dvc: remove dvc.data, fix imports

tests/func/data/db/test_index.py

@@ -1,9 +1,9 @@
 import pytest
 
-from dvc.data.db import get_index
 from dvc.exceptions import DownloadError, UploadError
 from dvc.fs import LocalFileSystem
 from dvc.utils.fs import remove
+from dvc_data.db import get_index
 from tests.utils import clean_staging
 
 

tests/func/test_add.py

@@ -11,6 +11,7 @@ import colorama
 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
@@ -103,7 +104,7 @@ def test_add_unsupported_file(dvc):
 
 
 def test_add_directory(tmp_dir, dvc):
-    from dvc.data import load
+    from dvc_data import load
 
     (stage,) = tmp_dir.dvc_gen({"dir": {"file": "file"}})
 
@@ -457,7 +458,7 @@ class TestAddCommit(TestDvc):
 
 def test_should_collect_dir_cache_only_once(mocker, tmp_dir, dvc):
     tmp_dir.gen({"data/data": "foo"})
-    counter = mocker.spy(dvc_module.data.stage, "_stage_tree")
+    counter = mocker.spy(dvc_data.stage, "_stage_tree")
     ret = main(["add", "data"])
     assert ret == 0
     assert counter.mock.call_count == 1
@@ -881,7 +882,7 @@ def test_add_with_cache_link_error(tmp_dir, dvc, mocker, capsys):
     tmp_dir.gen("foo", "foo")
 
     mocker.patch(
-        "dvc.data.checkout.test_links",
+        "dvc_data.checkout.test_links",
         return_value=[],
     )
     dvc.add("foo")

tests/func/test_checkout.py

@@ -84,7 +84,7 @@ class TestCheckoutCorruptedCacheFile(TestRepro):
 
 class TestCheckoutCorruptedCacheDir(TestDvc):
     def test(self):
-        from dvc.data import load
+        from dvc_data import load
 
         # NOTE: using 'copy' so that cache and link don't have same inode
         ret = main(["config", "cache.type", "copy"])

tests/func/test_data_cloud.py

@@ -7,13 +7,13 @@ from flaky.flaky_decorator import flaky
 
 import dvc_objects
 from dvc.cli import main
-from dvc.data.db.local import LocalObjectDB
 from dvc.external_repo import clean_repos
 from dvc.stage.exceptions import StageNotFound
 from dvc.testing.test_remote import (  # noqa, pylint: disable=unused-import
     TestRemote,
 )
 from dvc.utils.fs import remove
+from dvc_data.db.local import LocalObjectDB
 from dvc_objects.db import ObjectDB
 
 

tests/func/test_external_repo.py

@@ -3,11 +3,11 @@ from unittest.mock import ANY
 
 from scmrepo.git import Git
 
-from dvc.data.stage import stage
-from dvc.data.transfer import transfer
 from dvc.external_repo import CLONES, external_repo
 from dvc.utils import relpath
 from dvc.utils.fs import makedirs, remove
+from dvc_data.stage import stage
+from dvc_data.transfer import transfer
 from tests.unit.fs.test_dvc import make_subrepo
 from tests.utils import clean_staging
 

tests/func/test_gc.py

@@ -7,11 +7,11 @@ import pytest
 from git import Repo
 
 from dvc.cli import main
-from dvc.data.db.local import LocalObjectDB
 from dvc.exceptions import CollectCacheError
 from dvc.fs import LocalFileSystem
 from dvc.repo import Repo as DvcRepo
 from dvc.utils.fs import remove
+from dvc_data.db.local import LocalObjectDB
 from tests.basic_env import TestDir, TestDvcGit
 
 

tests/func/test_ignore.py

@@ -4,13 +4,13 @@ from pathlib import Path
 
 import pytest
 
-from dvc.data.stage import IgnoreInCollectedDirError
 from dvc.ignore import DvcIgnore, DvcIgnorePatterns
 from dvc.output import OutputIsIgnoredError
 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.stage import IgnoreInCollectedDirError
 from dvc_objects.utils import get_mtime_and_size
 
 

tests/func/test_import.py

@@ -529,7 +529,7 @@ def test_import_with_no_exec(tmp_dir, dvc, erepo_dir):
 
 
 def test_import_with_jobs(mocker, dvc, erepo_dir):
-    import dvc.data.transfer as otransfer
+    import dvc_data.transfer as otransfer
 
     with erepo_dir.chdir():
         erepo_dir.dvc_gen(

tests/func/test_odb.py

@@ -43,7 +43,7 @@ def test_cache(tmp_dir, dvc):
 
 
 def test_cache_load_bad_dir_cache(tmp_dir, dvc):
-    from dvc.data import load
+    from dvc_data import load
 
     dir_hash = "123.dir"
     fname = os.fspath(dvc.odb.local.hash_to_path(dir_hash))

tests/func/test_remote.py

@@ -139,8 +139,8 @@ def test_upper_case_remote(tmp_dir, dvc, local_cloud):
 
 
 def test_dir_hash_should_be_key_order_agnostic(tmp_dir, dvc):
-    from dvc.data.stage import stage
-    from dvc.data.tree import Tree
+    from dvc_data.stage import stage
+    from dvc_data.tree import Tree
 
     tmp_dir.gen({"data": {"1": "1 content", "2": "2 content"}})
 
@@ -150,7 +150,7 @@ def test_dir_hash_should_be_key_order_agnostic(tmp_dir, dvc):
         [{"relpath": "1", "md5": "1"}, {"relpath": "2", "md5": "2"}]
     )
     tree.digest()
-    with patch("dvc.data.stage._stage_tree", return_value=(None, tree)):
+    with patch("dvc_data.stage._stage_tree", return_value=(None, tree)):
         _, _, obj = stage(dvc.odb.local, path, dvc.odb.local.fs, "md5")
         hash1 = obj.hash_info
 
