def create_word_bucketer_from_profile(bucket_type,
freq_counts=None,
freq_count_file=None,
freq_corpus_file=None,
freq_data=None,
label_set=None,
bucket_cutoffs=None,
case_insensitive=False):
if type(bucket_cutoffs) == str:
bucket_cutoffs = [
arg_utils.parse_intfloat(x) for x in bucket_cutoffs.split(':')
]
if bucket_type == 'freq':
return FreqWordBucketer(freq_counts=freq_counts,
freq_count_file=freq_count_file,
freq_corpus_file=freq_corpus_file,
freq_data=freq_data,
bucket_cutoffs=bucket_cutoffs,
case_insensitive=case_insensitive)
if bucket_type == 'case':
return CaseWordBucketer()
elif bucket_type == 'label':
return LabelWordBucketer(label_set=label_set)
elif bucket_type == 'numlabel':
return NumericalLabelWordBucketer(bucket_cutoffs=bucket_cutoffs)
else:
raise ValueError(f'Illegal bucket type {bucket_type}')
def create_sentence_bucketer_from_profile(bucket_type,
score_type=None,
bucket_cutoffs=None,
label_set=None,
case_insensitive=False):
if type(bucket_cutoffs) == str:
bucket_cutoffs = [
arg_utils.parse_intfloat(x) for x in bucket_cutoffs.split(':')
]
if bucket_type == 'score':
return ScoreSentenceBucketer(score_type,
bucket_cutoffs=bucket_cutoffs,
case_insensitive=case_insensitive)
elif bucket_type == 'length':
return LengthSentenceBucketer(bucket_cutoffs=bucket_cutoffs)
elif bucket_type == 'lengthdiff':
return LengthDiffSentenceBucketer(bucket_cutoffs=bucket_cutoffs)
elif bucket_type == 'label':
return LabelSentenceBucketer(label_set=label_set)
elif bucket_type == 'multilabel':
return MultiLabelSentenceBucketer(label_set=label_set)
elif bucket_type == 'numlabel':
return NumericalLabelSentenceBucketer(bucket_cutoffs=bucket_cutoffs)
else:
raise NotImplementedError(f'Illegal bucket type {bucket_type}')