deps: bump dvc-data to 0.0.6; fix imports

dvc/cli/__init__.py

@@ -107,7 +107,7 @@ def _log_exceptions(exc: Exception) -> Optional[int]:
         )
         return 251
 
-    from dvc_objects.hashfile.cache import DiskError
+    from dvc_data.hashfile.cache import DiskError
 
     if isinstance(exc, DiskError):
         from dvc.utils import relpath

dvc/commands/dag.py

@@ -70,9 +70,9 @@ def _collect_targets(repo, target, outs):
             targets.extend([str(out) for out in stage.outs])
             continue
 
-        for out in outs_trie.itervalues(
+        for out in outs_trie.itervalues(  # noqa: B301
             prefix=repo.fs.path.parts(path)
-        ):  # noqa: B301
+        ):
             targets.extend(str(out))
 
     return targets

dvc/data_cloud.py

@@ -6,9 +6,9 @@ from typing import TYPE_CHECKING, Iterable, Optional
 from dvc_data.db import get_index
 
 if TYPE_CHECKING:
+    from dvc_data.hashfile.db import HashFileDB
+    from dvc_data.hashfile.hash_info import HashInfo
     from dvc_data.status import CompareStatusResult
-    from dvc_objects.db import ObjectDB
-    from dvc_objects.hashfile.hash_info import HashInfo
 
 logger = logging.getLogger(__name__)
 
@@ -31,7 +31,7 @@ class DataCloud:
         self,
         name: Optional[str] = None,
         command: str = "<command>",
-    ) -> "ObjectDB":
+    ) -> "HashFileDB":
         from dvc.config import NoRemoteError
 
         if not name:
@@ -76,8 +76,8 @@ class DataCloud:
 
     def transfer(
         self,
-        src_odb: "ObjectDB",
-        dest_odb: "ObjectDB",
+        src_odb: "HashFileDB",
+        dest_odb: "HashFileDB",
         objs: Iterable["HashInfo"],
         **kwargs,
     ):
@@ -94,7 +94,7 @@ class DataCloud:
         objs: Iterable["HashInfo"],
         jobs: Optional[int] = None,
         remote: Optional[str] = None,
-        odb: Optional["ObjectDB"] = None,
+        odb: Optional["HashFileDB"] = None,
     ):
         """Push data items in a cloud-agnostic way.
 
@@ -121,7 +121,7 @@ class DataCloud:
         objs: Iterable["HashInfo"],
         jobs: Optional[int] = None,
         remote: Optional[str] = None,
-        odb: Optional["ObjectDB"] = None,
+        odb: Optional["HashFileDB"] = None,
     ):
         """Pull data items in a cloud-agnostic way.
 
@@ -149,7 +149,7 @@ class DataCloud:
         objs: Iterable["HashInfo"],
         jobs: Optional[int] = None,
         remote: Optional[str] = None,
-        odb: Optional["ObjectDB"] = None,
+        odb: Optional["HashFileDB"] = None,
         log_missing: bool = True,
     ):
         """Check status of data items in a cloud-agnostic way.

dvc/dependency/param.py

@@ -9,7 +9,7 @@ from voluptuous import Any
 
 from dvc.exceptions import DvcException
 from dvc.utils.serialize import LOADERS, ParseError
-from dvc_objects.hashfile.hash_info import HashInfo
+from dvc_data.hashfile.hash_info import HashInfo
 
 from .base import Dependency
 

dvc/dependency/repo.py

@@ -10,9 +10,9 @@ from dvc.prompt import confirm
 from .base import Dependency
 
 if TYPE_CHECKING:
+    from dvc_data.hashfile.file import HashFile
+    from dvc_data.hashfile.hash_info import HashInfo
     from dvc_objects.db import ObjectDB
-    from dvc_objects.hashfile.file import HashFile
-    from dvc_objects.hashfile.hash_info import HashInfo
 
