def test_orderedgroup_export_order():
"""
See https://github.com/OCR-D/core/issues/475
"""
# arrange
with open(assets.path_to('gutachten/data/TEMP1/PAGE_TEMP1.xml'), 'r') as f:
pcgts = parseString(f.read().encode('utf8'), silence=True)
# act
og = pcgts.get_Page().get_ReadingOrder().get_OrderedGroup()
xml_before = to_xml(og)
children = og.get_AllIndexed()
# assert
assert len(children) == 22
assert [c.index for c in children] == list(range(0, 22))
# mix up the indexes
children[0].index = 11
children[11].index = 3
children[3].index = 0
assert [c.index for c in children] == [11, 1, 2, 0, 4, 5, 6, 7, 8, 9, 10, 3, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21]
assert [c.index for c in og.get_AllIndexed()] == list(range(0, 22))
assert og.get_AllIndexed()[1].__class__ == OrderedGroupIndexedType
# serialize and make sure the correct order was serialized
new_pcgts = parseString(to_xml(pcgts).encode('utf8'), silence=True)
new_og = new_pcgts.get_Page().get_ReadingOrder().get_OrderedGroup()
assert [c.index for c in new_og.get_AllIndexed()] == list(range(0, 22))
xml_after = to_xml(new_og)
def test_issue_269(self):
"""
@conf is parsed as str but should be float
https://github.com/OCR-D/core/issues/269
"""
# GIGO
self.pcgts.get_Page().get_TextRegion()[0].get_TextEquiv()[0].set_conf(1.0)
self.assertEqual(type(self.pcgts.get_Page().get_TextRegion()[0].get_TextEquiv()[0].get_conf()), float)
self.pcgts.get_Page().get_TextRegion()[0].get_TextEquiv()[0].set_conf('1.0')
self.assertEqual(type(self.pcgts.get_Page().get_TextRegion()[0].get_TextEquiv()[0].get_conf()), str)
# test with parseString that @conf in TextEquiv won't throw an error
parseString(simple_page, silence=True)
def test_simpletypes(self):
pcgts = parseString(simple_page, silence=True)
self.assertTrue(isinstance(pcgts.get_Page().imageWidth, int))
el = pcgts.get_Page().get_TextRegion()[0].get_TextLine()[0].get_Word(
)[0].get_TextEquiv()[0]
self.assertTrue(isinstance(el.conf, float))
# XXX no validation on setting attributes :-(
# c.f. https://www.davekuhlman.org/generateDS.html#simpletype
# el.set_conf('2.0987')
# self.assertTrue(isinstance(el.conf, float))
with self.assertRaisesRegex(TypeError, ''):
el.set_conf('I AM NOT A FLOAT DEAL WITH IT')
parseString(to_xml(pcgts).encode('utf8'))
def test_get_UnorderdGroupChildren(self):
with open(assets.path_to('gutachten/data/TEMP1/PAGE_TEMP1.xml'),
'r') as f:
pcgts = parseString(f.read().encode('utf8'), silence=True)
ug = pcgts.get_Page().get_ReadingOrder().get_OrderedGroup(
).get_UnorderedGroupIndexed()[0]
self.assertEqual(len(ug.get_UnorderedGroupChildren()), 1)
def setUp(self):
with open(
assets.path_to(
'glyph-consistency/data/OCR-D-GT-PAGE/FAULTY_GLYPHS.xml'),
'rb') as f:
self.xml_as_str = f.read()
self.pcgts = parseString(self.xml_as_str, silence=True)
def test_image_from_page_basic(self):
with pushd_popd(assets.path_to('gutachten/data')):
ws = self.resolver.workspace_from_url('mets.xml')
with open('TEMP1/PAGE_TEMP1.xml', 'r') as f:
pcgts = parseString(f.read().encode('utf8'), silence=True)
img, info, exif = ws.image_from_page(pcgts.get_Page(), page_id='PHYS_0017', feature_selector='clipped', feature_filter='cropped')
