qurator-spk/page2tsv
1
0
Fork 0
mirror of https://github.com/qurator-spk/page2tsv.git synced 2025-06-16 14:59:54 +02:00
Code Issues Releases Wiki Activity

support Qurator calamari PAGE xml

Browse source
This commit is contained in:
Kai Labusch 2020-03-11 12:56:09 +01:00
parent abdabbac4f
commit 05f49df6d2
1 changed files with 9 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

12
cli.py
View file

@ -175,10 +175,16 @@ def page2tsv(page_xml_file, tsv_out_file, image_url, ner_rest_endpoint, noproxy,
rgn_number += 1 rgn_number += 1
for text_line in region.findall('.//{%s}TextLine' % xmlns): for text_line in region.findall('.//{%s}TextLine' % xmlns):
line_number += 1 line_number += 1
for words in text_line.findall('.//{%s}Word' % xmlns):
for word in words.findall('.//{%s}Unicode' % xmlns): # import ipdb; ipdb.set_trace()
for words in text_line.findall('./{%s}Word' % xmlns):
# import ipdb;ipdb.set_trace()
for word in words.findall('./{%s}TextEquiv/{%s}Unicode' % (xmlns, xmlns)):
text = word.text text = word.text
for coords in words.findall('.//{%s}Coords' % xmlns): for coords in words.findall('./{%s}Coords' % xmlns):
# transform OCR coordinates using `scale_factor` to derive # transform OCR coordinates using `scale_factor` to derive
# correct coordinates for the web presentation image # correct coordinates for the web presentation image