Merge branch 'iterative:main' into api_fix_config_no_scm

.github/.test_durations

@@ -835,9 +835,9 @@
     "tests/func/test_dataset.py::test_dvc": 0.46001330000012786,
     "tests/func/test_dataset.py::test_dvc_dataset_pipeline": 0.7948393000000351,
     "tests/func/test_dataset.py::test_dvc_dump": 0.17996560000005957,
-    "tests/func/test_dataset.py::test_dvcx": 0.22482930000001033,
-    "tests/func/test_dataset.py::test_dvcx_dataset_pipeline": 0.43405649999999696,
-    "tests/func/test_dataset.py::test_dvcx_dump": 0.1858166000000665,
+    "tests/func/test_dataset.py::test_datachain": 0.22482930000001033,
+    "tests/func/test_dataset.py::test_datachain_dataset_pipeline": 0.43405649999999696,
+    "tests/func/test_dataset.py::test_datachain_dump": 0.1858166000000665,
     "tests/func/test_dataset.py::test_invalidation": 0.16291839999996682,
     "tests/func/test_dataset.py::test_parametrized": 0.19277699999997822,
     "tests/func/test_dataset.py::test_pipeline_when_not_in_sync": 0.2063928999999689,
@@ -1627,7 +1627,7 @@
     "tests/unit/command/test_data_sync.py::test_pull": 0.17549940000003517,
     "tests/unit/command/test_data_sync.py::test_push": 0.16576090000000931,
     "tests/unit/command/test_dataset.py::test_add[spec0-lock0-Adding ds (url:/path @ main)\\n]": 0.22054290000005494,
-    "tests/unit/command/test_dataset.py::test_add[spec1-lock1-Adding mydataset (dvcx://dataset @ v1)\\n]": 0.22979979999990974,
+    "tests/unit/command/test_dataset.py::test_add[spec1-lock1-Adding mydataset (dc://dataset @ v1)\\n]": 0.22979979999990974,
     "tests/unit/command/test_dataset.py::test_add[spec2-lock2-Adding mydataset (s3://bucket/path)\\n]": 0.20264819999988504,
     "tests/unit/command/test_dataset.py::test_add_already_exists": 0.2456899000001158,
     "tests/unit/command/test_dataset.py::test_update[spec0-old_lock0-new_lock0-expected_outputs0-missing]": 0.2354315999998562,

.github/workflows/tests.yaml

@@ -104,7 +104,7 @@ jobs:
         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
+      uses: codecov/codecov-action@v5
       with:
         file: ./coverage.xml
         fail_ci_if_error: false

.pre-commit-config.yaml

@@ -22,7 +22,7 @@ repos:
       - id: sort-simple-yaml
       - id: trailing-whitespace
   - repo: https://github.com/astral-sh/ruff-pre-commit
-    rev: 'v0.7.1'
+    rev: 'v0.7.3'
     hooks:
       - id: ruff
         args: [--fix, --exit-non-zero-on-fix]

dvc/api/dataset.py

@@ -1,8 +1,8 @@
 from typing import Literal, TypedDict, Union
 
 
-class DVCXDataset(TypedDict):
-    type: Literal["dvcx"]
+class DatachainDataset(TypedDict):
+    type: Literal["dc"]
     name: str
     version: int
 
@@ -20,7 +20,7 @@ class URLDataset(TypedDict):
     path: str
 
 
-def get(name: str) -> Union[DVCXDataset, DVCDataset, URLDataset]:
+def get(name: str) -> Union[DatachainDataset, DVCDataset, URLDataset]:
     from difflib import get_close_matches
 
     from dvc.fs import get_cloud_fs
@@ -46,9 +46,9 @@ def get(name: str) -> Union[DVCXDataset, DVCDataset, URLDataset]:
             path=dataset.lock.path,
             sha=dataset.lock.rev_lock,
         )
-    if dataset.type == "dvcx":
-        return DVCXDataset(
-            type="dvcx", name=dataset.name_version[0], version=dataset.lock.version
+    if dataset.type == "dc":
+        return DatachainDataset(
+            type="dc", name=dataset.name_version[0], version=dataset.lock.version
         )
     if dataset.type == "url":
         fs_cls, _, path = get_cloud_fs(repo.config, url=dataset.lock.url)

