def main(args):
corpus = '.'
outFile = 'lm.txt'
max_order = 3
gtNmin = [1, 1, 1]
for i in xrange(0, len(args)):
if args[i] == '-c':
corpus = args[i + 1]
elif args[i] == '-lm':
outFile = args[i + 1]
elif args[i] == '-o':
max_order = int(args[i + 1])
elif args[i] == '-gt':
nums = args[i + 1].split(',')
gtNmin = map(lambda n: int(n), nums)
if max_order <= 0:
print "Max order must be non-negative"
exit(1)
if len(gtNmin) != max_order:
print 'Using default gtNmin parameter, cause it does not fit the provided max_order'
gtNmin = [1 for i in xrange(0, max_order)]
print 'Parameters:'
print 'corpus:', corpus
print 'outFile:', outFile
print 'max_order:', max_order
print 'gtNmin:', gtNmin
reader = Reader(corpus)
ng_maker = NgramMaker(max_order)
for file in reader:
ng_maker.parse(Normalizer.normalize(file))
gt = GoodTuring(ng_maker.storage(), gtNmin)
ng_storage = gt.storage()
out_file = open(outFile, 'w')
out_file.write('\\data\\\n\n')
for ng_ord in xrange(1, ng_storage.max_order() + 1):
out_file.write("ngram %d=%d\n" %
(ng_ord, ng_storage.distinct_n_grams(ng_ord)))
for ng_ord in xrange(1, ng_storage.max_order() + 1):
out_file.write('\n\\%d-grams:\n' % ng_ord)
for ng in sorted(ng_storage.get_n_grams(ng_ord)):
ngram = ng_storage.get_n_gram(ng)
if ng_ord < ng_storage.max_order():
out_file.write("%.8f\t%s\t0\n" % (ngram.prob, ' '.join(ng)))
else:
out_file.write("%.8f\t%s\n" % (ngram.prob, ' '.join(ng)))
out_file.write('\n\\end\\\n')
out_file.close()
def main(args):
corpus = '.'
outFile = 'lm.txt'
max_order = 3
gtNmin = [1,1,1]
for i in xrange(0, len(args)):
if args[i] == '-c':
corpus = args[i + 1]
elif args[i] == '-lm':
outFile = args[i + 1]
elif args[i] == '-o':
max_order = int(args[i + 1])
elif args[i] == '-gt':
nums = args[i + 1].split(',')
gtNmin = map(lambda n: int(n), nums)
if max_order <= 0:
print "Max order must be non-negative"
exit(1)
if len(gtNmin) != max_order:
print 'Using default gtNmin parameter, cause it does not fit the provided max_order'
gtNmin = [1 for i in xrange(0, max_order)]
print 'Parameters:'
print 'corpus:', corpus
print 'outFile:', outFile
print 'max_order:', max_order
print 'gtNmin:', gtNmin
reader = Reader(corpus)
ng_maker = NgramMaker(max_order)
for file in reader:
ng_maker.parse(Normalizer.normalize(file))
gt = GoodTuring(ng_maker.storage(), gtNmin)
ng_storage = gt.storage()
out_file = open(outFile, 'w')
out_file.write('\\data\\\n\n')
for ng_ord in xrange(1, ng_storage.max_order() + 1):
out_file.write("ngram %d=%d\n" % (ng_ord, ng_storage.distinct_n_grams(ng_ord)))
for ng_ord in xrange(1, ng_storage.max_order() + 1):
out_file.write('\n\\%d-grams:\n' % ng_ord)
for ng in sorted(ng_storage.get_n_grams(ng_ord)):
ngram = ng_storage.get_n_gram(ng)
if ng_ord < ng_storage.max_order():
out_file.write("%.8f\t%s\t0\n" % (ngram.prob, ' '.join(ng)))
else:
out_file.write("%.8f\t%s\n" % (ngram.prob, ' '.join(ng)))
out_file.write('\n\\end\\\n')
out_file.close()
def test_ngrams(self):
maker = NgramMaker(3)
maker.parse(self.text)
# for ng in maker:
# print ng, " : ", maker.at(ng)
self.assertEqual(3, maker.at(('in',)).count)
self.assertEqual(2, maker.at(('the',)).count)
# check that we have exact number of unigrams (don't forget the <s>, </s> and <punc>
self.assertEqual(38 + 3, len(maker.storage().get_n_grams(1)))
