def spell(word):
"""most likely correction for everything up to a double typo"""
w = Word(word)
candidates = (common([word]) or exact([word]) or known([word])
or known(w.typos()) or common(w.double_typos()) or [word])
correction = max(candidates, key=lambda item: NLP_COUNTS[item])
return get_case(word, correction)
def autocorrect(misspelled):
"""most likely correction for everything up to a double typo"""
w = Word(misspelled)
candidates = (common([misspelled]) or exact([misspelled]) or known([misspelled]) or
known(w.typos()) or common(w.double_typos()) or [misspelled])
correction = max(candidates, key=NLP_COUNTS.get)
return get_case(misspelled, correction)
def spell(word):
"""most likely correction for everything up to a double typo"""
w = Word(word)
candidates = (
common([word]) or exact([word]) or known([word]) or known(w.typos())
or common(w.double_typos()) #or [word]
)
correction = '<UNK>'
if candidates:
correction = max(candidates, key=NLP_COUNTS.get)
correction = get_case(word, correction)
return correction
def spell(word, lang_sample, file_format='bz'):
from autocorrect.nlp_parser import parse
from autocorrect.word import Word, common, exact, known, get_case
"""most likely correction for everything up to a double typo"""
if file_format == 'bz':
NLP_WORDS, NLP_COUNTS = parse('big.txt', 'bz')
elif file_format == 'txt':
NLP_WORDS, NLP_COUNTS = parse(lang_sample, 'txt')
w = Word(word)
candidates = (common([word], NLP_WORDS) or exact([word]) or known([word])
or known(w.typos()) or common(w.double_typos()) or [word])
correction = max(candidates, key=NLP_COUNTS.get)
return get_case(word, correction)
def spell(word, language='en'):
"""The language parameter takes into account of the language.
most likely correction for everything up to a double typo"""
if (language == 'en'):
w = Word(word)
candidates = (common([word]) or exact([word]) or known([word])
or known(w.typos()) or common(w.double_typos())
or [word])
correction = max(candidates, key=NLP_COUNTS.get)
return get_case(word, correction)
elif (language == 'bn'):
w = Word(word)
candidates = (common([word]) or exact([word]) or known([word])
or known(w.typos()) or common(w.double_typos())
or [word])
correction = max(candidates, key=NLP_COUNTS_BN.get)
return get_case(word, correction)
else:
raise ValueError("This language is not supported")
def spelltest(tests, verbose=False):
n, bad, unknown, start = 0, 0, 0, time.clock()
for target, incorrect_spellings in tests.items():
for incorrect_spelling in incorrect_spellings.split():
n += 1
w = spell(incorrect_spelling)
if w != target:
bad += 1
if not known([target]):
unknown += 1
if verbose:
print(MSG.format(incorrect_spelling, w, NLP_COUNTS[w],
target, NLP_COUNTS[target]))
return RESULT.format(bad, n, int(100. - 100. * bad / n),
unknown, int(time.clock() - start))
def spelltest(tests, verbose=False):
n, bad, unknown, start = 0, 0, 0, time.clock()
for target, incorrect_spellings in tests.items():
for incorrect_spelling in incorrect_spellings.split():
n += 1
w = spell(incorrect_spelling)
if w != target:
bad += 1
if not known([target]):
unknown += 1
if verbose:
print(
MSG.format(incorrect_spelling, w, NLP_COUNTS[w],
target, NLP_COUNTS[target]))
return RESULT.format(bad, n, int(100. - 100. * bad / n), unknown,
(time.clock() - start))
