def from_opencorpora_int(open_tag):
"""
Convert OpenCorpora tag to www.ruscorpora.com tag::
>>> print(from_opencorpora_int('NOUN,inan,masc sing,nomn'))
S,inan,m=sg,nom
"""
# Whitespace is replaced with ",|,"
# then "|" is treated as token and replaced with "=",
# then commas around "=" are removed in result.
# This way space is converted to "=".
grammeme_list = open_tag.replace(' ', ',|,').split(',')
if _is_initials(grammeme_list):
return 'INIT=abbr'
result = rule_engine.apply_rules(FROM_OPENCORPORA, grammeme_list)
result = ','.join(result).replace(',=,', '=').replace(',=', '')
if result == '':
return 'NONLEX'
return result
def _transform_tag(tag, **kwargs):
rules = rule_engine.parse(RULES2)
return ','.join(
rule_engine.apply_rules(rules, tag.split(','), **kwargs)
)
def _transform_tag(tag, **kwargs):
rules = rule_engine.parse(RULES2)
return ','.join(rule_engine.apply_rules(rules, tag.split(','), **kwargs))
