def test_terminology_matcher(store0, terminology0):
for store in terminology0.stores.all():
for unit in store.units.all():
terminology.get(unit.__class__)(unit).stem()
unit = store0.units.first()
matcher = terminology_matcher.get(unit.__class__)(unit)
assert matcher.text == unit.source_f
assert (
matcher.split(matcher.text)
== re.split(u"[\W]+", matcher.text))
assert (
matcher.tokens
== [t.lower()
for t
in matcher.split(matcher.text)
if (len(t) > 2
and t not in matcher.stopwords)])
assert matcher.stems == set(matcher.stemmer(t) for t in matcher.tokens)
assert (
matcher.matches
== matcher.similar(
matcher.terminology_units.filter(
stems__root__in=matcher.stems).distinct()))
unit.source_f = "on the cycle home"
unit.save()
matches = []
matched = []
results = matcher.terminology_units.filter(
stems__root__in=matcher.stems).distinct()
for result in results:
target_pair = (
result.source_f.lower().strip(),
result.target_f.lower().strip())
if target_pair in matched:
continue
similarity = matcher.comparison.similarity(result.source_f)
if similarity > matcher.similarity_threshold:
matches.append((similarity, result))
matched.append(target_pair)
assert (
matcher.similar(results)
== sorted(matches, key=lambda x: -x[0])[:matcher.max_matches])
assert (
matcher.matches
== matcher.similar(results))
def get_terminology(self):
"""get terminology suggestions"""
results = terminology_matcher.get(self.__class__)(self).matches
return [m[1] for m in results]
def get_terminology(self):
"""get terminology suggestions"""
results = terminology_matcher.get(self.__class__)(self).matches
return [m[1] for m in results]