 
 class RepoDependency(Dependency):

dvc/fs/data.py

@@ -117,12 +117,12 @@ class _DataFileSystem(AbstractFileSystem):  # pylint:disable=abstract-method
         return bool(info.get("outs") if recurse else info.get("isout"))
 
     def info(self, path, **kwargs):
-        from dvc_objects.hashfile.meta import Meta
+        from dvc_data.hashfile.meta import Meta
 
         key = self._get_key(path)
 
         try:
-            outs = list(self.repo.index.tree.iteritems(key))
+            outs = list(self.repo.index.tree.iteritems(key))  # noqa: B301
         except KeyError as exc:
             raise FileNotFoundError from exc
 

dvc/output.py

@@ -21,12 +21,12 @@ from dvc_data import Tree
 from dvc_data import check as ocheck
 from dvc_data import load as oload
 from dvc_data.checkout import checkout
+from dvc_data.hashfile.hash_info import HashInfo
+from dvc_data.hashfile.istextfile import istextfile
+from dvc_data.hashfile.meta import Meta
 from dvc_data.stage import stage as ostage
 from dvc_data.transfer import transfer as otransfer
 from dvc_objects.errors import ObjectFormatError
-from dvc_objects.hashfile.hash_info import HashInfo
-from dvc_objects.hashfile.istextfile import istextfile
-from dvc_objects.hashfile.meta import Meta
 
 from .fs import (
     HDFSFileSystem,

dvc/repo/__init__.py

@@ -3,7 +3,7 @@ import os
 from collections import defaultdict
 from contextlib import contextmanager
 from functools import wraps
-from typing import TYPE_CHECKING, Callable, Optional, Set
+from typing import TYPE_CHECKING, Callable, Optional
 
 from funcy import cached_property
 
@@ -18,7 +18,6 @@ if TYPE_CHECKING:
     from dvc.fs import FileSystem
     from dvc.repo.scm_context import SCMContext
     from dvc.scm import Base
-    from dvc_objects.hashfile.file import HashFile
 
 logger = logging.getLogger(__name__)
 
@@ -165,7 +164,7 @@ class Repo:
         from dvc.repo.stage import StageLoad
         from dvc.scm import SCM
         from dvc.stage.cache import StageCache
-        from dvc_objects.hashfile.state import State, StateNoop
+        from dvc_data.hashfile.state import State, StateNoop
 
         self.url = url
         self._fs_conf = {"repo_factory": repo_factory}
@@ -411,16 +410,7 @@ class Repo:
         """
         used = defaultdict(set)
 
-        def _add_suffix(objs: Set["HashFile"], suffix: str) -> None:
-            from itertools import chain
-
-            from dvc_data import iterobjs
-
-            for obj in chain.from_iterable(map(iterobjs, objs)):
-                if obj.name is not None:
-                    obj.name += suffix
-
-        for branch in self.brancher(
+        for _ in self.brancher(
             revs=revs,
             all_branches=all_branches,
             all_tags=all_tags,
@@ -435,8 +425,6 @@ class Repo:
                 recursive=recursive,
                 with_deps=with_deps,
             ).items():
-                if branch:
-                    _add_suffix(objs, f" ({branch})")
                 used[odb].update(objs)
 
         if used_run_cache:

dvc/repo/index.py

@@ -26,8 +26,8 @@ if TYPE_CHECKING:
     from dvc.stage import Stage
     from dvc.types import StrPath, TargetType
     from dvc_data import Tree
+    from dvc_data.hashfile.hash_info import HashInfo
     from dvc_objects.db import ObjectDB
-    from dvc_objects.hashfile.hash_info import HashInfo
 
