Report regexes with conflicting inline flags as invalid, not an error

[yzxcj797]

Fixes #1558.

Root cause

re.compile raises ValueError — which is not an re.error subclass — when a pattern sets incompatible inline flags in separate groups ("(?u)(?a)" or "(?a)(?u)"). The regex format checker declared raises=re.error only, so the exception escaped FormatChecker.check and crashed the caller instead of reporting the string as an invalid regex:

fc.conforms("[unterminated", "regex")  # False — re.error, caught
fc.conforms("(?u)(?a)", "regex")       # ValueError: ASCII and UNICODE flags are incompatible

The single-group form ("(?ua)") fails compilation with re.error and was already reported correctly; only the two-group form escaped.

Fix

Declare raises=(re.error, ValueError) on is_regex, so the conflicting-flags ValueError becomes a FormatError with the original exception as its cause, exactly like any other invalid regex.

The related-but-distinct exception types from #1526 (OverflowError on an oversized repeat) and #1538 (RecursionError on deep nesting) are left to their own issues — this PR only widens the declaration for the ValueError family demonstrated in #1558.

Test

test_regex_conflicting_inline_flags_report_invalid_not_error in jsonschema/tests/test_format.py: both flag orders raise FormatError / conform False, the single-group form and a plain re.error case still report invalid, and a valid pattern still passes.

Differential: the new test errors with the escaped ValueError on main and passes with the change. Full suite: python -m unittest discover jsonschema.tests — 8455 tests, only a pre-existing local test_exceptions import error unrelated to this change (also present on stashed main).

[read-the-docs-community]

Documentation build overview

📚 python-jsonschema | 🛠️ Build #34176377 | 📁 Comparing f2a9514 against latest (0abde23)

  🔍 Preview build  

1 file changed

± _modules/jsonschema/_format/index.html