2Shot0ccurence
commented
5 months ago 2
The software could be improved in a major way, in terms of recognition. Right now it just extracts everything it can see as plain text.
- Hyperlinks are not recognized.
- mathematical equations are not being written to Latex math notation, rather, they are kept as plain text.
- It tries to mitigate this, by converting these elements to .jpg, but often clips the equations in the middle.
- It doesn't recognize chapters or tableofcontents but I think if it recognized all hyperlinks it wouldn't really need to recognize them seperately anyways.
1
Thank you @vinaykanigicherla,
Unlike the listed packages, dependency requirements should be characterized in a different and more detailed way. For example, the tesseract package is required and requested for OCR and the language packs supported by this package must be installed separately. In addition, when the location of the image files in the localstore is transferred to the tex file, it uses the --filepath extension for example,
/mnt/t/test/playground/localstore/x1assets/logo.JPGas well as if localstore is next to .tex file and exported with the./tag, fewer compile errors may be encountered. It is not a very essential problem, but it should be considered.I could not see Unicode support for the exported tex code, especially the commonly used symbols such as ° and ± were not supported.