def get_ending(self,new_word,n=7):
begin=self.nearest_begin(new_word)
words=self.full_words(begin)
words=tools.unique_list(words)
if(len(words)<n):
n=len(words)
nearest_words=knn.nearest_k(new_word,words,k=n)
nearest_words=[distance.decode_digraphs(word_i)
for word_i in nearest_words]
return nearest_words
def build_forms_histogram(filename, forms2basic, hist_size=0):
text = tools.read_text(filename, clean_txt=False)
words = tools.find_words(text)
words = [code_digraphs(word_i) for word_i in words]
forms = [forms2basic[word_i] for word_i in words if (word_i in forms2basic)]
# print(len(forms))
forms = tools.unique_list(forms) # list(forms)
# print(len(forms))
# print(forms2basic)
return build_histogram(forms, laplace_smoothing=True, size=hist_size)
def build_forms_histogram(filename,forms2basic,hist_size=0):
text=tools.read_text(filename,clean_txt=False)
words=tools.find_words(text)
words=[code_digraphs(word_i) for word_i in words]
forms=[ forms2basic[word_i] for word_i in words
if(word_i in forms2basic)]
#print(len(forms))
forms=tools.unique_list(forms)#list(forms)
#print(len(forms))
#print(forms2basic)
return build_histogram(forms,laplace_smoothing=True,size=hist_size)