🧹 dinglehopper: Remove warning when there is a non-TextRegion in the ReadingOrder

pull/38/head
Gerber, Mike 4 years ago
parent 0f3857d8d3
commit f626a2ebe6

@ -74,7 +74,7 @@ def page_extract(tree, *, textequiv_level='region'):
if region is not None:
regions.append(ExtractedText.from_text_segment(region, nsmap, textequiv_level=textequiv_level))
else:
warn('Not a TextRegion: "%s"' % region_id)
pass # Not a TextRegion
else:
raise NotImplementedError
else:

@ -124,8 +124,7 @@ def test_page_order():
def test_page_mixed_regions():
# This file contains ImageRegions and TextRegions in the ReadingOrder
tree = ET.parse(os.path.join(data_dir, 'mixed-regions.page.xml'))
with pytest.warns(UserWarning, match=r'Not a TextRegion'):
result = page_text(tree)
result = page_text(tree)
assert 'non exaudiam uos. Chriſtiani uero quia orant iuxta' in result

Loading…
Cancel
Save