def test_to_praat_ends_after(self):
mock_tier = mock.Mock()
mock_tier.xmin = 0
mock_tier.xmax = 1
tg = TextGrid(0, 0.5, [mock_tier])
with assert_raises(ValueError):
tg.to_praat()
def test_to_praat_starts_before(self):
mock_tier = mock.Mock()
mock_tier.xmin = 0
mock_tier.xmax = 1
tg = TextGrid(0.5, 1, [mock_tier])
with assert_raises(ValueError):
tg.to_praat()
def test_to_praat_starts_before(self):
mock_tier = mock.Mock()
mock_tier.xmin = 0
mock_tier.xmax = 1
tg = TextGrid(0.5, 1, [mock_tier])
with assert_raises(ValueError):
tg.to_praat()
def test_to_praat_ends_after(self):
mock_tier = mock.Mock()
mock_tier.xmin = 0
mock_tier.xmax = 1
tg = TextGrid(0, 0.5, [mock_tier])
with assert_raises(ValueError):
tg.to_praat()
def test_to_dict(self):
mock_tier = mock.Mock()
mock_tier.to_dict.return_value = {'a tier': 'data'}
tg = TextGrid(0, 1, [mock_tier])
res = tg.to_dict()
assert_equal(res, {'xmin': 0,
'xmax': 1,
'tiers': [{'a tier': 'data'}]})
def test_to_praat(self):
mock_tier = mock.Mock()
mock_tier.xmin = 0
mock_tier.xmax = 1
mock_tier.to_praat.return_value = 'a tier'
tg = TextGrid(0, 1, [mock_tier, mock_tier])
res = tg.to_praat()
assert_equal(res, ('"ooTextFile"\n"TextGrid"\n'
'0 to 1 seconds <exists>\n'
'2 tiers\n\na tier\n\na tier'))
def test_to_praat(self):
mock_tier = mock.Mock()
mock_tier.xmin = 0
mock_tier.xmax = 1
mock_tier.to_praat.return_value = 'a tier'
tg = TextGrid(0, 1, [mock_tier, mock_tier])
res = tg.to_praat()
assert_equal(res, ('"ooTextFile"\n"TextGrid"\n'
'0 to 1 seconds <exists>\n'
'2 tiers\n\na tier\n\na tier'))
def test_to_dict(self):
mock_tier = mock.Mock()
mock_tier.to_dict.return_value = {'a tier': 'data'}
tg = TextGrid(0, 1, [mock_tier])
res = tg.to_dict()
assert_equal(res, {
'xmin': 0,
'xmax': 1,
'tiers': [{
'a tier': 'data'
}]
})
def test_to_praat_with_path(self):
mock_tier = mock.Mock()
mock_tier.xmin = 0
mock_tier.xmax = 1
mock_tier.to_praat.return_value = 'a tier'
tg = TextGrid(0, 1, [mock_tier, mock_tier])
tg.to_praat('test/files/output.TextGrid')
with open('test/files/output.TextGrid') as output_file:
res = output_file.read()
assert_equal(res, ('"ooTextFile"\n"TextGrid"\n'
'0 to 1 seconds <exists>\n'
'2 tiers\n\na tier\n\na tier'))
def test_to_praat_with_path(self):
mock_tier = mock.Mock()
mock_tier.xmin = 0
mock_tier.xmax = 1
mock_tier.to_praat.return_value = 'a tier'
tg = TextGrid(0, 1, [mock_tier, mock_tier])
tg.to_praat('tests/files/output.TextGrid')
with open('tests/files/output.TextGrid') as output_file:
res = output_file.read()
assert_equal(res, ('"ooTextFile"\n"TextGrid"\n'
'0 to 1 seconds <exists>\n'
'2 tiers\n\na tier\n\na tier'))
def test_from_file(self):
tg_file = 'tests/files/doubled-quotes-in-text-and-mark.TextGrid'
tg = TextGrid.from_file(tg_file)
assert_equal(tg.xmin, 0.25)
assert_equal(tg.xmax, 1.5)
assert_equal(len(tg.tiers), 2)
assert_is(tg.tiers[0].__class__, IntervalTier)
assert_equal(tg.tiers[0].name, 'words')
assert_equal(tg.tiers[0].xmin, 0.1)
assert_equal(tg.tiers[0].xmax, 1)
assert_equal(len(tg.tiers[0]._items), 2)
assert_equal(tg.tiers[0]._items[0].xmin, 0.1)
assert_equal(tg.tiers[0]._items[0].xmax, 0.5)
assert_equal(tg.tiers[0]._items[0].text, '"Is anyone home?"')
