For a small project I have to parse pdf files and take a specific part of them (a simple chain of characters). I'd like to use python to do this and I've found several libraries that are capable of doing what I want in some ways.
But now after a few researches, I'm wondering what is the real structure of a pdf file, does anyone know if there is a spec or some explanations anywhere online? I've found a link on adobe but it seems that it's a dead link :(
Here is a link to Adobe's reference material
http://www.adobe.com/devnet/pdf/pdf_reference.html
You should know though that PDF is only about presentation, not structure. Parsing will not come easy.
Here's the raw reference of PDF 1.7, and here's an article describing the structure of a PDF file. If you use Vim, the pdftk plugin is a good way to explore the document in an ever-so-slightly less raw form, and the pdftk utility itself (and its GPL source) is a great way to tease documents apart.
When I first started working with PDF, I found the PDF reference very hard to navigate. It might help you to know that the overview of the file structure is found in syntax, and what Adobe call the document structure is the object structure and not the file structure. That is also found in Syntax. The description of operators is hidden away in Appendix A - very useful for understanding what is happening in content streams. If you ever have the pain of working with colour spaces you will find that hidden in Graphics! Hopefully these pointers will help you find things more quickly than I did.
If you are using windows, pdftron CosEdit allows you to browse the object structure to understand it. There is a free demo available that allows you to examine the file but not save it.
Extracting text from PDF is a hard problem because PDF has such a layout-oriented structure. You can see the docs and source code of my barely-successful attempt on CPAN (my implementation is in Perl). The PDF data structure is very cool and well designed, but it's easier to write than read.
I'm trying to do pretty much the same thing. The PDF reference is a very difficult document to read. This tutorial is a better start I think.
One way to get some clues is to create a PDF file consisting of a blank page. I have CutePDF Writer on my computer, and made a blank Wordpad document of one page. Printed to a .pdf file, and then opened the .pdf file using Notepad.
Next, use a copy of this file and eliminate lines or blocks of text that might be of interest, then reload in Acrobat Reader. You'd be surprised at how little information is needed to make a working one-page PDF document.
I'm trying to make up a spreadsheet to create a PDF form from code.