self.assertEquals(info['features'], 'binarized,clipped')
img, info, exif = ws.image_from_page(pcgts.get_Page(), page_id='PHYS_0017')
self.assertEquals(info['features'], 'binarized,clipped')
def test_to_xml_unicode_nsprefix(self):
with open(assets.path_to('kant_aufklaerung_1784-binarized/data/OCR-D-GT-WORD/INPUT_0020.xml'), 'rb') as f:
from_xml = f.read()
self.assertIn('<Unicode>', from_xml.decode('utf-8'), 'without NS prefix')
self.assertIn('<Created', from_xml.decode('utf-8'), 'without NS prefix')
pcgts = parseString(from_xml, silence=True)
as_xml = to_xml(pcgts)
self.assertIn('<pc:Unicode>', as_xml, 'with NS prefix')
self.assertIn('<pc:Created>', as_xml, 'with NS prefix')
def test_serialize_no_empty_readingorder():
"""
https://github.com/OCR-D/core/issues/602
"""
pcgts = page_from_image(create_ocrd_file_with_defaults(url=assets.path_to('kant_aufklaerung_1784/data/OCR-D-IMG/INPUT_0017.tif')))
pcgts.get_Page().set_ReadingOrder(ReadingOrderType())
assert pcgts.get_Page().get_ReadingOrder()
pcgts = parseString(to_xml(pcgts, skip_declaration=True))
assert not pcgts.get_Page().get_ReadingOrder()
def test_delete_region():
pcgts = parseString(simple_page, silence=True)
assert len(pcgts.get_Page().get_TextRegion()) == 1
# act
del pcgts.get_Page().get_TextRegion()[0]
# assert
assert len(pcgts.get_Page().get_TextRegion()) == 0
def test_get_unorderd_group_children():
# arrange
with open(assets.path_to('gutachten/data/TEMP1/PAGE_TEMP1.xml'), 'r') as f:
pcgts = parseString(f.read().encode('utf8'), silence=True)
# act
ug = pcgts.get_Page().get_ReadingOrder().get_OrderedGroup().get_UnorderedGroupIndexed()[0]
# assert
assert len(ug.get_UnorderedGroupChildren()) == 1
def test_get_all_regions_invalid_depth_raises_exeption():
# arrange
with open(assets.path_to('gutachten/data/TEMP1/PAGE_TEMP1.xml'), 'r') as f:
pg = parseString(f.read().encode('utf8'), silence=True).get_Page()
# act
with pytest.raises(Exception) as exc:
pg.get_AllRegions(depth=-1)
# assert
assert "Argument 'depth' must be an integer greater-or-equal 0, not '-1'" in str(exc.value)
def test_get_AllIndexed_classes(self):
with open(assets.path_to('gutachten/data/TEMP1/PAGE_TEMP1.xml'),
'r') as f:
og = parseString(
f.read().encode('utf8'),
silence=True).get_Page().get_ReadingOrder().get_OrderedGroup()
self.assertEqual(len(og.get_AllIndexed(classes=['RegionRef'])), 17)
self.assertEqual(len(og.get_AllIndexed(classes=['OrderedGroup'])),
3)
self.assertEqual(
len(og.get_AllIndexed(classes=['UnorderedGroup'])), 2)
def test_empty_groups_to_regionrefindexed(self):
"""
Corrolary See https://github.com/OCR-D/core/issues/475
"""
with open(assets.path_to('gutachten/data/TEMP1/PAGE_TEMP1.xml'),
'r') as f:
pcgts = parseString(f.read().encode('utf8'), silence=True)
og = pcgts.get_Page().get_ReadingOrder().get_OrderedGroup()
children = og.get_AllIndexed()
self.assertTrue(isinstance(children[1], OrderedGroupIndexedType))
self.assertTrue(isinstance(children[21],
UnorderedGroupIndexedType))
# empty all the elements in the first orederdGroupIndexed
children[1].set_RegionRefIndexed([])
