6 +4,7 @@ from typing import TYPE_CHECKING, Union
 
 from funcy import concat, first, lsplit, rpartial, without
 
+from dvc.annotations import ANNOTATION_FIELDS
 from dvc.exceptions import InvalidArgumentError
 from dvc_data.hashfile.meta import Meta
 
@@ -107,12 +108,16 @@ def _load_live_output(
     return outs
 
 
-def fill_stage_dependencies(stage, deps=None, erepo=None, params=None):
+def fill_stage_dependencies(
+    stage, deps=None, erepo=None, params=None, fs_config=None
+):
     from dvc.dependency import loads_from, loads_params
 
     assert not stage.deps
     stage.deps = []
-    stage.deps += loads_from(stage, deps or [], erepo=erepo)
+    stage.deps += loads_from(
+        stage, deps or [], erepo=erepo, fs_config=fs_config
+    )
     stage.deps += loads_params(stage, params or [])
 
 
@@ -194,9 +199,9 @@ def compute_md5(stage):
     return dict_md5(
         d,
         exclude=[
+            *ANNOTATION_FIELDS,
             stage.PARAM_LOCKED,  # backward compatibility
             stage.PARAM_FROZEN,
-            Output.PARAM_DESC,
             Output.PARAM_METRIC,
             Output.PARAM_PERSIST,
             Output.PARAM_CHECKPOINT,
@@ -221,7 +226,7 @@ def resolve_paths(fs, path, wdir=None):
     return path, wdir
 
 
-def get_dump(stage):
+def get_dump(stage, **kwargs):
     return {
         key: value
         for key, value in {
@@ -230,8 +235,8 @@ def get_dump(stage):
             stage.PARAM_CMD: stage.cmd,
             stage.PARAM_WDIR: resolve_wdir(stage.wdir, stage.path),
             stage.PARAM_FROZEN: stage.frozen,
-            stage.PARAM_DEPS: [d.dumpd() for d in stage.deps],
-            stage.PARAM_OUTS: [o.dumpd() for o in stage.outs],
+            stage.PARAM_DEPS: [d.dumpd(**kwargs) for d in stage.deps],
+            stage.PARAM_OUTS: [o.dumpd(**kwargs) for o in stage.outs],
             stage.PARAM_ALWAYS_CHANGED: stage.always_changed,
             stage.PARAM_META: stage.meta,
         }.items()

dvc/testing/api_tests.py

@@ -0,0 +1,125 @@
+import pytest
+
+from dvc import api
+from dvc.api import DVCFileSystem
+from dvc.utils.fs import remove
+
+
+class TestAPI:
+    def test_get_url(self, tmp_dir, dvc, remote):
+        tmp_dir.dvc_gen("foo", "foo")
+
+        expected_url = (remote / "ac/bd18db4cc2f85cedef654fccc4a4d8").url
+        assert api.get_url("foo") == expected_url
+
+    def test_open(self, tmp_dir, dvc, remote):
+        tmp_dir.dvc_gen("foo", "foo-text")
+        dvc.push()
+
+        # Remove cache to force download
+        remove(dvc.odb.local.path)
+
+        with api.open("foo") as fd:
+            assert fd.read() == "foo-text"
+
+    @pytest.mark.parametrize(
+        "clear_cache", [True, False], ids=["cache", "no_cache"]
+    )
+    @pytest.mark.parametrize(
+        "fs_kwargs",
+        [
+            {},
+            {"url": "{path}"},
+            {"url": "{path}", "rev": "master"},
+            {"url": "file://{posixpath}"},
+            {"url": "file://{posixpath}", "rev": "master"},
+        ],
+        ids=["current", "local", "local_rev", "git", "git_rev"],
+    )
+    def test_filesystem(
+        self,
+        M,  # from `pytest-test-utils`
+        tmp_dir,
+        make_tmp_dir,
+        scm,
+        dvc,
+        remote,
+        fs_kwargs,
+        clear_cache,
+    ):
+        fs_kwargs = fs_kwargs.copy()
+
+        tmp_dir.scm_gen({"scripts": {"script1": "script1"}}, commit="scripts")
+        tmp_dir.dvc_gen({"data": {"foo": "foo", "bar": "bar"}}, commit="data")
+        dvc.push()
+
+        if clear_cache:
+            remove(dvc.odb.repo.path)
+
+        if url := fs_kwargs.get("url"):
+            fs_kwargs["url"] = url.format(
+                path=tmp_dir, posixpath=tmp_dir.as_posix()
+            )
+
+        fs = DVCFileSystem(**fs_kwargs)
+
+        assert fs.ls("/", detail=False) == M.unordered(
+            "/.gitignore", "/scripts", "/data"
+        )
+        assert fs.ls("scripts", detail=False) == ["scripts/script1"]
+        assert fs.ls("data", detail=False) == M.unordered(
+            "data/foo", "data/bar"
+        )
+
+        data_info = M.dict(
+            name="/data",
+            type="directory",
+            dvc_info=M.dict(isdvc=True, isout=True),
+        )
+        scripts_info = M.dict(name="/scripts", type="directory", isexec=False)
+
+        assert sorted(fs.ls("/"), key=lambda i: i["name"]) == [
+            M.dict(name="/.gitignore", type="file", isexec=False),
+            data_info,
+            scripts_info,
+        ]
+
+        with pytest.raises(FileNotFoundError):
+            fs.info("/not-existing-path")
+
+        assert fs.info("/") == M.dict(name="/", isexec=False, type="directory")
+        assert fs.info("/data") == data_info
+        assert fs.info("/scripts") == scripts_info
+        assert fs.info("/data/foo") == M.dict(name="/data/foo", type="file")
+        assert fs.info("/scripts/script1") == M.dict(
+            name="/scripts/script1", type="file"
+        )
+
+        assert not fs.isdvc("/")
+        assert fs.isdvc("/data")
+        assert fs.isdvc("/data/foo")
+        assert not fs.isdvc("/scripts")
+        assert not fs.isdvc("/scripts/script1")
+
+        with pytest.raises((IsADirectoryError, PermissionError)):
+            fs.open("data")
+        with pytest.raises((IsADirectoryError, PermissionError)):
+            fs.open("scripts")
+        with fs.open("/data/foo") as fobj:
+            assert fobj.read() == b"foo"
+        with fs.open("/scripts/script1") as fobj:
+            assert fobj.read() == b"script1"
+
+        tmp = make_tmp_dir("temp-download")
+        fs.get_file("data/foo", tmp / "foo")
+        assert (tmp / "foo").read_text() == "foo"
+
+        fs.get_file("scripts/script1", tmp / "script1")
+        assert (tmp / "script1").read_text() == "script1"
+
+        fs.get("/", (tmp / "all").fs_path, recursive=True)
+        assert (tmp / "all").read_text() == {
+            ".gitignore": "/data\n",
+            "data": {"bar": "bar", "foo": "foo"},
+            "scripts": {"script1": "script1"},
+        }

