page extract with pdf4llm.to_markdown not extracting the first line of the page in specific scenarios

[leelaraj72]

I am using following simple python code to extract a page (and the corresponding text) from a pdf file

import pdf4llm
import pprint

data = pdf4llm.to_markdown("../docs/abc.pdf", pages=[13], page_chunks=True)
pprint.pprint(data[0])

The first line is not extracted always if the first line is a plain text. For Ex: if the page has following text

promotional modifiers where a service item gets added automatically when a certain
finished good is added to the order. This helps in cutting down order creation time, adds
efficiency and accuracy of order creator, and enables companies to implement service
...

then data[0]['text'] has

'finished good is added to the order. This helps in cutting down order '
'creation time, adds\n'
'efficiency and accuracy of order creator, and enables companies to implement '
'service\n'
...

Did anyone else face the same problem?

[JorjMcKie]

Try setting the margin parameter. Default is margins=(0, 50, 0, 50) which ignores stripes of height 50 at top and bottom.
Using margins=0 looks at the full page.

[Fianax]

With this code snippet it does return all the lines of code from the example pdf.

Although with the code that you have passed me it works for what I wanted (thanks for your code 💯 ) but the hdr_info would still not give all the text in the ‘span’ value of the method that is passed to it.

[Fianax]

I have also tried to use the flag=TEXTFLAGS_TEXT but I have not been able to import it from anywhere(fitz.).

What numerical value does it have to put it?

[JorjMcKie]

pymupdf.TEXTFLAGS_TEXT of course!!!

[JorjMcKie]

The check in hdr_info will only called for the first span of a line - not for every span.

[Fianax]

oki, thanks for everything