Ocrd processors #9
Conversation
|
Ist das nicht verkehrt herum?
Warum steht da 800? Wie wird der Skalierungsfaktor berechnet? |
Ja, so ist es gemeint, bzw. implementiert
In dem Fall keine Skalierung, sondern IIIF sollte die Breite von PAGE-XML/Bild haben. Unter der Annahme, dass das Seitenverhältnis stimmt. |
|
The outstanding issues are fixed, added a test to verify behavior. So this can be merged AFAICT. However, we should discuss (tomorrow) whether there is a more lightweight way to add the processors to ocrd_all without pulling in all the dependencies unrelated to the functionality here. |
tsvtools/ocrd_processors.py
Outdated
| pcgts = page_from_file(self.workspace.download_file(input_file)) | ||
| page = pcgts.get_Page() | ||
|
|
||
| iiif_url = iiif_url_template\ |
There was a problem hiding this comment.
Does that work universally? If so, we should probably write an IIIF image importer for OCR-D from scratch (instead of extending and using https://github.com/karkraeg/iiimets).
There was a problem hiding this comment.
To an extent. It is geared towards the ID conventions @StaatsbibliothekBerlin ({{ PPN }}-{{ page_no }}) but except from that, it is applicable to any IIIF URL scheme.
# Conflicts: # setup.py
…o ocrd-processors # Conflicts: # setup.py
Implement
page2tsvandtsv2pageas OCR-D processors, to be included in ocrd_all and then the OCR-D Butler.All is working fine except the IIIF URL. I consistently fail to produce the right previews in the neat HTML.
There are two variants:
scale_factor == 1.0. In this case, assuming width800the IIIF looks like this:https://<server>/<prefix>/<identifier>/left,top,width,height/800,/0/default.jpgscale_factor != 1.0by comparing with the images in another fileGrp likeMAX. URL looks like this:https://<server>/<prefix>/<identifier>/left,top,width,height/full/0/default.jpg@labusch If you have any idea what I am doing wrong, I'd appreciate any hints.