ePubLift/epublift web-v1.33.0

on GitHub

Added

• Import PDF reads scans by OCR, up to 50 pages, with the CLI's
  engine (Latin-script languages, Turkish letters included). The result says
  the book was read by OCR, and warns when its language is one OCR cannot read.
  A longer scan is refused before OCR starts, pointing to the CLI or Smart
  Import. The image carries the two OCR models (10.5 MB, checked by SHA-256 at
  build time), so the service downloads nothing and still runs with no route
  out.
• A file of the wrong type is refused in the browser before upload,
  in the page's language (a dropped file skipped the picker's filter), and the
  upload modes' errors are shown in the page's language where they are known.

Changed

• Import PDF and Import Markdown use the new engine (see cli-v5.2.0): an
  imported PDF has its cover, images are sized for readers, words broken by a
  soft hyphen are joined again, every file has a DOCTYPE and lang, the title
  and author come from the file name when the PDF has none, the book has a
  style sheet, and no "Section 1" is made up.

Run it

docker pull ghcr.io/epublift/epublift-web:1.33.0

Multi-arch (amd64 + arm64). The web service ships as a container image
only; for the command-line tool, see the latest cli-v* release.