def test_mark_jonhson(tmpdir, datadir):
# this is a transcription of the original "toy run" delivered with
# the original AG code (as a target in the Makefile)
assert os.path.isdir(datadir)
grammar_file = os.path.join(datadir, 'ag_testengger.lt')
text = list(
codecs.open(os.path.join(datadir, 'ag_testeng.yld'),
'r',
encoding='utf8'))
arguments = (
'-r 1234 -P -D -R -1 -d 100 -a 1e-2 -b 1 -e 1 -f 1 '
'-g 1e2 -h 1e-2 -n 10 -C -E -A {prs} -N 10 -F {trace} -G {wlt} '
# -X "cat > {X1}" -X "cat > {X2}" '
# -U "cat > {prs1}" -v {testeng2} -V "cat > {prs2}"'
'-u {testeng1} '.format(
testeng1=os.path.join(datadir, 'ag_testeng1.yld'),
# testeng2=os.path.join(datadir, 'ag_testeng2.yld'),
trace=tmpdir.join('trace'),
wlt=tmpdir.join('wlt'),
prs=tmpdir.join('prs')
# X1=tmpdir.join('X1'), X2=tmpdir.join('X2'),
# prs1=tmpdir.join('prs1'), prs2=tmpdir.join('prs2')))
))
# pc = ag.ParseCounter(len(text))
output = ag.segment(text,
grammar_file=grammar_file,
category='VP',
args=arguments,
ignore_first_parses=0,
nruns=1)
assert len(text) == len(output)
for i in range(len(text)):
assert text[i].strip().replace(' ', '') == output[i].replace(' ', '')
def test_default_grammar(prep):
segmented = ag.segment(prep, args=TEST_ARGUMENTS, nruns=1)
assert len(segmented) == len(prep)
segmented = ''.join(utt.replace(' ', '').strip() for utt in segmented)
prep = ''.join(utt.replace(' ', '').strip() for utt in prep)
assert segmented == prep
def test_ignore_first_parses(prep, ignore):
# we use the default test value -n 10 -x 2 (10 iterations yields to 6
# parses, initial one and 5 each 2 iterations)
if ignore < 6:
segmented = ag.segment(prep,
args=TEST_ARGUMENTS,
nruns=1,
ignore_first_parses=ignore)
assert len(segmented) == len(prep)
else:
# ignoring more than the extracted parses raises an error
with pytest.raises(RuntimeError):
ag.segment(prep,
args=TEST_ARGUMENTS,
nruns=1,
ignore_first_parses=ignore)
def test_grammars(prep, grammar, level):
grammar = os.path.join(GRAMMAR_DIR, grammar)
segmented = ag.segment(prep, grammar, level, TEST_ARGUMENTS, nruns=1)
assert len(segmented) == len(prep)
segmented = ''.join(utt.replace(' ', '').strip() for utt in segmented)
prep = ''.join(utt.replace(' ', '').strip() for utt in prep)
assert segmented == prep
def test_traintext_equal_testtext(prep, grammar, level):
grammar = os.path.join(GRAMMAR_DIR, grammar)
segmented = ag.segment(prep,
train_text=prep,
grammar_file=grammar,
category=level,
args=TEST_ARGUMENTS,
nruns=1)
assert len(segmented) == len(prep)
segmented = ''.join(utt.replace(' ', '').strip() for utt in segmented)
prep = ''.join(utt.replace(' ', '').strip() for utt in prep)
assert segmented == prep
def test_traintext_notequal_testtext(grammar, level):
# hello world
train_text = ['hh ax l ow w er l d'] * 10
# good morn, that dog is big (no phones shared with train)
test_text = ['g uh d m ao r n', 'dh ae t d ao g ih z b ih g']
grammar = os.path.join(GRAMMAR_DIR, grammar)
segmented = ag.segment(test_text,
train_text=train_text,
grammar_file=grammar,
category=level,
args=TEST_ARGUMENTS,
nruns=1)
assert len(segmented) == len(test_text)
segmented = ''.join(utt.replace(' ', '').strip() for utt in segmented)
prep = ''.join(utt.replace(' ', '').strip() for utt in test_text)
assert segmented == prep
sys.stdout.write(
'* Statistics\n\n' +
json.dumps(stats, indent=4) + '\n')
# prepare the input for segmentation
prepared = list(prepare(text))
# generate the gold text
gold = list(gold(text))
# segment the prepared text with different algorithms
segmented_baseline = baseline.segment(prepared, probability=0.2)
segmented_tp = tp.segment(prepared, threshold='relative')
segmented_puddle = puddle.segment(prepared, njobs=4, window=2)
segmented_dpseg = dpseg.segment(prepared, nfolds=1, args='--randseed 1')
segmented_ag = ag.segment(prepared, nruns=4, njobs=4, args='-n 100')
# we must provide a trained model to dibs (with stats on diphones)
model_dibs = dibs.CorpusSummary(text)
segmented_dibs = dibs.segment(prepared, model_dibs)
# evaluate them against the gold file
eval_baseline = evaluate(segmented_baseline, gold, units=prepared)
eval_tp = evaluate(segmented_tp, gold, units=prepared)
eval_puddle = evaluate(segmented_puddle, gold, units=prepared)
eval_dpseg = evaluate(segmented_dpseg, gold, units=prepared)
eval_ag = evaluate(segmented_ag, gold, units=prepared)
eval_dibs = evaluate(segmented_dibs, gold, units=prepared)
# a little function to display score with 4-digits precision