bugfix json/pure mixing escaped with literal unicode raises Encoding::CompatibilityError #483
Conversation
|
this would fix #232 (opened 7 years ago) |
|
@notEthan haha looks like you beat me to it: #484 Did you run into this in https://github.com/json-schema-org/JSON-Schema-Test-Suite/ as well? At least we came up with the same fix :) |
|
remarkable, the very same. my implementation is here: https://github.com/notEthan/jsi . I've just merged (but not yet released) support for validating on its own. it would be cool to have a chat someplace if you'd be interested. |
That sounds good to me. Feel free to send me an email (davishmcclurg@gmail.com) and we can figure something out. |
|
I also ran into this bug and came up with the same fix as you two. It would be great to get this merged 🤞 |
|
hey @hsbt! sorry for the noise but if you get a chance could this PR be considered? |
|
@notEthan Can you rebase this from the master branch? I'm not sure why this pr didn't invoke our CI. |
…rcing encoding back to ASCII_8BIT before gsubing back into parse_string result
notEthan
force-pushed
the
unicode_mixed
branch
from
9abafed
to
766f816
Compare
|
rebased. the CI doesn't run because I am a first-time contributor. I see the message:
|
|
@notEthan Thanks, I approved it now. |
Truffleruby's version of json/pure has a bug that's triggered by these files. It's been fixed but the fix isn't in Truffleruby yet: - flori/json#483 - flori/json#484 This skips the files when the error is raised and drops them before checking fixtures.
Truffleruby's version of json/pure has a bug that's triggered by these files. It's been fixed but the fix isn't in Truffleruby yet: - flori/json#483 - flori/json#484 This skips the files when the error is raised and drops them before checking fixtures.
The [fix](flori/json#483) must've made it fully into Truffleruby now, because this isn't failing anymore.
mixing \u-escaped unicode with literal unicode characters, like so:
results in this error:
which comes from here:
I think that forcing the encoding of the gsub result back to ASCII_8BIT is a good fix for this since the StringScanner string is forced to ASCII_8BIT. I can't be entirely certain though; the handling of encoding in this parser is a bit beyond me.