def test_dot_export_including_annotation(self):
witness_a = "<tei><s1>x y z</s1></tei>"
witness_b = "<tei><s2>x</s2>y<s3>z</s3></tei>"
tokens_a = convert_xml_string_into_tokens(witness_a)
tokens_b = convert_xml_string_into_tokens(witness_b)
superwitness = align_tokens_and_return_superwitness(tokens_a, tokens_b)
textgraph = convert_superwitness_to_textgraph(superwitness)
dot_export = export_as_dot(textgraph, annotations=True)
expected_out = """strict digraph TextGraph {
1 [label="x"]
2 [label="y"]
3 [label="z"]
1 -> 2
2 -> 3
{ rank=same; 1; 2; 3 }
a1 [label="tei"]
a2 [label="s1"]
a3 [label="s2"]
a4 [label="s3"]
a2 -> a1
a3 -> a1
1 -> a2
1 -> a3
a2 -> a1
2 -> a2
a2 -> a1
a4 -> a1
3 -> a2
3 -> a4
}"""
# TODO: There are some duplication annotation edges here that should be removed! (a2 - a1)
# NOTE: For now we work around the problem by adding the "strict" keyword to the DOT export.
self.assertEqual(expected_out, dot_export)
def test_3_textgraph_text(self):
witness_a = "<tei><p><s>a b<s>c</s>d</s></p></tei>"
witness_b = "<tei><div><p><s>a b d</s></p></div></tei>"
tokens_a = convert_xml_string_into_tokens(witness_a)
tokens_b = convert_xml_string_into_tokens(witness_b)
superwitness = align_tokens_and_return_superwitness(tokens_a, tokens_b)
textgraph = convert_superwitness_to_textgraph(superwitness)
text_tokens = textgraph.text_tokens
self.assertEquals("[a, b, -c, d]", str(text_tokens))
def test_textgraph_text(self):
witness_a = "<tei><s>x y z</s></tei>"
witness_b = "<tei><s>x</s>y<s>z</s></tei>"
tokens_a = convert_xml_string_into_tokens(witness_a)
tokens_b = convert_xml_string_into_tokens(witness_b)
superwitness = align_tokens_and_return_superwitness(tokens_a, tokens_b)
textgraph = convert_superwitness_to_textgraph(superwitness)
text_tokens = textgraph.text_tokens
self.assertEquals("[x, y, z]", str(text_tokens))
def test_textgraph_milestones_annotations(self):
witness_a = "<tei><lb/></tei>"
witness_b = "<tei><lb/></tei>"
tokens_a = convert_xml_string_into_tokens(witness_a)
tokens_b = convert_xml_string_into_tokens(witness_b)
superwitness = align_tokens_and_return_superwitness(tokens_a, tokens_b)
textgraph = convert_superwitness_to_textgraph(superwitness)
annotations = textgraph.annotations
self.assertEqual(0, len(annotations))
def collate_xml(witness_a, witness_b):
tokens_a = convert_xml_string_into_tokens(witness_a)
tokens_b = convert_xml_string_into_tokens(witness_b)
superwitness = align_tokens_and_return_superwitness(tokens_a, tokens_b)
textgraph = convert_superwitness_to_textgraph(superwitness)
dot_export = export_as_dot(textgraph, annotations=True)
dot = Source(dot_export, format="svg")
svg = dot.render()
return display(SVG(svg))
def test_dot1_textnodes_only(self):
witness_a = "<tei><s>x y z</s></tei>"
witness_b = "<tei><s>x</s>y<s>z</s></tei>"
tokens_a = convert_xml_string_into_tokens(witness_a)
tokens_b = convert_xml_string_into_tokens(witness_b)
superwitness = align_tokens_and_return_superwitness(tokens_a, tokens_b)
textgraph = convert_superwitness_to_textgraph(superwitness)
xml = convert_textgraph_to_xml(textgraph)
expected = '''<xml><tei><s cx:witness="a"><s cx:witness="b">x</s>y<s cx:witness="b">z</s></s></tei>
