pyproject.toml: remove stale comment from dependency (#10546)

dvcfs.get: simplify and separate files into respective filesystems while collecting/walking

.github/ISSUE_TEMPLATE/epic_story.md

@@ -1,60 +0,0 @@
----
-name: Epic/Story Issue Template
-about: Template for "top" level issues - Epics (>2 weeks) / Stories (<2 weeks)
-title: 'Epic: New Feature'
-labels: epic
-assignees:
-
----
-
-## Summary / Background
-What do you want to achieve, why? business context
-...
-
-## Scope
-
-What will be impacted and what won't be?
-What needs implementation and what is invariant?
-e.g.
-- user should be able to run workflow X from UI
-- enable workflow Y, Z from CLI
-
-## Assumptions
-Product / UX assumptions as well as technical assumptions / limitations
-e.g.
-* Support only Python Runtime
-* Focus on DVC experiments only
-* Deployment environments don't change often and can be picked up from shared configuration
-
-## Open Questions
-e.g.
-- How should access control work for shared artifacts (workflow X)
-- Python runtime assumption - is it really valid? in light of <...>
-
-## Blockers / Dependencies
-List issues or other conditions / blockers
-
-## General Approach
-Invocation example:
-```shell
-$ mapper-run task.tar.gz --ray-cluster <ip>:<port>
-```
-
-## Steps
-
-### Must have (p1)
-- [ ] subissue2
-- [ ] step 2
- - info
- - info
-
-### Optional / followup (p2)
-- [ ] ⌛ step 3 wip
-- [ ] step 4
-
-## Timelines
-
-Put your estimations here. Update once certainty changes
-- end of week (Feb 3) for prototype with workflows X, Y
-- Feb 15 - MVP in production
-- Low priority followups can be done later

.github/workflows/benchmarks.yaml

@@ -5,6 +5,10 @@ concurrency:
   group: ${{ github.workflow }}-${{ github.head_ref || github.run_id }}
   cancel-in-progress: true
 
+env:
+  FORCE_COLOR: "1"
+  PY_COLORS: "1"
+
 jobs:
   bench:
     name: run benchmarks
@@ -12,7 +16,25 @@ jobs:
     steps:
       - uses: actions/setup-python@v5
         with:
-            python-version: "3.11"
-      - uses: iterative/dvc-bench@main
+            python-version: "3.12"
+
+      - uses: hynek/setup-cached-uv@v2
+      - uses: actions/checkout@v4
         with:
-            pytest_options: "-k 'test_init or test_help'"
+          ref: ${{ github.event.pull_request.base.sha }}
+          fetch-depth: 0
+
+      - run: uv pip install '.[tests]' --system
+      - name: run benchmarks on base branch
+        run: pytest --benchmark-autosave dvc/testing/benchmarks/ -k 'test_init or test_help'
+
+      - uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+          clean: false
+      - run: uv pip install '.[tests]' --system
+      - name: run benchmarks for PR
+        run: >
+          pytest --benchmark-compare --benchmark-compare-fail=min:5%
+          --benchmark-group-by name
+          dvc/testing/benchmarks/ -k 'test_init or test_help'

.github/workflows/build.yaml

@@ -27,14 +27,15 @@ jobs:
       with:
         python-version: 3.12
 
-    - run: python -m pip install --upgrade pip setuptools_scm build twine
+    - uses: hynek/setup-cached-uv@v2
+    - run: uv pip install --upgrade setuptools_scm build twine --system
 
     - name: Force version for Test PyPI upload
       if: ${{ !startsWith(github.ref, 'refs/tags') }}
       run: echo version=$(python -m setuptools_scm | awk -F+ '{print $1}' | tail -1) >> $GITHUB_ENV
 
     - run: echo 'PKG = "pip"'>dvc/_build.py
-    - run: python -m build
+    - run: python -m build --installer=uv
       env:
         SETUPTOOLS_SCM_PRETEND_VERSION_FOR_DVC: ${{ env.version }}
     - run: twine check --strict dist/*

.github/workflows/plugin_tests.yaml

@@ -18,8 +18,8 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        os: [ubuntu-20.04]
-        pyv: ["3.10"]
+        os: [ubuntu-latest]
+        pyv: ["3.12"]
         plugin: ["dvc-s3"]
 
     steps:
@@ -37,16 +37,17 @@ jobs:
       uses: actions/setup-python@v5
       with:
         python-version: ${{ matrix.pyv }}
-        cache: pip
+
+    - uses: hynek/setup-cached-uv@v2
+      with:
         cache-dependency-path: |
           dvc/pyproject.toml
-          ${{ matrix.plugin }}/setup.cfg
+          ${{ matrix.plugin }}/pyproject.toml
 
     - name: Install plugin + DVC@PR
       run: |
-        pip install --upgrade pip wheel
-        pip install "./dvc[testing]"
-        pip install -e "./${{ matrix.plugin }}[tests]"
+        uv pip install "./dvc[testing]" --system
+        uv pip install -e "./${{ matrix.plugin }}[tests]" --system
 
     - name: Run plugin tests
       timeout-minutes: 15

.github/workflows/tests.yaml

@@ -16,15 +16,68 @@ permissions:
   contents: read
 
 jobs:
-  tests:
-    timeout-minutes: 50
+  lint:
+    timeout-minutes: 10
     runs-on: ${{ matrix.os }}
     strategy:
       fail-fast: false
       matrix:
-        os: [ubuntu-20.04, windows-latest, macos-latest]
-        pyv: ["3.9", "3.10", "3.11", "3.12"]
+        os: [ubuntu-latest, macos-latest, windows-latest]
+    steps:
+    - uses: actions/checkout@v4
+      with:
+        fetch-depth: 0
+    - name: Set up Python
+      uses: actions/setup-python@v5
+      with:
+        python-version: '3.12'
+    - uses: hynek/setup-cached-uv@v2
+      with:
+        cache-dependency-path: pyproject.toml
+    - run: uv pip install -e ".[dev]" --system
+      env:
+        UV_LINK_MODE: ${{ runner.os == 'Windows' && 'symlink' || 'hardlink' }}
 
+    - name: Cache mypy
+      uses: actions/cache@v4
+      with:
+        path: .mypy_cache
+        key: mypy-${{ runner.os }}-${{ env.pythonLocation }}-${{ hashFiles('pyproject.toml') }}
+    - uses: pre-commit/[email protected]
+
+  tests:
+    timeout-minutes: 20
+    runs-on: ${{ matrix.os }}
+    strategy:
+      fail-fast: false
+      matrix:
+        os: [ubuntu-latest, macos-latest]
+        pyv: ["3.9", "3.10", "3.11", "3.12"]
+        include:
+        - os: windows-latest
+          pyv: "3.9"
+          pytestargs: "--splitting-algorithm=least_duration --splits 4 --group 1"
+        - os: windows-latest
+          pyv: "3.9"
+          pytestargs: "--splitting-algorithm=least_duration --splits 4 --group 2"
+        - os: windows-latest
+          pyv: "3.9"
+          pytestargs: "--splitting-algorithm=least_duration --splits 4 --group 3"
+        - os: windows-latest
+          pyv: "3.9"
+          pytestargs: "--splitting-algorithm=least_duration --splits 4 --group 4"
+        - os: windows-latest
+          pyv: "3.12"
+          pytestargs: "--splitting-algorithm=least_duration --splits 4 --group 1"
+        - os: windows-latest
+          pyv: "3.12"
+          pytestargs: "--splitting-algorithm=least_duration --splits 4 --group 2"
+        - os: windows-latest
+          pyv: "3.12"
+          pytestargs: "--splitting-algorithm=least_duration --splits 4 --group 3"
+        - os: windows-latest
+          pyv: "3.12"
+          pytestargs: "--splitting-algorithm=least_duration --splits 4 --group 4"
     steps:
     - uses: actions/checkout@v4
       with:
@@ -33,20 +86,21 @@ jobs:
       uses: actions/setup-python@v5
       with:
         python-version: ${{ matrix.pyv }}
-        cache: pip
+    - uses: hynek/setup-cached-uv@v2
+      with:
+        cache-suffix: ${{ matrix.pyv }}
         cache-dependency-path: pyproject.toml
-    - name: install
-      run: |
-        pip install --upgrade pip wheel
-        pip install -e ".[dev]"
-    - uses: pre-commit/[email protected]
+    - run: uv pip install -e ".[dev]" pytest-split --system
+      env:
+        UV_LINK_MODE: ${{ runner.os == 'Windows' && 'symlink' || 'hardlink' }}
+
     - name: run tests
       timeout-minutes: 40
       env:
         PYTHONUTF8: 1
       run: >
-        pytest -n=logical --timeout=300 --durations=100
-        --cov --cov-report=xml --cov-report=term
+        pytest ${{ matrix.pytestargs }} -n=logical --dist=worksteal --timeout=300 --durations=0
+        --cov --cov-report=xml --cov-report=term  --durations-path=./.github/.test_durations
     - name: upload coverage report
       uses: codecov/codecov-action@v3
       with:

.pre-commit-config.yaml

@@ -22,13 +22,13 @@ repos:
       - id: sort-simple-yaml
       - id: trailing-whitespace
   - repo: https://github.com/astral-sh/ruff-pre-commit
-    rev: 'v0.4.4'
+    rev: 'v0.6.2'
     hooks:
       - id: ruff
         args: [--fix, --exit-non-zero-on-fix]
       - id: ruff-format
   - repo: https://github.com/codespell-project/codespell
-    rev: v2.2.6
+    rev: v2.3.0
     hooks:
       - id: codespell
         additional_dependencies: ["tomli"]

README.rst

@@ -1,4 +1,6 @@
-|Banner|
+🚀 Check out our new product `DataChain <https://github.com/iterative/datachain>`_ (and give it a ⭐!) if you need to version and process a large number of files. Contact us at [email protected] to discuss commercial solutions and support for AI reproducibility and data management scenarios.
+
+--------------------------
 
 `Website <https://dvc.org>`_
 • `Docs <https://dvc.org/doc>`_

dvc/api/data.py

@@ -285,7 +285,7 @@ def _open(
             if TYPE_CHECKING:
                 from dvc.fs import FileSystem
 
-            fs: Union["FileSystem", DataFileSystem, DVCFileSystem]
+            fs: Union[FileSystem, DataFileSystem, DVCFileSystem]
             if os.path.isabs(path):
                 fs = DataFileSystem(index=_repo.index.data["local"])
                 fs_path = path

dvc/cli/command.py

@@ -18,8 +18,8 @@ class CmdBase(ABC):
 
         os.chdir(args.cd)
 
-        self.repo: "Repo" = Repo(uninitialized=self.UNINITIALIZED)
-        self.config: "Config" = self.repo.config
+        self.repo: Repo = Repo(uninitialized=self.UNINITIALIZED)
+        self.config: Config = self.repo.config
         self.args = args
 
     def do_run(self):

dvc/commands/artifacts.py

@@ -13,6 +13,9 @@ class CmdArtifactsGet(CmdBaseNoRepo):
         from dvc.scm import CloneError
         from dvc.ui import ui
 
+        if self.args.show_url:
+            return self._show_url()
+
         try:
             count, out = Artifacts.get(
                 self.args.url,
@@ -36,6 +39,28 @@ class CmdArtifactsGet(CmdBaseNoRepo):
             )
             return 1
 
+    def _show_url(self):
+        from dvc.api import artifacts_show, get_url
+        from dvc.ui import ui
+
+        artifact = artifacts_show(
+            self.args.name,
+            version=self.args.rev,
+            stage=self.args.stage,
+            repo=self.args.url,
+        )
+
+        url = get_url(
+            artifact["path"],
+            repo=self.args.url,
+            rev=artifact["rev"],
+            remote=self.args.remote,
+            remote_config=self.args.remote_config,
+        )
+        ui.write(url, force=True)
+
+        return 0
+
 
 def add_parser(subparsers, parent_parser):
     ARTIFACTS_HELP = "DVC model registry artifact commands."
@@ -84,6 +109,14 @@ def add_parser(subparsers, parent_parser):
         help="Destination path to download artifact to",
         metavar="<path>",
     ).complete = completion.DIR
+    get_parser.add_argument(
+        "--show-url",
+        action="store_true",
+        help=(
+            "Print the storage location (URL) the target data would be "
+            "downloaded from, and exit."
+        ),
+    )
     get_parser.add_argument(
         "-j",
         "--jobs",

dvc/commands/dag.py

@@ -17,7 +17,7 @@ def _show_ascii(graph: "DiGraph"):
 
     ret = []
     for pipeline in pipelines:
-        ret.append(draw(pipeline.nodes, pipeline.edges))
+        ret.append(draw(pipeline.nodes, pipeline.edges))  # noqa: PERF401
 
     return "\n".join(ret)
 

dvc/commands/data_sync.py

@@ -194,7 +194,7 @@ def add_parser(subparsers, _parent_parser):
     )
     pull_parser.add_argument(
         "--run-cache",
-        action="store_true",
+        action=argparse.BooleanOptionalAction,
         default=False,
         help="Fetch run history for all stages.",
     )
@@ -262,7 +262,7 @@ def add_parser(subparsers, _parent_parser):
     )
     push_parser.add_argument(
         "--run-cache",
-        action="store_true",
+        action=argparse.BooleanOptionalAction,
         default=False,
         help="Push run history for all stages.",
     )
@@ -324,7 +324,7 @@ def add_parser(subparsers, _parent_parser):
     )
     fetch_parser.add_argument(
         "--run-cache",
-        action="store_true",
+        action=argparse.BooleanOptionalAction,
         default=False,
         help="Fetch run history for all stages.",
     )

dvc/commands/dataset.py

@@ -42,7 +42,7 @@ class CmdDatasetAdd(CmdBase):
             if rev := dataset.lock.rev:
                 ver = rev
 
-        ver_part: Optional["Text"] = None
+        ver_part: Optional[Text] = None
         if ver:
             ver_part = ui.rich_text.assemble(" @ ", (ver, "repr.number"))
         text = ui.rich_text.assemble("(", (url, "repr.url"), ver_part or "", ")")

dvc/commands/experiments/pull.py

@@ -1,3 +1,5 @@
+import argparse
+
 from dvc.cli import formatter
 from dvc.cli.command import CmdBase
 from dvc.cli.utils import append_doc_link
@@ -81,7 +83,7 @@ def add_parser(experiments_subparsers, parent_parser):
     )
     experiments_pull_parser.add_argument(
         "--run-cache",
-        action="store_true",
+        action=argparse.BooleanOptionalAction,
         default=False,
         help="Pull run history for all stages.",
     )

dvc/commands/experiments/push.py

@@ -1,3 +1,4 @@
+import argparse
 from typing import Any
 
 from dvc.cli import completion, formatter
@@ -120,7 +121,7 @@ def add_parser(experiments_subparsers, parent_parser):
     )
     experiments_push_parser.add_argument(
         "--run-cache",
-        action="store_true",
+        action=argparse.BooleanOptionalAction,
         default=False,
         help="Push run history for all stages.",
     )

dvc/commands/plots.py

@@ -72,7 +72,7 @@ class CmdPlots(CmdBase):
                 html_template_path = os.path.join(self.repo.dvc_dir, html_template_path)
         return html_template_path
 
-    def run(self) -> int:  # noqa: C901, PLR0911, PLR0912
+    def run(self) -> int:  # noqa: C901, PLR0911
         from pathlib import Path
 
         from dvc.render.match import match_defs_renderers

dvc/commands/repro.py

@@ -117,10 +117,7 @@ and then the stage name name.
         "--pull",
         action="store_true",
         default=False,
-        help=(
-            "Try automatically pulling missing cache for outputs restored "
-            "from the run-cache."
-        ),
+        help="Try automatically pulling missing data.",
     )
     repro_parser.add_argument(
         "--allow-missing",

dvc/commands/stage.py

@@ -68,7 +68,7 @@ def prepare_stages_data(
 class CmdStageList(CmdBase):
     def _get_stages(self) -> Iterable["Stage"]:
         if self.args.all:
-            stages: list["Stage"] = self.repo.index.stages
+            stages: list[Stage] = self.repo.index.stages
             logger.trace("%d no. of stages found", len(stages))
             return stages
 

dvc/commands/studio.py

@@ -18,11 +18,16 @@ class CmdStudioLogin(CmdConfig):
         from dvc.utils.studio import STUDIO_URL
         from dvc_studio_client.auth import StudioAuthError, get_access_token
 
+        studio = self.config.get("studio", {})
         name = self.args.name
-        hostname = self.args.hostname or os.environ.get(DVC_STUDIO_URL) or STUDIO_URL
+        hostname = (
+            self.args.hostname
+            or os.environ.get(DVC_STUDIO_URL)
+            or studio.get("url")
+            or STUDIO_URL
+        )
         scopes = self.args.scopes
 
-        studio = self.config.get("studio", {})
         if studio.get("url", hostname) == hostname and "token" in studio:
             raise DvcException(
                 "Token already exists. "

dvc/compare.py

@@ -160,7 +160,7 @@ class TabularData(MutableSequence[Sequence["CellT"]]):
             if key not in keys:
                 self.add_column(key)
 
-        row: list["CellT"] = [
+        row: list[CellT] = [
             with_value(d.get(key), self._fill_value) for key in self.keys()
         ]
         self.append(row)

dvc/daemon.py

@@ -44,10 +44,10 @@ def _win_detached_subprocess(args: Sequence[str], **kwargs) -> int:
 
     startupinfo = STARTUPINFO()
     startupinfo.dwFlags |= STARTF_USESHOWWINDOW
-    popen = subprocess.Popen(
+    popen = subprocess.Popen(  # noqa: S603
         args,
         close_fds=True,
-        shell=False,  # noqa: S603
+        shell=False,
         startupinfo=startupinfo,
         creationflags=creationflags,
         **kwargs,
@@ -149,7 +149,7 @@ def _spawn(
     env: Optional[Mapping[str, str]] = None,
     output_file: Optional[str] = None,
 ) -> int:
-    file: "AbstractContextManager[Any]" = nullcontext()
+    file: AbstractContextManager[Any] = nullcontext()
     kwargs = {}
     if output_file:
         file = open(output_file, "ab")  # noqa: SIM115

dvc/dependency/base.py

@@ -45,7 +45,7 @@ class Dependency(Output):
             self.fs_path = self.fs.version_path(self.fs_path, self.meta.version_id)
 
     def download(self, to, jobs=None):
-        fs_download(self.fs, self.fs_path, to.fs_path, jobs=jobs)
+        return fs_download(self.fs, self.fs_path, to.fs_path, jobs=jobs)
 
     def save(self):
         super().save()

dvc/dependency/db.py

@@ -37,13 +37,17 @@ class AbstractDependency(Dependency):
     """Dependency without workspace/fs/fs_path"""
 
     def __init__(self, stage: "Stage", info: dict[str, Any], *args, **kwargs):
-        self.repo: "Repo" = stage.repo
+        self.repo: Repo = stage.repo
         self.stage = stage
         self.fs = None
         self.fs_path = None
         self.def_path = None  # type: ignore[assignment]
         self.info = info or {}
 