assert_equal(tg.tiers[0]._items[1].xmin, 0.5)
assert_equal(tg.tiers[0]._items[1].xmax, 1)
assert_equal(tg.tiers[0]._items[1].text, 'asked "Pat"')
assert_is(tg.tiers[1].__class__, TextTier)
assert_equal(tg.tiers[1].name, 'points')
assert_equal(tg.tiers[1].xmin, 0.1)
assert_equal(tg.tiers[1].xmax, 1)
assert_equal(len(tg.tiers[1]._items), 2)
assert_equal(tg.tiers[1]._items[0].number, 0.25)
assert_equal(tg.tiers[1]._items[0].mark, '"event"')
assert_equal(tg.tiers[1]._items[1].number, 0.75)
assert_equal(tg.tiers[1]._items[1].mark, '"event" with quotes again')
def test_from_file_armenian(self):
tg = TextGrid.from_file('tests/files/numbers.TextGrid', encoding='utf_16')
assert_equal(tg.tiers[0][0].text, u'մեկ')
assert_equal(tg.tiers[0][1].text, u'երկու')
assert_equal(tg.tiers[0][2].text, u'երեք')
assert_equal(tg.tiers[0][3].text, u'չորս')
def test_from_file(self):
tg_file = 'test/files/doubled-quotes-in-text-and-mark.TextGrid'
tg = TextGrid.from_file(tg_file)
assert_equal(tg.xmin, 0.25)
assert_equal(tg.xmax, 1.5)
assert_equal(len(tg.tiers), 2)
assert_is(tg.tiers[0].__class__, IntervalTier)
assert_equal(tg.tiers[0].name, 'words')
assert_equal(tg.tiers[0].xmin, 0.1)
assert_equal(tg.tiers[0].xmax, 1)
assert_equal(len(tg.tiers[0]._items), 2)
assert_equal(tg.tiers[0]._items[0].xmin, 0.1)
assert_equal(tg.tiers[0]._items[0].xmax, 0.5)
assert_equal(tg.tiers[0]._items[0].text, '"Is anyone home?"')
assert_equal(tg.tiers[0]._items[1].xmin, 0.5)
assert_equal(tg.tiers[0]._items[1].xmax, 1)
assert_equal(tg.tiers[0]._items[1].text, 'asked "Pat"')
assert_is(tg.tiers[1].__class__, TextTier)
assert_equal(tg.tiers[1].name, 'points')
assert_equal(tg.tiers[1].xmin, 0.1)
assert_equal(tg.tiers[1].xmax, 1)
assert_equal(len(tg.tiers[1]._items), 2)
assert_equal(tg.tiers[1]._items[0].number, 0.25)
assert_equal(tg.tiers[1]._items[0].mark, '"event"')
assert_equal(tg.tiers[1]._items[1].number, 0.75)
assert_equal(tg.tiers[1]._items[1].mark, '"event" with quotes again')
def main(rootDir, outDir):
print("number of phones appeared in mapping:",
len(set(phonesTargetMapper.values())), ",",
"number of target phones:", len(targetPhones))
phonesLocations = {
} # {"ph1": [(fpath1, xmin1, xmax1), (fpath2, xmin2, xmax2), ...], "ph2": [(fpath1, xmin1, xmax1), (fpath2, xmin2, xmax2), ...], ...}
for targetPhone in targetPhones:
phonesLocations[targetPhone] = []
phonesLocations['others'] = []
for filePath in getFilesPaths(rootDir):
textGrid = TextGrid.from_file(filePath)
phonesTier = next(
filter(lambda interval: interval.name == "phones", textGrid), None)
if (not phonesTier):
print("the file has no phones tier !!")
