def test_normalize_parse_tree(self):
num_matches = 0
for parse_tree, s, e in TEILiteParser.pRegion.scanString(self.tei_data1):
num_matches = num_matches + 1
nparse_tree = TEILiteParser.normalizeParseTree(parse_tree.asList())
self.assertTrue(nparse_tree['type'] == 'p')
self.assertTrue(nparse_tree['id'] != None)
self.assertEqual(num_matches, 6)
num_matches = 0
for parse_tree, s, e in TEILiteParser.headRegion.scanString(self.tei_data1):
num_matches = num_matches + 1
nparse_tree = TEILiteParser.normalizeParseTree(parse_tree.asList())
self.assertTrue(nparse_tree['type'] == 'head')
self.assertTrue(nparse_tree['id'] != None)
self.assertEqual(num_matches, 5)
num_matches = 0
for parse_tree, s, e in TEILiteParser.sssRegion.scanString(self.tei_data1):
num_matches = num_matches + 1
nparse_tree = TEILiteParser.normalizeParseTree(parse_tree.asList())
self.assertTrue(nparse_tree['type'] == 'subsubsection')
self.assertTrue(nparse_tree['id'] != None)
self.assertEqual(num_matches, 2)
num_matches = 0
for parse_tree, s, e in TEILiteParser.ssRegion.scanString(self.tei_data1):
num_matches = num_matches + 1
nparse_tree = TEILiteParser.normalizeParseTree(parse_tree.asList())
self.assertTrue(nparse_tree['type'] == 'subsection')
self.assertTrue(nparse_tree['id'] != None)
self.assertTrue(num_matches, 1)
num_matches = 0
for parse_tree, s, e in TEILiteParser.sRegion.scanString(self.tei_data1):
num_matches = num_matches + 1
nparse_tree = TEILiteParser.normalizeParseTree(parse_tree.asList())
self.assertTrue(nparse_tree['type'] == 'section')
self.assertTrue(nparse_tree['id'] != None)
self.assertTrue(num_matches, 2)
def test_get_region_id(self):
# Write TEI tests for subsubsection, subsection, and section
parser = TEILiteParser()
expected_ids = {'subsubsection': ['1', '2'],
'subsection': ['1'],
'section': ['1', '9']}
for region_type in expected_ids.keys():
production = parser.getGrammarForUnit(region_type)
ids = []
match_forest = production.scanString(self.tei_data1)
subtree_idx = 0
for subtree, s, e in match_forest:
new_id = parser.getRegionID(subtree, region_type, subtree_idx)
ids.append(new_id)
subtree_idx = subtree_idx + 1
i = 0
for id in ids:
self.assertEqual(id, expected_ids[region_type][i])
i = i + 1
