qurator-spk/ocrd_repair_inconsistencies
1
0
Fork 0
mirror of https://github.com/qurator-spk/ocrd_repair_inconsistencies.git synced 2025-06-08 19:29:59 +02:00
Code Issues Releases Wiki Activity

backout gracefully when text annotation is missing

Browse source
This commit is contained in:
Robert Sachunsky 2019-11-29 18:06:37 +01:00
parent 9002606e1c
commit ad8f25666b
1 changed files with 13 additions and 11 deletions
Download patch file Download diff file Expand all files Collapse all files

22
ocrd_repair_inconsistencies/ocrd_repair_inconsistencies.py
View file

@ -96,23 +96,24 @@ class RepairInconsistencies(Processor):
content=to_xml(pcgts)) content=to_xml(pcgts))
def get_text(thing, joiner=None): def get_text(thing, joiner=''):
"""Get the text of the given thing, joining if necessary""" """Get the text of the given thing, joining if necessary"""
def _get_text_for_one(t): def _get_text_for_one(one):
if len(t.get_TextEquiv()) != 1:
raise NotImplementedError
try: try:
return t.get_TextEquiv()[0].get_Unicode() return one.get_TextEquiv()[0].get_Unicode()
except Exception: except Exception:
LOG.warning('element "%s" has no text', one.id)
return None return None
if isinstance(thing, Sequence): if isinstance(thing, Sequence):
text = joiner.join(_get_text_for_one(t) for t in thing) texts = [_get_text_for_one(part) for part in thing]
if all(texts):
return joiner.join(texts)
else: else:
text = _get_text_for_one(thing) return None
return text else:
return _get_text_for_one(thing)
def _fix_segment(segment, page_id, reverse=False): def _fix_segment(segment, page_id, reverse=False):
"""Fix order of child elements of (region/line/word) segment.""" """Fix order of child elements of (region/line/word) segment."""
@ -138,7 +139,8 @@ def _fix_segment(segment, page_id, reverse=False):
return return
segment_text = get_text(segment) segment_text = get_text(segment)
concat_text = get_text(children, joiner) concat_text = get_text(children, joiner)
if (segment_text != concat_text and if (segment_text and concat_text and
segment_text != concat_text and
segment_text.replace(joiner, '') != concat_text.replace(joiner, '')): segment_text.replace(joiner, '') != concat_text.replace(joiner, '')):
def polygon_position(child, horizontal=sort_horizontal): def polygon_position(child, horizontal=sort_horizontal):
polygon = Polygon(polygon_from_points(child.get_Coords().points)) polygon = Polygon(polygon_from_points(child.get_Coords().points))