# coding=utf-8
from collections import Counter
from plp import PLP
p = PLP()
def basic_form(word):
ids = p.rec(word)
return p.bform(ids[0]) if len(ids) > 0 else word
def stats_sorted(stats):
return sorted(stats, key=lambda x: x[1], reverse=True)
def ranking(words):
basic_forms = [basic_form(word) for word in words]
stats = Counter(basic_forms)
return stats_sorted(stats.items()), stats_sorted(stats.most_common(100))
def getForms(bodziec):
return map(lambda x: PLP().forms(x), PLP().orec(bodziec))[0]
# coding: utf-8
import codecs
from collections import defaultdict
from plp import PLP
__author__ = "Michał Ciołczyk"
_FILENAME = "data/odm.txt"
_ENCODING = "windows-1250"
_basic_forms = defaultdict(list)
_initialized = False
_plp = PLP()
_SIE = ' się'
def _load_flection_map():
global _initialized
if not _initialized:
with codecs.open(_FILENAME, 'r', encoding=_ENCODING) as f:
for line in f:
forms = line.rstrip('\n').split(', ')
bform = forms[0]
for form in forms:
_basic_forms[form].append(bform)
for form, bforms in _basic_forms.items():
_basic_forms[form] = list(set(bforms))
_initialized = True
def _strip_sie(form):
def setUp(self):
self.plp = PLP('/usr/local/clp/lib/libclp_2.6.so')
