Cannot input international chars with current version on main

[pbgc]

Hi,

I've been building and using the current version on main (for a lot of nice updates like the file manager integration and git app integration etc...) but with it I cannot input chars like 'ç' , 'á', 'ã' etc etc ....

Inputting those chars work without problems with release 0.7.1 (and always worked before).

Thanks in advance!

Pedro Costa

[jmacdonald]

Hey Pedro!

Thanks for raising this. Any chance you can help me pinpoint the specific commit that introduced this regression?

[pbgc]

Hey!

Thanks for raising this. Any chance you can help me pinpoint the specific commit that introduced this regression?

Yes, I will, as soon as I can, checkout backwards and try to identify when it was introduced.

[pbgc]

@jmacdonald it seems that it was cb3b258

[pbgc]

@jmacdonald Claude Opus 4.7 says:

Found the regression. At lines 50–51, the parser only handles ASCII printable bytes (0x20–0x7E). Any byte ≥ 0x80 falls to _ => return None, which:

1. Drops the input silently (UTF-8 multi-byte chars like 'ç' = 0xC3 0xA7, 'á' = 0xC3 0xA1, 'ã' = 0xC3 0xA3 all start ≥ 0x80).
2. Doesn't advance self.offset, so on the next call the same byte is hit again — though since the iterator returns None, the consumer typically stops polling until more data arrives.

The fix: detect a UTF-8 leading byte, wait until the full sequence is buffered, then decode it into a char.

I attach a patch file. With the change it works. At least it will give you some info about the regression and how to solve it (if you don't want to use the patch)

input_parser_utf8.patch