+    @property
+    def is_in_repo(self):
+        return False
+
 
 class DbDependency(AbstractDependency):
     PARAM_CONNECTION = "connection"
@@ -70,8 +74,8 @@ class DbDependency(AbstractDependency):
         return self.db_info.get(self.PARAM_QUERY) or self.db_info.get(self.PARAM_TABLE)
 
     def __repr__(self):
-        return "{}:{}".format(
-            self.__class__.__name__, "".join(f"{k}=={v}" for k, v in self.info.items())
+        return "{}: {}".format(
+            self.__class__.__name__, "".join(f"{k}={v}" for k, v in self.info.items())
         )
 
     def __str__(self):

dvc/dependency/repo.py

@@ -9,7 +9,9 @@ from .base import Dependency
 
 if TYPE_CHECKING:
     from dvc.fs import DVCFileSystem
+    from dvc.output import Output
     from dvc.stage import Stage
+    from dvc_data.hashfile.hash_info import HashInfo
 
 
 class RepoDependency(Dependency):
@@ -35,7 +37,7 @@ class RepoDependency(Dependency):
         super().__init__(stage, *args, **kwargs)
 
         self.fs = self._make_fs()
-        self.fs_path = as_posix(self.def_path)
+        self.fs_path = as_posix(self.fs.normpath(self.def_path))
 
     def _parse_path(self, fs, fs_path):  # noqa: ARG002
         return None
@@ -91,6 +93,27 @@ class RepoDependency(Dependency):
             self.PARAM_REPO: self._dump_def_repo(self.def_repo),
         }
 
+    def download(self, to: "Output", jobs: Optional[int] = None):
+        from dvc.fs import LocalFileSystem
+
+        files = super().download(to=to, jobs=jobs)
+        if not isinstance(to.fs, LocalFileSystem):
+            return
+
+        hashes: list[tuple[str, HashInfo, dict[str, Any]]] = []
+        for src_path, dest_path, maybe_info in files:
+            try:
+                info = maybe_info or self.fs.info(src_path)
+                hash_info = info["dvc_info"]["entry"].hash_info
+                dest_info = to.fs.info(dest_path)
+            except (KeyError, AttributeError):
+                # If no hash info found, just keep going and output will be hashed later
+                continue
+            if hash_info:
+                hashes.append((dest_path, hash_info, dest_info))
+        cache = to.cache if to.use_cache else to.local_cache
+        cache.state.save_many(hashes, to.fs)
+
     def update(self, rev: Optional[str] = None):
         if rev:
             self.def_repo[self.PARAM_REV] = rev

dvc/fs/__init__.py

@@ -1,4 +1,5 @@
 import glob
+from itertools import repeat
 from typing import Optional
 from urllib.parse import urlparse
 
@@ -47,28 +48,12 @@ known_implementations.update(
 
 def download(
     fs: "FileSystem", fs_path: str, to: str, jobs: Optional[int] = None
-) -> int:
+) -> list[tuple[str, str, Optional[dict]]]:
     from dvc.scm import lfs_prefetch
 
     from .callbacks import TqdmCallback
 
     with TqdmCallback(desc=f"Downloading {fs.name(fs_path)}", unit="files") as cb:
-        # NOTE: We use dvc-objects generic.copy over fs.get since it makes file
-        # download atomic and avoids fsspec glob/regex path expansion.
-        if fs.isdir(fs_path):
-            from_infos = [
-                path for path in fs.find(fs_path) if not path.endswith(fs.flavour.sep)
-            ]
-            if not from_infos:
-                localfs.makedirs(to, exist_ok=True)
-                return 0
-            to_infos = [
-                localfs.join(to, *fs.relparts(info, fs_path)) for info in from_infos
-            ]
-        else:
-            from_infos = [fs_path]
-            to_infos = [to]
-
         if isinstance(fs, DVCFileSystem):
             lfs_prefetch(
                 fs,
@@ -78,10 +63,29 @@ def download(
                     else glob.escape(fs_path)
                 ],
             )
+            if not glob.has_magic(fs_path):
+                return fs._get(fs_path, to, batch_size=jobs, callback=cb)
+
+        # NOTE: We use dvc-objects generic.copy over fs.get since it makes file
+        # download atomic and avoids fsspec glob/regex path expansion.
+        if fs.isdir(fs_path):
+            from_infos = [
+                path for path in fs.find(fs_path) if not path.endswith(fs.flavour.sep)
+            ]
+            if not from_infos:
+                localfs.makedirs(to, exist_ok=True)
+                return []
+            to_infos = [
+                localfs.join(to, *fs.relparts(info, fs_path)) for info in from_infos
+            ]
+        else:
+            from_infos = [fs_path]
+            to_infos = [to]
+
         cb.set_size(len(from_infos))
         jobs = jobs or fs.jobs
         generic.copy(fs, from_infos, localfs, to_infos, callback=cb, batch_size=jobs)
-        return len(to_infos)
+        return list(zip(from_infos, to_infos, repeat(None)))
 
 
 def parse_external_url(url, fs_config=None, config=None):

dvc/fs/dvc.py

@@ -4,22 +4,28 @@ import ntpath
 import os
 import posixpath
 import threading
-from collections import deque
-from contextlib import ExitStack, suppress
+from collections import defaultdict, deque
+from contextlib import ExitStack, nullcontext, suppress
+from glob import has_magic
 from typing import TYPE_CHECKING, Any, Callable, Optional, Union
 
-from fsspec.spec import AbstractFileSystem
+from fsspec.spec import DEFAULT_CALLBACK, AbstractFileSystem
 from funcy import wrap_with
 
 from dvc.log import logger
-from dvc_objects.fs.base import FileSystem
+from dvc.utils.threadpool import ThreadPoolExecutor
+from dvc_objects.fs.base import AnyFSPath, FileSystem
 
 from .data import DataFileSystem
 
 if TYPE_CHECKING:
+    from contextlib import AbstractContextManager
+
     from dvc.repo import Repo
     from dvc.types import DictStrAny, StrPath
 
+    from .callbacks import Callback
+
 logger = logger.getChild(__name__)
 
 RepoFactory = Union[Callable[..., "Repo"], type["Repo"]]
@@ -429,7 +435,7 @@ class _DVCFileSystem(AbstractFileSystem):
         ignore_subrepos = kwargs.get("ignore_subrepos", True)
         return self._info(key, path, ignore_subrepos=ignore_subrepos)
 
-    def _info(  # noqa: C901, PLR0912
+    def _info(  # noqa: C901
         self, key, path, ignore_subrepos=True, check_ignored=True
     ):
         repo, dvc_fs, subkey = self._get_subrepo_info(key)
@@ -474,9 +480,127 @@ class _DVCFileSystem(AbstractFileSystem):
         info["name"] = path
         return info
 
+    def get(
+        self,
+        rpath,
+        lpath,
+        recursive=False,
+        callback=DEFAULT_CALLBACK,
+        maxdepth=None,
+        batch_size=None,
+        **kwargs,
+    ):
+        self._get(
+            rpath,
+            lpath,
+            recursive=recursive,
+            callback=callback,
+            maxdepth=maxdepth,
+            batch_size=batch_size,
+            **kwargs,
+        )
+
+    def _get(  # noqa: C901, PLR0912
+        self,
+        rpath,
+        lpath,
+        recursive=False,
+        callback=DEFAULT_CALLBACK,
+        maxdepth=None,
+        batch_size=None,
+        **kwargs,
+    ) -> list[tuple[str, str, Optional[dict]]]:
+        if (
+            isinstance(rpath, list)
+            or isinstance(lpath, list)
+            or has_magic(rpath)
+            or not self.exists(rpath)
+            or not recursive
+        ):
+            super().get(
+                rpath,
+                lpath,
+                recursive=recursive,
+                callback=callback,
+                maxdepth=maxdepth,
+                **kwargs,
+            )
+            return []
+
+        if os.path.isdir(lpath) or lpath.endswith(os.path.sep):
+            lpath = self.join(lpath, os.path.basename(rpath))
+
+        if self.isfile(rpath):
+            with callback.branched(rpath, lpath) as child:
+                self.get_file(rpath, lpath, callback=child, **kwargs)
+                return [(rpath, lpath, None)]
+
+        result: list[tuple[str, str, Optional[dict]]] = []
+        _dirs: list[str] = []
+        _files: dict[FileSystem, list[tuple[str, str, Optional[dict]]]]
+        _files = defaultdict(list)
+
+        for root, dirs, files in self.walk(rpath, maxdepth=maxdepth, detail=True):
+            if files:
+                callback.set_size((callback.size or 0) + len(files))
+
+            parts = self.relparts(root, rpath)
+            if parts in ((os.curdir,), ("",)):
+                parts = ()
+            dest_root = os.path.join(lpath, *parts)
+            if not maxdepth or len(parts) < maxdepth - 1:
+                _dirs.extend(f"{dest_root}{os.path.sep}{d}" for d in dirs)
+
+            key = self._get_key_from_relative(root)
+            _, dvc_fs, _ = self._get_subrepo_info(key)
+
+            for name, info in files.items():
+                dvc_info = info.get("dvc_info")
+                fs_info = info.get("fs_info")
+                if dvc_fs and dvc_info and not fs_info:
+                    fs = dvc_fs
+                    fs_path = dvc_info["name"]
+                else:
+                    fs = self.repo.fs
+                    fs_path = fs_info["name"]
+
+                src_path = f"{root}{self.sep}{name}"
+                dest_path = f"{dest_root}{os.path.sep}{name}"
+                _files[fs].append((fs_path, dest_path, dvc_info))
+                result.append((src_path, dest_path, info))
+
+        os.makedirs(lpath, exist_ok=True)
+        for d in _dirs:
+            os.mkdir(d)
+
+        def get_file(arg: tuple[FileSystem, tuple[str, str, Optional[dict]]]):
+            fs, (src, dest, info) = arg
+            kw = kwargs
+            if isinstance(fs, DataFileSystem):
+                kw = kw | {"info": info}
+            return fs.get_file(src, dest, callback=callback, **kw)
+
+        if batch_size == 1:
+            ctx: AbstractContextManager = nullcontext()
+            map_fn: Callable = map
+        else:
+            ctx = ThreadPoolExecutor(max_workers=batch_size)
+            map_fn = ctx.imap_unordered
+
+        with ctx:
+            it = ((fs, f) for fs, files in _files.items() for f in files)
+            deque(map_fn(get_file, it), maxlen=0)
+        return result
+
     def get_file(self, rpath, lpath, **kwargs):
+        dvc_info = kwargs.pop("info", {}).pop("dvc_info", None)
         key = self._get_key_from_relative(rpath)
         fs_path = self._from_key(key)
+        dirpath = os.path.dirname(lpath)
+        if dirpath:
+            # makedirs raises error if the string is empty
+            os.makedirs(dirpath, exist_ok=True)
+
         try:
             return self.repo.fs.get_file(fs_path, lpath, **kwargs)
         except FileNotFoundError:
@@ -485,7 +609,7 @@ class _DVCFileSystem(AbstractFileSystem):
                 raise
 
         dvc_path = _get_dvc_path(dvc_fs, subkey)
-        return dvc_fs.get_file(dvc_path, lpath, **kwargs)
+        return dvc_fs.get_file(dvc_path, lpath, info=dvc_info, **kwargs)
 
     def du(self, path, total=True, maxdepth=None, withdirs=False, **kwargs):
         if maxdepth is not None:
@@ -553,6 +677,45 @@ class DVCFileSystem(FileSystem):
     def getcwd(self):
         return self.fs.getcwd()
 
+    def _get(
+        self,
+        from_info: Union[AnyFSPath, list[AnyFSPath]],
+        to_info: Union[AnyFSPath, list[AnyFSPath]],
+        callback: "Callback" = DEFAULT_CALLBACK,
+        recursive: bool = False,
+        batch_size: Optional[int] = None,
+        **kwargs,
+    ) -> list[tuple[str, str, Optional[dict]]]:
+        # FileSystem.get is non-recursive by default if arguments are lists
+        # otherwise, it's recursive.
+        recursive = not (isinstance(from_info, list) and isinstance(to_info, list))
+        return self.fs._get(
+            from_info,
+            to_info,
+            callback=callback,
+            recursive=recursive,
+            batch_size=batch_size,
+            **kwargs,
+        )
+
+    def get(
+        self,
+        from_info: Union[AnyFSPath, list[AnyFSPath]],
+        to_info: Union[AnyFSPath, list[AnyFSPath]],
+        callback: "Callback" = DEFAULT_CALLBACK,
+        recursive: bool = False,
+        batch_size: Optional[int] = None,
+        **kwargs,
+    ) -> None:
+        self._get(
+            from_info,
+            to_info,
+            callback=callback,
+            batch_size=batch_size,
+            recursive=recursive,
+            **kwargs,
+        )
+
     @property
     def fsid(self) -> str:
         return self.fs.fsid

dvc/info.py

@@ -123,7 +123,7 @@ def _get_supported_remotes():
         if not fs_cls.get_missing_deps():
             dependencies = []
             for requirement in fs_cls.REQUIRES:
-                dependencies.append(
+                dependencies.append(  # noqa: PERF401
                     f"{requirement} = {importlib_metadata.version(requirement)}"
                 )
 

dvc/lock.py

@@ -4,6 +4,7 @@ import hashlib
 import os
 from abc import ABC, abstractmethod
 from datetime import timedelta
+from typing import Optional, Union
 
 import flufl.lock
 import zc.lockfile
@@ -174,9 +175,9 @@ class HardlinkLock(flufl.lock.Lock, LockBase):
         self._owned = True
         self._retry_errnos = []
 
-    def lock(self):
+    def lock(self, timeout: Optional[Union[timedelta, int]] = None):
         try:
-            super().lock(timedelta(seconds=DEFAULT_TIMEOUT))
+            super().lock(timeout or timedelta(seconds=DEFAULT_TIMEOUT))
         except flufl.lock.TimeOutError:
             raise LockError(FAILED_TO_LOCK_MESSAGE)  # noqa: B904
 

dvc/output.py

@@ -390,7 +390,7 @@ class Output:
         self.can_push = push
 
         self.fs_path = self._parse_path(self.fs, fs_path)
-        self.obj: Optional["HashFile"] = None
+        self.obj: Optional[HashFile] = None
 
         self.remote = remote
 
@@ -542,7 +542,9 @@ class Output:
             desc=f"Collecting files and computing hashes in {self}",
             disable=no_progress_bar,
         ) as pb:
-            return build(*args, callback=pb.as_callback(), **kwargs)
+            kwargs["callback"] = pb.as_callback()
+            kwargs.setdefault("checksum_jobs", self.fs.hash_jobs)
+            return build(*args, **kwargs)
 
     def _get_hash_meta(self):
         if self.use_cache:
@@ -730,6 +732,7 @@ class Output:
         from dvc_data.hashfile.checkout import LinkError, PromptError
 
         kwargs.setdefault("ignore", self.dvcignore)
+        kwargs.setdefault("checksum_jobs", self.fs.hash_jobs)
         try:
             return checkout(*args, **kwargs)
         except PromptError as exc:
@@ -749,11 +752,8 @@ class Output:
             granular = (
                 self.is_dir_checksum and filter_info and filter_info != self.fs_path
             )
-            # NOTE: trying to use hardlink during transfer only if we will be
-            # relinking later
-            hardlink = relink
             if granular:
-                obj = self._commit_granular_dir(filter_info, hardlink)
+                obj = self._commit_granular_dir(filter_info, hardlink=False)
             else:
                 staging, _, obj = self._build(
                     self.cache,
@@ -771,7 +771,7 @@ class Output:
                         self.cache,
                         {obj.hash_info},
                         shallow=False,
-                        hardlink=hardlink,
+                        hardlink=False,
                         callback=cb,
                     )
             if relink:
@@ -786,6 +786,7 @@ class Output:
                         state=self.repo.state,
                         prompt=prompt.confirm,
                         progress_callback=cb,
+                        old=obj,
                     )
                 self.set_exec()
 
@@ -902,7 +903,7 @@ class Output:
     def get_obj(
         self, filter_info: Optional[str] = None, **kwargs
     ) -> Optional["HashFile"]:
-        obj: Optional["HashFile"] = None
+        obj: Optional[HashFile] = None
         if self.obj:
             obj = self.obj
         elif self.files:
@@ -1158,7 +1159,7 @@ class Output:
             logger.warning(msg)
             return {}
 
-        obj: Optional["HashFile"]
+        obj: Optional[HashFile]
         if self.is_dir_checksum:
             obj = self._collect_used_dir_cache(**kwargs)
         else:
@@ -1353,7 +1354,7 @@ class Output:
         cache = self.cache if self.use_cache else self.local_cache
         assert isinstance(cache, HashFileDB)
 
-        new: "HashFile"
+        new: HashFile
         try:
             assert self.hash_name
             staging, meta, obj = self._build(
@@ -1403,7 +1404,7 @@ class Output:
                 staging,
                 self.cache,
                 {obj.hash_info},
-                hardlink=relink,
+                hardlink=False,
                 shallow=False,
                 callback=cb,
             )
@@ -1421,6 +1422,7 @@ class Output:
                     state=self.repo.state,
                     prompt=prompt.confirm,
                     progress_callback=callback,
+                    old=obj,
                 )
             self.set_exec()
         return obj

dvc/parsing/__init__.py

@@ -563,10 +563,10 @@ class MatrixDefinition:
         iterable = self.resolved_iterable
         assert isinstance(iterable, Mapping)
 
-        ret: dict[str, "DictStrAny"] = {}
+        ret: dict[str, DictStrAny] = {}
         matrix = {key: enumerate(v) for key, v in iterable.items()}
         for combination in product(*matrix.values()):
-            d: "DictStrAny" = {}
+            d: DictStrAny = {}
             fragments: list[str] = []
             for k, (i, v) in zip(matrix.keys(), combination):
                 d[k] = v

dvc/progress.py

@@ -41,7 +41,7 @@ class Tqdm(tqdm):
         "miniters": 1,
     }
 
-    def __init__(  # noqa: PLR0913
+    def __init__(
         self,
         iterable=None,
         disable=None,

dvc/render/converter/vega.py

@@ -23,7 +23,7 @@ def _lists(blob: Union[dict, list]) -> Iterable[list]:
     if isinstance(blob, list):
         yield blob
     else:
-        for _, value in blob.items():
+        for value in blob.values():
             if isinstance(value, dict):
                 yield from _lists(value)
             elif isinstance(value, list):

dvc/repo/__init__.py

@@ -147,7 +147,7 @@ class Repo:
     ):
         from dvc.cachemgr import CacheManager
         from dvc.data_cloud import DataCloud
-        from dvc.fs import GitFileSystem, LocalFileSystem, localfs
+        from dvc.fs import GitFileSystem, LocalFileSystem
         from dvc.lock import LockNoop, make_lock
         from dvc.repo.artifacts import Artifacts
         from dvc.repo.datasets import Datasets
@@ -161,7 +161,7 @@ class Repo:
 
         self.url = url
         self._fs_conf = {"repo_factory": repo_factory}
-        self._fs = fs or localfs
+        self._fs = fs or LocalFileSystem()
         self._scm = scm
         self._config = config
         self._remote = remote
@@ -184,12 +184,12 @@ class Repo:
         # used by DVCFileSystem to determine if it should traverse subrepos
         self.subrepos = subrepos
 