# serialize apnd parse to see empty group converted
pcgts = parseString(to_xml(pcgts).encode('utf8'), silence=True)
og = pcgts.get_Page().get_ReadingOrder().get_OrderedGroup()
children = og.get_AllIndexed()
self.assertTrue(isinstance(children[1], RegionRefIndexedType))
self.assertTrue(isinstance(children[21], RegionRefIndexedType))
def test_get_all_regions_invalid_order_raises_exeption():
# arrange
with open(assets.path_to('gutachten/data/TEMP1/PAGE_TEMP1.xml'), 'r') as f:
pg = parseString(f.read().encode('utf8'), silence=True).get_Page()
# act
with pytest.raises(Exception) as exc:
pg.get_AllRegions(order='random')
# assert
assert "Argument 'order' must be either 'document', 'reading-order' or 'reading-order-only', not 'random'" in str(exc.value)
def test_get_all_indexed_classes():
# arrange
with open(assets.path_to('gutachten/data/TEMP1/PAGE_TEMP1.xml'), 'r') as f:
pcgts = parseString(f.read().encode('utf8'), silence=True)
# act
og = pcgts.get_Page().get_ReadingOrder().get_OrderedGroup()
# assert
assert len(og.get_AllIndexed(classes=['RegionRef'])) == 17
assert len(og.get_AllIndexed(classes=['OrderedGroup'])) == 3
assert len(og.get_AllIndexed(classes=['UnorderedGroup'])) == 2
def test_get_AllAlternativeImages():
with open(assets.path_to('kant_aufklaerung_1784-complex/data/OCR-D-OCR-OCRO-fraktur-SEG-LINE-tesseract-ocropy-DEWARP/OCR-D-OCR-OCRO-fraktur-SEG-LINE-tesseract-ocropy-DEWARP_0001.xml'), 'r') as f:
pcgts = parseString(f.read().encode('utf8'), silence=True)
page = pcgts.get_Page()
assert page.get_AllAlternativeImages(page=False, region=False, line=False) == []
assert [x.filename for x in page.get_AllAlternativeImages(page=True, region=False, line=False)] == [
'OCR-D-IMG-BIN/OCR-D-IMG-BINPAGE-sauvola_0001-BIN_sauvola-ms-split.png',
'OCR-D-IMG-CROP/OCR-D-IMG-CROP_0001.png',
'OCR-D-IMG-BIN/INPUT_0017-BIN_sauvola-ms-split.png',
'OCR-D-IMG-DESPECK/OCR-D-IMG-DESPECK_0001.png',
'OCR-D-IMG-DESKEW/OCR-D-IMG-DESKEW_0001.png',
'OCR-D-IMG-DESKEW/OCR-D-IMG-DESKEW_0001.png']
assert isinstance(page.get_AllAlternativeImages()[0], AlternativeImageType)
def test_extend_AllIndexed_validate_continuity(self):
with open(assets.path_to('gutachten/data/TEMP1/PAGE_TEMP1.xml'),
'r') as f:
og = parseString(
f.read().encode('utf8'),
silence=True).get_Page().get_ReadingOrder().get_OrderedGroup()
with self.assertRaisesRegex(Exception, "@index already used: 1"):
og.extend_AllIndexed([
RegionRefIndexedType(index=3, id='r3'),
RegionRefIndexedType(index=2, id='r2'),
RegionRefIndexedType(index=1, id='r1'),
],
validate_continuity=True)
def test_extend_AllIndexed_no_validation(self):
with open(assets.path_to('gutachten/data/TEMP1/PAGE_TEMP1.xml'),
'r') as f:
og = parseString(
f.read().encode('utf8'),
silence=True).get_Page().get_ReadingOrder().get_OrderedGroup()
og.extend_AllIndexed([
RegionRefIndexedType(index=3, id='r3'),
RegionRefIndexedType(index=2, id='r2'),
RegionRefIndexedType(index=1, id='r1'),
])
rrs = og.get_RegionRefIndexed()
self.assertEqual([x.index for x in rrs][-3:], [22, 23, 24])
def test_to_xml_unicode_nsprefix():
"""see https://github.com/OCR-D/core/pull/474#issuecomment-621477590"""
