How to get xref associated with a text span

[SirishaGorasa]

How do I get the xref associated with a text span?

I have seen in previous discussions, as we can't extract headings directly using pymupdf. Wanna know if there are any improvements!

[JorjMcKie]

Text components are not associated with xrefs.
Headers, footers, etc. are no identifyable information in PDF - has nothing to do with PyMuPDF. Neither are text blocks or spans.
The base library MuPDF does its bests to use a bunch of heuristics to identify blocks, lines and spans - but there is no materal foundation in the document for such interpretation.
You may for example encounter PDFs where all the text characters are coded in an arbitrary sequence - obviously with up to n! different possibilities of extraction sequences if n is the number of characters on a page. I have created such an awkward example myself: https://github.com/pymupdf/PyMuPDF-Utilities/blob/master/text-extraction/textmaker2.pdf.
Similarly it is very possible to extract a header line as the last text piece from a page.
So you need to use text positioning coordinates to establish a reading sequence you like. This can also not be generalized - think of non-Western reading sequences (right to left: Arabian, Hebrew) or top-down-left-right for some ancient Chinese or Japanese documents.

But there is help: use python -m fitz gettext textmaker2.pdf to extract and re-layout all the single characters in the intended reading sequence.

[JorjMcKie]

Is there a way to get bbox or position info of a xref? I use doc. xref_object(xref) only get a string as follow:

<<\n  /A 80 0 R\n  /Next 76 0 R\n  /Parent 40 0 R\n  /Title (2 | Flexible Scaling for Analytic Workloads)\n>>

This string maybe includes text and parent xref etc. But I don't know the introduction of each character. And this maybe don't include bbox info.

This xref looks like a PDF bookmark. This is not text or whatever, so does not have a bbox. It does not appear anywhere on the page.

[buptyyf]

doc.get_toc() sometimes only returns xref info and don't have point info. I just want to know where the toc point to, but I can't. Are there some alternative ways let me know that?

[JorjMcKie]

doc.get_toc() sometimes only returns xref info and don't have point info. I just want to know where the toc point to, but I can't. Are there some alternative ways let me know that?

This happens if the creator did not specify a target point. Did you confirm that that bookmark points to some place inside the same PDF - check the "kind" value. Also if the page is < 1 means that the target is not within this PDF.

[buptyyf]

Example PDF:
CB_7YGBwmBxZEhw6eK6cN4Ot6eb.pdf

doc.get_toc(False) get as follows:

[1, '上篇', 7, {'kind': 4, 'xref': 1033, 'name': 'page=7&zoom=nan,90,117', 'collapse': False, 'zoom': 0.0}]
[2, '第一章 实际GDP是假的', 7, {'kind': 4, 'xref': 1101, 'name': 'page=7&zoom=nan,268,171', 'collapse': False, 'zoom': 0.0}]
...

It's a named location toc. But I don't it's actual position.

[JorjMcKie]

You are not using the current version. There was an error before that did not recognize "&zoom=nan" as being equivalent to "&zoom=0".
Upgrading takes care of this, so that these links are normal internal GOTO links.