-        self.cloud: "DataCloud" = DataCloud(self)
-        self.stage: "StageLoad" = StageLoad(self)
+        self.cloud: DataCloud = DataCloud(self)
+        self.stage: StageLoad = StageLoad(self)
 
-        self.lock: "LockBase"
+        self.lock: LockBase
         self.cache: CacheManager
-        self.state: "StateBase"
+        self.state: StateBase
         if isinstance(self.fs, GitFileSystem) or not self.dvc_dir:
             self.lock = LockNoop()
             self.state = StateNoop()
@@ -206,6 +206,11 @@ class Repo:
                     friendly=True,
                 )
                 os.makedirs(self.site_cache_dir, exist_ok=True)
+                if not fs and (
+                    checksum_jobs := self.config["core"].get("checksum_jobs")
+                ):
+                    self.fs.hash_jobs = checksum_jobs
+
                 self.state = State(self.root_dir, self.site_cache_dir, self.dvcignore)
             else:
                 self.lock = LockNoop()

dvc/repo/artifacts.py

@@ -137,14 +137,12 @@ class Artifacts:
 
         assert not (version and stage)
         name = _reformat_name(name)
-        tags: list["GitTag"] = find(
-            name=name, version=version, stage=stage, scm=self.scm
-        )
+        tags: list[GitTag] = find(name=name, version=version, stage=stage, scm=self.scm)
         if not tags:
             raise ArtifactNotFoundError(name, version=version, stage=stage)
         if version or stage:
             return tags[-1].target
-        gto_tags: list["GTOTag"] = sort_versions(parse_tag(tag) for tag in tags)
+        gto_tags: list[GTOTag] = sort_versions(parse_tag(tag) for tag in tags)
         return gto_tags[0].tag.target
 
     @classmethod
@@ -222,7 +220,7 @@ class Artifacts:
 
             out = resolve_output(path, out, force=force)
             fs = self.repo.dvcfs
-            count = fs_download(fs, path, os.path.abspath(out), jobs=jobs)
+            count = len(fs_download(fs, path, os.path.abspath(out), jobs=jobs))
         return count, out
 
     @staticmethod
@@ -285,7 +283,7 @@ class Artifacts:
         return len(to_infos), relpath(localfs.commonpath(to_infos))
 
     @classmethod
-    def get(  # noqa: PLR0913
+    def get(
         cls,
         url: str,
         name: str,
@@ -314,6 +312,13 @@ class Artifacts:
 
         name = _reformat_name(name)
         saved_exc: Optional[Exception] = None
+
+        local_dvc_studio_config = Config().get("studio", {})
+        args_dvc_studio_config = {}
+        if config and not isinstance(config, dict):
+            config = Config.load_file(config)
+            args_dvc_studio_config = config.get("studio", {})
+
         try:
             logger.trace("Trying studio-only config")
             return cls._download_studio(
@@ -324,14 +329,13 @@ class Artifacts:
                 out=out,
                 force=force,
                 jobs=jobs,
+                dvc_studio_config=local_dvc_studio_config | args_dvc_studio_config,
             )
         except FileExistsLocallyError:
             raise
         except Exception as exc:  # noqa: BLE001
             saved_exc = exc
 
-        if config and not isinstance(config, dict):
-            config = Config.load_file(config)
         with Repo.open(
             url=url,
             subrepos=True,
@@ -341,7 +345,7 @@ class Artifacts:
             remote_config=remote_config,
         ) as repo:
             logger.trace("Trying repo [studio] config")
-            dvc_studio_config = dict(repo.config.get("studio"))
+            repo_dvc_studio_config = repo.config.get("studio", {})
             try:
                 return cls._download_studio(
                     url,
@@ -351,7 +355,9 @@ class Artifacts:
                     out=out,
                     force=force,
                     jobs=jobs,
-                    dvc_studio_config=dvc_studio_config,
+                    dvc_studio_config=local_dvc_studio_config
+                    | repo_dvc_studio_config
+                    | args_dvc_studio_config,
                 )
             except FileExistsLocallyError:
                 raise

dvc/repo/checkout.py

@@ -125,7 +125,7 @@ def checkout(  # noqa: C901
             (StageFileDoesNotExistError, StageFileBadNameError, NoOutputOrStageError),
         ):
             raise CheckoutErrorSuggestGit(target) from exc
-        raise
+        raise  # noqa: PLE0704
 
     view = self.index.targets_view(
         targets, recursive=recursive, with_deps=with_deps, onerror=onerror

dvc/repo/datasets.py

@@ -76,7 +76,7 @@ class SerDe:
         return asdict(self, filter=filter_defaults, value_serializer=value_serializer)
 
     @classmethod
-    def from_dict(cls: type["Self"], d: dict[str, Any]) -> "Self":
+    def from_dict(cls, d: dict[str, Any]) -> "Self":
         _fields = fields(cast("type[AttrsInstance]", cls))
         kwargs = {f.name: d[f.name] for f in _fields if f.name in d}
         return cls(**kwargs)
@@ -242,7 +242,7 @@ class DatasetNotFoundError(DvcException, KeyError):
 
 class Datasets(Mapping[str, Dataset]):
     def __init__(self, repo: "Repo") -> None:
-        self.repo: "Repo" = repo
+        self.repo: Repo = repo
 
     def __repr__(self):
         return repr(dict(self))

dvc/repo/diff.py

@@ -26,7 +26,7 @@ def _diff(old, new, data_keys, with_missing=False):
     from dvc_data.index.diff import ADD, DELETE, MODIFY, RENAME
     from dvc_data.index.diff import diff as idiff
 
-    ret: "dict[str, list[dict]]" = {
+    ret: dict[str, list[dict]] = {
         "added": [],
         "deleted": [],
         "modified": [],

dvc/repo/experiments/__init__.py

@@ -119,7 +119,7 @@ class Experiments:
         **kwargs,
     ):
         """Reproduce and checkout a single (standalone) experiment."""
-        exp_queue: "BaseStashQueue" = (
+        exp_queue: BaseStashQueue = (
             self.tempdir_queue if tmp_dir else self.workspace_queue
         )
         self.queue_one(exp_queue, **kwargs)
@@ -249,7 +249,7 @@ class Experiments:
         exec_results = queue.reproduce(copy_paths=copy_paths, message=message)
 
         results: dict[str, str] = {}
-        for _, exp_result in exec_results.items():
+        for exp_result in exec_results.values():
             results.update(exp_result)
         return results
 

dvc/repo/experiments/apply.py

@@ -25,7 +25,7 @@ def apply(repo: "Repo", rev: str, **kwargs):
     from dvc.repo.checkout import checkout as dvc_checkout
     from dvc.scm import RevError, resolve_rev
 
-    exps: "Experiments" = repo.experiments
+    exps: Experiments = repo.experiments
 
     is_stash: bool = False
 
@@ -57,7 +57,7 @@ def apply(repo: "Repo", rev: str, **kwargs):
 
 
 def _apply(repo: "Repo", rev: str, name: Optional[str] = None, is_stash: bool = False):
-    exps: "Experiments" = repo.experiments
+    exps: Experiments = repo.experiments
 
     with exps.apply_stash.preserve_workspace(rev, name=name):
         with repo.scm.detach_head(rev, force=True):

dvc/repo/experiments/collect.py

@@ -69,11 +69,11 @@ def collect_rev(
 
 def _collect_rev(
     repo: "Repo",
-    rev: str,
+    revision: str,
     param_deps: bool = False,
     **kwargs,
 ) -> SerializableExp:
-    with repo.switch(rev) as rev:
+    with repo.switch(revision) as rev:
         if rev == "workspace":
             timestamp: Optional[datetime] = None
         else:
@@ -174,7 +174,7 @@ def collect_active(
     """
     if not baseline_revs:
         return {}
-    result: dict[str, list["ExpRange"]] = {}
+    result: dict[str, list[ExpRange]] = {}
     exps = repo.experiments
     for queue in (exps.workspace_queue, exps.tempdir_queue, exps.celery_queue):
         for baseline, active_exps in queue.collect_active_data(
@@ -220,7 +220,7 @@ def collect_successful(
     Returns:
         Dict mapping baseline revision to successful experiments.
     """
-    result: dict[str, list["ExpRange"]] = {}
+    result: dict[str, list[ExpRange]] = {}
     for baseline_rev in baseline_revs:
         result[baseline_rev] = list(_collect_baseline(repo, baseline_rev, **kwargs))
     return result
@@ -304,7 +304,7 @@ def collect(
         )
 
     workspace_data = collect_rev(repo, "workspace", **kwargs)
-    result: list["ExpState"] = [workspace_data]
+    result: list[ExpState] = [workspace_data]
     queued = collect_queued(repo, baseline_revs, **kwargs) if not hide_queued else {}
     active = collect_active(repo, baseline_revs, **kwargs)
     failed = collect_failed(repo, baseline_revs, **kwargs) if not hide_failed else {}

dvc/repo/experiments/executor/base.py

@@ -199,7 +199,7 @@ class BaseExecutor(ABC):
     @classmethod
     def from_info(cls, info: "ExecutorInfo") -> "Self":
         if info.result_hash:
-            result: Optional["ExecutorResult"] = ExecutorResult(
+            result: Optional[ExecutorResult] = ExecutorResult(
                 info.result_hash,
                 (ExpRefInfo.from_ref(info.result_ref) if info.result_ref else None),
                 info.result_force,
@@ -288,7 +288,7 @@ class BaseExecutor(ABC):
             stages = []
             if targets:
                 for target in targets:
-                    stages.append(
+                    stages.append(  # noqa: PERF401
                         dvc.commit(
                             target, recursive=recursive, force=True, relink=False
                         )
@@ -470,7 +470,7 @@ class BaseExecutor(ABC):
             cls._set_log_level(log_level)
 
         exp_hash: Optional[str] = None
-        exp_ref: Optional["ExpRefInfo"] = None
+        exp_ref: Optional[ExpRefInfo] = None
         repro_force: bool = False
 
         if info.name:
@@ -559,7 +559,7 @@ class BaseExecutor(ABC):
             )
 
         ref: Optional[str] = dvc.scm.get_ref(EXEC_BRANCH, follow=False)
-        exp_ref: Optional["ExpRefInfo"] = ExpRefInfo.from_ref(ref) if ref else None
+        exp_ref: Optional[ExpRefInfo] = ExpRefInfo.from_ref(ref) if ref else None
         if cls.WARN_UNTRACKED:
             untracked = dvc.scm.untracked_files()
             if untracked:

dvc/repo/experiments/executor/local.py

@@ -201,7 +201,7 @@ class WorkspaceExecutor(BaseLocalExecutor):
     @classmethod
     def from_stash_entry(cls, repo: "Repo", entry: "ExpStashEntry", **kwargs):
         root_dir = repo.scm.root_dir
-        executor: "WorkspaceExecutor" = cls._from_stash_entry(
+        executor: WorkspaceExecutor = cls._from_stash_entry(
             repo, entry, root_dir, **kwargs
         )
         logger.debug("Init workspace executor in '%s'", root_dir)

dvc/repo/experiments/pull.py

@@ -30,7 +30,7 @@ def pull(  # noqa: C901
     pull_cache: bool = False,
     **kwargs,
 ) -> Iterable[str]:
-    exp_ref_set: set["ExpRefInfo"] = set()
+    exp_ref_set: set[ExpRefInfo] = set()
     if all_commits:
         exp_ref_set.update(exp_refs(repo.scm, git_remote))
     elif exp_names:
@@ -55,7 +55,7 @@ def pull(  # noqa: C901
         rev_dict = iter_revs(repo.scm, rev, num)
         rev_set = set(rev_dict.keys())
         ref_info_dict = exp_refs_by_baseline(repo.scm, rev_set, git_remote)
-        for _, ref_info_list in ref_info_dict.items():
+        for ref_info_list in ref_info_dict.values():
             exp_ref_set.update(ref_info_list)
 
     pull_result = _pull(repo, git_remote, exp_ref_set, force)
@@ -97,7 +97,7 @@ def _pull(
     def group_result(refspec):
         return results[str(refspec)]
 
-    pull_result: Mapping[SyncStatus, list["ExpRefInfo"]] = group_by(group_result, refs)
+    pull_result: Mapping[SyncStatus, list[ExpRefInfo]] = group_by(group_result, refs)
 
     return pull_result
 

dvc/repo/experiments/push.py

@@ -78,7 +78,7 @@ def exp_refs_from_rev(scm: "Git", rev: list[str], num: int = 1) -> set["ExpRefIn
     rev_dict = iter_revs(scm, rev, num)
     rev_set = set(rev_dict.keys())
     ref_info_dict = exp_refs_by_baseline(scm, rev_set)
-    for _, ref_info_list in ref_info_dict.items():
+    for ref_info_list in ref_info_dict.values():
         exp_ref_set.update(ref_info_list)
     return exp_ref_set
 
@@ -96,7 +96,7 @@ def push(
     push_cache: bool = False,
     **kwargs: Any,
 ) -> dict[str, Any]:
-    exp_ref_set: set["ExpRefInfo"] = set()
+    exp_ref_set: set[ExpRefInfo] = set()
     assert isinstance(repo.scm, Git)
     if all_commits:
         exp_ref_set.update(exp_refs(repo.scm))
@@ -162,7 +162,7 @@ def _push(
     def group_result(refspec):
         return results[str(refspec)]
 
-    pull_result: Mapping[SyncStatus, list["ExpRefInfo"]] = group_by(group_result, refs)
+    pull_result: Mapping[SyncStatus, list[ExpRefInfo]] = group_by(group_result, refs)
 
     return pull_result
 

dvc/repo/experiments/queue/celery.py

@@ -323,7 +323,7 @@ class LocalCeleryQueue(BaseStashQueue):
     def _get_running_task_ids(self) -> set[str]:
         running_task_ids: set[str] = set()
         active_workers = self.worker_status()
-        for _, tasks in active_workers.items():
+        for tasks in active_workers.values():
             task = first(tasks)
             if task:
                 running_task_ids.add(task["id"])
@@ -470,11 +470,11 @@ class LocalCeleryQueue(BaseStashQueue):
             exp_names = [exp_names]
         results: dict[str, ExpRefAndQueueEntry] = {}
 
-        exp_ref_match: dict[str, Optional["ExpRefInfo"]] = resolve_name(
+        exp_ref_match: dict[str, Optional[ExpRefInfo]] = resolve_name(
             self.scm, exp_names, git_remote
         )
         if not git_remote:
-            queue_entry_match: dict[str, Optional["QueueEntry"]] = (
+            queue_entry_match: dict[str, Optional[QueueEntry]] = (
                 self.match_queue_entry_by_name(
                     exp_names, self.iter_queued(), self.iter_done()
                 )
@@ -613,7 +613,7 @@ class LocalCeleryQueue(BaseStashQueue):
         """Map exp refs to any available successful executors."""
         from dvc.repo.experiments.serialize import ExpExecutor, LocalExpExecutor
 
-        result: dict[str, "ExpExecutor"] = {}
+        result: dict[str, ExpExecutor] = {}
         for entry, exec_result in self.iter_success():
             if baseline_revs and entry.baseline_rev not in baseline_revs:
                 continue

dvc/repo/experiments/queue/remove.py

@@ -21,11 +21,11 @@ def remove_tasks(  # noqa: C901, PLR0912
     """
     from celery.result import AsyncResult
 
-    stash_revs: dict[str, "ExpStashEntry"] = {}
-    failed_stash_revs: list["ExpStashEntry"] = []
-    done_entry_set: set["QueueEntry"] = set()
+    stash_revs: dict[str, ExpStashEntry] = {}
+    failed_stash_revs: list[ExpStashEntry] = []
+    done_entry_set: set[QueueEntry] = set()
     stash_rev_all = celery_queue.stash.stash_revs
-    failed_rev_all: dict[str, "ExpStashEntry"] = {}
+    failed_rev_all: dict[str, ExpStashEntry] = {}
     if celery_queue.failed_stash:
         failed_rev_all = celery_queue.failed_stash.stash_revs
     for entry in queue_entries:
@@ -90,17 +90,17 @@ def celery_clear(
     """
 
     removed: list[str] = []
-    entry_list: list["QueueEntry"] = []
+    entry_list: list[QueueEntry] = []
     if queued:
-        queue_entries: list["QueueEntry"] = list(self.iter_queued())
+        queue_entries: list[QueueEntry] = list(self.iter_queued())
         entry_list.extend(queue_entries)
         removed.extend(_get_names(queue_entries))
     if failed:
-        failed_tasks: list["QueueDoneResult"] = list(self.iter_failed())
+        failed_tasks: list[QueueDoneResult] = list(self.iter_failed())
         entry_list.extend([result.entry for result in failed_tasks])
         removed.extend(_get_names(failed_tasks))
     if success:
-        success_tasks: list["QueueDoneResult"] = list(self.iter_success())
+        success_tasks: list[QueueDoneResult] = list(self.iter_success())
         entry_list.extend([result.entry for result in success_tasks])
         removed.extend(_get_names(success_tasks))
 
@@ -125,7 +125,7 @@ def celery_remove(self: "LocalCeleryQueue", revs: Collection[str]) -> list[str]:
 
     remained: list[str] = []
     removed: list[str] = []
-    entry_to_remove: list["QueueEntry"] = []
+    entry_to_remove: list[QueueEntry] = []
     for name, entry in match_results.items():
         if entry:
             entry_to_remove.append(entry)

dvc/repo/experiments/queue/utils.py

@@ -25,7 +25,7 @@ def get_remote_executor_refs(scm: "Git", remote_url: str) -> list[str]:
     refs = []
     for ref in iter_remote_refs(scm, remote_url, base=EXPS_NAMESPACE):
         if not ref.startswith(EXEC_NAMESPACE) and ref != EXPS_STASH:
-            refs.append(ref)
+            refs.append(ref)  # noqa: PERF401
     return refs
 
