Symbols parsed as text

[pn12]

I am trying to extract text using pdfreader. Symbols for email, linkedin , phone are converted to text. Is there any way, I can restrict them to not to convert to text?

[stefan6419846]

Without seeing the actual PDF file, this is just guessing, but in general, pypdf just extracts the characters present there. You might be able to filter this by text position, but it sounds like this is not what you are looking for.

[pn12]

Hi @stefan6419846 , please see the attached file, icons for email, linkedin, phone are converted to text:
♂phone+9184208873 /linkedinwww.linkedin.com/in/priyaa-aora /envel⌢pepriyamand@gmail.com
19536802cc0f0a49 (3) Copy.pdf

[stefan6419846]

I just had a look at the file. This is because the corresponding cmaps have /Differences entries which map this accordingly, for example to avoid loosing the semantic meaning.

[pn12]

• ok, thanks for sharing the information.
• do you suggest, some fix/solution to parse these icons correctly?
• like, in all the similar pdf input data files, the icons for emails/linkedin/phone are parsed as text; which is hindering the process for extracting these entities - email, linkedin & phone.

[stefan6419846]

No, I do not have a reliable/direct solution for this without any side effects, although my experience with cmaps is rather limited. pypdf probably will not change this unless there is evidence we are violating the standard here and given that a fix will work for most of the other use cases.

Apart from this, we as pypdf are not responsible for your post-processing as perfect text extraction from PDF files is nearly impossible - PDFs are meant for printing, not for automated processing (except maybe with the PDF/UA standard). If all or most of your CVs look like this, you should easily be able to post-process this accordingly or in general account for such alternatives.

[pn12]

ok, Thank you for sharing this. I'll just try to look for an alternative solution.