text flow split in OCR-ed documents

[myrmoteras]

Why does GGI break OCRed text into chunks?

Fosberg and Bullock 1971 plantsOCR.pdf

[gsautter]

This is simply down to the distance between the words ... in the absence of a template, page structuring has to rely on a bunch of defaults, and those are optimized for printed text, not typewriter ... on top of that, the large line distance also means that lines don't cover each other's gaps in all too many places, and the few places where they do seem to work as intended ...

Meaning to say this simply is a very atypical document that defies the usual catches and defaults ... "Merge Blocks" should be an easy fix, though ... if we run into more typewriter documents, we might have to consider a dedicated set of defaults and somehow tell page structure detection to use that instead of the usual print-oriented defaults.