Find answers from the community

Updated 6 months ago

Hello all,

At a glance

Hello all,
I’m looking to expand past XML to PDFs, and the one big issue is the one issue everyone has—tables. Is there a recommended OSS way to read them? Specifically something you’d recommend be used with LlamaIndex?

15 comments

LLogan M

probably unstructured will be the best OSS solution

LLogan M

but overall tables are hard

LLogan M

marked is another OSS library that does ok-ish

iisaackogan

Is OCR an acceptable solution

LLogan M

OCR is really only half of the solution

iisaackogan

in what sense

LLogan M

Sure you can get the text -- but then you need to make sure its formatted nicely

iisaackogan

Oh of course yeah

iisaackogan

Oh crap

iisaackogan

And then there’s the issue of hyperlinks

iisaackogan

God I hate PDFs

LLogan M

it really is the worst file format possible lol

LLogan M

and the most used

sshawtyisaten

@isaackogan mind sharing the pdf file you're trying to read?

iisaackogan

no sorry I’m testing with my employee pay statement 💀

Add a reply