def _load_classes(fname, corpus, split_mapping=None):
return load_classes_txt(fname, corpus, split=split_mapping)
def _load_classes(fname, corpus, split_mapping=None):
return load_classes_txt(fname, corpus, split=split_mapping)
def test_collapse():
is1 = [Interval(0, 1), Interval(2,3),
Interval(4, 5)]
assert(collapse(is1) == is1)
is2 = [Interval(0, 1), Interval(1,2)]
assert(collapse(is2) == [Interval(0, 2)])
is3 = [Interval(0, 1), Interval(2, 3),
Interval(0, 1.5)]
assert(collapse(is3) == [Interval(0, 1.5),
Interval(2,3)])
is4 = [Interval(0, 1), Interval(2, 3),
Interval(0, 10)]
assert(collapse(is4) == [Interval(0.,10)])
corpus = load_corpus_txt('tests/mockdata/tiny.phn')
disc_clsdict = load_classes_txt('tests/mockdata/tiny.classes', corpus)
gold_clsdict = load_classes_txt('tests/mockdata/tiny.classes', corpus)
def test_cover():
assert(cover(disc_clsdict) == 8.0)
assert(cover(gold_clsdict) == 8.0)
def test_coverage():
assert(coverage(disc_clsdict, gold_clsdict) == 1.0)
def test_ued():
assert(ued('kitten', 'sitting') == 3)
def test_ned():
assert(ned('kitten', 'sitting') == 3/7)
def load_disc(fname, corpus, split_file, truncate, verbose):
with verb_print(' loading discovered classes',
verbose, True, True, True):
split_mapping = load_split(split_file)
#disc, errors = load_classes_txt(fname, corpus, split=split_mapping)
disc, errors = load_classes_txt(fname, corpus, split=None)
if not truncate:
errors_found = len(errors) > 0
if len(errors) > 100:
print 'There were more than 100 interval errors found.'
print 'Printing only the first 100.'
print
errors = errors[:100]
for fragment in sorted(errors, key=lambda x: (x.name, x.interval.start)):
print ' error: {0} [{1:.3f}, {2:.3f}]'.format(
fragment.name, fragment.interval.start, fragment.interval.end)
if not truncate and errors_found:
print 'There were errors in {0}. Use option -f to'\
' automatically skip invalid intervals.'.format(fname)
sys.exit()
if truncate:
with verb_print(' checking discovered classes and truncating'):
disc, filename_errors, interval_errors = \
truncate_intervals(disc, corpus,
split_mapping)
else:
with verb_print(' checking discovered classes', verbose, True,
True, True):
filename_errors, interval_errors = \
check_intervals(disc, split_mapping)
if not truncate:
filename_errors = sorted(filename_errors,
key=lambda x: (x.name, x.interval.start))
interval_errors = sorted(interval_errors,
key=lambda x: (x.name, x.interval.start))
interval_error = len(interval_errors) > 0
filename_error = len(filename_errors) > 0
errors_found = filename_error or interval_error
if interval_error:
print banner('intervals found in {0} outside of valid'
' splits'.format(fname))
if len(interval_errors) > 100:
print 'There were more than 100 interval errors found.'
print 'Printing only the first 100.'
print
interval_errors = interval_errors[:100]
for fragment in sorted(interval_errors,
key=lambda x: (x.name, x.interval.start)):
print ' error: {0} [{1:.3f}, {2:.3f}]'.format(
fragment.name,
fragment.interval.start, fragment.interval.end)
if filename_error:
print banner('unknown filenames found in {0}'
.format(fname))
if len(filename_errors) > 100:
print 'There were more than 100 filename errors found.'
print 'Printing only the first 100.'
print
filename_errors = filename_errors[:100]
for fragment in sorted(filename_errors,
key=lambda x: (x.name, x.interval.start)):
print ' error: {0}'.format(fragment.name)
if not truncate and errors_found:
print 'There were errors in {0}. Use option -f to automatically skip invalid intervals.'.format(fname)
sys.exit()
return disc
from tde.util.reader import load_corpus_txt, load_classes_txt
def test_collapse():
is1 = [Interval(0, 1), Interval(2, 3), Interval(4, 5)]
assert (collapse(is1) == is1)
is2 = [Interval(0, 1), Interval(1, 2)]
assert (collapse(is2) == [Interval(0, 2)])
is3 = [Interval(0, 1), Interval(2, 3), Interval(0, 1.5)]
assert (collapse(is3) == [Interval(0, 1.5), Interval(2, 3)])
is4 = [Interval(0, 1), Interval(2, 3), Interval(0, 10)]
assert (collapse(is4) == [Interval(0., 10)])
corpus = load_corpus_txt('tests/mockdata/tiny.phn')
disc_clsdict = load_classes_txt('tests/mockdata/tiny.classes', corpus)
gold_clsdict = load_classes_txt('tests/mockdata/tiny.classes', corpus)
def test_cover():
assert (cover(disc_clsdict) == 8.0)
assert (cover(gold_clsdict) == 8.0)
def test_coverage():
assert (coverage(disc_clsdict, gold_clsdict) == 1.0)
def test_ued():
assert (ued('kitten', 'sitting') == 3)