# arrange
with open(assets.path_to('kant_aufklaerung_1784-binarized/data/OCR-D-GT-WORD/INPUT_0020.xml'), 'rb') as f:
from_xml = f.read()
# assert
assert '<Unicode>' in from_xml.decode('utf-8'), 'without NS prefix'
assert '<Created' in from_xml.decode('utf-8'), 'without NS prefix'
pcgts = parseString(from_xml, silence=True)
as_xml = to_xml(pcgts)
assert '<pc:Unicode>' in as_xml, 'with NS prefix'
assert '<pc:Created>' in as_xml, 'with NS prefix'
def test_extend_all_indexed_validate_continuity():
# arrange
with open(assets.path_to('gutachten/data/TEMP1/PAGE_TEMP1.xml'), 'r') as f:
og = parseString(f.read().encode('utf8'), silence=True).get_Page().get_ReadingOrder().get_OrderedGroup()
# act
with pytest.raises(Exception) as index_exc:
og.extend_AllIndexed([
RegionRefIndexedType(index=3, id='r3'),
RegionRefIndexedType(index=2, id='r2'),
RegionRefIndexedType(index=1, id='r1'),
], validate_continuity=True)
assert "@index already used: 1" in str(index_exc.value)
def test_image_from_page_basic(workspace_gutachten_data):
# arrange
with open(assets.path_to('gutachten/data/TEMP1/PAGE_TEMP1.xml'), 'r') as f:
pcgts = parseString(f.read().encode('utf8'), silence=True)
# act + assert
_, info, _ = workspace_gutachten_data.image_from_page(
pcgts.get_Page(),
page_id='PHYS_0017',
feature_selector='clipped',
feature_filter='cropped')
assert info['features'] == 'binarized,clipped'
_, info, _ = workspace_gutachten_data.image_from_page(pcgts.get_Page(),
page_id='PHYS_0017')
assert info['features'] == 'binarized,clipped'
def test_get_all_indexed_index_sort():
# arrange
with open(assets.path_to('gutachten/data/TEMP1/PAGE_TEMP1.xml'), 'r') as f:
og = parseString(f.read().encode('utf8'), silence=True).get_Page().get_ReadingOrder().get_OrderedGroup()
# act
unogs = og.get_UnorderedGroupIndexed()
# assert
assert [x.index for x in unogs] == [20, 21]
unogs[0].index = 21
unogs[1].index = 20
assert [x.index for x in og.get_AllIndexed(classes=['UnorderedGroup'], index_sort=True)] == [20, 21]
assert [x.index for x in og.get_AllIndexed(classes=['UnorderedGroup'], index_sort=False)] == [21, 20]
og.sort_AllIndexed()
assert [x.index for x in og.get_AllIndexed(classes=['UnorderedGroup'], index_sort=False)] == [20, 21]
def test_get_all_alternative_image_paths():
# arrange
with open(assets.path_to('kant_aufklaerung_1784-complex/data/OCR-D-OCR-OCRO-fraktur-SEG-LINE-tesseract-ocropy-DEWARP/OCR-D-OCR-OCRO-fraktur-SEG-LINE-tesseract-ocropy-DEWARP_0001.xml'), 'r') as f:
pcgts = parseString(f.read().encode('utf8'), silence=True)
# assert
assert pcgts.get_AllAlternativeImagePaths(page=False, region=False, line=False) == []
assert pcgts.get_AllAlternativeImagePaths(page=True, region=False, line=False) == [
'OCR-D-IMG-BIN/OCR-D-IMG-BINPAGE-sauvola_0001-BIN_sauvola-ms-split.png',
'OCR-D-IMG-CROP/OCR-D-IMG-CROP_0001.png',
'OCR-D-IMG-BIN/INPUT_0017-BIN_sauvola-ms-split.png',
'OCR-D-IMG-DESPECK/OCR-D-IMG-DESPECK_0001.png',
'OCR-D-IMG-DESKEW/OCR-D-IMG-DESKEW_0001.png',
'OCR-D-IMG-DESKEW/OCR-D-IMG-DESKEW_0001.png']
assert len(pcgts.get_AllAlternativeImagePaths(page=True, region=True, line=False)) == 12
assert len(pcgts.get_AllAlternativeImagePaths(page=True, region=True, line=False)) == 12
