fix(contract): identify virtualenv roots by marker, not by name

loopx/contract.py

@@ -119,6 +119,20 @@
     r"[A-Za-z_][A-Za-z0-9_]*(?:\.[A-Za-z_][A-Za-z0-9_]*)*\s*[,)\]}]"
 )
 
+# A virtualenv is identified by its marker file rather than by its name.
+# `DEFAULT_SKIP_DIRS` can only list the names it already knows, so a project
+# that names its environment `.venv-conn`, `venv311`, or anything else would
+# have `site-packages` scanned as repository content, producing credential and
+# private-IP hits from vendored third-party source.
+VIRTUALENV_MARKER_FILE = "pyvenv.cfg"
+
+
+def is_virtualenv_root(path: Path) -> bool:
+    try:
+        return (path / VIRTUALENV_MARKER_FILE).is_file()
+    except OSError:
+        return False
+
 
 def _credential_match_is_reference(line: str, match: re.Match[str]) -> bool:
     value = line[match.start() :]
@@ -752,14 +766,18 @@ def iter_scan_files(scan_root: Path) -> list[Path]:
     root_parts = set(scan_root.parts)
     if any(part in DEFAULT_SKIP_DIRS or part.endswith(".egg-info") for part in root_parts):
         return sorted(tracked_files)
+    if is_virtualenv_root(scan_root):
+        return sorted(tracked_files)
 
     for dir_path, dir_names, file_names in os.walk(scan_root):
+        current_dir = Path(dir_path)
         dir_names[:] = [
             name
             for name in dir_names
-            if name not in DEFAULT_SKIP_DIRS and not name.endswith(".egg-info")
+            if name not in DEFAULT_SKIP_DIRS
+            and not name.endswith(".egg-info")
+            and not is_virtualenv_root(current_dir / name)
         ]
-        current_dir = Path(dir_path)
         for file_name in file_names:
             path = (current_dir / file_name).resolve()
             if path.name.endswith(".local.json"):

tests/test_contract_scan_unreadable_files.py

@@ -33,6 +33,44 @@ def test_scan_public_boundary_survives_dangling_symlink(tmp_path: Path) -> None:
     assert payload["unreadable_files"] == []
 
 
+def test_iter_scan_files_skips_virtualenv_roots_by_marker(tmp_path: Path) -> None:
+    """A virtualenv is skipped by its marker file, not by its directory name.
+
+    The skip list can only name environments it already knows. A project whose
+    environment is called `.venv-conn` or `venv311` would otherwise have
+    `site-packages` scanned as repository content, and vendored third-party
+    source would be reported as credential and private-IP hits. The payload is
+    assembled from split literals so this fixture does not trip the scan it is
+    describing.
+    """
+
+    keyword = "tok" + "en="
+    payload = f'AUTH = "{keyword}syntheticsyntheticsynthetic"\n'
+    (tmp_path / "module.py").write_text(payload, encoding="utf-8")
+
+    venv = tmp_path / ".venv-conn"
+    vendored = venv / "lib" / "site-packages" / "vendored.py"
+    vendored.parent.mkdir(parents=True)
+    (venv / "pyvenv.cfg").write_text("home = /usr/bin\n", encoding="utf-8")
+    vendored.write_text(payload, encoding="utf-8")
+
+    unmarked = tmp_path / "vendor-copy" / "lib" / "site-packages" / "vendored.py"
+    unmarked.parent.mkdir(parents=True)
+    unmarked.write_text(payload, encoding="utf-8")
+
+    scanned = iter_scan_files(tmp_path)
+
+    assert tmp_path / "module.py" in scanned
+    assert vendored not in scanned
+    # The marker is what decides: the same layout without `pyvenv.cfg` is a
+    # normal directory and is still scanned.
+    assert unmarked in scanned
+    # The same bytes outside a virtualenv are still scanned and still hit, so
+    # the marker skip cannot be passing by reading nothing at all.
+    assert scan_public_boundary([venv])["hits"] == []
+    assert scan_public_boundary([tmp_path / "module.py"])["hits"]
+
+
 def test_tracked_product_runtime_source_is_not_private_state(
     tmp_path: Path,
     monkeypatch: pytest.MonkeyPatch,