Skip to content

v1.3.0

Choose a tag to compare

@bbertucc bbertucc released this 18 Oct 21:54
7dfd134

New

  • Convert from the command line: npm run convert -- <file> turns a PDF or image into HTML without the server. It checks uploads the same way the API does (#576, #582). Thanks @salorajan.

Fixes

  • The page agent is told what each specialist covers, and the verifier is told what a merged specialist is for, so it stops deleting specialist output as invented (#575, #579).
  • Floor plans without labels go to the floor-plan agent, and spectra go to the spectroscopy agent (#580).
  • The demo's PDF button now reads "Download Newly Tagged PDF" (#573).

Tagged PDFs (iris-pdf acb8248)

  • Math characters such as 𝑥 are kept whole. Before, they came out as Korean characters (#28, fixing #581).
  • Each Formula's Alt text is read from its MathML, and the MathML is attached to the Formula as an associated file (#23, #30, #34).
  • Headings shift up together when the first heading is not h1 (#25).
  • List labels follow the list's type, start, value and reversed (#29).
  • Copied text keeps a space between table cells, headings and paragraphs (#33).
  • An optional content configuration without a name now gets one (#26).

Upgrading

  • Nothing to change. Deployments that build iris-pdf should pin acb8248.

Full changelog: v1.2.0...v1.3.0