Modernize tests/utils style and dedupe test_git_url.py

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Author: J03D03

tests/utils/test_datasets_utils.py

@@ -79,17 +79,8 @@ def test_resolves_tag_to_commit_sha(self) -> None:
     def test_nonexistent_ref_returns_none(self) -> None:
         """Non-existent refs should return None without raising."""
         result = _resolve_ref_local("this-tag-does-not-exist-12345", "https://github.com/curl/curl")
-        # Either None (repo exists, ref not found) or None (repo not cloned)
         assert result is None
 
-    @pytest.mark.integration
-    def test_invalid_repo_directory_returns_none(self, tmp_path: pytest.TempPathFactory) -> None:
-        """Return None when the repo path exists but is not a valid git repo."""
-        # _resolve_ref_local checks url_to_pathname, so we can't easily point it
-        # at tmp_path. Instead, verify it handles InvalidGitRepositoryError gracefully
-        # by testing with a URL whose local path doesn't contain a valid repo.
-        assert _resolve_ref_local("main", "https://github.com/nonexistent/repo-xyz-999") is None
-
 
 class TestResolveSymbolicRef:
     """Tests for _resolve_symbolic_ref function."""
@@ -105,14 +96,9 @@ def test_resolves_curl_tag_to_commit_sha(self) -> None:
         This tests both the API call logic AND the SHA normalization (lowercase).
         """
         sha = _resolve_symbolic_ref("curl-7_50_0", "https://github.com/curl/curl")
-
-        # Known stable commit for curl-7_50_0 tag
-        expected_sha = "79e63a53bb9598af863b0afe49ad662795faeef4"
-
-        if sha:
-            assert sha == expected_sha, f"Expected {expected_sha}, got {sha}"
-        else:
+        if sha is None:
             pytest.skip("Could not resolve (no local repo and API unavailable)")
+        assert sha == "79e63a53bb9598af863b0afe49ad662795faeef4"
 
     @pytest.mark.integration
     def test_nonexistent_ref_returns_none(self) -> None:

tests/utils/test_extensions.py

@@ -4,20 +4,20 @@
 
 
 class TestExtensionsFor:
-    def test_single_language(self):
+    def test_single_language(self) -> None:
         exts = extensions_for("python")
         assert ".py" in exts
         assert ".pyw" in exts
 
-    def test_multiple_languages(self):
+    def test_multiple_languages(self) -> None:
         exts = extensions_for("c", "cpp")
         assert ".c" in exts
         assert ".cpp" in exts
         assert ".h" in exts
 
-    def test_unknown_language_returns_empty(self):
+    def test_unknown_language_returns_empty(self) -> None:
         assert extensions_for("fortran") == set()
 
-    def test_all_extensions_map_to_known_language(self):
+    def test_all_extensions_map_to_known_language(self) -> None:
         for ext, lang in EXTENSION_TO_LANGUAGE.items():
             assert ext in extensions_for(lang)

tests/utils/test_git_repository_strategy.py

@@ -1,7 +1,5 @@
 """Tests for CloneStrategy helpers in utils.git.repository."""
 
-from __future__ import annotations
-
 import shutil
 from pathlib import Path
 
@@ -43,28 +41,28 @@ def _full_clone(bare: Path, dest: Path) -> Repo:
 
 @pytest.mark.integration
 class TestIsPartialClone:
-    def test_blobless_clone_detected(self, tmp_path):
+    def test_blobless_clone_detected(self, tmp_path: Path) -> None:
         bare = _init_bare_source(tmp_path)
         repo = _blobless_clone(bare, tmp_path / "clone")
         assert _is_partial_clone(repo) is True
 
-    def test_full_clone_not_partial(self, tmp_path):
+    def test_full_clone_not_partial(self, tmp_path: Path) -> None:
         bare = _init_bare_source(tmp_path)
         repo = _full_clone(bare, tmp_path / "clone")
         assert _is_partial_clone(repo) is False
 
 
 @pytest.mark.integration
 class TestUpgradeToFull:
-    def test_unsets_filter_config(self, tmp_path):
+    def test_unsets_filter_config(self, tmp_path: Path) -> None:
         bare = _init_bare_source(tmp_path)
         repo = _blobless_clone(bare, tmp_path / "clone")
         assert _is_partial_clone(repo)
 
         assert _upgrade_to_full(repo, timeout=60) is True
         assert not _is_partial_clone(repo)
 
-    def test_returns_false_when_remote_gone(self, tmp_path):
+    def test_returns_false_when_remote_gone(self, tmp_path: Path) -> None:
         """If fetch fails, caller needs False so it can recover."""
         bare = _init_bare_source(tmp_path)
         repo = _blobless_clone(bare, tmp_path / "clone")