 
 ObjectContainer = Dict[Optional["ObjectDB"], Set["HashInfo"]]

dvc/schema.py

@@ -7,7 +7,7 @@ from dvc.output import CHECKSUMS_SCHEMA, Output
 from dvc.parsing import DO_KWD, FOREACH_KWD, VARS_KWD
 from dvc.parsing.versions import SCHEMA_KWD, lockfile_version_schema
 from dvc.stage.params import StageParams
-from dvc_objects.hashfile.meta import Meta
+from dvc_data.hashfile.meta import Meta
 
 STAGES = "stages"
 SINGLE_STAGE_SCHEMA = {

dvc/stage/__init__.py

@@ -36,8 +36,8 @@ from .utils import (
 
 if TYPE_CHECKING:
     from dvc.dvcfile import DVCFile
+    from dvc_data.hashfile.hash_info import HashInfo
     from dvc_objects.db.base import ObjectDB
-    from dvc_objects.hashfile.hash_info import HashInfo
 
 logger = logging.getLogger(__name__)
 # Disallow all punctuation characters except hyphen and underscore

dvc/stage/cache.py

@@ -27,7 +27,7 @@ class RunCacheNotSupported(DvcException):
 
 
 def _get_cache_hash(cache, key=False):
-    from dvc_objects.hashfile.meta import Meta
+    from dvc_data.hashfile.meta import Meta
 
     if key:
         cache["outs"] = [out["path"] for out in cache.get("outs", [])]

dvc/stage/loader.py

@@ -8,8 +8,8 @@ from funcy import cached_property, get_in, lcat, once, project
 from dvc import dependency, output
 from dvc.parsing import FOREACH_KWD, JOIN, DataResolver, EntryNotFound
 from dvc.parsing.versions import LOCKFILE_VERSION
-from dvc_objects.hashfile.hash_info import HashInfo
-from dvc_objects.hashfile.meta import Meta
+from dvc_data.hashfile.hash_info import HashInfo
+from dvc_data.hashfile.meta import Meta
 
 from . import PipelineStage, Stage, loads_from
 from .exceptions import StageNameUnspecified, StageNotFound

dvc/stage/utils.py

@@ -5,7 +5,7 @@ from typing import TYPE_CHECKING, Union
 from funcy import concat, first, lsplit, rpartial, without
 
 from dvc.exceptions import InvalidArgumentError
-from dvc_objects.hashfile.meta import Meta
+from dvc_data.hashfile.meta import Meta
 
 from .exceptions import (
     MissingDataSource,

setup.cfg

@@ -69,7 +69,7 @@ install_requires =
     scmrepo==0.0.24
     dvc-render==0.0.6
     dvclive>=0.7.3
-    dvc-data==0.0.5
+    dvc-data==0.0.6
 
 [options.extras_require]
 all =

tests/func/test_add.py

@@ -12,7 +12,6 @@ import pytest
 
 import dvc as dvc_module
 import dvc_data
-import dvc_objects
 from dvc.cli import main
 from dvc.dvcfile import DVC_FILE_SUFFIX
 from dvc.exceptions import (
@@ -38,8 +37,8 @@ from dvc.testing.test_workspace import TestAdd
 from dvc.utils import LARGE_DIR_SIZE, relpath
 from dvc.utils.fs import path_isin
 from dvc.utils.serialize import YAMLFileCorruptedError, load_yaml
-from dvc_objects.hashfile.hash import file_md5
-from dvc_objects.hashfile.hash_info import HashInfo
+from dvc_data.hashfile.hash import file_md5
+from dvc_data.hashfile.hash_info import HashInfo
 from tests.basic_env import TestDvc
 from tests.utils import get_gitignore_content
 