'''
self.assertEqual(expected, xml)
def test_2_textgraph_annotations(self):
witness_a = "<tei><p><s>a b<s>c</s>d</s></p></tei>"
witness_b = "<tei><p><s>a b d</s></p></tei>"
tokens_a = convert_xml_string_into_tokens(witness_a)
tokens_b = convert_xml_string_into_tokens(witness_b)
superwitness = align_tokens_and_return_superwitness(tokens_a, tokens_b)
textgraph = convert_superwitness_to_textgraph(superwitness)
annotations = textgraph.annotations
self.assertIn(Annotation("s", ["A"], 2, 2, 3), annotations)
self.assertIn(Annotation("s", ["A", "B"], 0, 3, 2), annotations)
self.assertIn(Annotation("p", ["A", "B"], 0, 3, 1), annotations)
self.assertIn(Annotation("tei", ["A", "B"], 0, 3, 0), annotations)
self.assertEqual(4, len(annotations))
def collate_xml_svg(limit=1000):
# convert XML files into tokens
tokens1 = convert_xml_file_into_tokens("xml_source_transcriptions/ts-fol-test-small.xml")
tokens2 = convert_xml_file_into_tokens("xml_source_transcriptions/tsq-test-small.xml")
superwitness = align_tokens_and_return_superwitness(tokens1, tokens2)
print(superwitness[0:20])
textgraph = convert_superwitness_to_textgraph(superwitness)
dot_export = export_as_dot(textgraph, annotations=True, limit=limit)
# print(dot_export)
# render dot_export as SVG
dot = Source(dot_export, format="svg")
svg = dot.render()
return display(SVG(svg))
def test_sort_annotations_based_on_positions_and_level(self):
witness_a = "<tei><s>x y z</s></tei>"
witness_b = "<tei><s>x</s>y<s>z</s></tei>"
tokens_a = convert_xml_string_into_tokens(witness_a)
tokens_b = convert_xml_string_into_tokens(witness_b)
superwitness = align_tokens_and_return_superwitness(tokens_a, tokens_b)
textgraph = convert_superwitness_to_textgraph(superwitness)
# sort on range start, then on range end, then on level
a1 = Annotation("tei", ["A", "B"], 0, 2, 0)
a2 = Annotation("s", ["A"], 0, 2, 1)
a3 = Annotation("s", ["B"], 0, 0, 1)
a4 = Annotation("s", ["B"], 2, 2, 1)
expected_annotations = [a1, a2, a3, a4]
annotations = textgraph.annotations_sorted
self.assertEqual(expected_annotations, annotations)
def collate_xml_example():
# convert XML files into tokens
tokens1 = convert_xml_file_into_tokens("../xml_source_transcriptions/ts-fol-test-small.xml")
tokens2 = convert_xml_file_into_tokens("../xml_source_transcriptions/tsq-test-small.xml")
# log
print(tokens1)
print(tokens2)
# end
superwitness = align_tokens_and_return_superwitness(tokens1, tokens2)
# log
print(superwitness)
# end
textgraph = convert_superwitness_to_textgraph(superwitness)
dot_export = export_as_dot(textgraph, annotations=True)
print(dot_export)
def test_dot1_textnodes_only(self):
witness_a = "<tei><s>x y z</s></tei>"
witness_b = "<tei><s>x</s>y<s>z</s></tei>"
tokens_a = convert_xml_string_into_tokens(witness_a)
tokens_b = convert_xml_string_into_tokens(witness_b)
superwitness = align_tokens_and_return_superwitness(tokens_a, tokens_b)
textgraph = convert_superwitness_to_textgraph(superwitness)
dot_export = export_as_dot(textgraph)
expected_out = """strict digraph TextGraph {
1 [label="x"]
2 [label="y"]
3 [label="z"]
1 -> 2
2 -> 3
{ rank=same; 1; 2; 3 }
}"""
self.assertEqual(expected_out, dot_export)