JorjMcKie
commented
3 months ago Thank you for your interest and appreciation! No issue is irrelevant 😉.
A few words on the specifics for some of the many supported document types:
There are documents with a fixed page layout like PDF or XPS. And then there are others with a variable layout which we call "reflowable".
Examples for reflowable types are e-books (EPUB, MOBI), but also HTML, TEXT and Office documents. For these types, PyMuPDF is forced to start with standard assumptions about page size (400 x 600) because there exist no recognizable page breaks and page widths.
However, PyMuPDF is capable to change these standard assumptions after open via its method Document.layout().
We are planning to offer additional arguments for method to_markdown like page_width together with the option to assume that the whole reflowable document consists of just one page with that width. Maybe page_height=infinite with some suitable expression.
Based on this, you wouldn't ever see again those page separator "-----".
Hello, and thanks a lot for this encouraging project.
I want to bring to your attention the issue arising when importing text files. myselectdoc.txt
When applying the to_markdown function, the text is chunked into pieces with the separator "-----". Is it possible to disable chunking?
When I specify input margins=[0, 0, 0, 0], some text is not lost, but chunking still does not happen.
The issue may be irrelevant to your project goal, but it might be interesting because you allow for non-pdf files.