 

dvc/repo/experiments/remove.py

@@ -34,17 +34,17 @@ def remove(  # noqa: C901, PLR0912
     removed: list[str] = []
     if not any([exp_names, queue, all_commits, rev]):
         return removed
-    celery_queue: "LocalCeleryQueue" = repo.experiments.celery_queue
+    celery_queue: LocalCeleryQueue = repo.experiments.celery_queue
 
     if queue:
         removed.extend(celery_queue.clear(queued=True))
 
     assert isinstance(repo.scm, Git)
 
-    exp_ref_list: list["ExpRefInfo"] = []
-    queue_entry_list: list["QueueEntry"] = []
+    exp_ref_list: list[ExpRefInfo] = []
+    queue_entry_list: list[QueueEntry] = []
     if exp_names:
-        results: dict[str, "ExpRefAndQueueEntry"] = (
+        results: dict[str, ExpRefAndQueueEntry] = (
             celery_queue.get_ref_and_entry_by_names(exp_names, git_remote)
         )
         remained: list[str] = []
@@ -94,11 +94,11 @@ def _resolve_exp_by_baseline(
 ) -> dict[str, "ExpRefInfo"]:
     assert isinstance(repo.scm, Git)
 
-    commit_ref_dict: dict[str, "ExpRefInfo"] = {}
+    commit_ref_dict: dict[str, ExpRefInfo] = {}
     rev_dict = iter_revs(repo.scm, rev, num)
     rev_set = set(rev_dict.keys())
     ref_info_dict = exp_refs_by_baseline(repo.scm, rev_set, git_remote)
-    for _, ref_info_list in ref_info_dict.items():
+    for ref_info_list in ref_info_dict.values():
         for ref_info in ref_info_list:
             commit_ref_dict[ref_info.name] = ref_info
     return commit_ref_dict

dvc/repo/experiments/serialize.py

@@ -50,7 +50,12 @@ class SerializableExp:
         Params, metrics, deps, outs are filled via repo fs/index, all other fields
         should be passed via kwargs.
         """
-        from dvc.dependency import ParamsDependency, RepoDependency
+        from dvc.dependency import (
+            DatasetDependency,
+            DbDependency,
+            ParamsDependency,
+            RepoDependency,
+        )
 
         rev = rev or repo.get_rev()
         assert rev
@@ -68,7 +73,10 @@ class SerializableExp:
                     nfiles=dep.meta.nfiles if dep.meta else None,
                 )
                 for dep in repo.index.deps
-                if not isinstance(dep, (ParamsDependency, RepoDependency))
+                if not isinstance(
+                    dep,
+                    (ParamsDependency, RepoDependency, DatasetDependency, DbDependency),
+                )
             },
             outs={
                 relpath(out.fs_path, repo.root_dir): ExpOut(

dvc/repo/experiments/show.py

@@ -117,7 +117,7 @@ def _build_rows(
     **kwargs,
 ) -> Iterator[tuple["CellT", ...]]:
     for baseline in baseline_states:
-        row: dict[str, "CellT"] = dict.fromkeys(all_headers, fill_value)
+        row: dict[str, CellT] = dict.fromkeys(all_headers, fill_value)
         row["Experiment"] = ""
         if baseline.name:
             row["rev"] = baseline.name
@@ -232,7 +232,7 @@ def _exp_range_rows(
         logger.debug("Returning tip commit for legacy checkpoint exp")
     exp = first(exp_range.revs)
     if exp:
-        row: dict[str, "CellT"] = dict.fromkeys(all_headers, fill_value)
+        row: dict[str, CellT] = dict.fromkeys(all_headers, fill_value)
         row["Experiment"] = exp.name or ""
         row["rev"] = exp.rev[:7] if Git.is_sha(exp.rev) else exp.rev
         row["typ"] = "branch_base" if is_base else "branch_commit"

dvc/repo/experiments/utils.py

@@ -199,7 +199,7 @@ def exp_commits(
     scm: "Git", ref_infos: Optional[Iterable[ExpRefInfo]] = None
 ) -> Iterable[str]:
     """Iterate over all experiment commits."""
-    shas: set["str"] = set()
+    shas: set[str] = set()
     refs = ref_infos if ref_infos else exp_refs(scm)
     for ref_info in refs:
         shas.update(scm.branch_revs(str(ref_info), ref_info.baseline_sha))

dvc/repo/fetch.py

@@ -2,6 +2,7 @@ from typing import TYPE_CHECKING
 
 from dvc.exceptions import DownloadError
 from dvc.log import logger
+from dvc.stage.cache import RunCacheNotSupported
 from dvc.ui import ui
 from dvc_data.index import DataIndex, FileStorage
 
@@ -136,6 +137,8 @@ def fetch(  # noqa: PLR0913
     try:
         if run_cache:
             self.stage_cache.pull(remote)
+    except RunCacheNotSupported as e:
+        logger.debug("failed to pull run cache: %s", e)
     except DownloadError as exc:
         failed_count += exc.amount
 
@@ -200,7 +203,7 @@ def fetch(  # noqa: PLR0913
 
 
 def _log_unversioned(data: list["DataIndex"]) -> tuple[list["DataIndex"], int]:
-    ret: list["DataIndex"] = []
+    ret: list[DataIndex] = []
     unversioned: list[str] = []
     for fs_index in data:
         remote = fs_index.storage_map[()].remote

dvc/repo/gc.py

@@ -96,7 +96,7 @@ def gc(  # noqa: C901, PLR0912, PLR0913
         repos = []
     all_repos = [Repo(path) for path in repos]
 
-    odb_to_obj_ids: "ObjectContainer" = {}
+    odb_to_obj_ids: ObjectContainer = {}
     with ExitStack() as stack:
         for repo in all_repos:
             stack.enter_context(repo.lock)

dvc/repo/get.py

@@ -54,7 +54,7 @@ def get(
         from dvc.fs import download
         from dvc.fs.data import DataFileSystem
 
-        fs: Union[DataFileSystem, "DVCFileSystem"]
+        fs: Union[DataFileSystem, DVCFileSystem]
         if os.path.isabs(path):
             fs = DataFileSystem(index=repo.index.data["local"])
             fs_path = fs.from_os_path(path)

dvc/repo/graph.py

@@ -22,7 +22,7 @@ def check_acyclic(graph: "DiGraph") -> None:
     except nx.NetworkXNoCycle:
         return
 
-    stages: set["Stage"] = set()
+    stages: set[Stage] = set()
     for from_node, to_node, _ in edges:
         stages.add(from_node)
         stages.add(to_node)

dvc/repo/imp_db.py

@@ -4,7 +4,6 @@ from funcy import compact
 
 from dvc.exceptions import OutputDuplicationError
 from dvc.repo.scm_context import scm_context
-from dvc.ui import ui
 from dvc.utils import resolve_output, resolve_paths
 
 if TYPE_CHECKING:
@@ -25,11 +24,6 @@ def imp_db(
     force: bool = False,
     connection: Optional[str] = None,
 ):
-    ui.warn("WARNING: import-db is an experimental feature.")
-    ui.warn(
-        "The functionality may change or break without notice, "
-        "which could lead to unexpected behavior."
-    )
     assert sql or table
     assert output_format in ("csv", "json")
 

dvc/repo/index.py

@@ -300,7 +300,7 @@ class Index:
         self._artifacts = artifacts or {}
         self._datasets: dict[str, list[dict[str, Any]]] = datasets or {}
         self._datasets_lock: dict[str, list[dict[str, Any]]] = datasets_lock or {}
-        self._collected_targets: dict[int, list["StageInfo"]] = {}
+        self._collected_targets: dict[int, list[StageInfo]] = {}
 
     @cached_property
     def rev(self) -> Optional[str]:
@@ -416,7 +416,7 @@ class Index:
 
     @cached_property
     def out_data_keys(self) -> dict[str, set["DataIndexKey"]]:
-        by_workspace: dict[str, set["DataIndexKey"]] = defaultdict(set)
+        by_workspace: dict[str, set[DataIndexKey]] = defaultdict(set)
 
         by_workspace["repo"] = set()
         by_workspace["local"] = set()
@@ -470,7 +470,7 @@ class Index:
 
     @cached_property
     def data_keys(self) -> dict[str, set["DataIndexKey"]]:
-        by_workspace: dict[str, set["DataIndexKey"]] = defaultdict(set)
+        by_workspace: dict[str, set[DataIndexKey]] = defaultdict(set)
 
         by_workspace["repo"] = set()
         by_workspace["local"] = set()
@@ -488,7 +488,7 @@ class Index:
     def metric_keys(self) -> dict[str, set["DataIndexKey"]]:
         from .metrics.show import _collect_top_level_metrics
 
-        by_workspace: dict[str, set["DataIndexKey"]] = defaultdict(set)
+        by_workspace: dict[str, set[DataIndexKey]] = defaultdict(set)
 
         by_workspace["repo"] = set()
 
@@ -509,7 +509,7 @@ class Index:
     def param_keys(self) -> dict[str, set["DataIndexKey"]]:
         from .params.show import _collect_top_level_params
 
-        by_workspace: dict[str, set["DataIndexKey"]] = defaultdict(set)
+        by_workspace: dict[str, set[DataIndexKey]] = defaultdict(set)
         by_workspace["repo"] = set()
 
         param_paths = _collect_top_level_params(self.repo)
@@ -525,7 +525,7 @@ class Index:
 
     @cached_property
     def plot_keys(self) -> dict[str, set["DataIndexKey"]]:
-        by_workspace: dict[str, set["DataIndexKey"]] = defaultdict(set)
+        by_workspace: dict[str, set[DataIndexKey]] = defaultdict(set)
 
         by_workspace["repo"] = set()
 
@@ -548,7 +548,7 @@ class Index:
 
     @cached_property
     def data(self) -> "dict[str, DataIndex]":
-        prefix: "DataIndexKey"
+        prefix: DataIndexKey
         loaded = False
 
         index = self.repo.data_index
@@ -600,7 +600,7 @@ class Index:
         if not onerror:
 
             def onerror(_target, _exc):
-                raise
+                raise  # noqa: PLE0704
 
         targets = ensure_list(targets)
         if not targets:
@@ -627,7 +627,7 @@ class Index:
         jobs: Optional[int] = None,
         push: bool = False,
     ) -> "ObjectContainer":
-        used: "ObjectContainer" = defaultdict(set)
+        used: ObjectContainer = defaultdict(set)
         pairs = self.collect_targets(targets, recursive=recursive, with_deps=with_deps)
         for stage, filter_info in pairs:
             for odb, objs in stage.get_used_objs(
@@ -748,7 +748,7 @@ class IndexView:
 
     @cached_property
     def out_data_keys(self) -> dict[str, set["DataIndexKey"]]:
-        by_workspace: dict[str, set["DataIndexKey"]] = defaultdict(set)
+        by_workspace: dict[str, set[DataIndexKey]] = defaultdict(set)
 
         by_workspace["repo"] = set()
         by_workspace["local"] = set()
@@ -764,7 +764,7 @@ class IndexView:
 
     @cached_property
     def _data_prefixes(self) -> dict[str, "_DataPrefixes"]:
-        prefixes: dict[str, "_DataPrefixes"] = defaultdict(
+        prefixes: dict[str, _DataPrefixes] = defaultdict(
             lambda: _DataPrefixes(set(), set())
         )
         for out, filter_info in self._filtered_outs:
@@ -781,7 +781,7 @@ class IndexView:
 
     @cached_property
     def data_keys(self) -> dict[str, set["DataIndexKey"]]:
-        ret: dict[str, set["DataIndexKey"]] = defaultdict(set)
+        ret: dict[str, set[DataIndexKey]] = defaultdict(set)
 
         for out, filter_info in self._filtered_outs:
             if not out.use_cache:
@@ -811,7 +811,7 @@ class IndexView:
             except KeyError:
                 return False
 
-        data: dict[str, Union["DataIndex", "DataIndexView"]] = {}
+        data: dict[str, Union[DataIndex, DataIndexView]] = {}
         for workspace, data_index in self._index.data.items():
             if self.stages:
                 data[workspace] = view(data_index, partial(key_filter, workspace))

dvc/repo/ls.py

@@ -77,7 +77,7 @@ def _ls(
     recursive: Optional[bool] = None,
     dvc_only: bool = False,
 ):
-    fs: "DVCFileSystem" = repo.dvcfs
+    fs: DVCFileSystem = repo.dvcfs
     fs_path = fs.from_os_path(path)
 
     fs_path = fs.info(fs_path)["name"]

dvc/repo/open_repo.py

@@ -146,7 +146,7 @@ def _cached_clone(url, rev):
 
 
 @wrap_with(threading.Lock())
-def _clone_default_branch(url, rev):  # noqa: PLR0912
+def _clone_default_branch(url, rev):
     """Get or create a clean clone of the url.
 
     The cloned is reactualized with git pull unless rev is a known sha.

dvc/repo/push.py

@@ -1,10 +1,14 @@
 from contextlib import suppress
 
 from dvc.exceptions import InvalidArgumentError, UploadError
+from dvc.log import logger
+from dvc.stage.cache import RunCacheNotSupported
 from dvc.ui import ui
 
 from . import locked
 
+logger = logger.getChild(__name__)
+
 
 def _rebuild(idx, path, fs, cb):
     from dvc_data.index import DataIndex, DataIndexEntry, Meta
@@ -95,8 +99,11 @@ def push(  # noqa: PLR0913
                 "Multiple rev push is unsupported for cloud versioned remotes"
             )
 
-    used_run_cache = self.stage_cache.push(remote) if run_cache else []
-    transferred_count += len(used_run_cache)
+    try:
+        used_run_cache = self.stage_cache.push(remote) if run_cache else []
+        transferred_count += len(used_run_cache)
+    except RunCacheNotSupported as e:
+        logger.debug("failed to push run cache: %s", e)
 
     if isinstance(targets, str):
         targets = [targets]

dvc/repo/reproduce.py

@@ -29,7 +29,7 @@ def collect_stages(
     recursive: bool = False,
     glob: bool = False,
 ) -> list["Stage"]:
-    stages: list["Stage"] = []
+    stages: list[Stage] = []
     for target in targets:
         stages.extend(repo.stage.collect(target, recursive=recursive, glob=glob))
     return ldistinct(stages)
@@ -166,10 +166,10 @@ def _reproduce(
 ) -> list["Stage"]:
     assert on_error in ("fail", "keep-going", "ignore")
 
-    result: list["Stage"] = []
-    failed: list["Stage"] = []
-    to_skip: dict["Stage", "Stage"] = {}
-    ret: Optional["Stage"] = None
+    result: list[Stage] = []
+    failed: list[Stage] = []
+    to_skip: dict[Stage, Stage] = {}
+    ret: Optional[Stage] = None
 
     force_state = dict.fromkeys(stages, force)
 
@@ -228,7 +228,7 @@ def reproduce(
     if not kwargs.get("interactive", False):
         kwargs["interactive"] = self.config["core"].get("interactive", False)
 
-    stages: list["Stage"] = []
+    stages: list[Stage] = []
     if not all_pipelines:
         targets_list = ensure_list(targets or PROJECT_FILE)
         stages = collect_stages(self, targets_list, recursive=recursive, glob=glob)

dvc/repo/scm_context.py

@@ -21,7 +21,7 @@ class SCMContext:
     def __init__(self, scm: "Base", config: Optional[dict[str, Any]] = None) -> None:
         from funcy import get_in
 
-        self.scm: "Base" = scm
+        self.scm: Base = scm
         self.autostage: bool = get_in(
             config or {}, ["core", "autostage"], default=False
         )

dvc/repo/stage.py

@@ -108,7 +108,7 @@ def locked(f):
 
 class StageLoad:
     def __init__(self, repo: "Repo") -> None:
-        self.repo: "Repo" = repo
+        self.repo: Repo = repo
 
     @property
     def fs(self):

dvc/repo/update.py

@@ -34,7 +34,7 @@ def update(  # noqa: C901
         raise InvalidArgumentError("--remote can't be used without --to-remote")
 
     import_stages = set()
-    other_stage_infos: list["StageInfo"] = []
+    other_stage_infos: list[StageInfo] = []
 
     for stage_info in self.index.collect_targets(targets, recursive=recursive):
         if stage_info.stage.is_import:

dvc/repo/worktree.py

@@ -166,7 +166,7 @@ def update_worktree_stages(repo: "Repo", stage_infos: Iterable["StageInfo"]):
 
     view = IndexView(repo.index, stage_infos, outs_filter=outs_filter)
     local_index = view.data["repo"]
-    remote_indexes: dict[str, tuple["Remote", "DataIndex"]] = {}
+    remote_indexes: dict[str, tuple[Remote, DataIndex]] = {}
     for stage in view.stages:
         for out in stage.outs:
             _update_worktree_out(repo, out, local_index, remote_indexes)

dvc/stage/__init__.py

@@ -290,7 +290,7 @@ class Stage(params.StageParams):
         )
 
     def short_description(self) -> Optional["str"]:
-        desc: Optional["str"] = None
+        desc: Optional[str] = None
         if self.desc:
             with suppress(ValueError):
                 # try to use first non-empty line as a description
@@ -339,7 +339,7 @@ class Stage(params.StageParams):
         for out in self.outs:
             status = out.status()
             if status:
-                if allow_missing and status[str(out)] == "not in cache":
+                if allow_missing and status[str(out)] in ["not in cache", "deleted"]:
                     continue
                 logger.debug(
                     "Output '%s' of %s changed because it is '%s'.",
@@ -418,17 +418,28 @@ class Stage(params.StageParams):
 
     @rwlocked(read=["deps"], write=["outs"])
     def reproduce(self, interactive=False, **kwargs) -> Optional["Stage"]:
-        if not (
-            kwargs.get("force", False)
-            or self.changed(
-                kwargs.get("allow_missing", False), kwargs.pop("upstream", None)
-            )
-        ):
+        force = kwargs.get("force", False)
+        allow_missing = kwargs.get("allow_missing", False)
+        pull = kwargs.get("pull", False)
+        upstream = kwargs.pop("upstream", None)
+        if force:
+            pass
+        # Skip stages with missing data if otherwise unchanged
+        elif not self.changed(allow_missing, upstream):
             if not isinstance(self, PipelineStage) and self.is_data_source:
                 logger.info("'%s' didn't change, skipping", self.addressing)
             else:
                 logger.info("Stage '%s' didn't change, skipping", self.addressing)
             return None
+        # Pull stages with missing data if otherwise unchanged
+        elif not self.changed(True, upstream) and pull:
+            try:
+                logger.info("Pulling data for %s", self)
+                self.repo.pull(self.addressing, jobs=kwargs.get("jobs", None))
+                self.checkout()
+                return None
+            except CheckoutError:
+                logger.info("Unable to pull data for %s", self)
 
         msg = f"Going to reproduce {self}. Are you sure you want to continue?"
         if interactive and not prompt.confirm(msg):
@@ -597,16 +608,10 @@ class Stage(params.StageParams):
         if (self.cmd or self.is_import) and not self.frozen and not dry:
             self.remove_outs(ignore_remove=False, force=False)
 