for interval in phonesTier:
actualPhone = interval.text.upper()
targetPhone = phonesTargetMapper.get(actualPhone, 'others')
relDir = getRelDir(filePath, back=3)
phonesLocations[targetPhone].append(
(relDir, interval.xmin, interval.xmax))
for targetPhone in targetPhones:
np.savetxt(f"{os.path.join(outDir, targetPhone)}.txt",
phonesLocations[targetPhone],
fmt='%a')
def test_from_file_armenian(self):
tg = TextGrid.from_file('test/files/numbers.TextGrid',
encoding='utf_16')
assert_equal(tg.tiers[0][0].text, u'մեկ')
assert_equal(tg.tiers[0][1].text, u'երկու')
assert_equal(tg.tiers[0][2].text, u'երեք')
assert_equal(tg.tiers[0][3].text, u'չորս')
def test_from_reader(self, TierFromReaderMock):
TierFromReaderMock.return_value = 'tier'
stream = iter([0, 1, 2])
tg = TextGrid.from_reader(stream)
assert_equal(tg.xmin, 0)
assert_equal(tg.xmax, 1)
assert_equal(tg.tiers, ['tier', 'tier'])
def test_from_reader(self, TierFromReaderMock):
TierFromReaderMock.return_value = 'tier'
stream = iter([0, 1, 2])
tg = TextGrid.from_reader(stream)
assert_equal(tg.xmin, 0)
assert_equal(tg.xmax, 1)
assert_equal(tg.tiers, ['tier', 'tier'])
def test_from_file_missing_header(self):
with assert_raises(ValueError):
TextGrid.from_file('tests/files/intervals-no-filetype.TextGrid')
with assert_raises(ValueError):
TextGrid.from_file('tests/files/intervals-no-object-class.TextGrid')
def test_utf8_bom(self):
# BOM should be skipped by regex
tg = TextGrid.from_file('tests/files/intervals-utf8-bom.TextGrid')
assert_equal(tg.tiers[1][1].text, 'ciao')
def test_repr(self):
tg = TextGrid(0, 1, [1, 2, 3])
assert_equal(repr(tg), "TextGrid(0, 1, [1, 2, 3])")
def test_str(self):
tg = TextGrid(0, 1, [1, 2, 3])
assert_equal(str(tg), '<TextGrid from 0 to 1 seconds with 3 tiers>')
def test_get(self):
tg = TextGrid(0, 1, ['tier 0', 'tier 1'])
assert_equal(tg[1], 'tier 1')
def test_set(self):
tg = TextGrid(0, 1, ['tier 0', 'tier 1'])
tg[1] = 'new tier'
assert_equal(tg[1], 'new tier')
def test_len(self):
tg = TextGrid(0, 1, ['tier 0', 'tier 1'])
assert_equal(len(tg), 2)
def test_from_reader_wrong_size(self, TierFromReaderMock):
TierFromReaderMock.return_value = 'tier'
stream = iter([0, 1, 2, 'extra_value'])
with assert_raises(ValueError):
TextGrid.from_reader(stream)
def test_reversed(self):
tg = TextGrid(0, 1, ['tier 0', 'tier 1'])
assert_equal(list(reversed(tg)), ['tier 1', 'tier 0'])
def test_from_file_missing_header(self):
with assert_raises(ValueError):
TextGrid.from_file('test/files/intervals-no-filetype.TextGrid')
with assert_raises(ValueError):
TextGrid.from_file('test/files/intervals-no-object-class.TextGrid')
def test_utf8_bom(self):
# BOM should be skipped by regex
tg = TextGrid.from_file('test/files/intervals-utf8-bom.TextGrid')
assert_equal(tg.tiers[1][1].text, 'ciao')
def test_from_reader_wrong_size(self, TierFromReaderMock):
TierFromReaderMock.return_value = 'tier'
stream = iter([0, 1, 2, 'extra_value'])
with assert_raises(ValueError):
TextGrid.from_reader(stream)
def test_del(self):
tg = TextGrid(0, 1, ['tier 0', 'tier 1'])
del tg[0]
assert_equal(tg[0], 'tier 1')
def test_init(self):
tg = TextGrid(0, 1, [])
assert_equal(tg.xmin, 0)
assert_equal(tg.xmax, 1)
assert_equal(tg.tiers, [])
def test_iter(self):
tg = TextGrid(0, 1, ['tier 0', 'tier 1'])
assert_equal(list(iter(tg)), ['tier 0', 'tier 1'])
