def main(argv):
if len(argv) != 4:
print >> sys.stderr, 'Usage: hoccorpus2posneg INDIR LABEL OUTFN'
return 1
indir, label, outfn = argv[1:]
documents = load_hoccorpus(indir)
posfn, negfn = outfn + '.pos', outfn + '.neg'
with io.open(posfn, 'wt', encoding='utf-8') as posout:
with io.open(negfn, 'wt', encoding='utf-8') as negout:
for d in documents:
for s in d.sentences:
out = posout if label in s.labels else negout
out.write(s.text + u'\n')
def main(argv):
if len(argv) != 4:
print >> sys.stderr, 'Usage: hoccorpus2posneg INDIR LABEL OUTFN'
return 1
indir, label, outfn = argv[1:]
documents = load_hoccorpus(indir)
posfn, negfn = outfn + '.pos', outfn + '.neg'
with io.open(posfn, 'wt', encoding='utf-8') as posout:
with io.open(negfn, 'wt', encoding='utf-8') as negout:
for d in documents:
for s in d.sentences:
out = posout if label in s.labels else negout
out.write(s.text + u'\n')
def main(argv):
if len(argv) != 3:
print >> sys.stderr, 'Usage: hoccorpus2ann INDIR OUTDIR'
return 1
indir, outdir = argv[1:]
if not path.isdir(outdir):
print >> sys.stderr, '%s is not a directory' % outdir
return 1
documents = load_hoccorpus(indir)
for d in documents:
txtout = path.join(outdir, d.id+'.txt')
with io.open(txtout, 'wt', encoding='utf-8') as out:
out.write(d.text)
annout = path.join(outdir, d.id+'.ann')
with io.open(annout, 'wt', encoding='utf-8') as out:
out.write(u'\n'.join(d.to_standoff()))