Page getText() returns blank contents

[gerardnll]

• PHP Version: 8.4.14
• PDFParser Version: 2.12.3

Description:

Parsing in version 2.12.2 works fine, updating to the last one returns empty content when doing ->getText() to a page. I haven't touched the implementation.

PDF input

I can provide the PDF privately.

Expected output & actual output

Code

[k00ni]

I hope its not a regression. It is most likely due to the changes of #783. @rupertj do have an idea?

@gerardnll Could you go into the vendor folder in your local installation and in smalot/pdfparser change the file src/Smalot/PdfParser/PDFObject.php by reverting this line and try if the problem persists?

[gerardnll]

I've tried reverting the line and that fixes the problem.

[rupertj]

@gerardnll If you could provide me with the PDF you've found the issue with and which text is missing from the extract, I'll see if I can fix this a different way.

[rupertj]

@gerardnll I've got it - thanks!

[MrMitch]

@rupertj Hi ! I'm in a similar situation: part of the text in the PDF is not included in the returned value of Document::getText().

I've created a private reproducer repository that I can open to you. I can also provide a sample PDF privately.

Best regards

[rupertj]

Hi @MrMitch - Do you get the text you're expecting if you use v2.12.2 of this library? (Or revert the line that kooni advised above?)

[MrMitch]

@rupertj yes, absolutely.

v2.12.2 works.

Reverting the line mentionned above in v2.12.3 works as well.

[rupertj]

Thanks. Could you attach the PDF to the issue? Or email it to rupert@wearechicken.co.uk if you'd rather do that. If the missing text isn't obvious, please let me know what it is. Thanks!

[MrMitch]

@rupertj PDF sent. I've added you as a collaborator on the minimal reproducer repository since i already built-it beforehand. There's a full description in the README, including the missing text. The PDF is in the repository as well.

[rupertj]

Ive had a look at this now, and I think the correct approach is when assembling the text for the Text Array, check if any Forms return meaningful text and include it when they do. IE, when a Form's getText() method returns " ", don't include that.

This approach returns the text that was previously missing from @gerardnll and @MrMitch's PDFs, and also lets the tests that I have that need accurate text positioning pass.

Next step is to do a thorough check on the extracted text from 2.12.2 and after the change to check it doesn't have any other effects.

[MrMitch]

@rupertj Thanks for the feedback, feel free to reach out if necessary