@@ -161,7 +161,7 @@ def test_dir_hash_should_be_key_order_agnostic(tmp_dir, dvc):
         [{"md5": "1", "relpath": "1"}, {"md5": "2", "relpath": "2"}]
     )
     tree.digest()
-    with patch("dvc.data.stage._stage_tree", return_value=(None, tree)):
+    with patch("dvc_data.stage._stage_tree", return_value=(None, tree)):
         _, _, obj = stage(dvc.odb.local, path, dvc.odb.local.fs, "md5")
         hash2 = obj.hash_info
 

tests/func/test_stage.py

@@ -342,8 +342,8 @@ def test_stage_run_checkpoint(tmp_dir, dvc, mocker, checkpoint):
 def test_stage_dir_optimization(
     tmp_dir, dvc, mocker, dry_run, expected_staging_contents
 ):
-    from dvc.data import stage
-    from dvc.data.tree import Tree
+    from dvc_data import stage
+    from dvc_data.tree import Tree
 
     tmp_dir.dvc_gen(
         {

tests/unit/cli/test_main.py

@@ -4,7 +4,7 @@ import pytest
 from funcy import raiser
 
 from dvc.cli import main
-from dvc.data.stage import IgnoreInCollectedDirError
+from dvc_data.stage import IgnoreInCollectedDirError
 from dvc_objects.cache import DiskError
 from dvc_objects.fs.base import FileSystem, RemoteMissingDepsError
 

tests/unit/data/db/test_local.py

@@ -3,8 +3,8 @@ import os
 
 import pytest
 
-from dvc.data.db.local import LocalObjectDB
 from dvc.fs import LocalFileSystem
+from dvc_data.db.local import LocalObjectDB
 from dvc_objects.hash_info import HashInfo
 
 
@@ -13,7 +13,7 @@ def test_status_download_optimization(mocker, dvc):
     And the desired files to fetch are already on the local cache,
     Don't check the existence of the desired files on the remote cache
     """
-    from dvc.data.status import compare_status
+    from dvc_data.status import compare_status
 
     odb = LocalObjectDB(LocalFileSystem(), os.getcwd())
     obj_ids = {
@@ -85,9 +85,9 @@ def test_set_exec_ignore_errors(tmp_dir, dvc, mocker, err):
 
 
 def test_staging_file(tmp_dir, dvc):
-    from dvc.data import check
-    from dvc.data.stage import stage
-    from dvc.data.transfer import transfer
+    from dvc_data import check
+    from dvc_data.stage import stage
+    from dvc_data.transfer import transfer
 
     tmp_dir.gen("foo", "foo")
     fs = LocalFileSystem()
@@ -113,9 +113,9 @@ def test_staging_file(tmp_dir, dvc):
 
 
 def test_staging_dir(tmp_dir, dvc):
-    from dvc.data import check
-    from dvc.data.stage import stage
-    from dvc.data.transfer import transfer
+    from dvc_data import check
+    from dvc_data.stage import stage
+    from dvc_data.transfer import transfer
 
     tmp_dir.gen({"dir": {"foo": "foo", "bar": "bar"}})
     fs = LocalFileSystem()

tests/unit/data/test_tree.py

@@ -2,7 +2,7 @@ from operator import itemgetter
 
 import pytest
 
-from dvc.data.tree import Tree, _merge
+from dvc_data.tree import Tree, _merge
 from dvc_objects.hash_info import HashInfo
 from dvc_objects.meta import Meta
 

tests/unit/fs/test_data.py

@@ -5,9 +5,9 @@ import pytest
 
 import dvc_objects
 from dvc.config import NoRemoteError
-from dvc.data.stage import stage
 from dvc.fs.data import DataFileSystem
 from dvc.utils.fs import remove
+from dvc_data.stage import stage
 from dvc_objects.hash_info import HashInfo
 
 

tests/unit/fs/test_dvc.py

@@ -5,8 +5,8 @@ from unittest import mock
 
 import pytest
 
-from dvc.data.stage import stage
 from dvc.fs.dvc import DvcFileSystem
+from dvc_data.stage import stage
 from dvc_objects.hash_info import HashInfo
 from tests.utils import clean_staging
 
@@ -575,7 +575,7 @@ 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 import check
     from dvc_objects.errors import ObjectFormatError
     from dvc_objects.hash import hash_file
 

tests/unit/remote/test_index.py

@@ -3,7 +3,7 @@ import os
 import pytest
 from funcy import first
 
-from dvc.data.db.index import ObjectDBIndex
+from dvc_data.db.index import ObjectDBIndex
 
 
 @pytest.fixture

tests/unit/remote/test_remote.py

@@ -2,8 +2,8 @@ import os
 
 import pytest
 
-from dvc.data.db import get_index
 from dvc.fs import GSFileSystem, S3FileSystem, get_cloud_fs
+from dvc_data.db import get_index
 
 
 def test_remote_with_hash_jobs(dvc):

tests/utils/__init__.py

@@ -45,8 +45,8 @@ def dump_sv(stream, metrics, delimiter=",", header=True):
 
 
 def clean_staging():
-    from dvc.data.stage import _STAGING_MEMFS_PATH
     from dvc.fs import MemoryFileSystem
+    from dvc_data.stage import _STAGING_MEMFS_PATH
 
     try:
         MemoryFileSystem().fs.rm(