def test_two_sentences(self) -> None:
"""Two sentences should be linked by 'mais'."""
self.assertEqual(
"Vous avez de l'expérience mais vous êtes vieux.",
french.join_sentences_properly(
["vous avez de l'expérience", 'vous êtes vieux']))
def test_double_sentences(self) -> None:
"""Internal punctation/capitalization should be kept."""
self.assertEqual(
'Votre recherche est sur la bonne voie mais votre projet est compliqué. \
Il faut attendre.',
french.join_sentences_properly([
'votre recherche est sur la bonne voie',
'votre projet est compliqué. Il faut attendre',
]))
def test_empty_list(self) -> None:
"""Empty list should return empty string."""
self.assertEqual('', french.join_sentences_properly([]))
def test_one_sentence(self) -> None:
"""One sentence list should be stripped, capitalized and punctuated."""
self.assertEqual(
'Bonjour, le monde.',
french.join_sentences_properly(['bonjour, le monde ']))
def _compute_diagnostic_topic_score_and_text(topic, scorers, scoring_project):
"""Create the score and text for a given diagnostic submetric on a given project.
Args:
topic: the diagnostic topic we wish to evaluate
scorers: a list of scorers for the given topic, with their template sentences
scoring_project: the project we want to score
Returns:
the populated subdiagnostic protobuf.
"""
topic_score = 0
topic_weight = 0
min_score = None
max_score = None
max_scorer = None
min_scorer = None
for scorer in scorers:
model = scoring.get_scoring_model(scorer.trigger_scoring_model)
if not model:
logging.error(
'Diagnostic for topic "%s" uses the scoring model "%s" which does not exist.',
diagnostic_pb2.DiagnosticTopic.Name(topic),
scorer.trigger_scoring_model)
continue
try:
score = model.score(scoring_project)
except scoring.NotEnoughDataException:
continue
# Use default weight of 1
weight = scorer.weight or 1
weighted_score = score * weight
topic_score += weighted_score
topic_weight += weight
# Use positive sentence only for scores above average.
if score > _SCORE_AVERAGE:
positive_score = (score - _SCORE_AVERAGE) * weight
if max_score is None or positive_score > max_score:
max_score = positive_score
max_scorer = scorer
# Use negative sentence only for scores below average.
else:
negative_score = (_SCORE_AVERAGE - score) * weight
if min_score is None or negative_score > min_score:
min_score = negative_score
min_scorer = scorer
if not topic_weight:
return None
sub_diagnostic = diagnostic_pb2.SubDiagnostic(
topic=topic, score=round(topic_score / topic_weight * 100 / 3))
sentences = []
def _append_sentence(template):
translated_template = scoring_project.translate_string(template)
sentences.append(
scoring_project.populate_template(translated_template))
# Do not put positive sentence if score is below 40.
if max_scorer and sub_diagnostic.score > 40:
_append_sentence(max_scorer.positive_sentence_template)
# Do not put negative sentence if score is above 80.
if min_scorer and sub_diagnostic.score < 80:
_append_sentence(min_scorer.negative_sentence_template)
sub_diagnostic.text = french.join_sentences_properly(sentences)
return sub_diagnostic