-        if (
-            self.is_import and (not self.frozen or kwargs.get("pull"))
-        ) or self.is_partial_import:
+        if (self.is_import and not self.frozen) 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)
-        elif kwargs.get("pull"):
-            logger.info("Pulling data for %s", self)
-            self.repo.pull(self.addressing, jobs=kwargs.get("jobs", None))
-            self.checkout()
         elif not dry:
             args = ("outputs", "frozen ") if self.frozen else ("data sources", "")
             logger.info("Verifying %s in %s%s", *args, self)
@@ -653,7 +658,7 @@ class Stage(params.StageParams):
     def checkout(
         self, allow_missing: bool = False, **kwargs
     ) -> dict[str, list["StrPath"]]:
-        stats: dict[str, list["StrPath"]] = defaultdict(list)
+        stats: dict[str, list[StrPath]] = defaultdict(list)
         if self.is_partial_import:
             return stats
 

dvc/stage/cache.py

@@ -229,7 +229,7 @@ class StageCache:
         if not dry:
             cached_stage.checkout()
 
-    def transfer(self, from_odb, to_odb):
+    def transfer(self, from_odb, to_odb, force=True):
         from dvc.fs import HTTPFileSystem, LocalFileSystem
         from dvc.fs.callbacks import TqdmCallback
 
@@ -258,10 +258,11 @@ class StageCache:
 
             dst = to_fs.join(to_odb.path, rel)
             key = to_fs.parent(dst)
+
             # check if any build cache already exists for this key
             # TODO: check if MaxKeys=1 or something like that applies
             # or otherwise this will take a lot of time!
-            if to_fs.exists(key) and first(to_fs.find(key)):
+            if not force and to_fs.exists(key) and first(to_fs.find(key)):
                 continue
 
             src_name = from_fs.name(src)

dvc/stage/loader.py

@@ -50,7 +50,7 @@ class StageLoader(Mapping):
         from dvc.output import Output, merge_file_meta_from_cloud
 
         assert isinstance(lock_data, dict)
-        items: Iterable[tuple[str, "Output"]] = chain(
+        items: Iterable[tuple[str, Output]] = chain(
             ((StageParams.PARAM_DEPS, dep) for dep in stage.deps),
             ((StageParams.PARAM_OUTS, out) for out in stage.outs),
         )

dvc/stage/run.py

@@ -2,6 +2,9 @@ import os
 import signal
 import subprocess
 import threading
+from functools import cache
+
+from packaging.version import InvalidVersion, Version
 
 from dvc.log import logger
 from dvc.utils import fix_env
@@ -12,25 +15,59 @@ from .exceptions import StageCmdFailedError
 logger = logger.getChild(__name__)
 
 
+@cache
+def _fish_supports_no_config(executable) -> bool:
+    """
+    Check if the fish shell supports the --no-config option.
+
+    Parameters:
+    executable (str): The path to the fish shell executable.
+
+    Returns:
+    bool: True if the fish version is greater than 3.3.0, False otherwise.
+    """
+    try:
+        output = subprocess.check_output(  # noqa: S603
+            [executable, "--version"],
+            text=True,
+        )
+        version = Version(output.split(" ")[-1].strip())
+        version_to_check = Version("3.3.0")
+        return version >= version_to_check
+    except (subprocess.CalledProcessError, IndexError, InvalidVersion):
+        logger.trace("could not check fish version, defaulting to False")
+        return False
+
+
+def _warn_if_fish(executable):
+    if (
+        executable is None
+        or os.path.basename(executable) != "fish"
+        or _fish_supports_no_config(executable)
+    ):
+        return
+    logger.warning(
+        "DVC detected that you are using a version of fish shell below 3.3.0 "
+        "Be aware that it might cause problems by overwriting "
+        "your current environment variables with values defined "
+        "in 'config.fish', which might affect your command. See "
+        "https://github.com/iterative/dvc/issues/1307. "
+    )
+
+
 def _make_cmd(executable, cmd):
     if executable is None:
         return cmd
-    opts = {"zsh": ["--no-rcs"], "bash": ["--noprofile", "--norc"]}
+    opts = {
+        "zsh": ["--no-rcs"],
+        "bash": ["--noprofile", "--norc"],
+        "fish": [],
+    }
     name = os.path.basename(executable).lower()
-    return [executable, *opts.get(name, []), "-c", cmd]
-
-
-def warn_if_fish(executable):
-    if executable is None or os.path.basename(os.path.realpath(executable)) != "fish":
-        return
-
-    logger.warning(
-        "DVC detected that you are using fish as your default "
-        "shell. Be aware that it might cause problems by overwriting "
-        "your current environment variables with values defined "
-        "in '.fishrc', which might affect your command. See "
-        "https://github.com/iterative/dvc/issues/1307. "
-    )
+    opt = opts.get(name, [])
+    if name == "fish" and _fish_supports_no_config(executable):
+        opt.append("--no-config")
+    return [executable, *opt, "-c", cmd]
 
 
 def _enforce_cmd_list(cmd):
@@ -106,7 +143,7 @@ def cmd_run(stage, dry=False, run_env=None):
     executable = get_executable()
 
     if not dry:
-        warn_if_fish(executable)
+        _warn_if_fish(executable)
 
     for cmd in commands:
         display_command(cmd)
@@ -117,7 +154,11 @@ def cmd_run(stage, dry=False, run_env=None):
 
 
 def _pull_missing_deps(stage):
+    from dvc.dependency import DatasetDependency, DbDependency
+
     for dep in stage.deps:
+        if isinstance(dep, (DatasetDependency, DbDependency)):
+            continue
         if not dep.exists:
             stage.repo.pull(dep.def_path)
 

dvc/stage/utils.py

@@ -302,9 +302,9 @@ def _get_stage_files(stage: "Stage") -> list[str]:
             and dep.is_in_repo
             and not stage.repo.dvcfs.isdvc(stage.repo.dvcfs.from_os_path(str(dep)))
         ):
-            ret.append(relpath(dep.fs_path))
+            ret.append(relpath(dep.fs_path))  # noqa: PERF401
 
     for out in stage.outs:
         if not out.use_scm_ignore and out.is_in_repo:
-            ret.append(relpath(out.fs_path))
+            ret.append(relpath(out.fs_path))  # noqa: PERF401
     return ret

dvc/testing/benchmarks/cli/commands/test_add.py

@@ -1,2 +1,19 @@
-def test_add(bench_dvc, tmp_dir, dvc, dataset):
-    bench_dvc("add", dataset)
+from .test_checkout import _skip_unsupported_link
+
+
+def generate_test(*, link_type="copy"):
+    def _test_add(bench_dvc, tmp_dir, dvc, dataset):
+        _skip_unsupported_link((tmp_dir / ".dvc" / "cache"), tmp_dir, link_type)
+
+        with dvc.config.edit() as conf:
+            conf["cache"]["type"] = link_type
+
+        bench_dvc("add", dataset)
+        bench_dvc("add", dataset, name="noop")
+
+    return _test_add
+
+
+test_add_copy = generate_test(link_type="copy")
+test_add_symlink = generate_test(link_type="symlink")
+test_add_hardlink = generate_test(link_type="hardlink")

dvc/testing/benchmarks/cli/commands/test_data_status.py

@@ -1,9 +1,5 @@
 from shutil import rmtree
 
-import pytest
-
-pytestmark = pytest.mark.requires(minversion=(2, 15, 0), reason="new command")
-
 
 def test_data_status(bench_dvc, tmp_dir, scm, dvc, make_dataset):
     args = ("data", "status")

dvc/testing/benchmarks/cli/commands/test_exp_show.py

@@ -1,7 +1,3 @@
-import pytest
-
-
-@pytest.mark.requires(minversion=(2, 28, 0))
 def test_exp_show(make_project, monkeypatch, bench_dvc, dvc_bin):
     url = "https://github.com/iterative/example-get-started"
     rev = "main"

dvc/testing/benchmarks/cli/commands/test_import.py

@@ -2,6 +2,12 @@ import pytest
 
 
 @pytest.mark.flaky(reruns=3)
+@pytest.mark.requires(
+    "!=3.53.*,!=3.54.0",
+    reason="Takes 10 mins to run. Regression in 3.53.0, fixed in 3.54.1",
+)
+# Introduced in https://github.com/iterative/dvc/pull/10388.
+# Fixed in https://github.com/iterative/dvc/pull/10531.
 def test_import(bench_dvc, tmp_dir, scm, dvc, make_dataset, remote):
     dataset = make_dataset(
         cache=False, files=False, dvcfile=True, commit=True, remote=True

dvc/testing/benchmarks/cli/commands/test_plots.py

@@ -1,10 +1,7 @@
-import pytest
-
 from dvc.repo import Repo
 from dvc.testing.benchmarks.fixtures import _pull
 
 
-@pytest.mark.requires(minversion=(2, 34, 0), reason="top-level plots not supported")
 def test_plots(project, bench_dvc):
     with Repo() as dvc:
         _pull(dvc)

dvc/testing/benchmarks/cli/stories/use_cases/test_sharing.py

@@ -1,9 +1,8 @@
 import shutil
 
 
-def test_sharing(bench_dvc, tmp_dir, dvc, dataset, remote):
-    bench_dvc("add", dataset)
-    bench_dvc("add", dataset, name="noop")
+def test_sharing(bench_dvc, tmp_dir, dvc, make_dataset, remote):
+    dataset = make_dataset(cache=True, dvcfile=True)
 
     bench_dvc("push")
     bench_dvc("push", name="noop")
@@ -13,6 +12,3 @@ def test_sharing(bench_dvc, tmp_dir, dvc, dataset, remote):
 
     bench_dvc("fetch")
     bench_dvc("fetch", name="noop")
-
-    bench_dvc("checkout")
-    bench_dvc("checkout", name="noop")

dvc/testing/benchmarks/conftest.py

@@ -1 +1,3 @@
 from .fixtures import *  # noqa: F403
+
+pytest_plugins = ["dvc.testing.plugin"]

dvc/testing/benchmarks/fixtures.py

@@ -1,7 +1,8 @@
 import os
 import shutil
+import sys
 from pathlib import Path
-from subprocess import check_output
+from subprocess import check_call, check_output
 from typing import Optional
 
 import pytest
@@ -23,13 +24,14 @@ class VirtualEnv:
         self.bin = self.path / ("Scripts" if os.name == "nt" else "bin")
 
     def create(self) -> None:
-        import virtualenv
+        check_call([sys.executable, "-m", "uv", "venv", self.path])  # noqa:S603
 
-        virtualenv.cli_run([os.fspath(self.path)])
+    def install(self, *packages: str) -> None:
+        check_call([sys.executable, "-m", "uv", "pip", "install", *packages])  # noqa: S603
 
     def run(self, cmd: str, *args: str, env: Optional[dict[str, str]] = None) -> None:
         exe = self.which(cmd)
-        check_output([exe, *args], env=env)  # noqa: S603
+        check_call([exe, *args], env=env)  # noqa: S603
 
     def which(self, cmd: str) -> str:
         assert self.bin.exists()
@@ -58,8 +60,8 @@ def dvc_venvs():
 
 
 @pytest.fixture(scope="session")
-def dvc_git_repo(tmp_path_factory, bench_config):
-    url = bench_config.dvc_git_repo
+def dvc_repo(tmp_path_factory, bench_config):
+    url = bench_config.dvc_repo
 
     if os.path.isdir(url):
         return url
@@ -71,8 +73,10 @@ def dvc_git_repo(tmp_path_factory, bench_config):
 
 
 @pytest.fixture(scope="session")
-def dvc_bench_git_repo(tmp_path_factory, bench_config):
-    url = bench_config.dvc_bench_git_repo
+def dvc_bench_repo(tmp_path_factory, bench_config):
+    url = bench_config.dvc_bench_repo
+    if url is None:
+        pytest.skip("--dvc-bench-repo is not set")
 
     if os.path.isdir(url):
         return Path(url)
@@ -88,56 +92,55 @@ def make_dvc_bin(
     dvc_rev,
     dvc_venvs,
     make_dvc_venv,
-    dvc_git_repo,
+    dvc_repo,
     bench_config,
     request,
 ):
     if dvc_rev:
-        venv = dvc_venvs.get(dvc_rev)
+        venv: VirtualEnv = dvc_venvs.get(dvc_rev)
         if not venv:
             venv = make_dvc_venv(dvc_rev)
-            venv.run("pip", "install", "-U", "pip")
             if bench_config.dvc_install_deps:
-                egg = f"dvc[{bench_config.dvc_install_deps}]"
+                pkg = f"dvc[{bench_config.dvc_install_deps}]"
             else:
-                egg = "dvc"
-            venv.run("pip", "install", f"git+file://{dvc_git_repo}@{dvc_rev}#egg={egg}")
-            if dvc_rev in ["2.18.1", "2.11.0", "2.6.3"]:
-                venv.run("pip", "install", "fsspec==2022.11.0")
+                pkg = "dvc"
+            packages = [f"{pkg} @ git+file://{dvc_repo}@{dvc_rev}"]
+            try:
+                if version.Version(dvc_rev) < version.Version("3.50.3"):
+                    packages.append("pygit2==1.14.1")
+            except version.InvalidVersion:
+                pass
+            venv.install(*packages)
+
             dvc_venvs[dvc_rev] = venv
         dvc_bin = venv.which("dvc")
     else:
         dvc_bin = bench_config.dvc_bin
 
     def _dvc_bin(*args):
-        return check_output([dvc_bin, *args], text=True)  # noqa: S603
+        check_call([dvc_bin, *args])  # noqa: S603
 
-    _dvc_bin.version = parse_tuple(_dvc_bin("--version"))  # type: ignore[attr-defined]
+    _dvc_bin.version = check_output([dvc_bin, "--version"], text=True)  # type: ignore[attr-defined]  # noqa: S603
     return _dvc_bin
 
 
-def parse_tuple(version_string):
-    from packaging.version import Version
-
-    parsed = version.parse(version_string)
-    assert isinstance(parsed, Version)
-    return (parsed.major, parsed.minor, parsed.micro)
-
-
 @pytest.fixture
 def dvc_bin(request, make_dvc_bin):
     if marker := request.node.get_closest_marker("requires"):
-        minversion = marker.kwargs.get("minversion") or first(marker.args)
-        assert minversion, (
-            "'minversion' needs to be specified as"
-            " a positional or a keyword argument"
-        )
-        reason = marker.kwargs.get("reason", "")
-        if isinstance(minversion, str):
-            minversion = parse_tuple(minversion)
-        if make_dvc_bin.version < minversion:
-            version_repr = ".".join(map(str, minversion))
-            pytest.skip(f"requires dvc>={version_repr}: {reason}")
+        from packaging.specifiers import SpecifierSet
+        from packaging.version import Version, parse
+
+        spec = first(marker.args)
+        assert spec is not None
+        spec = SpecifierSet(spec) if isinstance(spec, str) else spec
+        reason = marker.kwargs["reason"]
+        dvc_version = make_dvc_bin.version
+        version = Version(parse(dvc_version).base_version)
+        if version not in spec:
+            pytest.skip(
+                f"Version {dvc_version} "
+                f"does not satisfy requirement {spec!r}: {reason}"
+            )
     return make_dvc_bin
 
 
@@ -164,11 +167,19 @@ def make_bench(request):
 
 
 @pytest.fixture
-def bench_dvc(dvc_bin, make_bench):
+def bench_dvc(request, dvc_bin, make_bench):
     def _bench_dvc(*args, **kwargs):
         name = kwargs.pop("name", None)
         name = f"-{name}" if name else ""
         bench = make_bench(args[0] + name)
+        if request.config.getoption("--benchmark-cprofile-dump") or kwargs.pop(
+            "cprofile", False
+        ):
+            cprofile_results = request.config.invocation_params.dir / "prof"
+            cprofile_results.mkdir(exist_ok=True)
+            stats_file = cprofile_results / f"{bench.name}.prof"
+            args = (*args, "--cprofile-dump", stats_file)
+
         return bench.pedantic(dvc_bin, args=args, **kwargs)
 
     return _bench_dvc
@@ -185,14 +196,14 @@ def _pull(repo, *args):
 
 
 @pytest.fixture
-def make_dataset(request, bench_config, tmp_dir, dvc_bench_git_repo):
+def make_dataset(request, bench_config, tmp_dir, dvc_bench_repo):
     def _make_dataset(
         dvcfile=False, files=True, cache=False, commit=False, remote=False
     ):
         from dvc.repo import Repo
 
         path = tmp_dir / "dataset"
-        root = dvc_bench_git_repo
+        root = dvc_bench_repo
         src = root / "data" / bench_config.dataset / "dataset"
         src_dvc = src.with_suffix(".dvc")
 
@@ -251,7 +262,7 @@ def make_project(tmp_path_factory):
 @pytest.fixture
 def project(bench_config, monkeypatch, make_project):
     rev = bench_config.project_rev
-    url = bench_config.project_git_repo
+    url = bench_config.project_repo
 
     if os.path.isdir(url):
         path = url

dvc/testing/benchmarks/plugin.py

@@ -1,57 +1,84 @@
+import os
+from dataclasses import dataclass, fields
+from pathlib import Path
+from typing import Optional
+
 DEFAULT_DVC_BIN = "dvc"
-DEFAULT_DVC_GIT_REPO = "https://github.com/iterative/dvc"
-DEFAULT_DVC_BENCH_GIT_REPO = "https://github.com/iterative/dvc-bench"
-DEFAULT_PROJECT_GIT_REPO = "https://github.com/iterative/example-get-started"
+DEFAULT_DVC_REPO = os.fspath(Path(__file__).parents[3])
+DEFAULT_PROJECT_REPO = "https://github.com/iterative/example-get-started"
 
 
 def pytest_report_header(config):
     bconf = config.bench_config
-    return f"dvc-bench: (dataset: {bconf.dataset!r}, revs: {bconf.dvc_revs!r})"
+    return f"dvc-bench: {bconf}"
 
 
 def pytest_generate_tests(metafunc):
-    str_revs = metafunc.config.getoption("--dvc-revs")
-    revs = str_revs.split(",") if str_revs else [None]
+    revs = metafunc.config.getoption("--dvc-revs")
+    if not revs:
+        revs = [None]
     if "dvc_rev" in metafunc.fixturenames:
         metafunc.parametrize("dvc_rev", revs, scope="session")
 