dvc/commands/dataset.py

@@ -34,7 +34,7 @@ def display(cls, name: str, dataset: "Dataset", action: str = "Adding"):
 
         url = dataset.spec.url
         ver: str = ""
-        if dataset.type == "dvcx":
+        if dataset.type == "dc":
             ver = f"v{dataset.lock.version}"
         if dataset.type == "dvc":
             if dataset.lock.path:
@@ -55,7 +55,7 @@ def run(self):
             raise DvcException("--path can't be used without --dvc")
 
         d = vars(self.args)
-        for key in ["dvc", "dvcx", "url"]:
+        for key in ["dvc", "dc", "url"]:
             if url := d.pop(key, None):
                 d.update({"type": key, "url": url})
                 break
@@ -88,8 +88,8 @@ def display(self, name: str, dataset: "Dataset", new: "Dataset"):
         assert new.lock
 
         v: Optional[tuple[str, str]] = None
-        if dataset.type == "dvcx":
-            assert new.type == "dvcx"
+        if dataset.type == "dc":
+            assert new.type == "dc"
             if new.lock.version < dataset.lock.version:
                 action = "Downgrading"
 
@@ -168,7 +168,7 @@ def add_parser(subparsers, parent_parser):
 
     url_exclusive_group = ds_add_parser.add_mutually_exclusive_group(required=True)
     url_exclusive_group.add_argument(
-        "--dvcx", metavar="name", help="Name of the dvcx dataset to track"
+        "--dc", metavar="name", help="Name of the DataChain dataset to track"
     )
     url_exclusive_group.add_argument(
         "--dvc",
@@ -219,7 +219,7 @@ def add_parser(subparsers, parent_parser):
         "--rev",
         "--version",
         nargs="?",
-        help="DVCX dataset version or Git revision (e.g. SHA, branch, tag)",
+        help="DataChain dataset version or Git revision (e.g. SHA, branch, tag)",
         metavar="<version>",
     )
     ds_update_parser.set_defaults(func=CmdDatasetUpdate)

dvc/dirs.py

@@ -21,6 +21,19 @@ def global_config_dir():
 
 
 def site_cache_dir():
+    from platformdirs import PlatformDirs
+    from platformdirs.unix import Unix
+
+    if issubclass(Unix, PlatformDirs):
+        # Return the cache directory shared by users, e.g. `/var/tmp/$appname`
+        # NOTE: platformdirs>=5 changed `site_cache_dir` to return /var/cache/$appname.
+        # as the following path is considered insecure.
+        # For details, see: https://github.com/tox-dev/platformdirs/pull/239
+
+        # FIXME: keeping the old behavior temporarily to avoid dependency conflict.
+        #        In the future, consider migrating to a more secure directory.
+        return f"/var/tmp/{APPNAME}"  # noqa: S108
+
     return os.getenv(env.DVC_SITE_CACHE_DIR) or platformdirs.site_cache_dir(
         APPNAME, APPAUTHOR, opinion=True
     )

dvc/repo/datasets.py

@@ -16,7 +16,7 @@
 from dvc_data.hashfile.meta import Meta
 
 if TYPE_CHECKING:
-    from dvcx.dataset import DatasetRecord, DatasetVersion  # type: ignore[import]
+    from datachain.dataset import DatasetRecord, DatasetVersion  # type: ignore[import]
     from typing_extensions import Self
 
     from dvc.repo import Repo
@@ -29,10 +29,10 @@ def _get_dataset_record(name: str) -> "DatasetRecord":
     from dvc.exceptions import DvcException
 
     try:
-        from dvcx.catalog import get_catalog  # type: ignore[import]
+        from datachain.catalog import get_catalog  # type: ignore[import]
 
     except ImportError as exc:
-        raise DvcException("dvcx is not installed") from exc
+        raise DvcException("datachain is not installed") from exc
 
     catalog = get_catalog()
     return catalog.get_remote_dataset(name)
@@ -86,7 +86,7 @@ def from_dict(cls, d: dict[str, Any]) -> "Self":
 class DatasetSpec(SerDe):
     name: str
     url: str
-    type: Literal["dvc", "dvcx", "url"]
+    type: Literal["dvc", "dc", "url"]
 
