feat: parse .epub files by reading the XHTML inside (closes #21)

[qiaobochi040726-source]

Closes #21.

An .epub is just a zip of XHTML, so parsers.parse now pulls the html/xhtml entries out with zipfile and reuses the existing tag stripping, instead of falling through to the binary best-effort read. Also let the importer accept .epub, otherwise import_folder skips it. A broken or non-zip file returns "" like the PDF path does.

Tried it: tests/test_core.py builds a tiny epub in tmp_path and runs it through import + ingest, checking the text lands in the source page. Didn't bother with spine/OPF ordering — sorted filenames are good enough for a flat read of the chapters.

Drafted with an AI coding assistant and gone over before sending, as CONTRIBUTING asks.

[francescobianco]

Thank you @qiaobochi040726-source for this contribution! 🎉

It's a clean, focused change: stdlib zipfile, reusing the existing tag stripping, failing gracefully like the PDF path, and an end-to-end test from import through ingest. We also appreciate the honest note on AI assistance and the trade-off on spine ordering. Sorted filenames are a reasonable first step, and OPF ordering can come later if anyone needs it.

You're warmly invited to become an official Openapi contributor: https://github.com/openapi/contributors 🙌