EPUB Reader¶

For .epub files Octa renders the book chapter-by-chapter as flowing text, using the same Markdown renderer the Markdown view uses. Each chapter's XHTML is converted to Markdown once at load time and cached on the tab, so there's no per-frame conversion overhead.

When the EPUB Reader appears¶

The EPUB Reader is the default view for .epub files. The Table view is still available: every chapter's paragraphs become rows with chapter, paragraph, text columns, which is useful for full-text searching the book via the search bar or SQL.

Switch with View → Table (or the F4 cycle shortcut).

The top toolbar shows:

Book title (from the EPUB's <dc:title> metadata).
◀ Previous / Next ▶ buttons step through chapters.
Chapter combo is the full chapter list with N. <label> (i/total). Pick any chapter to jump straight to it.

The chapter label is the filename of the chapter's XHTML file (EPUBs don't carry guaranteed chapter titles, so this is the most reliable identifier without parsing TOC structures).

Chapter rendering¶

Each chapter's body renders via pulldown_cmark, the same pipeline the Markdown view uses. That means:

Bold / italic / strong / em runs render with weight differences.
Headings (h1-h6) render at varying sizes.
Lists, blockquotes, code blocks all supported.
Inline links are clickable where they make sense (internal chapter links don't yet navigate, but external https://... links open in the default browser).

The reading width is capped at clamp(200.0, 900.0) pixels, matching the Markdown view, so on wide displays the text doesn't sprawl.

Embedded images¶

Images embedded in the EPUB (covers, figures, illustrations) are decoded once and shown as a thumbnail strip beneath the chapter text. Each thumbnail is capped at 200 px on its longest axis. The strip persists across chapter switches, since Octa caches decoded images on the tab so you don't re-decode on every page flip.

Why a thumbnail strip, not inline

EPUB chapters use HTML <img> tags to position images mid-text. The egui pulldown_cmark walker doesn't expose a clean way to weave widgets into paragraph layout, so v1 of the EPUB Reader collects all chapter images into a strip below the body. This is a known v1 trade-off; inline-in-paragraph positioning is deferred.

Image resolution falls back via three strategies:

Exact href match against the EPUB's manifest.
/-prefixed match (some EPUBs normalise paths with a leading /).
Filename-only match, where images/cover.jpg and OEBPS/images/cover.jpg share cover.jpg. First match wins on duplicates.

Image formats supported¶

PNG, JPEG, GIF, and WebP, which cover the common EPUB image set. Anything else (TIFF, BMP, SVG) renders as alt-text only.

Table view fallback¶

Switch to View → Table to inspect the book's text as a flat table. Three columns:

Column	Type	Notes
`chapter`	Int64	1-based chapter index
`paragraph`	Int64	1-based paragraph index within the chapter
`text`	Utf8	Paragraph text in Markdown form

This is the lens to use for:

Searching the book: type into the toolbar search box to filter to matching paragraphs.
Running SQL queries: SELECT * FROM data WHERE text LIKE '%love%' in the SQL panel.
Counting words per chapter, finding the longest paragraph, etc.

Limitations¶

Read-only. No edit-and-save-back. EPUB writing isn't on the roadmap.
No TOC navigation. The combo shows chapter filenames, not the book's declared TOC structure. Following an internal #anchor link doesn't navigate yet.
No font preservation. EPUBs that bundle custom fonts (sci-fi novels with unusual typography, etc.) render in Octa's default font.
No DRM. Octa uses rbook which doesn't support encrypted EPUBs, so Kindle / Adobe Digital Editions files won't open.