 
 @frozen(kw_only=True)
@@ -108,7 +108,7 @@ class DVCDatasetLock(DVCDatasetSpec):
 
 
 @frozen(kw_only=True)
-class DVCXDatasetLock(DatasetSpec):
+class DatachainDatasetLock(DatasetSpec):
     version: int
     created_at: datetime = field(converter=to_datetime)
 
@@ -160,13 +160,13 @@ def update(self, repo, rev: Optional[str] = None, **kwargs) -> "Self":
 
 
 @frozen(kw_only=True)
-class DVCXDataset:
+class DatachainDataset:
     manifest_path: str
     spec: "DatasetSpec"
-    lock: "Optional[DVCXDatasetLock]" = field(default=None)
+    lock: "Optional[DatachainDatasetLock]" = field(default=None)
     _invalidated: bool = field(default=False, eq=False, repr=False)
 
-    type: ClassVar[Literal["dvcx"]] = "dvcx"
+    type: ClassVar[Literal["dc"]] = "dc"
 
     @property
     def pinned(self) -> bool:
@@ -193,7 +193,7 @@ def update(
         name, _version = self.name_version
         version = version if version is not None else _version
         version_info = _get_dataset_info(name, record=record, version=version)
-        lock = DVCXDatasetLock(
+        lock = DatachainDatasetLock(
             **self.spec.to_dict(),
             version=version_info.version,
             created_at=version_info.created_at,
@@ -226,9 +226,9 @@ def update(self, repo, **kwargs):
         return evolve(self, lock=lock)
 
 
-Lock = Union[DVCDatasetLock, DVCXDatasetLock, URLDatasetLock]
+Lock = Union[DVCDatasetLock, DatachainDatasetLock, URLDatasetLock]
 Spec = Union[DatasetSpec, DVCDatasetSpec]
-Dataset = Union[DVCDataset, DVCXDataset, URLDataset]
+Dataset = Union[DVCDataset, DatachainDataset, URLDataset]
 
 
 class DatasetNotFoundError(DvcException, KeyError):
@@ -307,13 +307,13 @@ def _spec_from_info(spec: dict[str, Any]) -> Spec:
             raise ValueError("type should be present in spec")
         if typ == "dvc":
             return DVCDatasetSpec.from_dict(spec)
-        if typ in {"dvcx", "url"}:
+        if typ in {"dc", "url"}:
             return DatasetSpec.from_dict(spec)
         raise ValueError(f"unknown dataset type: {spec.get('type', '')}")
 
     @staticmethod
     def _lock_from_info(lock: Optional[dict[str, Any]]) -> Optional[Lock]:
-        kl = {"dvc": DVCDatasetLock, "dvcx": DVCXDatasetLock, "url": URLDatasetLock}
+        kl = {"dvc": DVCDatasetLock, "dc": DatachainDatasetLock, "url": URLDatasetLock}
         if lock and (cls := kl.get(lock.get("type", ""))):  # type: ignore[assignment]
             return cls.from_dict(lock)  # type: ignore[attr-defined]
         return None
@@ -356,9 +356,9 @@ def _build_dataset(
                 lock=lock,
                 invalidated=_invalidated,
             )
-        if spec.type == "dvcx":
-            assert lock is None or isinstance(lock, DVCXDatasetLock)
-            return DVCXDataset(
+        if spec.type == "dc":
+            assert lock is None or isinstance(lock, DatachainDatasetLock)
+            return DatachainDataset(
                 manifest_path=manifest_path,
                 spec=spec,
                 lock=lock,
@@ -374,7 +374,7 @@ def add(
         manifest_path: StrPath = "dvc.yaml",
         **kwargs: Any,
     ) -> Dataset:
-        assert type in {"dvc", "dvcx", "url"}
+        assert type in {"dvc", "dc", "url"}
         kwargs.update({"name": name, "url": url, "type": type})
         dataset = self._build_dataset(os.path.abspath(manifest_path), kwargs)
         dataset = dataset.update(self.repo)
@@ -389,13 +389,16 @@ def update(self, name, **kwargs) -> tuple[Dataset, Dataset]:
 
