Having a hard time parsing documents from Curitiba's town hall

Hi

I have started a little personal project to collect, monitor and consolidate governmental data (specially for the town where I live - Curitiba), and some interesting numbers are available but in PDF format.

I have tried many different parameters in excalibur, but I am having a hard time making it break the rows correctly.
Since there was a link on the WebUI to report extraction problems, here I am :)
The final implementation will actually use camelot directly from python, but first I need to figure out if and how to parse those documents.

This is the document I was actually trying to parse:
https://mid.curitiba.pr.gov.br/contaspublicas/2020/01/An7_RP_1B20.pdf

But that's not the only format. I have also found one that has tables across multiple pages:
https://mid.curitiba.pr.gov.br/contaspublicas/2020/02/An2_Fun_RREO_2B20.pdf

Could you please give me some guidance on what to try to get these documents parsed correctly or how I can help debugging what is causing the parsing to fail?

Thank you very much

Start by reproducing the extraction attempts against the linked Curitiba PDFs through Excalibur's WebUI, focusing on row boundaries and the multi-page table example. Compare the results with direct Camelot use in Python and document the parameters or parsing limitation that explains the failure. Done means providing actionable guidance or a reproducible diagnosis for these document formats.

Written by the indexing model from the issue text.