@@ -380,7 +379,7 @@ class TestDoubleAddUnchanged(TestDvc):
 
 
 def test_should_update_state_entry_for_file_after_add(mocker, dvc, tmp_dir):
-    file_md5_counter = mocker.spy(dvc_objects.hashfile.hash, "file_md5")
+    file_md5_counter = mocker.spy(dvc_data.hashfile.hash, "file_md5")
     tmp_dir.gen("foo", "foo")
 
     ret = main(["config", "cache.type", "copy"])
@@ -411,7 +410,7 @@ def test_should_update_state_entry_for_file_after_add(mocker, dvc, tmp_dir):
 def test_should_update_state_entry_for_directory_after_add(
     mocker, dvc, tmp_dir
 ):
-    file_md5_counter = mocker.spy(dvc_objects.hashfile.hash, "file_md5")
+    file_md5_counter = mocker.spy(dvc_data.hashfile.hash, "file_md5")
 
     tmp_dir.gen({"data/data": "foo", "data/data_sub/sub_data": "foo"})
 

tests/func/test_data_cloud.py

@@ -5,7 +5,7 @@ import shutil
 import pytest
 from flaky.flaky_decorator import flaky
 
-import dvc_objects
+import dvc_data
 from dvc.cli import main
 from dvc.external_repo import clean_repos
 from dvc.stage.exceptions import StageNotFound
@@ -14,7 +14,7 @@ from dvc.testing.test_remote import (  # noqa, pylint: disable=unused-import
 )
 from dvc.utils.fs import remove
 from dvc_data.db.local import LocalHashFileDB
-from dvc_objects.hashfile.db import HashFileDB
+from dvc_data.hashfile.db import HashFileDB
 
 
 def test_cloud_cli(tmp_dir, dvc, remote, mocker):
@@ -148,7 +148,7 @@ def test_warn_on_outdated_stage(tmp_dir, dvc, local_remote, caplog):
 
 def test_hash_recalculation(mocker, dvc, tmp_dir, local_remote):
     tmp_dir.gen({"foo": "foo"})
-    test_file_md5 = mocker.spy(dvc_objects.hashfile.hash, "file_md5")
+    test_file_md5 = mocker.spy(dvc_data.hashfile.hash, "file_md5")
     ret = main(["config", "cache.type", "hardlink"])
     assert ret == 0
     ret = main(["add", "foo"])
@@ -210,7 +210,7 @@ def test_verify_hashes(
     remove("dir")
     remove(dvc.odb.local.cache_dir)
 
-    hash_spy = mocker.spy(dvc_objects.hashfile.hash, "file_md5")
+    hash_spy = mocker.spy(dvc_data.hashfile.hash, "file_md5")
 
     dvc.pull()
     assert hash_spy.call_count == 0

tests/func/test_ignore.py

@@ -10,8 +10,8 @@ from dvc.pathspec_math import PatternInfo, merge_patterns
 from dvc.repo import Repo
 from dvc.testing.tmp_dir import TmpDir
 from dvc.types import List
+from dvc_data.hashfile.utils import get_mtime_and_size
 from dvc_data.stage import IgnoreInCollectedDirError
-from dvc_objects.hashfile.utils import get_mtime_and_size
 
 
 def _to_pattern_info_list(str_list: List):

tests/func/test_install.py

@@ -6,7 +6,7 @@ import pytest
 from git import GitCommandError
 
 from dvc.exceptions import DvcException
-from dvc_objects.hashfile.hash import file_md5
+from dvc_data.hashfile.hash import file_md5
 from tests.func.parsing.test_errors import escape_ansi
 
 

tests/func/test_lockfile.py

@@ -10,7 +10,7 @@ from dvc.stage.utils import split_params_deps
 from dvc.utils.fs import remove
 from dvc.utils.serialize import dumps_yaml, parse_yaml_for_update
 from dvc.utils.strictyaml import YAMLValidationError, make_relpath
-from dvc_objects.hashfile.hash_info import HashInfo
+from dvc_data.hashfile.hash_info import HashInfo
 from tests.func.test_run_multistage import supported_params
 
