qurator-spk/page2tsv
1
0
Fork 0
mirror of https://github.com/qurator-spk/page2tsv.git synced 2025-06-09 19:39:54 +02:00
Code Issues Releases Wiki Activity

fix repeated text rows

Browse source
This commit is contained in:
Kai 2021-03-10 15:15:24 +01:00
parent a6008b83b5
commit de575037e6
1 changed files with 7 additions and 5 deletions
Download patch file Download diff file Expand all files Collapse all files

12
tsvtools/cli.py
View file

@ -125,19 +125,21 @@ def page2tsv(page_xml_file, tsv_out_file, purpose, image_url, ner_rest_endpoint,
for text_equiv in word.findall('./{%s}TextEquiv/{%s}Unicode' % (xmlns, xmlns)):
text = text_equiv.text
points = []
for coords in word.findall('./{%s}Coords' % xmlns):
# transform OCR coordinates using `scale_factor` to derive
# correct coordinates for the web presentation image
points = [int(scale_factor * float(pos))
points += [int(scale_factor * float(pos))
for p in coords.attrib['points'].split(' ') for pos in p.split(',')]
x_points, y_points = points[0::2], points[1::2]
x_points, y_points = points[0::2], points[1::2]
left, right, top, bottom = min(x_points), max(x_points), min(y_points), max(y_points)
left, right, top, bottom = min(x_points), max(x_points), min(y_points), max(y_points)
tsv.append((rgn_number, line_number, left + (right - left) / 2.0, text,
len(urls), left, right, top, bottom))
tsv.append((rgn_number, line_number, left + (right - left) / 2.0, text,
len(urls), left, right, top, bottom))
line_info = pd.DataFrame(line_info, columns=['line', 'url_id', 'left', 'right', 'top', 'bottom', 'conf'])