Enable PDF.js CMap font translation to properly render digital PDFs with non-native (in this case Japanese) text in the A2I review UI.
Update dependencies of the A2I UI, including some notable major version increments:
PDF.js 2->3
NodeJS 16->18
vite 2->4
typescript 4->5
The PDF.js upgrade caused some issues because PDFViewer layout now appears to size only the <canvas> correctly for each page - with the containing div.page potentially including additional space. Had to refactor the way our annotation overlay manages its sizing with a bit of an ugly hack to accommodate this. The viewer overall was also not properly detecting itself as visible, and therefore not initiating page renders, hence adding a little transparent border to force the viewer div to have some initial on-screen size.
Also (as noted in tsconfig.json) because of an open issue with vue-tsc, we've been able to upgrade to TypeScript v5 but are currently ignoring deprecation warnings and pinned to <5.5.
Testing done:
Built and deployed the UI to an existing pipeline (using Tesseract OCR) and tested in Amazon A2I in Firefox ESR. Page also tested locally in Chrome.
By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.
Issue #, if available: Fixes #30
Description of changes:
The PDF.js upgrade caused some issues because PDFViewer layout now appears to size only the
<canvas>correctly for each page - with the containingdiv.pagepotentially including additional space. Had to refactor the way our annotation overlay manages its sizing with a bit of an ugly hack to accommodate this. The viewer overall was also not properly detecting itself as visible, and therefore not initiating page renders, hence adding a little transparent border to force the viewer div to have some initial on-screen size.Also (as noted in tsconfig.json) because of an open issue with
vue-tsc, we've been able to upgrade to TypeScript v5 but are currently ignoring deprecation warnings and pinned to <5.5.Testing done:
Built and deployed the UI to an existing pipeline (using Tesseract OCR) and tested in Amazon A2I in Firefox ESR. Page also tested locally in Chrome.
By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.