def get_parts_of_speech(word):
"""
Returns a list strings describing parts of speech the given russian word could be.
The enums are derived from pymorphy2.
>>> get_parts_of_speech('рогалик')
['NOUN']
>>> get_parts_of_speech('постовой')
['ADJF', 'NOUN']
>>> 'NOUN' in get_parts_of_speech('правил')
True
>>> 'ADJF' in get_parts_of_speech('правил')
False
:param word: a russian word
:return: list of pymorphy2 POS enums.
"""
met = set()
ret = []
for p in morph.parse(word):
# if p.score < .1:
# continue
pos = p.tag.POS
if pos not in met:
ret.append(pos)
met.add(pos)
return ret
def get_initial_forms(form: str, part_filter=None)->list:
"""
Gets all possible initial forms (there are several of them sometimes) of a given word.
Optional argument part_filter allows to prune unnecessary ambiguity with part of speech.
>>> get_initial_forms('Дядя')
['дядя']
>>> get_initial_forms('самых')
['самый']
>>> get_initial_forms('честных')
['честной', 'честный']
>>> get_initial_forms('правил')
['правило', 'править']
>>> get_initial_forms('правил', 'NOUN')
['правило']
>>> get_initial_forms('правил', ['VERB'])
['править']
:param form: a russian word
:param part_filter: something that supports `in' operator: str, list, set etc. If it is a container,
it should contain only Part-of-speech names according to pymorphy2 enumerations
:return: a list of possible initial forms of the given word in lowercase.
It's guaranteed that there are no repetitions.
Variants are generated in the order of descending certainty.
"""
met = set()
ret = []
for p in morph.parse(form):
# if p.score < .1:
# continue
if part_filter is None or p.tag.POS in part_filter:
norm = p.normal_form
if norm not in met:
ret.append(norm)
met.add(norm)
return ret
def get_valid_noun_initial_form(word: str)->str:
possible_forms = [p for p in morph.parse(word) if _is_valid_noun(p)]
if len(possible_forms) == 0:
return None
else:
return possible_forms[0].normal_form