 
+@dataclass
 class DVCBenchConfig:
-    def __init__(self):
-        self.dataset = "small"
-        self.dvc_bin = DEFAULT_DVC_BIN
-        self.dvc_revs = None
-        self.dvc_git_repo = DEFAULT_DVC_GIT_REPO
-        self.dvc_install_deps = None
-        self.dvc_bench_git_repo = DEFAULT_DVC_BENCH_GIT_REPO
-        self.project_rev = None
-        self.project_git_repo = DEFAULT_PROJECT_GIT_REPO
+    dataset: str = "tiny"
+    dvc_repo: str = DEFAULT_DVC_REPO
+    dvc_bench_repo: Optional[str] = None
+    project_repo: str = DEFAULT_PROJECT_REPO
+    project_rev: Optional[str] = None
+    dvc_bin: str = DEFAULT_DVC_BIN
+    dvc_revs: Optional[list[str]] = None
+    dvc_install_deps: Optional[str] = None
+
+    def __repr__(self):
+        args = ", ".join(
+            f"{f.name}={val!r}"
+            for f in fields(self)
+            if (val := getattr(self, f.name)) != f.default
+        )
+        return f"{self.__class__.__name__}({args})"
 
 
 def pytest_configure(config):
     config.addinivalue_line(
-        "markers", "requires(minversion): mark a test as requiring minimum DVC version"
+        "markers",
+        "requires(spec): mark a test to run only on versions that satisfy the spec",
+    )
+    config.bench_config = DVCBenchConfig(
+        dataset=config.getoption("--dataset"),
+        dvc_repo=config.getoption("--dvc-repo"),
+        dvc_bench_repo=config.getoption("--dvc-bench-repo"),
+        project_repo=config.getoption("--project-repo"),
+        project_rev=config.getoption("--project-rev"),
+        dvc_bin=config.getoption("--dvc-bin"),
+        dvc_revs=config.getoption("--dvc-revs"),
+        dvc_install_deps=config.getoption("--dvc-install-deps"),
     )
 
-    config.bench_config = DVCBenchConfig()
-    config.bench_config.dataset = config.getoption("--dataset")
-    config.bench_config.dvc_bin = config.getoption("--dvc-bin")
-    config.bench_config.dvc_revs = config.getoption("--dvc-revs")
-    config.bench_config.dvc_git_repo = config.getoption("--dvc-git-repo")
-    config.bench_config.dvc_install_deps = config.getoption("--dvc-install-deps")
-    config.bench_config.dvc_bench_git_repo = config.getoption("--dvc-bench-git-repo")
-    config.bench_config.project_rev = config.getoption("--project-rev")
-    config.bench_config.project_git_repo = config.getoption("--project-git-repo")
+
+def resolve_path(path):
+    if os.path.isdir(path):
+        return os.path.abspath(path)
+    return path
 
 
 def pytest_addoption(parser):
     parser.addoption(
         "--dataset",
         type=str,
-        default="small",
+        default="tiny",
         help="Dataset name to use in tests (e.g. tiny/small/large/mnist/etc)",
     )
 
+    parser.addoption(
+        "--benchmark-cprofile-dump",
+        action="store_true",
+        default=False,
+        help="Save cprofile results",
+    )
+
     parser.addoption(
         "--dvc-bin",
         type=str,
@@ -61,36 +88,34 @@ def pytest_addoption(parser):
 
     parser.addoption(
         "--dvc-revs",
-        type=str,
+        type=lambda revs: revs.split(","),
         help=("Comma-separated list of DVC revisions to test (overrides `--dvc-bin`)"),
     )
 
     parser.addoption(
-        "--dvc-git-repo",
-        type=str,
-        default=DEFAULT_DVC_GIT_REPO,
+        "--dvc-repo",
+        type=resolve_path,
+        default=DEFAULT_DVC_REPO,
         help="Path or url to dvc git repo",
     )
 
     parser.addoption(
         "--dvc-install-deps",
         type=str,
-        default="",
         help="Comma-separated list of DVC installation packages",
     )
 
     parser.addoption(
-        "--dvc-bench-git-repo",
-        type=str,
-        default=DEFAULT_DVC_BENCH_GIT_REPO,
+        "--dvc-bench-repo",
+        type=resolve_path,
+        default=None,
         help="Path or url to dvc-bench git repo (for loading benchmark datasets)",
     )
 
     parser.addoption("--project-rev", type=str, help="Project revision to test")
-
     parser.addoption(
-        "--project-git-repo",
-        type=str,
-        default=DEFAULT_PROJECT_GIT_REPO,
+        "--project-repo",
+        type=resolve_path,
+        default=DEFAULT_PROJECT_REPO,
         help="Path or url to dvc project",
     )

dvc/testing/fixtures.py

@@ -228,10 +228,10 @@ def docker_services(tmp_path_factory, request):
         pytest.skip("disabled for Windows on CI")
 
     try:
-        subprocess.check_output(
+        subprocess.check_output(  # noqa: S602
             "docker ps",  # noqa: S607
             stderr=subprocess.STDOUT,
-            shell=True,  # noqa: S602
+            shell=True,
         )
     except subprocess.CalledProcessError as err:
         out = (err.output or b"").decode("utf-8")

dvc/testing/path_info.py

@@ -32,7 +32,7 @@ class PathInfo(pathlib.PurePath, _BasePath):
 
         def __new__(cls, *args):
             if cls is PathInfo:
-                cls = WindowsPathInfo if os.name == "nt" else PosixPathInfo
+                cls = WindowsPathInfo if os.name == "nt" else PosixPathInfo  # noqa: PLW0642
 
             return cls._from_parts(args)  # type: ignore[attr-defined]
 
@@ -79,13 +79,13 @@ class PathInfo(pathlib.PurePath, _BasePath):
             and self._cparts[:n] == other._cparts  # type: ignore[attr-defined]
         )
 
-    def relative_to(self, other):
+    def relative_to(self, other, *args, **kwargs):
         # pathlib relative_to raises exception when one path is not a direct
         # descendant of the other when os.path.relpath would return abspath.
         # For DVC PathInfo we only need the relpath behavior.
         # See: https://bugs.python.org/issue40358
         try:
-            path = super().relative_to(other)
+            path = super().relative_to(other, *args, **kwargs)
         except ValueError:
             path = relpath(self, other)
         return self.__class__(path)

dvc/testing/plugin.py

@@ -18,3 +18,9 @@ def pytest_configure(config):
     from .benchmarks.plugin import pytest_configure as bench_configure
 
     bench_configure(config)
+
+
+def pytest_report_header(config):
+    from .benchmarks.plugin import pytest_report_header as bench_report_header
+
+    return bench_report_header(config)

dvc/testing/remote_tests.py

@@ -3,7 +3,6 @@ import shutil
 
 import pytest
 
-from dvc.stage.cache import RunCacheNotSupported
 from dvc.utils.fs import remove
 from dvc_data.hashfile.tree import Tree
 
@@ -97,8 +96,10 @@ class TestRemote:
         status_dir = dvc.cloud.status(dir_hashes)
         _check_status(status_dir, ok=dir_hashes)
 
-    @pytest.mark.xfail(raises=RunCacheNotSupported, strict=False)
     def test_stage_cache_push_pull(self, tmp_dir, dvc, remote):
+        if remote.scheme in ("http", "https"):
+            pytest.skip("HTTP remote does not support stage cache")
+
         tmp_dir.gen("foo", "foo")
         stage = dvc.stage.add(
             deps=["foo"], outs=["bar"], name="copy-foo-bar", cmd="cp foo bar"

dvc/testing/tmp_dir.py

@@ -71,7 +71,7 @@ class TmpDir(pathlib.Path):
 
         def __new__(cls, *args, **kwargs):
             if cls is TmpDir:
-                cls = WindowsTmpDir if os.name == "nt" else PosixTmpDir
+                cls = WindowsTmpDir if os.name == "nt" else PosixTmpDir  # noqa: PLW0642
 
             # init parameter and `_init` method has been removed in Python 3.10.
             kw = {"init": False} if sys.version_info < (3, 10) else {}

dvc/ui/__init__.py

@@ -93,7 +93,7 @@ class Console:
             styled=styled,
         )
 
-    def write_json(  # noqa: PLR0913
+    def write_json(
         self,
         data: Any,
         indent: Optional[int] = None,

dvc/utils/__init__.py

@@ -243,7 +243,7 @@ def env2bool(var, undefined=False):
     var = os.getenv(var, None)
     if var is None:
         return undefined
-    return bool(re.search("1|y|yes|true", var, flags=re.I))
+    return bool(re.search("1|y|yes|true", var, flags=re.IGNORECASE))
 
 
 def resolve_output(inp: str, out: Optional[str], force=False) -> str:

dvc/utils/studio.py

@@ -23,7 +23,7 @@ if TYPE_CHECKING:
 
 logger = logger.getChild(__name__)
 
-STUDIO_URL = "https://studio.iterative.ai"
+STUDIO_URL = "https://studio.dvc.ai"
 
 
 def post(

pyproject.toml

@@ -38,15 +38,15 @@ dependencies = [
     "distro>=1.3",
     "dpath<3,>=2.1.0",
     "dulwich",
-    "dvc-data>=3.15,<3.16",
+    "dvc-data>=3.16.2,<3.17",
     "dvc-http>=2.29.0",
     "dvc-objects",
     "dvc-render>=1.0.1,<2",
-    "dvc-studio-client>=0.20,<1",
+    "dvc-studio-client>=0.21,<1",
     "dvc-task>=0.3.0,<1",
     "flatten_dict<1,>=0.4.1",
-    "flufl.lock>=5,<8", # https://github.com/iterative/dvc/issues/9654
-    "fsspec",
+    "flufl.lock>=8.1.0,<9",
+    "fsspec>=2024.2.0",
     "funcy>=1.14",
     "grandalf<1,>=0.7",
     "gto>=1.6.0,<2",
@@ -65,7 +65,7 @@ dependencies = [
     "requests>=2.22",
     "rich>=12",
     "ruamel.yaml>=0.17.11",
-    "scmrepo>=3.3.2,<4",
+    "scmrepo>=3.3.4,<4",
     "shortuuid>=0.5",
     "shtab<2,>=1.3.4",
     "tabulate>=0.8.7",
@@ -83,7 +83,7 @@ gdrive = ["dvc-gdrive>=3,<4"]
 gs = ["dvc-gs>=3,<4"]
 hdfs = ["dvc-hdfs>=3,<4"]
 lint = [
-    "mypy==1.10.0",
+    "mypy==1.11.2",
     "pandas-stubs",
     "types-colorama",
     "types-psutil",
@@ -101,7 +101,7 @@ ssh_gssapi = ["dvc-ssh[gssapi]>=4,<5"]
 testing = [
   "pytest-benchmark[histogram]",
   "pytest-test-utils",
-  "virtualenv",
+  "uv",
 ]
 tests = [
     "dvc[testing]",
@@ -159,7 +159,7 @@ namespaces = false
 write_to = "dvc/_dvc_version.py"
 
 [tool.pytest.ini_options]
-addopts = "-ra --cov-config pyproject.toml --dist worksteal --benchmark-skip"
+addopts = "-ra --cov-config pyproject.toml"
 filterwarnings = [
     "error::ResourceWarning",
     "error::pytest.PytestUnraisableExceptionWarning",
@@ -256,6 +256,7 @@ module = [
 
 [tool.codespell]
 ignore-words-list = "ba,datas,fo,uptodate,cachable,falsy"
+skip = "CODE_OF_CONDUCT.md"
 
 [tool.ruff]
 output-format = "full"
@@ -263,21 +264,19 @@ show-fixes = true
 
 [tool.ruff.lint]
 ignore = [
-    "N818", "S101", "ISC001", "PT004", "PT007", "RET502", "RET503", "SIM105", "SIM108", "SIM117",
-    "TRY003", "TRY300", "PLR2004", "PLW2901", "LOG007",
+    "N818", "S101", "ISC001", "PT007", "RET502", "RET503", "SIM105", "SIM108", "SIM117",
+    "TRY003", "TRY300", "PERF203", "PLR2004", "PLW2901", "LOG007",
 ]
 select = [
     "F", "E", "W", "C90", "I", "N", "UP", "YTT", "ASYNC", "S", "BLE", "B", "A", "C4", "DTZ", "T10",
-    "EXE", "ISC", "ICN", "G", "INP", "PIE", "T20", "PYI", "PT", "Q", "RSE", "RET",
+    "EXE", "ISC", "ICN", "LOG", "G", "INP", "PIE", "T20", "PYI", "PT", "Q", "RSE", "RET",
     "SLOT", "SIM", "TID", "TCH", "ARG", "PGH", "PLC", "PLE", "PLR", "PLW", "TRY",
-    "FLY", "PERF101", "LOG", "RUF", "RUF022", "RUF023", "RUF024", "RUF025", "RUF026", "RUF027"
+    "FLY", "PERF", "FURB", "RUF", "RUF022", "RUF023", "RUF024", "RUF026", "RUF027", "RUF029", "RUF030",
 ]
 preview = true
 explicit-preview-rules = true
 
 [tool.ruff.lint.flake8-pytest-style]
-fixture-parentheses = false
-mark-parentheses = false
 parametrize-names-type = "csv"
 raises-extend-require-match-for = ["dvc.exceptions.DvcException", "dvc.scm.SCMError", "scmrepo.exceptions.SCMError"]
 
@@ -304,4 +303,4 @@ max-args = 10
 "dvc/commands/**" = ["N806"]
 "dvc/testing/**" = ["ARG002"]
 "dvc/testing/benchmarks/**" = ["ARG001"]
-"tests/**" = ["S", "ARG001", "ARG002", "TRY002", "TRY301"]
+"tests/**" = ["S", "ARG001", "ARG002", "TRY002", "TRY301", "PERF"]

tests/__main__.py

@@ -3,4 +3,4 @@ if __name__ == "__main__":
 
     import pytest
 
-    sys.exit(pytest.main(["-v", "-n=logical", *sys.argv[1:]]))
+    sys.exit(pytest.main(["-v", "-n=logical", "--dist=worksteal", *sys.argv[1:]]))

tests/func/artifacts/test_artifacts.py

@@ -281,3 +281,12 @@ def test_artifacts_download_subrepo(tmp_dir, scm, sub):
     assert Artifacts.get(subdir.fs_path, name, force=True) == result
     with subdir.chdir():
         assert Artifacts.get(".", name, force=True) == result
+
+
+def test_artifacts_download_studio(tmp_dir, dvc, mocker):
+    with dvc.config.edit("global") as conf:
+        conf["studio"]["token"] = "mytoken"
+
+    download_studio = mocker.patch("dvc.repo.artifacts.Artifacts._download_studio")
+    Artifacts.get("myart.pkl", "myart.pkl")
+    assert download_studio.call_args.kwargs["dvc_studio_config"]["token"] == "mytoken"

tests/func/repro/test_repro_allow_missing.py

@@ -55,3 +55,16 @@ def test_repro_allow_missing_upstream_stage_modified(
     ret = dvc.reproduce(pull=True, allow_missing=True)
     # create-foo is skipped ; copy-foo pulls modified dep
     assert len(ret) == 1
+
+
+def test_repro_allow_missing_cached(tmp_dir, dvc):
+    tmp_dir.gen("fixed", "fixed")
+    dvc.stage.add(name="create-foo", cmd="echo foo > foo", deps=["fixed"], outs=["foo"])
+    dvc.stage.add(name="copy-foo", cmd="cp foo bar", deps=["foo"], outs=["bar"])
+    dvc.reproduce()
+
+    remove("foo")
+
+    ret = dvc.reproduce(allow_missing=True)
+    # both stages are skipped
+    assert not ret

tests/func/repro/test_repro_pull.py

@@ -1,5 +1,7 @@
 import os
 
+import pytest
+
 from dvc.stage.cache import RunCacheNotSupported
 from dvc.utils.fs import remove
 
@@ -49,11 +51,14 @@ def test_repro_skip_pull_if_no_run_cache_is_passed(tmp_dir, dvc, mocker, local_r
     (foo,) = tmp_dir.dvc_gen("foo", "foo")
 
     dvc.push()
-    spy_pull = mocker.spy(dvc.stage_cache, "pull")
     dvc.stage.add(name="copy-foo", cmd="cp foo bar", deps=["foo"], outs=["bar"])
+
+    assert dvc.reproduce(pull=True)
+
     remove("foo")
     remove(foo.outs[0].cache_path)
-
+    remove("dvc.lock")
+    spy_pull = mocker.spy(dvc.stage_cache, "pull")
     assert dvc.reproduce(pull=True, run_cache=False)
     assert not spy_pull.called
 
@@ -68,3 +73,46 @@ def test_repro_skip_pull_if_single_item_is_passed(tmp_dir, dvc, mocker, local_re
     remove(foo.outs[0].cache_path)
 
     assert dvc.reproduce(pull=True, single_item=True)
+
+
+def test_repro_pulls_persisted_output(tmp_dir, dvc, mocker, local_remote):
+    tmp_dir.dvc_gen("foo", "foo")
+
+    dvc.push()
+
+    dvc.stage.add(name="copy-foo", cmd="cp foo bar", deps=["foo"], outs_persist=["bar"])
+    dvc.reproduce()
+    remove("bar")
+
+    # stage is skipped
+    assert not dvc.reproduce(pull=True)
+
+
+@pytest.mark.parametrize("allow_missing", [True, False])
+def test_repro_pulls_allow_missing(tmp_dir, dvc, mocker, local_remote, allow_missing):
+    tmp_dir.dvc_gen("foo", "foo")
+
+    dvc.push()
+
+    dvc.stage.add(name="copy-foo", cmd="cp foo bar", deps=["foo"], outs=["bar"])
+    dvc.reproduce()
+    remove("foo")
+
+    # stage is skipped
+    assert not dvc.reproduce(pull=True, allow_missing=allow_missing)
+    # data only pulled if allow_missing is false
+    assert (tmp_dir / "foo").exists() != allow_missing
+
+
+def test_repro_pull_fails(tmp_dir, dvc, mocker, local_remote):
+    tmp_dir.dvc_gen("foo", "foo")
+
+    dvc.stage.add(
+        name="concat-foo", cmd="cat foo foo > bar", deps=["foo"], outs=["bar"]
+    )
+    stages = dvc.reproduce()
+    remove("bar")
+    remove(stages[0].outs[0].cache_path)
+
+    # stage is run
+    assert dvc.reproduce(pull=True)

tests/func/test_add.py

@@ -305,20 +305,20 @@ def test_should_update_state_entry_for_directory_after_add(mocker, dvc, tmp_dir)
 
     ret = main(["add", "data"])
     assert ret == 0
-    assert file_md5_counter.mock.call_count == 4
+    assert file_md5_counter.mock.call_count == 3
 
     ret = main(["status"])
     assert ret == 0
-    assert file_md5_counter.mock.call_count == 5
+    assert file_md5_counter.mock.call_count == 4
 
     os.rename("data", "data.back")
     ret = main(["checkout"])
     assert ret == 0
-    assert file_md5_counter.mock.call_count == 6
+    assert file_md5_counter.mock.call_count == 5
 
     ret = main(["status"])
     assert ret == 0
-    assert file_md5_counter.mock.call_count == 7
+    assert file_md5_counter.mock.call_count == 6
 
 
 def test_add_commit(tmp_dir, dvc):
@@ -339,16 +339,16 @@ def test_should_collect_dir_cache_only_once(mocker, tmp_dir, dvc):
     counter = mocker.spy(dvc_data.hashfile.build, "_build_tree")
     ret = main(["add", "data"])
     assert ret == 0
+    assert counter.mock.call_count == 1
+
+    ret = main(["status"])
+    assert ret == 0
     assert counter.mock.call_count == 2
 
     ret = main(["status"])
     assert ret == 0
     assert counter.mock.call_count == 3
 
