fix(epub): parse manifest attributes in any order

[ojspace]

Problem

EPUB conversion returned epub-failed for valid EPUBs (e.g., Project Gutenberg books) even when unzip was available.

Root Cause

The manifest item regex required id= to appear before href= in <item> tags:

/<item\s+[^>]*id="([^"]+)"[^>]*href="([^"]+)"[^>]*/gi

Project Gutenberg EPUBs (and many others) use href first:

<item href="chapter1.html" id="item1" media-type="application/xhtml+xml"/>

This caused zero manifest items to be found, zero spine items, and epub-failed for every valid Gutenberg EPUB.

Fix

Match the full <item> element and extract id and href separately, independent of attribute order.

Test

• Alice's Adventures in Wonderland (Project Gutenberg EPUB) → extraction: epub-native, usefulness_score: 1.00, 179 chunks ✅

🤖 Generated with Claude Code

Summary by CodeRabbit

• Bug Fixes
  • Fixed EPUB manifest parsing to correctly handle variations in attribute ordering, enabling previously unrecognized e-books to be properly read.

[coderabbitai]

Caution

Review failed

The pull request is closed.

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 66957d42-1987-484e-a535-bcab214ccc99

📥 Commits

Reviewing files that changed from the base of the PR and between 8f34a24 and 5fc698c.

📒 Files selected for processing (1)

• src/providers/epub.ts

---

📝 Walkthrough

Walkthrough

The EPUB OPF manifest parsing logic was refactored to robustly extract <item> elements by matching the entire tag and separately extracting id and href attributes via case-insensitive regexes, accommodating any attribute ordering. The population of manifestItems is now conditional on both attributes being present.

Changes

Cohort / File(s)	Summary
EPUB Manifest Parsing src/providers/epub.ts	Refactored <item> element extraction from single sequential regex to separate case-insensitive regexes for id and href, enabling attribute order independence and improved robustness. Conditional population based on presence of both attributes.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

Poem

🐰 With whiskers twitched and nose held high,
The EPUB's items now fly,
No order matters, left or right,
Our parsing's now more sturdy and bright! ✨

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix/epub-manifest-attr-order

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

Tip

You can disable sequence diagrams in the walkthrough.

Disable the reviews.sequence_diagrams setting to disable sequence diagrams in the walkthrough.