TST: Test no-file for source

.github/workflows/label-check.yml

@@ -4,8 +4,10 @@ on:
   pull_request:
     types:
       - opened
+      - repoened
       - labeled
       - unlabeled
+      - synchronize
 
 env:
   LABELS: ${{ join( github.event.pull_request.labels.*.name, ' ' ) }}

numpydoc/tests/test_validate.py

@@ -41,6 +41,16 @@ def test_get_validation_checks_validity(checks):
         _ = validate.get_validation_checks(checks)
 
 
+class _DummyList(list):
+    """Dummy list class to test validation."""
+
+
+def test_no_file():
+    """Test that validation can be done on functions made on the fly."""
+    # Just a smoke test for now, <list> will have a None filename
+    validate.validate("numpydoc.tests.test_validate._DummyList.clear")
+
+
 @pytest.mark.parametrize(
     ["file_contents", "expected"],
     [

numpydoc/validate.py

@@ -7,9 +7,10 @@
 """
 
 from copy import deepcopy
-from typing import Dict, List, Set
+from typing import Dict, List, Set, Optional
 import ast
 import collections
+import functools
 import importlib
 import inspect
 import os
@@ -111,7 +112,15 @@
 IGNORE_COMMENT_PATTERN = re.compile("(?:.* numpydoc ignore[=|:] ?)(.+)")
 
 
-def extract_ignore_validation_comments(filepath: os.PathLike) -> Dict[int, List[str]]:
+# This function gets called once per function/method to be validated.
+# We have to balance memory usage with performance here. It shouldn't be too
+# bad to store these `dict`s (they should be rare), but to be safe let's keep
+# the limit low-ish. This was set by looking at scipy, numpy, matplotlib,
+# and pandas and they had between ~500 and ~1300 .py files as of 2023-08-16.
+@functools.lru_cache(maxsize=2000)
+def extract_ignore_validation_comments(
+    filepath: Optional[os.PathLike],
+) -> Dict[int, List[str]]:
     """
     Extract inline comments indicating certain validation checks should be ignored.
 
@@ -125,8 +134,12 @@ def extract_ignore_validation_comments(filepath: os.PathLike) -> Dict[int, List[
     dict[int, list[str]]
         Mapping of line number to a list of checks to ignore.
     """
-    with open(filepath) as file:
-        numpydoc_ignore_comments = {}
+    numpydoc_ignore_comments = {}
+    try:
+        file = open(filepath)
+    except (OSError, TypeError):  # can be None, nonexistent, or unreadable
+        return numpydoc_ignore_comments
+    with file:
         last_declaration = 1
         declarations = ["def", "class"]
         for token in tokenize.generate_tokens(file.readline):