3.1.0 Release fails to validate strings via regex

[amisevsk]

A CI build of ours pulled the new release (3.1.0) and all validation is failing with messages like

u'go' does not match u'^[^\\s]+$'

Building with <3.1.0 does not have this issue.

Reproducer:

test.json:

{
  "test": "testString"
}

schema.json;

{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "title": "Reproducer",
  "type": "object",
  "properties": {
    "test": {
      "type": "string",
      "pattern": "^[^\\s]+$"
    }
  }
}

dockerfile (or just install 3.1.0 locally)

FROM alpine:3.10
RUN apk add --no-cache py-pip jq && pip install jsonschema

COPY . /build
RUN jsonschema --version
RUN jsonschema /build/schema.json -i /build/test.json

[Julian]

Having a look...

[amisevsk]

Thanks @Julian, from a somewhat hasty check, looks to be related to character classes ("^[^ \t]+$" does not have the issue)

[dslagle]

This appears to be related to the switch to js_regex. Specifically in relation to character classes where something like [\w0-9] gets translated to [[a-zA-Z]0-9] inside js_regex. At least that's what happened in my case.

[Julian]

OK, given that there's 2 separate issues here (#612 and #611) and that this is fairly fundamental functionality, I've temporarily pushed out v3.1.1 reverting the switchover.

We'll add some tests for the two cases here (and anything else we missed), and then there likely will be a v3.1.2 that goes back again to js-regex again.

[Zac-HD]

This appears to be related to the switch to js_regex. Specifically in relation to character classes where something like [\w0-9] gets translated to [[a-zA-Z]0-9] inside js_regex.

Yep, that's a bug in js-regex. Sorry for the trouble this has caused everyone 😕

[amisevsk]

Thanks all, FWIW I re-read the jsonschema draft docs and it does suggest not using anything outside of pretty limited range of the regex spec, so my issues are partly my own doing ;)

[tschmidtb51]

@Julian: What is the current status on that issue? Does jsonschema still use js_regex?

[Julian]

No, newer releases are not using js-regex. We still may explore JS regexes in the future, but yeah this issue shouldn't be present.

[tschmidtb51]

The problem that js_regex tried to solve is still there, correct?

Did you know that regular expressions may vary between programming languages? For example, let's consider the pattern "^abc$", which matches the string "abc". But what about the string "abc\n"? It's also matched in Python, but not in Javascript!

[Julian]

Yes, today we support only Python regular expressions.

[tschmidtb51]

Thanks for clarifying.