FEATURE: Extract Paragraphs

[judy]

We're using PDF::Reader at Zipline for parsing content out of PDFs. (I also forked this project on our team repo here.) We have a number of cases where we want to pull out all of the text from multi-column PDF layouts, but PDF::Reader's visually-aligned output via page.text was still difficult for us to programmatically parse.

I saw that borb has a paragraph extraction feature (code here), and this is my attempt to implement something similar in Ruby.

I'm leaving this in Draft until I can finish the remaining todo items below. Any feedback is appreciated.

To-do:

• Implement type checking via Sorbet
• Add tests for DisjointSet

[judy]

I deleted an earlier comment about misunderstanding a Sorbet error. All good for review now!

[yob]

Thanks for the well written submission ❤️

I've been flat out, but I'll try and take a look soon.

It's definitely the case that Page#text isn't very versatile, returning all the text in plain text with no markup.

There is Page#runs which returns a lower level view of the text of a page, including positioning data. I've flip flopped over time on how much I want to add to PDF::Reader directly, and how much I want to encourage folks to build their own code on top of Page#runs. I'll take this for a spin and see how it feels though, cheers!

[Cosmo]

@yob I hope you're well!
Any updates on this PR?

Thanks! :)

[rgarner]

@judy I was interested in this but it looks like the branch has gone from the upstream?

EDIT: it's ok, found github's special refs for this repo's PR:

git fetch origin pull/524/head:group_paragraphs