Reading contents of a PDF

[santiagomed]

Is there an example on how to simply read the contents of a PDF successfully? I tried looking into read.rs but it seems to be outdated so I can't run it. Any way to read a PDF?

[s3bk]

What content do you want?
There is a lot in there.

• Content stream? You can get that from the page object.
• Text? See the pdf_render and pdf_text crates.

You can use the pdf crate in two version:

• from crates.io, then use the example that match it: https://github.com/pdf-rs/pdf/tree/a6e2abc96b23b64aa1051966bb000aabf1275d9f
• master with the latest fixes.

The pdf_render and pdf_text crates only work with the latest master.

[vjau]

What are the pdf_render and pdf_text crates ? crates.io doesn't know anything about that.

[s3bk]

They are not on crates.io because they do not meet my stability requirements for publishing there.
pdf_render … renders pdfs.
pdf_text extracts text.

[alexis779]

pdf-extract crate exists, but depends on lopdf, not pdf. This video benchmarks it against poppler, a C library.

I'd be curious to see a C/Rust comparison but with poppler against pdf_text.

[Gisbert12843]

Any chance for an easy example that just converts a PDF file to a String?

I need to search through valid utf8 text of a pdf and not panic if the pdf is formatted in any unexpected way..

Documentation found regarding this seems so scarce..

[s3bk]

If pdf_text does not do what you need, then no, there is no easy example.
This is not an easy problem. I have been working on this multiple years now and thrown many algorithms at it, and still it is not perfect.
pdf_render renders the pdf and allows you to capture the drawn strings. Thats as good as it gets.

[Gisbert12843]

Ahh thank you for clarifying that!

Unrelated to this project i was working with lopdf on that task.
Everything worked up until a pdf file does not follow regular encoding aka is corrupted or chinese xd

Sadly lopdf just panics in every case and does not error instead.
Weird behaviour from my pov.

[s3bk]

Oh sure. If everything is in standard encoding, it is easy.
And yes, I tried to not panic in the pdf crate. pdf_render might panic, but that would be a bug and needs fixing.

pdf_render is used in production with "random" PDFs. And it's not great for a server to crash from a user supplied PDF.

[acro5piano]

@santiagomed
I wanted to do the same thing as you did. Thank you for flagging this issue!

@alexis779
Thank you, pdf-extract works like a charm!