def main(args):
if os.path.isdir(args.input):
input_path = os.path.join(args.input, 'PragerFeats')
else:
input_path = args.input
if args.output:
output_path = args.output
else:
output_path = input_path + '.scanner'
# display paths
logger.info("input path: {0}".format(input_path))
logger.info("output path: {0}".format(output_path))
nb_features = read_features(input_path)
tk_nextmove, tk_output = build_scanner(nb_features)
scanner = tk_nextmove, tk_output, nb_features
with open(output_path, 'w') as f:
cPickle.dump(scanner, f)
logger.info("wrote scanner to {0}".format(output_path))
args = parser.parse_args()
if not (args.domain or args.lang) or (args.domain and args.lang):
parser.error("exactly one of domain(-d) or language (-l) must be specified")
if args.features:
feature_path = args.features
else:
feature_path = os.path.join(args.model, "DFfeats")
bucketlist_path = os.path.join(args.model, "bucketlist")
if not os.path.exists(feature_path):
parser.error("{0} does not exist".format(feature_path))
bucketlist = map(str.strip, open(bucketlist_path))
features = read_features(feature_path)
if args.domain:
index_path = os.path.join(args.model, "domain_index")
suffix = ".domain"
elif args.lang:
index_path = os.path.join(args.model, "lang_index")
suffix = ".lang"
else:
raise ValueError("no event specified")
if args.weights:
weights_path = args.weights
else:
weights_path = os.path.join(args.model, "IGweights" + suffix + (".bin" if args.binarize else ""))
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument("input", metavar="INPUT", help="build a scanner for INPUT. If input is a directory, read INPUT/LDfeats")
parser.add_argument("-o","--output", help="output scanner to OUTFILE", metavar="OUTFILE")
args = parser.parse_args()
if os.path.isdir(args.input):
input_path = os.path.join(args.input, 'LDfeats')
else:
input_path = args.input
if args.output:
output_path = args.output
else:
output_path = input_path + '.scanner'
# display paths
if not SILENT:
print "input path:", input_path
print "output path:", output_path
nb_features = read_features(input_path)
tk_nextmove, tk_output = build_scanner(nb_features)
scanner = tk_nextmove, tk_output, nb_features
with open(output_path, 'w') as f:
cPickle.dump(scanner, f)
if not SILENT:
print "wrote scanner to {0}".format(output_path)
if args.output:
makedir(args.output)
out_dir = args.output
else:
out_dir = model_dir
langs = sorted(all_langs)
# display paths
print "languages({1}): {0}".format(langs, len(langs))
print "model path:", model_dir
print "feature path:", feat_path
print "output path:", out_dir
print "temp (buckets) path:", buckets_dir
feats = read_features(feat_path)
indexer = CorpusIndexer(args.corpus, langs = langs)
items = [ (d,l,p) for (d,l,n,p) in indexer.items ]
if len(items) == 0:
raise ValueError("found no files!")
print "will process {0} features across {1} paths".format(len(feats), len(items))
# produce a scanner over all the features
tk_nextmove, tk_output = build_scanner(feats)
# Generate a class map over all the languages we are dealing with
cm = generate_cm([ (l,p) for d,l,p in items], len(langs))
# Compute P(t|C)
"exactly one of domain(-d) or language (-l) must be specified")
if args.features:
feature_path = args.features
else:
feature_path = os.path.join(args.model, 'DFfeats')
if args.buckets:
bucketlist_paths = args.buckets
else:
bucketlist_paths = [os.path.join(args.model, 'bucketlist')]
if not os.path.exists(feature_path):
parser.error('{0} does not exist'.format(feature_path))
features = read_features(feature_path)
if args.domain:
index_path = os.path.join(args.model, 'domain_index')
suffix = '.domain'
elif args.lang:
index_path = os.path.join(args.model, 'lang_index')
suffix = '.lang'
else:
raise ValueError("no event specified")
if args.weights:
weights_path = args.weights
else:
weights_path = os.path.join(
args.model,
with open(index_path, "w") as f:
writer = csv.writer(f)
writer.writerows(items)
if args.temp:
buckets_dir = args.temp
else:
buckets_dir = os.path.join(model_dir, "buckets")
makedir(buckets_dir)
bucketlist_path = os.path.join(model_dir, "bucketlist")
index_path = os.path.join(model_dir, "paths")
if args.ld_feats:
# LD features are pre-specified. We are basically just building the NB model.
LDfeats = read_features(args.ld_feats)
else:
# LD features not pre-specified, so we compute them.
# Tokenize
DFfeats = None
print "will tokenize %d files" % len(items)
# TODO: Custom tokenizer if doing custom first-pass features
if args.df_feats:
print "reading custom features from:", args.df_feats
DFfeats = read_features(args.df_feats)
print "building tokenizer for custom list of {0} features".format(len(DFfeats))
tk = Scanner(DFfeats)
elif args.word:
print "using word tokenizer"
if args.output:
makedir(args.output)
out_dir = args.output
else:
out_dir = model_dir
langs = sorted(all_langs)
# display paths
print "languages({1}): {0}".format(langs, len(langs))
print "model path:", model_dir
print "feature path:", feat_path
print "output path:", out_dir
print "temp (buckets) path:", buckets_dir
feats = read_features(feat_path)
indexer = CorpusIndexer(args.corpus, langs = langs)
items = [ (d,l,p) for (d,l,n,p) in indexer.items ]
if len(items) == 0:
raise ValueError("found no files!")
print "will process {0} features across {1} paths".format(len(feats), len(items))
# produce a scanner over all the features
tk_nextmove, tk_output = build_scanner(feats)
# Generate a class map over all the languages we are dealing with
cm = generate_cm([ (l,p) for d,l,p in items], len(langs))
# Compute P(t|C)
help=
"build a scanner for INPUT. If input is a directory, read INPUT/LDfeats"
)
parser.add_argument("-o",
"--output",
help="output scanner to OUTFILE",
metavar="OUTFILE")
args = parser.parse_args()
if os.path.isdir(args.input):
input_path = os.path.join(args.input, 'LDfeats')
else:
input_path = args.input
if args.output:
output_path = args.output
else:
output_path = input_path + '.scanner'
# display paths
print "input path:", input_path
print "output path:", output_path
nb_features = read_features(input_path)
tk_nextmove, tk_output = build_scanner(nb_features)
scanner = tk_nextmove, tk_output, nb_features
with open(output_path, 'w') as f:
cPickle.dump(scanner, f)
print "wrote scanner to {0}".format(output_path)
with open(index_path, 'w') as f:
writer = csv.writer(f)
writer.writerows(items)
if args.temp:
buckets_dir = args.temp
else:
buckets_dir = os.path.join(model_dir, 'buckets')
makedir(buckets_dir)
bucketlist_path = os.path.join(model_dir, 'bucketlist')
index_path = os.path.join(model_dir, 'paths')
if args.ld_feats:
# LD features are pre-specified. We are basically just building the NB model.
LDfeats = read_features(args.ld_feats)
else:
# LD features not pre-specified, so we compute them.
# Tokenize
DFfeats = None
print "will tokenize %d files" % len(items)
# TODO: Custom tokenizer if doing custom first-pass features
if args.df_feats:
print "reading custom features from:", args.df_feats
DFfeats = read_features(args.df_feats)
print "building tokenizer for custom list of {0} features".format(
len(DFfeats))
tk = Scanner(DFfeats)
elif args.word: