schemafile was not valid: is not a regex

[misery]

Hi there,

since 0.31.0 our regex isn't accepted anymore. I don't see a problem in the regex.

Error: schemafile was not valid

SchemaError: "^[\\säÄöÖüÜßẞ\\w\\_\\'\\/\\-\\+\\*\\&\\<\\>\\(\\)\\,\\:\\.\\?\\!]+$" is not a 'regex'

Failed validating 'format' in metaschema['properties']['definitions']['additionalProperties']['$dynamicRef']['allOf'][1]['properties']['properties']['additionalProperties']['$dynamicRef']['allOf'][3]['properties']['pattern']:
    {'type': 'string', 'format': 'regex'}

On schema['definitions']['languageString']['properties']['']['pattern']:
    "^[\\säÄöÖüÜßẞ\\w\\_\\'\\/\\-\\+\\*\\&\\<\\>\\(\\)\\,\\:\\.\\?\\!]+$"
  in "/tmp/bla/lib/python3.13/site-packages/check_jsonschema/checker.py", line 56
  >>> return self._schema_loader.get_validator(

0.30.0 works without problems.

[edgarrmondragon]

@misery have you tried 0.31.1?

[sirosen]

Under v0.31.0 the default regex behavior changed from ECMAScript regexes (regress engine) to ECMAScript in Unicode Mode. (https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/RegExp/unicode#description)

Unicode Mode regexes are stricter about invalid escape sequences.
Net-net, I think this is generally a good thing, as it helps you write more correct code.
But it does mean that in cases like this one, where there are invalid escapes, those escapes will cause the regex to fail validation.

You can either remove the extra escapes or run with --regex-variant nonunicode to get the old behavior.

I walked through each escape in your expression and the following are not valid:

• \_
• \'
• \&
• \<
• \>
• \,
• \:
• \!

Therefore, your expression can be corrected to

"^[\\säÄöÖüÜßẞ\\w_'\\/\\-\\+\\*&<>\\(\\),:\\.\\?!]+$"

[misery]

Thanks! I will go with the unicode version.