def charify(line):
text = line.txt
if text is not None:
chars = list(normalize_tweet(text))
line.chars = chars
return chars
else:
return None
def charify(line):
text = line.txt
if text is not None:
chars = list(normalize_tweet(text))
line.chars = chars
return chars
else:
return None
def preprocess_semeval(infile, outfile, vocab=None, label_dict=None):
tweets = load_from_tsv(infile, subtask_id='a')
tweets = filter(lambda t: t.label != 'neutral', tweets)
print('loaded %d tweets from %s' % (len(tweets), infile))
texts = map(lambda t: t.raw_text, tweets)
labels = map(lambda t: t.label, tweets)
mk_label_map = False
if label_dict is None:
mk_label_map = True
label_set = list(set(labels))
label_dict = {x: label_set.index(x) for x in label_set}
label_ints = map(lambda t: label_dict[t.label], tweets)
mk_vocab = False
if vocab is None:
mk_vocab = True
vocab = defaultdict(int)
ntexts = []
ct = 0
for text in texts:
ntext = normalize_tweet(text)
ntexts.append(ntext)
if mk_vocab:
chars = list(ntext)
for c in chars:
if c not in vocab:
vocab[c] = ct
ct += 1
print(vocab)
print
lines = []
for i, text in enumerate(ntexts):
ints = map(lambda c: vocab[c], text)
ints_str = ' '.join([str(c) for c in ints])
label = str(label_ints[i])
line = label + "\t" + ints_str
lines.append(line)
if mk_vocab:
vocab_filename = '%s.vocab.pkl' % outfile
print('writing vocab to %s' % vocab_filename)
cPickle.dump(vocab, open(vocab_filename, 'w'))
if mk_label_map:
label_filename = '%s.labels.pkl' % outfile
print('writing label map to %s' % label_filename)
cPickle.dump(label_dict, open(label_filename, 'w'))
print('writing output to %s' % outfile)
with open(outfile, 'w') as f:
for line in lines:
f.write(line + '\n')
def preprocess_semeval(infile, outfile, vocab=None, label_dict=None):
tweets = load_from_tsv(infile, subtask_id='a')
tweets = filter(lambda t: t.label != 'neutral', tweets)
print('loaded %d tweets from %s' % (len(tweets), infile))
texts = map(lambda t: t.raw_text, tweets)
labels = map(lambda t: t.label, tweets)
mk_label_map = False
if label_dict is None:
mk_label_map = True
label_set = list(set(labels))
label_dict = {x: label_set.index(x) for x in label_set}
label_ints = map(lambda t: label_dict[t.label], tweets)
mk_vocab = False
if vocab is None:
mk_vocab = True
vocab = defaultdict(int)
ntexts = []
ct = 0
for text in texts:
ntext = normalize_tweet(text)
ntexts.append(ntext)
if mk_vocab:
chars = list(ntext)
for c in chars:
if c not in vocab:
vocab[c] = ct
ct += 1
print(vocab)
print
lines = []
for i, text in enumerate(ntexts):
ints = map(lambda c: vocab[c], text)
ints_str = ' '.join([str(c) for c in ints])
label = str(label_ints[i])
line = label + "\t" + ints_str
lines.append(line)
if mk_vocab:
vocab_filename = '%s.vocab.pkl' % outfile
print('writing vocab to %s' % vocab_filename)
cPickle.dump(vocab, open(vocab_filename, 'w'))
if mk_label_map:
label_filename = '%s.labels.pkl' % outfile
print('writing label map to %s' % label_filename)
cPickle.dump(label_dict, open(label_filename, 'w'))
print('writing output to %s' % outfile)
with open(outfile, 'w') as f:
for line in lines:
f.write(line + '\n')
