DocumentSplitter type error

[ImVexed]

https://github.com/snexus/llm-search/blob/main/src/llmsearch/parsers/splitter.py#L53 and many other places in the file seem to access extension like it's a string, but it's an object. Causing the splitter to find no files. This was causing the examples to not work for me.

[snexus]

Thanks for reporting. Can you elaborate please on the issue and your setup? It seems the conversion happens implicitly on my system, it also works on Google Colab which is a similar setup (Ubuntu 22.04)

[ImVexed]

I'm on windows 11, When logging logger.info(f"Scanning path for extension: {extension}") for example I would see Scanning path for extension: DocumentExtension.pdf and I assume it also caused list(docs_path.glob(f"**/*.{extension.value}")) to be list(docs_path.glob(f"**/*.DocumentExtension.pdf"))

Considering that DocumentExtension is an Enum, It seems to have a key and value when used as a explicit type.

[ImVexed]

And for me, Python 3.11's enum.__str__ is:

    def __str__(self):
        return "%s.%s" % (self.__class__.__name__, self._name_, )

[snexus]

Thanks for that! I understand now, looks the problem is in different behaviour of Enum on Python 3.10 (which was used to develop this package) vs 3.11. In 3.10 has an implicit conversion to string, which doesn’t happen in 3.11.

I will work on compatibility with 3.11. If you have an access to virtualenv with 3.10, it should work hopefully.

[snexus]

Fixed in version 0.3.2