def test_ocrd_file_equality():
mets = OcrdMets.empty_mets()
f1 = mets.add_file('FOO', ID='FOO_1', mimetype='image/tiff')
f2 = mets.add_file('FOO', ID='FOO_2', mimetype='image/tiff')
assert f1 != f2
f3 = create_ocrd_file_with_defaults(ID='TEMP_1', mimetype='image/tiff')
f4 = create_ocrd_file_with_defaults(ID='TEMP_1', mimetype='image/tif')
# be tolerant of different equivalent mimetypes
assert f3 == f4
f5 = mets.add_file('TEMP', ID='TEMP_1', mimetype='image/tiff')
assert f3 == f5
def test_ocrd_file_eq(self):
mets = OcrdMets.empty_mets()
f1 = mets.add_file('FOO', ID='FOO_1', mimetype='image/tiff')
self.assertEqual(f1 == f1, True)
self.assertEqual(f1 != f1, False)
f2 = mets.add_file('FOO', ID='FOO_2', mimetype='image/tiff')
self.assertEqual(f1 == f2, False)
f3 = create_ocrd_file_with_defaults(ID='TEMP_1', mimetype='image/tiff')
f4 = create_ocrd_file_with_defaults(ID='TEMP_1', mimetype='image/tif')
# be tolerant of different equivalent mimetypes
self.assertEqual(f3 == f4, True)
f5 = mets.add_file('TEMP', ID='TEMP_1', mimetype='image/tiff')
self.assertEqual(f3 == f5, True)
def test_create_ocrd_file_with_defaults_basename_wo_extension(
local_filename, wo_extension):
"""Behavior for ocrd_file_with_defaults
"""
f = create_ocrd_file_with_defaults(local_filename=local_filename)
assert f.basename_without_extension == wo_extension
def test_page_from_file(self):
f = create_ocrd_file_with_defaults(mimetype='image/tiff',
local_filename=SAMPLE_IMG,
ID='file1')
self.assertEqual(f.mimetype, 'image/tiff')
p = page_from_file(f)
self.assertEqual(p.pcGtsId, f.ID)
self.assertEqual(p.get_Page().imageWidth, 1457)
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_hashable():
"""
https://github.com/OCR-D/ocrd_segment/issues/45
"""
pcgts = page_from_image(create_ocrd_file_with_defaults(url=assets.path_to('kant_aufklaerung_1784/data/OCR-D-IMG/INPUT_0017.tif')))
page = pcgts.get_Page()
testset = set()
testset.add(pcgts)
testset.add(page)
def test_loctype(self):
f = create_ocrd_file_with_defaults()
self.assertEqual(f.loctype, 'OTHER')
self.assertEqual(f.otherloctype, 'FILE')
f.otherloctype = 'foo'
self.assertEqual(f.otherloctype, 'foo')
f.loctype = 'URN'
self.assertEqual(f.loctype, 'URN')
self.assertEqual(f.otherloctype, None)
f.otherloctype = 'foo'
self.assertEqual(f.loctype, 'OTHER')
def test_set_loctype():
f = create_ocrd_file_with_defaults()
assert f.loctype == 'OTHER'
assert f.otherloctype == 'FILE'
f.otherloctype = 'foo'
assert f.otherloctype == 'foo'
f.loctype = 'URN'
assert f.loctype == 'URN'
assert f.otherloctype == None
f.otherloctype = 'foo'
assert f.loctype, 'OTHER'
def test_page_from_file_no_local_filename(self):
with self.assertRaisesRegex(
ValueError, "input_file must have 'local_filename' property"):
page_from_file(
create_ocrd_file_with_defaults(mimetype='image/tiff'))
def test_basename_without_extension_tar(self):
f = create_ocrd_file_with_defaults(
local_filename='/tmp/foo/bar/foo.tar.gz')
self.assertEqual(f.basename_without_extension, 'foo')
def test_extension(self):
f = create_ocrd_file_with_defaults(
local_filename='/tmp/foo/bar/foo.bar')
self.assertEqual(f.extension, '.bar')
def test_basename_from_url(self):
f = create_ocrd_file_with_defaults(url="http://foo.bar/quux")
self.assertEqual(f.basename, 'quux')
def test_basename(self):
f = create_ocrd_file_with_defaults(
local_filename='/tmp/foo/bar/foo.bar')
self.assertEqual(f.basename, 'foo.bar')
def test_set_id_none(self):
f = create_ocrd_file_with_defaults()
f.ID = 'foo12'
self.assertEqual(f.ID, 'foo12')
f.ID = None
self.assertEqual(f.ID, 'foo12')
def test_constructor_url(self):
f = create_ocrd_file_with_defaults(url="foo")
self.assertEqual(f.url, 'foo')
self.assertEqual(f.local_filename, 'foo')
def test_page_from_file_page(self):
f = create_ocrd_file_with_defaults(mimetype=MIMETYPE_PAGE,
local_filename=SAMPLE_PAGE)
p = page_from_file(f)
self.assertEqual(p.get_Page().imageWidth, 1457)
def test_set_url():
f = create_ocrd_file_with_defaults()
f.url = None
f.url = 'http://foo'
f.url = 'http://bar'
assert f.url == 'http://bar'
def test_create_ocrd_file_with_defaults_extension(local_filename, extension):
"""Behavior for ocrd_file_with_defaults
"""
f = create_ocrd_file_with_defaults(local_filename=local_filename)
assert f.extension == extension
def test_basename_from_url():
f = create_ocrd_file_with_defaults(url="http://foo.bar/quux")
assert f.basename == 'quux'
def test_basename():
f = create_ocrd_file_with_defaults(local_filename='/tmp/foo/bar/foo.bar')
assert f.basename == 'foo.bar'
def test_set_id_none():
f = create_ocrd_file_with_defaults()
f.ID = 'foo12'
assert f.ID == 'foo12'
f.ID = None
assert f.ID == 'foo12'
def test_constructor_url():
f = create_ocrd_file_with_defaults(url="foo")
assert f.url == 'foo'
assert f.local_filename == 'foo'
def test_set_url(self):
f = create_ocrd_file_with_defaults()
f.url = None
f.url = 'http://foo'
f.url = 'http://bar'
self.assertEqual(f.url, 'http://bar')
def test_page_from_file_unsupported_mimetype(self):
with self.assertRaisesRegex(ValueError, "Unsupported mimetype"):
page_from_file(
create_ocrd_file_with_defaults(local_filename=__file__,
mimetype='foo/bar'))