         if dataset.type == "url" and (version or kwargs.get("rev")):
             raise ValueError("cannot update version/revision for a url")
-        if dataset.type == "dvcx" and version is not None:
+        if dataset.type == "dc" and version is not None:
             if not isinstance(version, int):
                 raise TypeError(
-                    f"dvcx version has to be an integer, got {type(version).__name__!r}"
+                    "DataChain dataset version has to be an integer, "
+                    f"got {type(version).__name__!r}"
                 )
             if version < 1:
-                raise ValueError(f"dvcx version should be >=1, got {version}")
+                raise ValueError(
+                    f"DataChain dataset version should be >=1, got {version}"
+                )
 
         new = dataset.update(self.repo, **kwargs)
 

dvc/testing/benchmarks/fixtures.py

@@ -150,7 +150,8 @@ def _make_bench(name):
         import pytest_benchmark.plugin
 
         # hack from https://github.com/ionelmc/pytest-benchmark/issues/166
-        bench = pytest_benchmark.plugin.benchmark.__pytest_wrapped__.obj(request)
+        bench_gen = pytest_benchmark.plugin.benchmark.__pytest_wrapped__.obj(request)
+        bench = next(bench_gen)
 
         suffix = f"-{name}"
 
@@ -172,7 +173,7 @@ 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(
+        if request.config.getoption("--dvc-benchmark-cprofile-dump") or kwargs.pop(
             "cprofile", False
         ):
             cprofile_results = request.config.invocation_params.dir / "prof"

dvc/testing/benchmarks/plugin.py

@@ -73,7 +73,7 @@ def pytest_addoption(parser):
     )
 
     parser.addoption(
-        "--benchmark-cprofile-dump",
+        "--dvc-benchmark-cprofile-dump",
         action="store_true",
         default=False,
         help="Save cprofile results",

pyproject.toml

@@ -57,7 +57,7 @@ dependencies = [
     "omegaconf",
     "packaging>=19",
     "pathspec>=0.10.3",
-    "platformdirs<4,>=3.1.1",
+    "platformdirs<5,>=3.1.1",
     "psutil>=5.8",
     "pydot>=1.2.4",
     "pygtrie>=2.3.2",
@@ -99,7 +99,7 @@ s3 = ["dvc-s3>=3.2.0,<4"]
 ssh = ["dvc-ssh>=4,<5"]
 ssh_gssapi = ["dvc-ssh[gssapi]>=4,<5"]
 testing = [
-  "pytest-benchmark[histogram]",
+  "pytest-benchmark[histogram]>=5,<6",
   "pytest-test-utils",
   "uv",
 ]

tests/func/parsing/test_top_level.py

@@ -121,7 +121,7 @@ def test_artifacts(tmp_dir, dvc):
 def test_datasets(tmp_dir, dvc):
     template = {
         "datasets": [
-            {"name": "${ds1.name}", "url": "${ds1.url}", "type": "dvcx"},
+            {"name": "${ds1.name}", "url": "${ds1.url}", "type": "dc"},
             {
                 "name": "${ds2.name}",
                 "url": "${ds2.url}",
@@ -138,7 +138,7 @@ def test_datasets(tmp_dir, dvc):
 
     (tmp_dir / "params.yaml").dump(
         {
-            "ds1": {"name": "dogs", "url": "dvcx://dogs"},
+            "ds1": {"name": "dogs", "url": "dc://dogs"},
             "ds2": {
                 "name": "example-get-started",
                 "url": "[email protected]:iterative/example-get-started.git",
@@ -153,7 +153,7 @@ def test_datasets(tmp_dir, dvc):
 
     resolver = DataResolver(dvc, tmp_dir, template)
     assert resolver.resolve_datasets() == [
-        {"name": "dogs", "url": "dvcx://dogs", "type": "dvcx"},
+        {"name": "dogs", "url": "dc://dogs", "type": "dc"},
         {
             "name": "example-get-started",
             "url": "[email protected]:iterative/example-get-started.git",