-    ret = main(["status"])
-    assert ret == 0
-    assert counter.mock.call_count == 4
-
 
 def test_should_place_stage_in_data_dir_if_repository_below_symlink(
     mocker, tmp_dir, dvc
@@ -754,6 +754,7 @@ def test_add_file_in_symlink_dir(make_tmp_dir, tmp_dir, dvc):
 def test_add_with_cache_link_error(tmp_dir, dvc, mocker, capsys):
     tmp_dir.gen("foo", "foo")
 
+    dvc.cache.local.cache_types = ["symlink", "hardlink"]
     mocker.patch("dvc_data.hashfile.checkout.test_links", return_value=[])
     dvc.add("foo")
     err = capsys.readouterr()[1]

tests/func/test_daemon.py

@@ -6,7 +6,7 @@ import sys
 from collections import defaultdict
 from collections.abc import Iterator
 from contextlib import contextmanager, suppress
-from http.server import HTTPServer, SimpleHTTPRequestHandler
+from http.server import BaseHTTPRequestHandler, HTTPServer
 from pathlib import Path
 from threading import Thread
 from typing import ClassVar
@@ -41,7 +41,7 @@ UPDATER_INFO_STR = json.dumps(UPDATER_INFO).encode("utf8")
 
 
 def make_request_handler():
-    class RequestHandler(SimpleHTTPRequestHandler):
+    class RequestHandler(BaseHTTPRequestHandler):
         # save requests count for each method
         hits: ClassVar[dict[str, int]] = defaultdict(int)
 
@@ -80,7 +80,10 @@ def server():
         thread = Thread(target=httpd.serve_forever)
         thread.daemon = True
         thread.start()
-        yield httpd
+        try:
+            yield httpd
+        finally:
+            httpd.shutdown()
 
 
 def test_analytics(tmp_path, server):
@@ -94,6 +97,11 @@ def test_analytics(tmp_path, server):
     }
     env.pop("DVC_TEST", None)
     env.pop("DVC_NO_ANALYTICS", None)
+    # The `iterative-telemetry` package calls `gh api` to generate a CI id.
+    # This might hang especially on Windows,
+    # possibly due to system load from the running tests.
+    # Removing the GITHUB_ACTIONS env var avoids calling `gh api`.
+    env.pop("GITHUB_ACTIONS", None)
 
     output = subprocess.check_output(
         [*_get_dvc_args(), "config", "-l", "-vv"],
@@ -101,18 +109,24 @@ def test_analytics(tmp_path, server):
         text=True,
     )
 
-    match = re.search(r".*Saving analytics report to (.*)", output, flags=re.M)
+    match = re.search(r".*Saving analytics report to (.*)", output, flags=re.MULTILINE)
     assert match, "no match for the report file"
     report_file = match.group(1).strip()
 
-    match = re.search(r".*Spawned .*analytics.* with pid (.*)", output, flags=re.M)
+    match = re.search(
+        r".*Spawned .*analytics.* with pid (.*)", output, flags=re.MULTILINE
+    )
     assert match, "no match for the pid"
     pid = int(match.group(1).strip())
 
     with suppress(psutil.NoSuchProcess):
         psutil.Process(pid).wait(timeout=10)
+
+    log_contents = logfile.read_text(encoding="utf8")
+    expected_line = (f"Process {pid} " if os.name != "nt" else "") + "exiting with 0"
+    assert expected_line in log_contents
+
     assert not os.path.exists(report_file)
-    assert f"Process {pid} exiting with 0" in logfile.read_text(encoding="utf8")
     assert server.RequestHandlerClass.hits == {"POST": 1}
 
 
@@ -136,13 +150,19 @@ def test_updater(tmp_dir, dvc, server):
         text=True,
     )
 
-    match = re.search(r".*Spawned .*updater.* with pid (.*)", output, flags=re.M)
+    match = re.search(
+        r".*Spawned .*updater.* with pid (.*)", output, flags=re.MULTILINE
+    )
     assert match, "no match for the pid"
     pid = int(match.group(1).strip())
 
     with suppress(psutil.NoSuchProcess):
         psutil.Process(pid).wait(timeout=10)
-    assert f"Process {pid} exiting with 0" in logfile.read_text(encoding="utf8")
+
+    log_contents = logfile.read_text(encoding="utf8")
+    expected_line = (f"Process {pid} " if os.name != "nt" else "") + "exiting with 0"
+    assert expected_line in log_contents
+
     assert server.RequestHandlerClass.hits == {"GET": 1}
     # check that the file is saved correctly
     updater_file = Path(dvc.tmp_dir) / Updater.UPDATER_FILE

tests/func/test_import.py

@@ -12,7 +12,8 @@ from dvc.scm import Git
 from dvc.stage.exceptions import StagePathNotFoundError
 from dvc.testing.tmp_dir import make_subrepo
 from dvc.utils.fs import remove
-from dvc_data.index.index import DataIndexDirError
+from dvc_data.hashfile import hash
+from dvc_data.index.index import DataIndex, DataIndexDirError
 
 
 def test_import(tmp_dir, scm, dvc, erepo_dir):
@@ -305,11 +306,12 @@ def test_cache_type_is_properly_overridden(tmp_dir, scm, dvc, erepo_dir):
     assert scm.is_ignored("foo_imported")
 
 
-def test_pull_imported_directory_stage(tmp_dir, dvc, erepo_dir):
+@pytest.mark.parametrize("dirpath", ["dir", "dir/"])
+def test_pull_imported_directory_stage(tmp_dir, dvc, erepo_dir, dirpath):
     with erepo_dir.chdir():
         erepo_dir.dvc_gen({"dir": {"foo": "foo content"}}, commit="create dir")
 
-    dvc.imp(os.fspath(erepo_dir), "dir", "dir_imported")
+    stage = dvc.imp(os.fspath(erepo_dir), dirpath, "dir_imported")
 
     remove("dir_imported")
     dvc.cache.local.clear()
@@ -317,6 +319,7 @@ def test_pull_imported_directory_stage(tmp_dir, dvc, erepo_dir):
     dvc.pull(["dir_imported.dvc"])
 
     assert (tmp_dir / "dir_imported").read_text() == {"foo": "foo content"}
+    assert stage.deps[0].fs_path == "dir"
 
 
 def test_pull_wildcard_imported_directory_stage(tmp_dir, dvc, erepo_dir):
@@ -720,3 +723,43 @@ def test_import_invalid_configs(tmp_dir, scm, dvc, erepo_dir):
             remote="myremote",
             remote_config={"key": "value"},
         )
+
+
+@pytest.mark.parametrize(
+    "files,expected_info_calls",
+    [
+        ({"foo": "foo"}, {("foo",)}),
+        (
+            {
+                "dir": {
+                    "bar": "bar",
+                    "subdir": {"lorem": "ipsum", "nested": {"lorem": "lorem"}},
+                }
+            },
+            # info calls should be made for only directories
+            {("dir",), ("dir", "subdir"), ("dir", "subdir", "nested")},
+        ),
+    ],
+)
+def test_import_no_hash(
+    tmp_dir, scm, dvc, erepo_dir, mocker, files, expected_info_calls
+):
+    with erepo_dir.chdir():
+        erepo_dir.dvc_gen(files, commit="create foo")
+
+    file_md5_spy = mocker.spy(hash, "file_md5")
+    index_info_spy = mocker.spy(DataIndex, "info")
+    name = next(iter(files))
+
+    dvc.imp(os.fspath(erepo_dir), name, "out")
+
+    local_hashes = [
+        call.args[0]
+        for call in file_md5_spy.call_args_list
+        if call.args[1].protocol == "local"
+    ]
+    # no files should be hashed, should use existing metadata
+    assert not local_hashes
+    assert {
+        call.args[1] for call in index_info_spy.call_args_list
+    } == expected_info_calls

tests/func/test_run_cache.py

@@ -1,3 +1,4 @@
+import logging
 import os
 
 import pytest
@@ -231,13 +232,15 @@ def test_restore_pull(tmp_dir, dvc, run_copy, mocker, local_remote):
     assert (tmp_dir / LOCK_FILE).exists()
 
 
-def test_push_pull_unsupported(tmp_dir, dvc, mocker, run_copy):
+def test_push_pull_unsupported(tmp_dir, dvc, mocker, run_copy, local_remote, caplog):
     tmp_dir.gen("foo", "foo")
     run_copy("foo", "bar", name="copy-foo-bar")
     mocker.patch.object(
         dvc.cloud, "get_remote_odb", side_effect=RunCacheNotSupported("foo")
     )
-    with pytest.raises(RunCacheNotSupported):
+    with caplog.at_level(logging.DEBUG):
         dvc.push(run_cache=True)
-    with pytest.raises(RunCacheNotSupported):
+        assert "failed to push run cache" in caplog.text
+    with caplog.at_level(logging.DEBUG):
         dvc.pull(run_cache=True)
+        assert "failed to pull run cache" in caplog.text

