6.0.10 v1.23.0

medium risk

minitap-ai/mobile-useSep 21, 2026 → Sep 22, 20260 days between releases

What's New

Releases cover distinct components: Emscripten updates and PDF text-layer repairs

Breaking Changes

  • CLI tools now require the PDF path as the first non-option argument.
  • Built-in CMaps are compiled into the binary, affecting `cargo install` binaries.

New Features

  • New API field `cmap_gaps` to list interpolated and unmapped codes.
  • New OCR reason `invisible_text_layer` for detector classification.
  • Improved text extraction for PDFs with large BBox numerals in Form XObjects.
  • Improved right-to-left text rendering by voting for visual storage.
  • Handles re-encoded fonts with stale ToUnicode CMaps.

Improvements

  • Interpolates single-glyph gaps in ToUnicode CMaps when unambiguous.
  • Composes spacing accents with the letter they are painted over.
  • Classifies pages with invisible text layers.

Bug Fixes

  • Fixes text extraction for PDFs with large BBox numerals.
  • Fixes right-to-left text mirroring.
  • Handles re-encoded fonts with stale ToUnicode CMaps.
  • Fixes stray accents appearing as separate words.

Upgrade Notes

CLI tools (`pdf2md`, `detect-pdf`) now require the PDF path as the first non-option argument, regardless of flag order. Built-in CMaps are compiled into the binary, so `cargo install` binaries are self-contained.

The Two Releases

6.0.10

6.0.10

Sep 21, 2026· 1 month ago

Updated Emscripten to fix SDL3 status, deprecate WASM flags, and improve performance.

Full release notes
v1.23.0

1.23.0 — Text-layer repairs

Sep 22, 2026· 1 month ago

This release focuses on text-layer repairs for PDF documents, addressing issues with malformed Form XObjects, right-to-left text rendering, and ToUnicode CMap interpolation.

Full release notes