dvc/testing/remote_tests.py

@@ -71,7 +71,7 @@ class TestRemote:
         _check_status(status_dir, ok=dir_hashes)
 
         # Remove and check status
-        remove(dvc.odb.local.path)
+        dvc.odb.local.clear()
 
         status = dvc.cloud.status(foo_hashes)
         _check_status(status, deleted={foo_hash})

dvc/testing/test_api.py

@@ -1,20 +0,0 @@
-from dvc import api
-from dvc.utils.fs import remove
-
-
-class TestAPI:
-    def test_get_url(self, tmp_dir, dvc, remote):
-        tmp_dir.dvc_gen("foo", "foo")
-
-        expected_url = (remote / "ac/bd18db4cc2f85cedef654fccc4a4d8").url
-        assert api.get_url("foo") == expected_url
-
-    def test_open(self, tmp_dir, dvc, remote):
-        tmp_dir.dvc_gen("foo", "foo-text")
-        dvc.push()
-
-        # Remove cache to force download
-        remove(dvc.odb.local.path)
-
-        with api.open("foo") as fd:
-            assert fd.read() == "foo-text"

dvc/testing/workspace_tests.py

@@ -2,6 +2,9 @@ import os
 
 import pytest
 
+from dvc.exceptions import URLMissingError
+from dvc.repo.ls_url import ls_url, parse_external_url
+
 
 class TestImport:
     def test_import(self, tmp_dir, dvc, workspace):
@@ -134,3 +137,62 @@ class TestAdd:
         assert (
             workspace / "cache" / dir_hash_value[:2] / dir_hash_value[2:]
         ).is_file()
+
+
+def match_files(fs, entries, expected):
+    entries_content = {
+        (fs.path.normpath(d["path"]), d["isdir"]) for d in entries
+    }
+    expected_content = {
+        (fs.path.normpath(d["path"]), d["isdir"]) for d in expected
+    }
+    assert entries_content == expected_content
+
+
+class TestLsUrl:
+    @pytest.mark.parametrize("fname", ["foo", "foo.dvc", "dir/foo"])
+    def test_file(self, cloud, fname):
+        cloud.gen({fname: "foo contents"})
+        fs, fs_path = parse_external_url(cloud.url, cloud.config)
+        result = ls_url(str(cloud / fname), config=cloud.config)
+        match_files(
+            fs,
+            result,
+            [{"path": fs.path.join(fs_path, fname), "isdir": False}],
+        )
+
+    def test_dir(self, cloud):
+        cloud.gen(
+            {"dir/foo": "foo contents", "dir/subdir/bar": "bar contents"}
+        )
+        fs, _ = parse_external_url(cloud.url, cloud.config)
+        result = ls_url(str(cloud / "dir"), config=cloud.config)
+        match_files(
+            fs,
+            result,
+            [
+                {"path": "foo", "isdir": False},
+                {"path": "subdir", "isdir": True},
+            ],
+        )
+
+    def test_recursive(self, cloud):
+        cloud.gen(
+            {"dir/foo": "foo contents", "dir/subdir/bar": "bar contents"}
+        )
+        fs, _ = parse_external_url(cloud.url, cloud.config)
+        result = ls_url(
+            str(cloud / "dir"), config=cloud.config, recursive=True
+        )
+        match_files(
+            fs,
+            result,
+            [
+                {"path": "foo", "isdir": False},
+                {"path": "subdir/bar", "isdir": False},
+            ],
+        )
+
+    def test_nonexistent(self, cloud):
+        with pytest.raises(URLMissingError):
+            ls_url(str(cloud / "dir"), config=cloud.config)