My current understanding of --raise-on-error is that it should act as a method for deeply "debugging" a set of document or Kraken directly. Unfortunately, the blla.segment has some raising at the line level which prevent the full document to be processed:
I proposed a first potential patch in #442 but if I were to propose a "better" patch, I would probably add a raise_on_error parameter to blla.segment and segmentation.calculate_polygonal_environment to decide between logging and raising at line 730.
I'll propose a second patch, so that you don't have to code it yourself, and choose the version you prefer.
mittagessen
commented
1 year ago
My current understanding of
--raise-on-erroris that it should act as a method for deeply "debugging" a set of document or Kraken directly. Unfortunately, theblla.segmenthas some raising at the line level which prevent the full document to be processed:https://github.com/mittagessen/kraken/blob/89adb865a419c87d3b17b75c799f5b8ad76aa6ca/kraken/lib/segmentation.py#L729-L732
I proposed a first potential patch in #442 but if I were to propose a "better" patch, I would probably add a
raise_on_errorparameter toblla.segmentandsegmentation.calculate_polygonal_environmentto decide between logging and raising at line 730.I'll propose a second patch, so that you don't have to code it yourself, and choose the version you prefer.