[Feature Request] Obsidian Document Loader

JordieB / lippy

MIT License

0 stars 0 forks source link

[Feature Request] Obsidian Document Loader #14

Open JerrickB opened 12 months ago

JerrickB commented 12 months ago

Switch to ObisidanLoader to maintain metadata. Can feed text splitter type like MarkDownHeader #13 (loader.load_and_split(text_splitter=MarkdownHeaderTextSplitter(listHeaders)))

JerrickB commented 12 months ago

Created custom Obsidian Loader to merge MarkdownHeaderTextSplitter due to it's weird implementation. Not complete. Currently returns list of lists of documents split by header.

Need to collapse list of lists and merge metadata.

JerrickB commented 12 months ago

List collapsed. Change behavior to either add one header path string to front of page content instead of adding individual metadata entries for headers. Metadata appears to be mainly useful for filtering results instead of aiding search.

JerrickB commented 12 months ago

Done.

Need to check that docs aren't being added as multiple entires