ePubLift/epublift cli-v1.7.0

on GitHub

Added

• [EXPERIMENTAL] PDF → EPUB import (epublift import) — turns a PDF into a
  reflowable EPUB. Handles PDFs that carry a text layer: born-digital books and
  already-searchable scans (archive.org / Google Books), including modern
  Type0/CID fonts decoded via the font's own /ToUnicode CMap. Accurate
  word spacing from real glyph widths (/Widths, /W) — a born-digital test
  book matched the publisher's own EPUB at ~99.8% word overlap. Structure
  recovery strips running heads/footers + page numbers, de-hyphenates line
  breaks, reassembles paragraphs, and detects chapter headings → spine + ToC.
  Scanned PDFs with no text layer are detected and report that OCR is coming in a
  later release (no broken output). Pure-Rust, behind the opt-in pdf build
  feature (the pre-built binary includes it). See docs/pdf-import.md.
• Reader compatibility doc (docs/device-compatibility.md) — which readers
  render which image formats, with an Our recommendation column mapping each
  reader to the right flags. A Verified by us table (first-hand: Kobo Forma/Sage
  WebP blank; Apple Books WebP OK) is kept separate from a Reported table
  (vendor-documented support for Kindle, PocketBook, Onyx Boox, and RMSDK-based
  readers — WebP generally not listed; AVIF/JXL nowhere). Plus the durable
  "reading appliance vs Android e-ink slate" split that explains why modern
  formats lag. Linked from the README.

Full Changelog: cli-v1.6.1...cli-v1.7.0