Images

[Scoutink]

Great work!
Certainly the best I came across and I've been searching...
Can't you make it also extract images from the pdfs? I gave it a complex pdf with text tables and images and it was impressive beside missing on extracting the images.

[NastyBoget]

Thank you for your feedback!
Extracting images from PDFs should work with parameter with_attachments="true".
You can get more details in the documentation:

1. If you use dedoc API, parameters description and return format description may help.
2. If you use dedoc library, this page can be useful.

Feel free to ask any questions if something doesn't work properly.

[Scoutink]

@NastyBoget Thank you for the feedback. For some reason it is not working for me. Not sure if this is appropriate to ask here, but are you open to help me using this in a project I am working on (for a fee we can discuss, of course). I may even point you to another repo with an interesting approach to this same field that may improve the results.

[NastyBoget]

If you have an example PDF file, I can try to process it myself. Maybe we have a bug on some specific cases.

[Scoutink]

[sample.pdf]
here is a sample I downloaded from the web. I am testing here, btw: https://dedoc-readme.hf.space/

[NastyBoget]

Did you try to set return_format into json? In this case, images are enlisted in attachments field (tried it myself). Currently, we don't show attachments in the HTML representation (it will be added in the next version of dedoc). We use HTML representation for better understanding of dedoc's parsing results, but in practice (in our projects) we use json representation.

By the way, I would recommend to work with attached images using dedoc library (not API), or use return_base64 parameter for storing images content in metadata.

[Scoutink]

That makes sense, and very helpful. Thanks.