tests/unit/fs/test_dvcfs.py

@@ -0,0 +1,805 @@
+import os
+import posixpath
+from hashlib import md5
+from itertools import product
+
+import pytest
+from fsspec.implementations.local import LocalFileSystem, make_path_posix
+from fsspec.tests.abstract.common import GLOB_EDGE_CASES_TESTS
+
+from dvc.api import DVCFileSystem
+
+
+class DVCFixtures:
+    """The fixtures imitate the fsspec.tests.abstract.AbstractFixtures.
+
+    This has been modified to use dvc's fixture, as DVCFileSystem is a read-only
+    filesystem, and cannot be used to create directories or files.
+
+    The `Output.ignore()` is mocked to avoid `.gitignore` files in the directories,
+    as we can reuse the tests from fsspec with minimal modifications.
+    `.gitignore` file is manually created with required patterns at the root of the
+    repository.
+    """
+
+    @pytest.fixture
+    def fs_bulk_operations_scenario_0(self, tmp_dir):
+        """
+        Scenario that is used for many cp/get/put tests. Creates the following
+        directory and file structure:
+
+        📁 source
+        ├── 📄 file1
+        ├── 📄 file2
+        └── 📁 subdir
+            ├── 📄 subfile1
+            ├── 📄 subfile2
+            └── 📁 nesteddir
+                └── 📄 nestedfile
+        """
+        source = tmp_dir / "source"
+        source.mkdir()
+        tmp_dir.scm_gen(
+            ".gitignore", "/source/file2/\nsource/subdir", commit="add .gitignore"
+        )
+
+        tmp_dir.scm_gen("source/file1", "file1", commit="add file1")
+        tmp_dir.dvc_gen("source/file2", "file2", commit="add file2")
+        tmp_dir.dvc_gen(
+            {
+                "source/subdir": {
+                    "subfile1": "subfile1",
+                    "subfile2": "subfile2",
+                    "nesteddir": {"nestedfile": "nestedfile"},
+                }
+            },
+            commit="add subdir",
+        )
+        return "/source"
+
+    @pytest.fixture
+    def fs_10_files_with_hashed_names(self, tmp_dir, local_fs, local_join, local_path):
+        """
+        Scenario that is used to check cp/get/put files order when source and
+        destination are lists. Creates the following directory and file structure:
+
+        📁 source
+        └── 📄 {hashed([0-9])}.txt
+        """
+        dir_contents = {
+            md5(str(i).encode("utf-8")).hexdigest() + ".txt": str(i) for i in range(10)
+        }
+        tmp_dir.dvc_gen({"source": dir_contents}, commit="add source")
+        tmp_dir.scm_gen(".gitignore", "/source", commit="add .gitignore")
+        return "/source"
+
+    @pytest.fixture
+    def src_directory(self, tmp_dir):
+        # https://github.com/fsspec/filesystem_spec/issues/1062
+        # Recursive cp/get/put of source directory into non-existent target directory.
+        tmp_dir.dvc_gen({"src": {"file": "file"}}, commit="add source")
+        return "/src"
+
+    @pytest.fixture
+    def fs_dir_and_file_with_same_name_prefix(self, tmp_dir):
+        """
+        Scenario that is used to check cp/get/put on directory and file with
+        the same name prefixes. Creates the following directory and file structure:
+
+        📁 source
+        ├── 📄 subdir.txt
+        └── 📁 subdir
+            └── 📄 subfile.txt
+        """
+        source = tmp_dir / "source"
+        source.mkdir()
+
+        tmp_dir.scm_gen(".gitignore", "/source/subdir", commit="add .gitignore")
+        tmp_dir.scm_gen("source/subdir.txt", "subdir.txt", commit="add subdir.txt")
+        tmp_dir.dvc_gen(
+            {"source/subdir": {"subfile.txt": "subfile.txt"}}, commit="add subdir"
+        )
+        return "/source"
+
+    @pytest.fixture
+    def fs_glob_edge_cases_files(self, tmp_dir):
+        """
+        Scenario that is used for glob edge cases cp/get/put tests.
+        Creates the following directory and file structure:
+
+        📁 source
+        ├── 📄 file1
+        ├── 📄 file2
+        ├── 📁 subdir0
+        │   ├── 📄 subfile1
+        │   ├── 📄 subfile2
+        │   └── 📁 nesteddir
+        │       └── 📄 nestedfile
+        └── 📁 subdir1
+            ├── 📄 subfile1
+            ├── 📄 subfile2
+            └── 📁 nesteddir
+                └── 📄 nestedfile
+        """
+        source = tmp_dir / "source"
+        source.mkdir()
+
+        tmp_dir.scm_gen(
+            ".gitignore", "/source/file1\n/source/subdir1", commit="add .gitignore"
+        )
+        tmp_dir.scm_gen("source/file1", "file1", commit="add file1")
+        tmp_dir.dvc_gen("source/file2", "file2", commit="add file2")
+
+        dir_contents = {
+            "subfile1": "subfile1",
+            "subfile2": "subfile2",
+            "nesteddir": {"nestedfile": "nestedfile"},
+        }
+        tmp_dir.scm_gen({"source/subdir0": dir_contents}, commit="add subdir0")
+        tmp_dir.dvc_gen({"source/subdir1": dir_contents}, commit="add subdir1")
+        return "/source"
+
+    @pytest.fixture(params=[{"rev": "HEAD"}, {}])
+    def fs(self, request, tmp_dir, dvc, scm):
+        return DVCFileSystem(tmp_dir, **request.param)
+
+    @pytest.fixture(autouse=True)
+    def mock_ignore(self, mocker):
+        mocker.patch("dvc.output.Output.ignore")
+
+    @pytest.fixture
+    def fs_join(self):
+        return posixpath.join
+
+    @pytest.fixture
+    def fs_path(self, fs):
+        return fs.root_marker
+
+    @pytest.fixture(scope="class")
+    def local_fs(self):
+        # Maybe need an option for auto_mkdir=False?  This is only relevant
+        # for certain implementations.
+        return LocalFileSystem(auto_mkdir=True)
+
+    @pytest.fixture
+    def local_join(self):
+        """
+        Return a function that joins its arguments together into a path, on
+        the local filesystem.
+        """
+        return os.path.join
+
+    @pytest.fixture
+    def local_path(self, tmpdir):
+        return tmpdir
+
+    @pytest.fixture
+    def local_target(self, local_fs, local_join, local_path):
+        """
+        Return name of local directory that does not yet exist to copy into.
+
+        Cleans up at the end of each test it which it is used.
+        """
+        target = local_join(local_path, "target")
+        yield target
+        if local_fs.exists(target):
+            local_fs.rm(target, recursive=True)
+
+
+class TestDVCFileSystemGet(DVCFixtures):
+    """
+    This test is adapted from `fsspec.tests.abstract.get.AbstractGetTests`
+    with minor modifications to work with DVCFixtures and DVCFileSystem.
+    """
+
+    def test_get_file_to_existing_directory(
+        self,
+        fs,
+        fs_join,
+        fs_bulk_operations_scenario_0,
+        local_fs,
+        local_join,
+        local_target,
+    ):
+        # Copy scenario 1a
+        source = fs_bulk_operations_scenario_0
+
+        target = local_target
+        local_fs.mkdir(target)
+        assert local_fs.isdir(target)
+
+        target_file2 = local_join(target, "file2")
+        target_subfile1 = local_join(target, "subfile1")
+
+        # Copy from source directory
+        fs.get(fs_join(source, "file2"), target)
+        assert local_fs.isfile(target_file2)
+
+        # Copy from sub directory
+        fs.get(fs_join(source, "subdir", "subfile1"), target)
+        assert local_fs.isfile(target_subfile1)
+
+        # Remove copied files
+        local_fs.rm([target_file2, target_subfile1])
+        assert not local_fs.exists(target_file2)
+        assert not local_fs.exists(target_subfile1)
+
+        # Repeat with trailing slash on target
+        fs.get(fs_join(source, "file2"), target + "/")
+        assert local_fs.isdir(target)
+        assert local_fs.isfile(target_file2)
+
+        fs.get(fs_join(source, "subdir", "subfile1"), target + "/")
+        assert local_fs.isfile(target_subfile1)
+
+    def test_get_file_to_new_directory(
+        self,
+        fs,
+        fs_join,
+        fs_bulk_operations_scenario_0,
+        local_fs,
+        local_join,
+        local_target,
+    ):
+        # Copy scenario 1b
+        source = fs_bulk_operations_scenario_0
+
+        target = local_target
+        local_fs.mkdir(target)
+
+        fs.get(
+            fs_join(source, "subdir", "subfile1"), local_join(target, "newdir/")
+        )  # Note trailing slash
+
+        assert local_fs.isdir(target)
+        assert local_fs.isdir(local_join(target, "newdir"))
+        assert local_fs.isfile(local_join(target, "newdir", "subfile1"))
+
+    def test_get_file_to_file_in_existing_directory(
+        self,
+        fs,
+        fs_join,
+        fs_bulk_operations_scenario_0,
+        local_fs,
+        local_join,
+        local_target,
+    ):
+        # Copy scenario 1c
+        source = fs_bulk_operations_scenario_0
+
+        target = local_target
+        local_fs.mkdir(target)
+
+        fs.get(fs_join(source, "subdir", "subfile1"), local_join(target, "newfile"))
+        assert local_fs.isfile(local_join(target, "newfile"))
+
+    def test_get_file_to_file_in_new_directory(
+        self,
+        fs,
+        fs_join,
+        fs_bulk_operations_scenario_0,
+        local_fs,
+        local_join,
+        local_target,
+    ):
+        # Copy scenario 1d
+        source = fs_bulk_operations_scenario_0
+
+        target = local_target
+        local_fs.mkdir(target)
+
+        fs.get(
+            fs_join(source, "subdir", "subfile1"),
+            local_join(target, "newdir", "newfile"),
+        )
+        assert local_fs.isdir(local_join(target, "newdir"))
+        assert local_fs.isfile(local_join(target, "newdir", "newfile"))
+
+    def test_get_directory_to_existing_directory(
+        self,
+        fs,
+        fs_join,
+        fs_bulk_operations_scenario_0,
+        local_fs,
+        local_join,
+        local_target,
+    ):
+        # Copy scenario 1e
+        source = fs_bulk_operations_scenario_0
+
+        target = local_target
+        local_fs.mkdir(target)
+        assert local_fs.isdir(target)
+
+        for source_slash, target_slash in zip([False, True], [False, True]):
+            s = fs_join(source, "subdir")
+            if source_slash:
+                s += "/"
+            t = target + "/" if target_slash else target
+
+            # Without recursive does nothing
+            fs.get(s, t)
+            assert local_fs.ls(target) == []
+
+            # With recursive
+            fs.get(s, t, recursive=True)
+            if source_slash:
+                assert local_fs.isfile(local_join(target, "subfile1"))
+                assert local_fs.isfile(local_join(target, "subfile2"))
+                assert local_fs.isdir(local_join(target, "nesteddir"))
+                assert local_fs.isfile(local_join(target, "nesteddir", "nestedfile"))
+                assert not local_fs.exists(local_join(target, "subdir"))
+
+                local_fs.rm(
+                    [
+                        local_join(target, "subfile1"),
+                        local_join(target, "subfile2"),
+                        local_join(target, "nesteddir"),
+                    ],
+                    recursive=True,
+                )
+            else:
+                assert local_fs.isdir(local_join(target, "subdir"))
+                assert local_fs.isfile(local_join(target, "subdir", "subfile1"))
+                assert local_fs.isfile(local_join(target, "subdir", "subfile2"))
+                assert local_fs.isdir(local_join(target, "subdir", "nesteddir"))
+                assert local_fs.isfile(
+                    local_join(target, "subdir", "nesteddir", "nestedfile")
+                )
+
+                local_fs.rm(local_join(target, "subdir"), recursive=True)
+            assert local_fs.ls(target) == []
+
+            # Limit recursive by maxdepth
+            fs.get(s, t, recursive=True, maxdepth=1)
+            if source_slash:
+                assert local_fs.isfile(local_join(target, "subfile1"))
+                assert local_fs.isfile(local_join(target, "subfile2"))
+                assert not local_fs.exists(local_join(target, "nesteddir"))
+                assert not local_fs.exists(local_join(target, "subdir"))
+
+                local_fs.rm(
+                    [
+                        local_join(target, "subfile1"),
+                        local_join(target, "subfile2"),
+                    ],
+                    recursive=True,
+                )
+            else:
+                assert local_fs.isdir(local_join(target, "subdir"))
+                assert local_fs.isfile(local_join(target, "subdir", "subfile1"))
+                assert local_fs.isfile(local_join(target, "subdir", "subfile2"))
+                assert not local_fs.exists(local_join(target, "subdir", "nesteddir"))
+
+                local_fs.rm(local_join(target, "subdir"), recursive=True)
+            assert local_fs.ls(target) == []
+
+    def test_get_directory_to_new_directory(
+        self,
+        fs,
+        fs_join,
+        fs_bulk_operations_scenario_0,
+        local_fs,
+        local_join,
+        local_target,
+    ):
+        # Copy scenario 1f
+        source = fs_bulk_operations_scenario_0
+
+        target = local_target
+        local_fs.mkdir(target)
+
+        for source_slash, target_slash in zip([False, True], [False, True]):
+            s = fs_join(source, "subdir")
+            if source_slash:
+                s += "/"
+            t = local_join(target, "newdir")
+            if target_slash:
+                t += "/"
+
+            # Without recursive does nothing
+            fs.get(s, t)
+            assert local_fs.ls(target) == []
+
+            # With recursive
+            fs.get(s, t, recursive=True)
+            assert local_fs.isdir(local_join(target, "newdir"))
+            assert local_fs.isfile(local_join(target, "newdir", "subfile1"))
+            assert local_fs.isfile(local_join(target, "newdir", "subfile2"))
+            assert local_fs.isdir(local_join(target, "newdir", "nesteddir"))
+            assert local_fs.isfile(
+                local_join(target, "newdir", "nesteddir", "nestedfile")
+            )
+            assert not local_fs.exists(local_join(target, "subdir"))
+
+            local_fs.rm(local_join(target, "newdir"), recursive=True)
+            assert local_fs.ls(target) == []
+
+            # Limit recursive by maxdepth
+            fs.get(s, t, recursive=True, maxdepth=1)
+            assert local_fs.isdir(local_join(target, "newdir"))
+            assert local_fs.isfile(local_join(target, "newdir", "subfile1"))
+            assert local_fs.isfile(local_join(target, "newdir", "subfile2"))
+            assert not local_fs.exists(local_join(target, "newdir", "nesteddir"))
+            assert not local_fs.exists(local_join(target, "subdir"))
+
+            local_fs.rm(local_join(target, "newdir"), recursive=True)
+            assert not local_fs.exists(local_join(target, "newdir"))
+
+    def test_get_glob_to_existing_directory(
+        self,
+        fs,
+        fs_join,
+        fs_bulk_operations_scenario_0,
+        local_fs,
+        local_join,
+        local_target,
+    ):
+        # Copy scenario 1g
+        source = fs_bulk_operations_scenario_0
+
+        target = local_target
+        local_fs.mkdir(target)
+
+        for target_slash in [False, True]:
+            t = target + "/" if target_slash else target
+
+            # Without recursive
+            fs.get(fs_join(source, "subdir", "*"), t)
+            assert local_fs.isfile(local_join(target, "subfile1"))
+            assert local_fs.isfile(local_join(target, "subfile2"))
+            assert not local_fs.isdir(local_join(target, "nesteddir"))
+            assert not local_fs.exists(local_join(target, "nesteddir", "nestedfile"))
+            assert not local_fs.exists(local_join(target, "subdir"))
+
+            local_fs.rm(
+                [
+                    local_join(target, "subfile1"),
+                    local_join(target, "subfile2"),
+                ],
+                recursive=True,
+            )
+            assert local_fs.ls(target) == []
+
+            # With recursive
+            for glob, recursive in zip(["*", "**"], [True, False]):
+                fs.get(fs_join(source, "subdir", glob), t, recursive=recursive)
+                assert local_fs.isfile(local_join(target, "subfile1"))
+                assert local_fs.isfile(local_join(target, "subfile2"))
+                assert local_fs.isdir(local_join(target, "nesteddir"))
+                assert local_fs.isfile(local_join(target, "nesteddir", "nestedfile"))
+                assert not local_fs.exists(local_join(target, "subdir"))
+
+                local_fs.rm(
+                    [
+                        local_join(target, "subfile1"),
+                        local_join(target, "subfile2"),
+                        local_join(target, "nesteddir"),
+                    ],
+                    recursive=True,
+                )
+                assert local_fs.ls(target) == []
+
+                # Limit recursive by maxdepth
+                fs.get(
+                    fs_join(source, "subdir", glob), t, recursive=recursive, maxdepth=1
+                )
+                assert local_fs.isfile(local_join(target, "subfile1"))
+                assert local_fs.isfile(local_join(target, "subfile2"))
+                assert not local_fs.exists(local_join(target, "nesteddir"))
+                assert not local_fs.exists(local_join(target, "subdir"))
+
+                local_fs.rm(
+                    [
+                        local_join(target, "subfile1"),
+                        local_join(target, "subfile2"),
+                    ],
+                    recursive=True,
+                )
+                assert local_fs.ls(target) == []
+
+    def test_get_glob_to_new_directory(
+        self,
+        fs,
+        fs_join,
+        fs_bulk_operations_scenario_0,
+        local_fs,
+        local_join,
+        local_target,
+    ):
+        # Copy scenario 1h
+        source = fs_bulk_operations_scenario_0
+
+        target = local_target
+        local_fs.mkdir(target)
+
+        for target_slash in [False, True]:
+            t = fs_join(target, "newdir")
+            if target_slash:
+                t += "/"
+
+            # Without recursive
+            fs.get(fs_join(source, "subdir", "*"), t)
+            assert local_fs.isdir(local_join(target, "newdir"))
+            assert local_fs.isfile(local_join(target, "newdir", "subfile1"))
+            assert local_fs.isfile(local_join(target, "newdir", "subfile2"))
+            assert not local_fs.exists(local_join(target, "newdir", "nesteddir"))
+            assert not local_fs.exists(
+                local_join(target, "newdir", "nesteddir", "nestedfile")
+            )
+            assert not local_fs.exists(local_join(target, "subdir"))
+            assert not local_fs.exists(local_join(target, "newdir", "subdir"))
+
+            local_fs.rm(local_join(target, "newdir"), recursive=True)
+            assert local_fs.ls(target) == []
+
+            # With recursive
+            for glob, recursive in zip(["*", "**"], [True, False]):
+                fs.get(fs_join(source, "subdir", glob), t, recursive=recursive)
+                assert local_fs.isdir(local_join(target, "newdir"))
+                assert local_fs.isfile(local_join(target, "newdir", "subfile1"))
+                assert local_fs.isfile(local_join(target, "newdir", "subfile2"))
+                assert local_fs.isdir(local_join(target, "newdir", "nesteddir"))
+                assert local_fs.isfile(
+                    local_join(target, "newdir", "nesteddir", "nestedfile")
+                )
+                assert not local_fs.exists(local_join(target, "subdir"))
+                assert not local_fs.exists(local_join(target, "newdir", "subdir"))
+
+                local_fs.rm(local_join(target, "newdir"), recursive=True)
+                assert not local_fs.exists(local_join(target, "newdir"))
+
+                # Limit recursive by maxdepth
+                fs.get(
+                    fs_join(source, "subdir", glob), t, recursive=recursive, maxdepth=1
+                )
+                assert local_fs.isdir(local_join(target, "newdir"))
+                assert local_fs.isfile(local_join(target, "newdir", "subfile1"))
+                assert local_fs.isfile(local_join(target, "newdir", "subfile2"))
+                assert not local_fs.exists(local_join(target, "newdir", "nesteddir"))
+                assert not local_fs.exists(local_join(target, "subdir"))
+                assert not local_fs.exists(local_join(target, "newdir", "subdir"))
+
+                local_fs.rm(local_fs.ls(target, detail=False), recursive=True)
+                assert not local_fs.exists(local_join(target, "newdir"))
+
+    @pytest.mark.parametrize(
+        GLOB_EDGE_CASES_TESTS["argnames"],
+        GLOB_EDGE_CASES_TESTS["argvalues"],
+    )
+    def test_get_glob_edge_cases(
+        self,
+        path,
+        recursive,
+        maxdepth,
+        expected,
+        fs,
+        fs_join,
+        fs_glob_edge_cases_files,
+        local_fs,
+        local_join,
+        local_target,
+    ):
+        # Copy scenario 1g
+        source = fs_glob_edge_cases_files
+
+        target = local_target
+
+        for new_dir, target_slash in product([True, False], [True, False]):
+            local_fs.mkdir(target)
+
+            t = local_join(target, "newdir") if new_dir else target
+            t = t + "/" if target_slash else t
+
+            fs.get(fs_join(source, path), t, recursive=recursive, maxdepth=maxdepth)
+
+            output = local_fs.find(target)
+            if new_dir:
+                prefixed_expected = [
+                    make_path_posix(local_join(target, "newdir", p)) for p in expected
+                ]
+            else:
+                prefixed_expected = [
+                    make_path_posix(local_join(target, p)) for p in expected
+                ]
+            assert sorted(output) == sorted(prefixed_expected)
+
+            try:
+                local_fs.rm(target, recursive=True)
+            except FileNotFoundError:
+                pass
+
+    def test_get_list_of_files_to_existing_directory(
+        self,
+        fs,
+        fs_join,
+        fs_bulk_operations_scenario_0,
+        local_fs,
+        local_join,
+        local_target,
+    ):
+        # Copy scenario 2a
+        source = fs_bulk_operations_scenario_0
+
+        target = local_target
+        local_fs.mkdir(target)
+
+        source_files = [
+            fs_join(source, "file1"),
+            fs_join(source, "file2"),
+            fs_join(source, "subdir", "subfile1"),
+        ]
+
+        for target_slash in [False, True]:
+            t = target + "/" if target_slash else target
+
+            fs.get(source_files, t)
+            assert local_fs.isfile(local_join(target, "file1"))
+            assert local_fs.isfile(local_join(target, "file2"))
+            assert local_fs.isfile(local_join(target, "subfile1"))
+
+            local_fs.rm(
+                [
+                    local_join(target, "file1"),
+                    local_join(target, "file2"),
+                    local_join(target, "subfile1"),
+                ],
+                recursive=True,
+            )
+            assert local_fs.ls(target) == []
+
+    def test_get_list_of_files_to_new_directory(
+        self,
+        fs,
+        fs_join,
+        fs_bulk_operations_scenario_0,
+        local_fs,
+        local_join,
+        local_target,
+    ):
+        # Copy scenario 2b
+        source = fs_bulk_operations_scenario_0
+
+        target = local_target
+        local_fs.mkdir(target)
+
+        source_files = [
+            fs_join(source, "file1"),
+            fs_join(source, "file2"),
+            fs_join(source, "subdir", "subfile1"),
+        ]
+
+        fs.get(source_files, local_join(target, "newdir") + "/")  # Note trailing slash
+        assert local_fs.isdir(local_join(target, "newdir"))
+        assert local_fs.isfile(local_join(target, "newdir", "file1"))
+        assert local_fs.isfile(local_join(target, "newdir", "file2"))
+        assert local_fs.isfile(local_join(target, "newdir", "subfile1"))
+
+    def test_get_directory_recursive(
+        self, src_directory, fs, fs_join, fs_path, local_fs, local_join, local_target
+    ):
+        target = local_target
+        src = src_directory
+
+        # get without slash
+        assert not local_fs.exists(target)
+        for loop in range(2):
+            fs.get(src, target, recursive=True)
+            assert local_fs.isdir(target)
+
+            if loop == 0:
+                assert local_fs.isfile(local_join(target, "file"))
+                assert not local_fs.exists(local_join(target, "src"))
+            else:
+                assert local_fs.isfile(local_join(target, "file"))
+                assert local_fs.isdir(local_join(target, "src"))
+                assert local_fs.isfile(local_join(target, "src", "file"))
+
+        local_fs.rm(target, recursive=True)
+
+        # get with slash
+        assert not local_fs.exists(target)
+        for _ in range(2):
+            fs.get(src + "/", target, recursive=True)
+            assert local_fs.isdir(target)
+            assert local_fs.isfile(local_join(target, "file"))
+            assert not local_fs.exists(local_join(target, "src"))
+
+    def test_get_directory_without_files_with_same_name_prefix(
+        self,
+        fs,
+        fs_join,
+        local_fs,
+        local_join,
+        local_target,
+        fs_dir_and_file_with_same_name_prefix,
+    ):
+        # Create the test dirs
+        source = fs_dir_and_file_with_same_name_prefix
+        target = local_target
+
+        # Test without glob
+        fs.get(fs_join(source, "subdir"), target, recursive=True)
+
+        assert local_fs.isfile(local_join(target, "subfile.txt"))
+        assert not local_fs.isfile(local_join(target, "subdir.txt"))
+
+        local_fs.rm([local_join(target, "subfile.txt")])
+        assert local_fs.ls(target) == []
+
+        # Test with glob
+        fs.get(fs_join(source, "subdir*"), target, recursive=True)
+
+        assert local_fs.isdir(local_join(target, "subdir"))
+        assert local_fs.isfile(local_join(target, "subdir", "subfile.txt"))
+        assert local_fs.isfile(local_join(target, "subdir.txt"))
+
+    def test_get_with_source_and_destination_as_list(
+        self,
+        fs,
+        fs_join,
+        local_fs,
+        local_join,
+        local_target,
+        fs_10_files_with_hashed_names,
+    ):
+        # Create the test dir
+        source = fs_10_files_with_hashed_names
+        target = local_target
+
+        # Create list of files for source and destination
+        source_files = []
+        destination_files = []
+        for i in range(10):
+            hashed_i = md5(str(i).encode("utf-8")).hexdigest()
+            source_files.append(fs_join(source, f"{hashed_i}.txt"))
+            destination_files.append(
+                make_path_posix(local_join(target, f"{hashed_i}.txt"))
+            )
+
+        # Copy and assert order was kept
+        fs.get(rpath=source_files, lpath=destination_files)
+
+        for i in range(10):
+            file_content = local_fs.cat(destination_files[i]).decode("utf-8")
+            assert file_content == str(i)
+
+
+def test_maxdepth(tmp_dir, dvc, scm):
+    tmp_dir.dvc_gen(
+        {
+            "dir": {
+                "file1": "file1",
+                "subdir": {
+                    "file2": "file2",
+                    "subdir2": {"file3": "file3", "subdir3": {"file4": "file4"}},
+                },
+            }
+        },
+        commit="add dir",
+    )
+
+    fs = DVCFileSystem(url=tmp_dir)
+    fs.get("dir", "dir1", recursive=True, maxdepth=1)
+    assert (tmp_dir / "dir1").read_text() == {"file1": "file1"}
+
+    fs.get("dir", "dir2", recursive=True, maxdepth=2)
+    assert (tmp_dir / "dir2").read_text() == {
+        "file1": "file1",
+        "subdir": {"file2": "file2"},
+    }
+
+    fs.get("dir", "dir3", recursive=True, maxdepth=3)
+    assert (tmp_dir / "dir3").read_text() == {
+        "file1": "file1",
+        "subdir": {"file2": "file2", "subdir2": {"file3": "file3"}},
+    }
+
+    fs.get("dir", "dir4", recursive=True, maxdepth=4)
+    assert (tmp_dir / "dir4").read_text() == {
+        "file1": "file1",
+        "subdir": {
+            "file2": "file2",
+            "subdir2": {"file3": "file3", "subdir3": {"file4": "file4"}},
+        },
+    }