fix(tests): escape the regex in test_verify_pdf's docstring and compile-check tests/*.py (#503)

FindNonAsciiDateRangesTests' docstring (from #498) quotes the regex `\s*`
in a non-raw string. `\s` is an invalid escape sequence, so Python 3.12+
prints a SyntaxWarning for it (CPython gh-98401) when
tests/test_verify_pdf.py is compiled. It is in this repo's own CI log
for a45bbd1 on 3.12, 3.13 and 3.14, at tests/test_verify_pdf.py:75, while
the run stays green. Under `-W error::SyntaxWarning` the module fails to
import and the whole test_verify_pdf suite drops out of the run.

Escaped in the docstring. ToolsCompileWithoutWarnings (#480) only compiled
tools/*.py, which is why this got through; it now compiles tests/*.py too.
With the unescaped docstring it fails naming tests/test_verify_pdf.py, and
it passes with the fix.

Verification on Python 3.10, 3.11, 3.12, 3.13 and 3.14, from a clean
bytecode cache: `python -W error::SyntaxWarning -m unittest discover -s
tests -t .` OK (501 tests, the same count as master),
`tools/lint_skills.py`, `tools/check_framework_version.py`,
`tools/security_guards.py`.

CHANGELOG.md

@@ -105,6 +105,14 @@ per-file diff commands.
 
 ### Fixed
 
+- **The compile-warning check covers `tests/*.py` too** (`tests/test_verify_pdf.py`,
+  `tests/test_verify_layout.py`) - `FindNonAsciiDateRangesTests`' docstring quotes the regex
+  `\s*` unescaped, so Python 3.12+ prints a `SyntaxWarning` for the invalid escape `\s` when
+  the test module is compiled. It shows up in this repo's own CI log on 3.12, 3.13 and 3.14 while
+  the run stays green, and under `-W error::SyntaxWarning` the module fails to import. Escaped
+  in the docstring; `ToolsCompileWithoutWarnings` now compiles `tests/*.py` as well as
+  `tools/*.py` and fails on the unescaped version.
+
 - **`linkedin-search detail` no longer fetches an unrelated posting for a URL on another
   host** - `normalizeId` took the first 6+-digit path segment from *any* URL, so a
   Greenhouse or Lever apply link (the kind a posting's own page hands out, and what a user

tests/test_verify_layout.py

@@ -177,15 +177,19 @@ class ToolsCompileWithoutWarnings(unittest.TestCase):
     the warning lands in the middle of a verification report. Compiling is a
     pure `compile()` over the source text - no cache file, no temp file - so the
     check costs nothing and covers every `tools/*.py`, guarding the next
-    docstring that quotes a macro too.
+    docstring that quotes a macro too. It covers `tests/*.py` as well: a test
+    docstring that quotes a regex (`\\s*` in `test_verify_pdf.py`) warned in CI
+    on every matrix Python while the run stayed green.
     """
 
     def test_sources_have_no_invalid_escape_sequences(self):
         import warnings
 
         tools_dir = Path(verify_layout.__file__).resolve().parent
-        sources = sorted(tools_dir.glob("*.py"))
+        tests_dir = Path(__file__).resolve().parent
+        sources = sorted(tools_dir.glob("*.py")) + sorted(tests_dir.glob("*.py"))
         self.assertIn(Path(verify_layout.__file__).resolve(), sources)
+        self.assertIn(Path(__file__).resolve(), sources)
         offenders: dict[str, list[str]] = {}
         for source in sources:
             with warnings.catch_warnings(record=True) as caught:
@@ -197,8 +201,8 @@ class ToolsCompileWithoutWarnings(unittest.TestCase):
                 if issubclass(w.category, (SyntaxWarning, DeprecationWarning))
             ]
             if syntax:
-                offenders[source.name] = syntax
-        self.assertEqual({}, offenders, "tools/*.py warn about invalid escapes when compiled")
+                offenders[f"{source.parent.name}/{source.name}"] = syntax
+        self.assertEqual({}, offenders, "tools/*.py or tests/*.py warn about invalid escapes when compiled")
 
 
 if __name__ == "__main__":

tests/test_verify_pdf.py

@@ -72,7 +72,7 @@ class FindNonAsciiDateRangesTests(unittest.TestCase):
     The fold that makes `--contains "2016-2024"` pass on that layer (#458) is
     what makes `--contains` unable to detect it - so this check never folds.
 
-    The newline cases are yang2632's finding on the PR: `\s*` between year and
+    The newline cases are yang2632's finding on the PR: `\\s*` between year and
     dash also matched a line break, so a heading ending in an ASCII date joined
     a dash that merely opened the next line.
     """