def main(args):
set_quiet(args.quiet)
if args.corpus:
unigrams = count_unigrams(args.corpus)
vocab, outfname = decide_vocab(unigrams, args.cutoff,
args.vocab, args.want)
if args.rewrite:
rewrite_corpus(args.corpus, vocab, outfname)
else:
extract_stat(args.corpus, vocab, outfname, args.window)
if args.stat:
assert(args.m is not None and args.kappa is not None)
if args.no_matlab:
C = canon()
C.set_params(args.m, args.kappa)
C.get_stat(args.stat)
C.start_logging()
C.approx_cca()
C.end_logging()
C.write_result()
else:
call_matlab(args.stat, args.m, args.kappa)
if args.clean: clean()
def main(args):
set_quiet(args.quiet)
if args.corpus:
unigrams = count_unigrams(args.corpus)
vocab, outfname = decide_vocab(unigrams, args.cutoff, args.vocab,
args.want)
if args.rewrite:
rewrite_corpus(args.corpus, vocab, outfname)
else:
extract_stat(args.corpus, vocab, outfname, args.window)
if args.stat:
assert (args.m is not None and args.kappa is not None)
if args.no_matlab:
C = canon()
C.set_params(args.m, args.kappa)
C.get_stat(args.stat)
C.start_logging()
C.approx_cca()
C.end_logging()
C.write_result()
else:
call_matlab(args.stat, args.m, args.kappa)
if args.clean: clean()
def check():
unigrams = count_unigrams(corpus)
vocab, outfname = decide_vocab(unigrams, cutoff, None, None)
XYcount, Xcount, Ycount, stat = extract_stat(corpus, vocab,
outfname, window)
for x in Xcount: assert(Xcount[x] == gold_Xcount[x])
for y in Ycount: assert(Ycount[y] == gold_Ycount[y])
for x, y in XYcount: assert(XYcount[x,y] == gold_XYcount[x,y])
return stat
def check():
unigrams = count_unigrams(corpus)
vocab, outfname = decide_vocab(unigrams, cutoff, None, None)
XYcount, Xcount, Ycount, stat = extract_stat(corpus, vocab, outfname,
window)
for x in Xcount:
assert (Xcount[x] == gold_Xcount[x])
for y in Ycount:
assert (Ycount[y] == gold_Ycount[y])
for x, y in XYcount:
assert (XYcount[x, y] == gold_XYcount[x, y])
return stat