 FS_STRUCTURE = {

tests/func/test_odb.py

@@ -7,8 +7,8 @@ import pytest
 from dvc.cli import main
 from dvc.odbmgr import ODBManager
 from dvc.utils import relpath
+from dvc_data.hashfile.hash_info import HashInfo
 from dvc_objects.errors import ObjectFormatError
-from dvc_objects.hashfile.hash_info import HashInfo
 
 
 def test_cache(tmp_dir, dvc):

tests/func/test_repo_index.py

@@ -258,7 +258,7 @@ def get_index(dvc, rev):
 
 @pytest.mark.parametrize("rev", ["workspace", "HEAD"])
 def test_used_objs(tmp_dir, scm, dvc, run_copy, rev):
-    from dvc_objects.hashfile.hash_info import HashInfo
+    from dvc_data.hashfile.hash_info import HashInfo
 
     dvc.config["core"]["autostage"] = True
     tmp_dir.dvc_gen({"dir": {"subdir": {"file": "file"}}, "foo": "foo"})

tests/func/test_repro.py

@@ -21,7 +21,7 @@ from dvc.stage.exceptions import StageFileDoesNotExistError
 from dvc.utils import relpath
 from dvc.utils.fs import remove
 from dvc.utils.serialize import dump_yaml, load_yaml
-from dvc_objects.hashfile.hash import file_md5
+from dvc_data.hashfile.hash import file_md5
 from tests.basic_env import TestDvc
 
 
@@ -817,7 +817,7 @@ class TestReproAllPipelines(SingleStageRun, TestDvc):
             ),
         ]
 
-        from dvc_objects.hashfile.state import StateNoop
+        from dvc_data.hashfile.state import StateNoop
 
         self.dvc.state = StateNoop()
 

tests/func/test_run_single_stage.py

@@ -31,7 +31,7 @@ from dvc.stage.exceptions import (
     StagePathOutsideError,
 )
 from dvc.utils.serialize import load_yaml
-from dvc_objects.hashfile.hash import file_md5
+from dvc_data.hashfile.hash import file_md5
 from tests.basic_env import TestDvc, TestDvcGit
 
 

tests/func/test_state.py

@@ -2,9 +2,9 @@ import os
 import re
 
 from dvc.repo import Repo
-from dvc_objects.hashfile.hash import file_md5
-from dvc_objects.hashfile.hash_info import HashInfo
-from dvc_objects.hashfile.state import State
+from dvc_data.hashfile.hash import file_md5
+from dvc_data.hashfile.hash_info import HashInfo
+from dvc_data.hashfile.state import State
 
 
 def test_state(tmp_dir, dvc):

tests/unit/cli/test_main.py

@@ -4,9 +4,9 @@ import pytest
 from funcy import raiser
 
 from dvc.cli import main
+from dvc_data.hashfile.cache import DiskError
 from dvc_data.stage import IgnoreInCollectedDirError
 from dvc_objects.fs.base import FileSystem, RemoteMissingDepsError
-from dvc_objects.hashfile.cache import DiskError
 
 
 def test_state_pickle_errors_are_correctly_raised(tmp_dir, caplog, mocker):

tests/unit/data/db/test_local.py

@@ -5,7 +5,7 @@ import pytest
 
 from dvc.fs import LocalFileSystem
 from dvc_data.db.local import LocalHashFileDB
-from dvc_objects.hashfile.hash_info import HashInfo
+from dvc_data.hashfile.hash_info import HashInfo
 
 
 def test_status_download_optimization(mocker, dvc):

tests/unit/fs/test_data.py

@@ -3,12 +3,12 @@ import shutil
 
 import pytest
 
-import dvc_objects
+import dvc_data
 from dvc.config import NoRemoteError
 from dvc.fs.data import DataFileSystem
 from dvc.utils.fs import remove
+from dvc_data.hashfile.hash_info import HashInfo
 from dvc_data.stage import stage
-from dvc_objects.hashfile.hash_info import HashInfo
 