assert len(pcgts.get_AllAlternativeImagePaths(page=True, region=True, line=True)) == 36
def test_all_regions_with_reading_order(self):
"""
https://github.com/OCR-D/core/pull/479
https://github.com/OCR-D/core/issues/240#issuecomment-493135797
"""
with open(assets.path_to('gutachten/data/TEMP1/PAGE_TEMP1.xml'),
'r') as f:
pg = parseString(f.read().encode('utf8'), silence=True).get_Page()
with self.assertRaisesRegex(
Exception,
"Argument 'order' must be either 'document', 'reading-order' or 'reading-order-only', not 'random'"
):
pg.get_AllRegions(order='random')
with self.assertRaisesRegex(
Exception,
"Argument 'depth' must be an integer greater-or-equal 0, not '-1'"
):
pg.get_AllRegions(depth=-1)
self.assertEqual(
len(pg.get_AllRegions(order='reading-order-only')), 40)
self.assertEqual(
len(pg.get_AllRegions(order='reading-order-only', depth=1)),
20)
self.assertEqual(
len(pg.get_AllRegions(order='reading-order-only', depth=2)),
40)
self.assertEqual(
len(pg.get_AllRegions(order='reading-order', depth=0)), 65)
self.assertEqual(
len(pg.get_AllRegions(order='reading-order', depth=1)), 45)
self.assertEqual(
len(pg.get_AllRegions(order='reading-order', depth=2)), 65)
self.assertEqual(
len(pg.get_AllRegions(classes=['Table'],
order='reading-order')), 3)
self.assertEqual(
len(pg.get_AllRegions(classes=['Text'],
order='reading-order')), 37)
self.assertEqual(
len(
pg.get_AllRegions(classes=['Text'],
order='reading-order',
depth=1)), 17)
def test_image_feature_selectoro(self):
with pushd_popd('tests/data/sample-features'):
ws = self.resolver.workspace_from_url('mets.xml')
with open('image_features.page.xml', 'r') as f:
pcgts = parseString(f.read().encode('utf8'), silence=True)
# richest feature set is not last:
img, info, exif = ws.image_from_page(pcgts.get_Page(),
page_id='page1',
feature_selector='dewarped')
# recropped because foo4 contains cropped+deskewed but not recropped yet:
self.assertEqual(
info['features'],
'cropped,dewarped,binarized,despeckled,deskewed,recropped')
# richest feature set is also last:
img, info, exif = ws.image_from_page(pcgts.get_Page(),
page_id='page1',
feature_selector='dewarped',
feature_filter='binarized')
# no deskewing here, thus no recropping:
self.assertEqual(info['features'], 'cropped,dewarped,despeckled')
def test_image_feature_selectoro(workspace_sample_features):
# arrange
with open(
join(str(workspace_sample_features.directory),
'image_features.page.xml'), 'r') as f:
pcgts = parseString(f.read().encode('utf8'))
# richest feature set is not last:
_, info, _ = workspace_sample_features.image_from_page(
pcgts.get_Page(), page_id='page1', feature_selector='dewarped')
# recropped because foo4 contains cropped+deskewed but not recropped yet:
assert info['features'] == 'cropped,dewarped,binarized,despeckled,deskewed'
# richest feature set is also last:
_, info, _ = workspace_sample_features.image_from_page(
pcgts.get_Page(),
page_id='page1',
feature_selector='dewarped',
feature_filter='binarized')
# no deskewing here, thus no recropping:
assert info['features'] == 'cropped,dewarped,despeckled'
def test_all_regions_with_reading_order():
"""
