def generate_typos(word, distance, d=0):
yield (word, d)
if d == distance:
return
# add letter
for pos in range(len(word)):
for letter in ALPHABET:
yield from generate_typos(word[:pos] + letter + word[pos:], distance, d + 1)
# remove letter
for pos in range(len(word)):
yield from generate_typos(word[:pos] + word[pos + 1:], distance, d + 1)
# change letter
for pos in range(len(word)):
for letter in generate_letter_typos(word[pos]):
yield from generate_typos(word[:pos] + letter + word[pos + 1:], distance, d + 1)
# swap letters
for pos in range(len(word) - 1):
yield from generate_typos(word[:pos] + word[pos + 1] + word[pos] + word[pos + 2:], distance, d + 1)
if __name__ == '__main__':
grams1 = grams.load_grams('../1grams_cleaned', 1)
pool = multiprocessing.Pool()
data = pool.map(possible_typos, grams1[0])
with open('typos.dat', 'wb') as f:
pickle.dump(data, f)
# remove letter
for pos in range(len(word)):
yield from generate_typos(word[:pos] + word[pos + 1:], distance, d + 1)
# change letter
for pos in range(len(word)):
for letter in ALPHABET:
if letter != word[pos]:
yield from generate_typos(word[:pos] + letter + word[pos + 1:], distance, d + 1)
# swap letters
for pos in range(len(word) - 1):
yield from generate_typos(word[:pos] + word[pos + 1] + word[pos] + word[pos + 2:], distance, d + 1)
def fix_line(line):
return fix_typos(line.strip(), dictionary, unigrams, bigrams)
if __name__ == '__main__':
#generate_dictionary('../slownik_do_literowek.txt')
dictionary = load_dictionary('../slownik_do_literowek.dat')
unigrams = [grams.load_grams('../1grams_min_cleaned', 1)]
bigrams = [grams.load_grams('../2grams_min_cleaned', 2)]
unigrams.append(sum(unigrams[0][3][0]))
bigrams.append(sum(bigrams[0][3][0]))
print('Loaded!')
pool = multiprocessing.Pool()
print('\n'.join(pool.map(fix_line, sys.stdin)))
if a in ALT and ALT[a] == b:
break
elif a in REVALT and REVALT[a] == b:
break
elif i + 1 < len(word) and (a, word[i + 1]) == (typo[i + 1],
b):
trans_ab[(a, b)] += count
break
else:
break
grams1 = grams.load_grams('../1grams_cleaned', 1)
paired = defaultdict(lambda: 0)
for word in grams1[0]:
if word not in DICTIONARY:
continue
count = grams.find_ngram([word], *grams1)
for i in range(len(word)):
if i + 1 < len(word):
paired[(word[i], word[i + 1])] += count
for (key, value) in trans_ab.items():
trans_ab[key] = value / (value + paired[key])
for (key, value) in sorted(trans_ab.items(), key=lambda x: -x[1]):
print(''.join(key) + ':', value)
from collections import defaultdict
import grams
import random
import re
import sys
WORD = 1
TAG = 2
WORDS = {}
WORDS2 = defaultdict(lambda: dict())
TAGS = defaultdict(lambda: list())
grams1 = grams.load_grams('../1grams_min_cleaned', 1)
grams2 = grams.load_grams('../2grams_min_cleaned', 2)
for idx in range(len(grams1[3][0])):
WORDS[grams1[0][grams1[3][1][idx]]] = grams1[3][0][idx]
for idx in range(len(grams2[3][0])):
WORDS2[grams2[0][grams2[3][1][idx]]][grams2[0][grams2[3][2][idx]]] = grams2[3][0][idx]
regex = re.compile(r'[^\w ]', re.UNICODE | re.IGNORECASE)
with open('../morfeuszTagsAndBasesForNKJP.txt', 'r') as f:
for line in f:
word, base, *tags = line.strip().split()
word = regex.sub('', word).strip()
if word not in WORDS:
continue
for tag in tags:
TAGS[tag].append(word)
#!/bin/env python3
import grams
import random
import sys
words, words_index, words_position, connections, connections_index = grams.load_grams(
'../2grams', 2)
def upper_bound(val, arr, s=0, e=None, key=lambda x: x):
if e is None:
e = len(arr)
while s < e:
mid = (s + e) // 2
if val >= key(arr[mid]):
s = mid + 1
else:
e = mid
return s
def choose_simple(i, s, e):
if s == e:
return None
cid = connections_index[random.randint(s, e - 1)]
wid = connections[i + 1][cid]
return wid
#!/bin/env python3
import grams
import itertools
import math
import random
WINDOW = 2
words, words_index, words_position, connections, connections_index = grams.load_grams(
'../{}grams'.format(WINDOW), WINDOW)
cache = {}
def process_sentence(sentence, done=None):
if done is None:
done = []
if not sentence:
yield done
return
for i in range(len(sentence)):
if sentence[i].startswith('-'):
break
done.append(sentence[i])
else:
yield done
return
#!/bin/env python3
import grams
import random
import sys
words, words_index, words_position, connections, connections_index = grams.load_grams('../2grams', 2)
def upper_bound(val, arr, s=0, e=None, key=lambda x: x):
if e is None:
e = len(arr)
while s < e:
mid = (s + e) // 2
if val >= key(arr[mid]):
s = mid + 1
else:
e = mid
return s
def choose_simple(i, s, e):
if s == e:
return None
cid = connections_index[random.randint(s, e - 1)]
wid = connections[i+1][cid]
return wid
def choose_ranked(i, s, e):
if s == e:
#!/bin/env python3
import grams
import itertools
import random
WINDOW = 2
words, words_index, words_position, connections, connections_index = grams.load_grams('../{}grams'.format(WINDOW), WINDOW)
def upper_bound(val, arr, s=0, e=None, key=lambda x: x):
if e is None:
e = len(arr)
while s < e:
mid = (s + e) // 2
if val >= key(arr[mid]):
s = mid + 1
else:
e = mid
return s
def find_ngram(ngram):
s = 0
e = len(connections_index)
for i, word in enumerate(ngram):
word_id = words_index[upper_bound(word, words_index, key=lambda idx: words[idx]) - 1]
if words[word_id] != word:
return 0
word_position = words_position[word_id]