 
 @pytest.mark.parametrize(
@@ -231,7 +231,7 @@ def test_get_hash_dir(tmp_dir, dvc, mocker):
         {"dir": {"foo": "foo", "bar": "bar", "subdir": {"data": "data"}}}
     )
     fs = DataFileSystem(repo=dvc)
-    hash_file_spy = mocker.spy(dvc_objects.hashfile.hash, "hash_file")
+    hash_file_spy = mocker.spy(dvc_data.hashfile.hash, "hash_file")
     assert fs.info("dir")["md5"] == "8761c4e9acad696bee718615e23e22db.dir"
     assert not hash_file_spy.called
 

tests/unit/fs/test_dvc.py

@@ -6,8 +6,8 @@ from unittest import mock
 import pytest
 
 from dvc.fs.dvc import DvcFileSystem
+from dvc_data.hashfile.hash_info import HashInfo
 from dvc_data.stage import stage
-from dvc_objects.hashfile.hash_info import HashInfo
 from tests.utils import clean_staging
 
 
@@ -576,8 +576,8 @@ def test_get_hash_mixed_dir(tmp_dir, scm, dvc):
 
 def test_get_hash_dirty_file(tmp_dir, dvc):
     from dvc_data import check
+    from dvc_data.hashfile.hash import hash_file
     from dvc_objects.errors import ObjectFormatError
-    from dvc_objects.hashfile.hash import hash_file
 
     tmp_dir.dvc_gen("file", "file")
     file_hash_info = HashInfo("md5", "8c7dd922ad47494fc02c388e12c00eac")

tests/unit/output/test_local.py

@@ -4,8 +4,8 @@ from unittest.mock import patch
 from dvc.output import Output
 from dvc.stage import Stage
 from dvc.utils import relpath
-from dvc_objects.hashfile.hash_info import HashInfo
-from dvc_objects.hashfile.meta import Meta
+from dvc_data.hashfile.hash_info import HashInfo
+from dvc_data.hashfile.meta import Meta
 from tests.basic_env import TestDvc
 
 

tests/unit/remote/test_base.py

@@ -3,7 +3,7 @@ import posixpath
 from unittest import mock
 
 from dvc.fs import FileSystem
-from dvc_objects.hashfile.db import ObjectDB
+from dvc_data.hashfile.db import ObjectDB
 
 
 class _CallableOrNone:

tests/unit/repo/test_repo.py

@@ -5,7 +5,7 @@ import pytest
 
 from dvc.exceptions import OutputDuplicationError
 from dvc.repo import NotDvcRepoError, Repo, locked
-from dvc_objects.hashfile.hash_info import HashInfo
+from dvc_data.hashfile.hash_info import HashInfo
 
 
 def test_is_dvc_internal(dvc):

tests/unit/stage/test_loader_pipeline_file.py

@@ -8,7 +8,7 @@ from dvc.dvcfile import PIPELINE_FILE, Dvcfile
 from dvc.stage import PipelineStage, create_stage
 from dvc.stage.loader import StageLoader
 from dvc.stage.serialize import split_params_deps
-from dvc_objects.hashfile.hash_info import HashInfo
+from dvc_data.hashfile.hash_info import HashInfo
 
 
 @pytest.fixture

tests/unit/stage/test_serialize_pipeline_lock.py

@@ -11,7 +11,7 @@ from dvc.stage.serialize import (
     to_single_stage_lockfile as _to_single_stage_lockfile,
 )
 from dvc.stage.utils import split_params_deps
-from dvc_objects.hashfile.hash_info import HashInfo
+from dvc_data.hashfile.hash_info import HashInfo
 
 kwargs = {"name": "something", "cmd": "command", "path": PIPELINE_FILE}
 Schema = _Schema(LOCK_FILE_STAGE_SCHEMA)

tests/unit/test_hashinfo.py

@@ -1,4 +1,4 @@
-from dvc_objects.hashfile.hash_info import HashInfo
+from dvc_data.hashfile.hash_info import HashInfo
 
 
 def test_as_raw():