def test_metadata(self):
doc = ElementTree(file='fixtures/metadata.xml')
parser = MorParser()
metadata = parser.parse_metadata(doc)
self.assertEqual(metadata, {
'lang': 'eng',
'corpus': 'manchester',
'date': '1984-01-01',
'participants': [
{
'id': 'INV',
'name': 'Caroline',
'role': 'Investigator',
'language': 'eng'
},
{
'id': 'CHI',
'name': 'Nicole',
'role': 'Target_Child',
'language': 'eng',
'age': 'P2Y10M8D',
'sex': 'female',
'group': 'typical',
'SES' :'MC'
},
{
'id': 'MOT',
'name': 'Mother',
'language': 'eng'
}]})
def test_missing_pos(self):
parser = MorParser()
for uid, speaker, tokens in parser.parse("fixtures/missing_pos.xml"):
for token in tokens:
#print(token.word + '/' + token.pos + '|' + token.stem)
self.assertNotEqual(token.pos, 'unk',
'failed to parse known tag')
def test_compound(self):
comps = self.compounds.findall("w")
self.assertEqual(5, len(comps))
for word in comps:
parser = MorParser()
parser.namespace = ""
parts = parser.parse_mor_element(word, word.find('mor'))
self.assertGreaterEqual(parts[0].stem.count("_"), 1)
def test_clitics(self):
parser = MorParser()
for uid, speaker, tokens in parser.parse("fixtures/clitics.xml"):
self.assertGreater(len(tokens), 1,
"failed splitting {0} into clitics".format(tokens))
self.assertNotIn("?", [w.word for w in tokens])
self.assertEqual(' '.join(map(str, tokens)),
("hidden/part|hide&PERF away/adv|away where/adv:wh|where "
"nobody/pro:indef|nobody 'd/mod|genmod be/v:cop|be ./.|."))
head, tail = parser.split_clitic_wordform("that's")
self.assertEqual(head, "that")
self.assertEqual(tail, ["'s"])
def test_commas(self):
parser = MorParser()
for uid, speaker, tokens in parser.parse("fixtures/commas.xml"):
self.assertIn(',', [word.stem for word in tokens])
def test_document(self):
parser = MorParser()
for i in parser.parse("fixtures/test_doc.xml"):
# iterate through an ensure no exceptions are thrown
pass
from talkbank_parser import MorParser
import glob
import illegal_utterance_filter
parser = MorParser()
def age_transform(metadata):
'''transform the original age into age in months'''
ageday = 0
for i in metadata: # get the location of 'Y', 'M', 'D'
if i == 'Y':
loc_Y = metadata.index(i)
elif i == 'M':
loc_M = metadata.index(i)
elif i == 'D':
loc_D = metadata.index(i)
for i in metadata: # get the number before 'Y', 'M', and 'D'
if i == 'Y':
age_year = int(metadata[loc_Y - 1])
elif i == 'M':
age_month = int(metadata[loc_Y + 1:loc_M])
elif i == 'D':
age_day = int(metadata[loc_M + 1:loc_D])
if age_day >= 15:
ageday = 1
else:
ageday = 0
age_transformed = age_year * 12 + age_month + ageday
return age_transformed
def test_compound(self):
for word in self.compounds.findall("w/mor"):
parser = MorParser()
parser.namespace = ""
parts = parser.parse_mor_element(None, word)
self.assertGreaterEqual(parts[0].stem.count("_"), 1)
