fix(perf): only load a stage/.dvc file if the target is an address to that stage (#10838)

tests/func/test_checkout.py

@@ -8,7 +8,7 @@ import pytest
 from dulwich.porcelain import remove as git_rm
 
 from dvc.cli import main
-from dvc.dvcfile import PROJECT_FILE, load_file
+from dvc.dvcfile import PROJECT_FILE, FileMixin, SingleStageFile, load_file
 from dvc.exceptions import CheckoutError, CheckoutErrorSuggestGit, NoOutputOrStageError
 from dvc.fs import system
 from dvc.stage.exceptions import StageFileDoesNotExistError
@@ -754,3 +754,20 @@ def test_checkout_cleanup_properly_on_untracked_nested_directories(tmp_dir, scm,
     dvc.checkout(force=True)
 
     assert (tmp_dir / "datasets").read_text() == {"dir1": {"file1": "file1"}}
+
+
+def test_checkout_loads_specific_file(tmp_dir, dvc, mocker):
+    tmp_dir.dvc_gen("foo", "foo")
+    tmp_dir.dvc_gen("bar", "bar")
+
+    (tmp_dir / "bar").unlink()
+    (tmp_dir / "foo").unlink()
+
+    f = SingleStageFile(dvc, "foo.dvc")
+
+    spy = mocker.spy(FileMixin, "_load")
+    assert dvc.checkout("foo.dvc") == {"added": ["foo"], "deleted": [], "modified": []}
+
+    spy.assert_called_with(f)
+    assert (tmp_dir / "foo").exists()
+    assert not (tmp_dir / "bar").exists()

tests/func/test_data_cloud.py

@@ -7,6 +7,7 @@ import pytest
 
 import dvc_data
 from dvc.cli import main
+from dvc.dvcfile import FileMixin, SingleStageFile
 from dvc.exceptions import CheckoutError
 from dvc.repo.open_repo import clean_repos
 from dvc.scm import CloneError
@@ -678,3 +679,38 @@ def test_pull_granular_excluding_import_that_cannot_be_pulled(
         dvc.pull()
     with pytest.raises(CloneError, match="SCM error"):
         dvc.pull(imp_stage.addressing)
+
+
+def test_loads_single_file(tmp_dir, dvc, local_remote, mocker):
+    tmp_dir.dvc_gen("foo", "foo")
+    tmp_dir.dvc_gen("bar", "bar")
+
+    foo_dvcfile = SingleStageFile(dvc, "foo.dvc")
+    bar_dvcfile = SingleStageFile(dvc, "bar.dvc")
+
+    spy = mocker.spy(FileMixin, "_load")
+    assert dvc.push("foo.dvc") == 1
+    spy.assert_called_with(foo_dvcfile)
+    spy.reset_mock()
+
+    assert dvc.push("bar.dvc") == 1
+    spy.assert_called_with(bar_dvcfile)
+    spy.reset_mock()
+
+    dvc.cache.local.clear()
+    (tmp_dir / "bar").unlink()
+    (tmp_dir / "foo").unlink()
+
+    assert dvc.pull("foo.dvc") == {
+        "added": ["foo"],
+        "deleted": [],
+        "fetched": 1,
+        "modified": [],
+    }
+    spy.assert_called_with(foo_dvcfile)
+    assert (tmp_dir / "foo").exists()
+    assert not (tmp_dir / "bar").exists()
+    spy.reset_mock()
+
+    assert dvc.fetch("bar.dvc") == 1
+    spy.assert_called_with(bar_dvcfile)

tests/func/test_repo_index.py

@@ -4,7 +4,8 @@ from itertools import chain
 import pytest
 from pygtrie import Trie
 
-from dvc.repo.index import Index
+from dvc.exceptions import NoOutputOrStageError
+from dvc.repo.index import Index, index_from_targets
 from dvc.stage import Stage
 
 
@@ -402,3 +403,31 @@ def test_data_index(tmp_dir, dvc, local_cloud, erepo_dir):
 
     assert data.storage_map[("ifoo_partial",)].remote.read_only
     assert data.storage_map[("idir_partial",)].remote.read_only
+
+
+def test_index_from_targets(tmp_dir, dvc):
+    stage1 = dvc.stage.add(name="stage1", cmd="echo hello")
+    stage2 = dvc.stage.add(name="stage2", cmd="echo hello world")
+
+    (foo_stage,) = tmp_dir.dvc_gen("foo", "foo")
+
+    index = index_from_targets(dvc, ["stage1"])
+    assert index.stages == [stage1]
+
+    index = index_from_targets(dvc, ["stage2"])
+    assert index.stages == [stage2]
+
+    index = index_from_targets(dvc, ["dvc.yaml"])
+    assert set(index.stages) == {stage1, stage2}
+
+    index = index_from_targets(dvc, ["dvc.yaml:stage2"])
+    assert index.stages == [stage2]
+
+    index = index_from_targets(dvc, ["foo.dvc"])
+    assert index.stages == [foo_stage]
+
+    index = index_from_targets(dvc, ["stage1", "foo.dvc"])
+    assert set(index.stages) == {foo_stage, stage1}
+
+    with pytest.raises(NoOutputOrStageError):
+        index = index_from_targets(dvc, ["not-existing-stage"])