https://github.com/OCR-D/core/pull/479
https://github.com/OCR-D/core/issues/240#issuecomment-493135797
"""
# arrange
with open(assets.path_to('gutachten/data/TEMP1/PAGE_TEMP1.xml'), 'r') as f:
pg = parseString(f.read().encode('utf8'), silence=True).get_Page()
# assert
assert len(pg.get_AllRegions(order='reading-order-only')) == 40
assert len(pg.get_AllRegions(order='reading-order-only', depth=1)) == 20
assert len(pg.get_AllRegions(order='reading-order-only', depth=2)) == 40
assert len(pg.get_AllRegions(order='reading-order', depth=0)) == 65
assert len(pg.get_AllRegions(order='reading-order', depth=1)) == 45
assert len(pg.get_AllRegions(order='reading-order', depth=2)) == 65
assert len(pg.get_AllRegions(classes=['Table'], order='reading-order')) == 3
assert len(pg.get_AllRegions(classes=['Text'], order='reading-order')) == 37
assert len(pg.get_AllRegions(classes=['Text'], order='reading-order', depth=1)) == 17
def get_lines(fname, flist=False):
with open(fname, 'r') as fd:
rawlines = [line.rstrip('\r\n') for line in fd.readlines()]
try:
# PAGE-XML case
if rawlines and rawlines[0].startswith('<?xml'):
rawlines[0] = rawlines[0][rawlines[0].index('?>') + 2:]
pcgts = parseString(''.join(rawlines))
#pcgts = parse(fname)
lines = page_get_lines(pcgts)
except Exception:
# plaintext case
lines = rawlines
if flist:
# ocropy style (e.g. -F <(ls -1 *.gt.txt) <(ls -1 *.ocr.txt))
files = lines
lines = []
for fname in files:
with open(fname, 'r') as fd:
lines.append(fd.readline())
return lines
def test_all_regions_without_reading_order(self):
"""
https://github.com/OCR-D/core/pull/479
https://github.com/OCR-D/core/issues/240#issuecomment-493135797
"""
with open(assets.path_to('gutachten/data/TEMP1/PAGE_TEMP1.xml'),
'r') as f:
pcgts = parseString(f.read().encode('utf8'), silence=True)
pg = pcgts.get_Page()
self.assertEqual(len(pg.get_AllRegions()), 65)
self.assertEqual(len(pg.get_AllRegions(depth=0)), 65)
self.assertEqual(len(pg.get_AllRegions(depth=1)), 45)
self.assertEqual(len(pg.get_AllRegions(depth=2)), 65)
self.assertEqual(len(pg.get_AllRegions(depth=3)), 65)
self.assertEqual(len(pg.get_AllRegions(classes=['Separator'])), 25)
self.assertEqual(len(pg.get_AllRegions(classes=['Table'])), 3)
self.assertEqual(len(pg.get_AllRegions(classes=['Text'])), 37)
self.assertEqual(len(pg.get_AllRegions(classes=['Text'], depth=1)),
17)
self.assertEqual(len(pg.get_AllRegions(classes=['Text'], depth=2)),
37)
def test_get_AllIndexed_index_sort(self):
with open(assets.path_to('gutachten/data/TEMP1/PAGE_TEMP1.xml'),
'r') as f:
og = parseString(
f.read().encode('utf8'),
silence=True).get_Page().get_ReadingOrder().get_OrderedGroup()
unogs = og.get_UnorderedGroupIndexed()
self.assertEqual([x.index for x in unogs], [20, 21])
unogs[0].index = 21
unogs[1].index = 20
self.assertEqual([
x.index for x in og.get_AllIndexed(classes=['UnorderedGroup'],
index_sort=True)
], [20, 21])
self.assertEqual([
x.index for x in og.get_AllIndexed(classes=['UnorderedGroup'],
index_sort=False)
], [21, 20])
og.sort_AllIndexed()
self.assertEqual([
x.index for x in og.get_AllIndexed(classes=['UnorderedGroup'],
index_sort=False)
], [20, 21])