def write_sequence(out, sequence):
num_features = len(dicts)
is_beginning = True
for features in sequence:
assert len(features) == num_features, \
"Wrong number of features: " + line
sample = DataFormat.DataSample()
for i in xrange(num_original_columns):
id = dicts[i].get(features[i], -1)
if id != -1:
sample.id_slots.append(id)
elif oov_policy[i] == OOV_POLICY_IGNORE:
sample.id_slots.append(0xffffffff)
elif oov_policy[i] == OOV_POLICY_ERROR:
logger.fatal("Unknown token: %s" % features[i])
else:
sample.id_slots.append(0)
if patterns:
dim = 0
vec = sample.vector_slots.add()
for i in xrange(num_original_columns, num_features):
id = dicts[i].get(features[i], -1)
if id != -1:
vec.ids.append(dim + id)
elif oov_policy[i] == OOV_POLICY_IGNORE:
pass
elif oov_policy[i] == OOV_POLICY_ERROR:
logger.fatal("Unknown token: %s" % features[i])
else:
vec.ids.append(dim + 0)
dim += len(dicts[i])
sample.is_beginning = is_beginning
is_beginning = False
write_proto(out, sample)
@return True success, False for end of file
"""
buf = file.read(8)
if not buf:
return False
result, pos = _DecodeVarint(buf, 0)
buf = buf[pos:] + file.read(result - len(buf) + pos)
message.ParseFromString(buf)
return True
def usage():
print >> sys.stderr, "Usage: python show_pb.py PROTO_DATA_FILE"
exit(1)
if __name__ == '__main__':
if len(sys.argv) < 2:
usage()
f = open(sys.argv[1])
header = DataFormat.DataHeader()
read_proto(f, header)
print header
sample = DataFormat.DataSample()
while read_proto(f, sample):
print sample
def gen_proto_file(
input_file,
dicts,
oov_policy,
output_file):
def write_sequence(out, sequence):
num_features = len(dicts)
is_beginning = True
for features in sequence:
assert len(features) == num_features, \
"Wrong number of features: " + line
sample = DataFormat.DataSample()
for i in xrange(num_original_columns):
id = dicts[i].get(features[i], -1)
if id != -1:
sample.id_slots.append(id)
elif oov_policy[i] == OOV_POLICY_IGNORE:
sample.id_slots.append(0xffffffff)
elif oov_policy[i] == OOV_POLICY_ERROR:
logger.fatal("Unknown token: %s" % features[i])
else:
sample.id_slots.append(0)
if patterns:
dim = 0
vec = sample.vector_slots.add()
for i in xrange(num_original_columns, num_features):
id = dicts[i].get(features[i], -1)
if id != -1:
vec.ids.append(dim + id)
elif oov_policy[i] == OOV_POLICY_IGNORE:
pass
elif oov_policy[i] == OOV_POLICY_ERROR:
logger.fatal("Unknown token: %s" % features[i])
else:
vec.ids.append(dim + 0)
dim += len(dicts[i])
sample.is_beginning = is_beginning
is_beginning = False
write_proto(out, sample)
num_features = len(dicts)
f = open(input_file, 'rb')
out = open(output_file, 'wb')
header = DataFormat.DataHeader()
if patterns:
slot_def = header.slot_defs.add()
slot_def.type = DataFormat.SlotDef.VECTOR_SPARSE_NON_VALUE
slot_def.dim = sum([len(dicts[i])
for i in xrange(num_original_columns, len(dicts))])
logger.info("feature_dim=%s" % slot_def.dim)
for i in xrange(num_original_columns):
slot_def = header.slot_defs.add()
slot_def.type = DataFormat.SlotDef.INDEX
slot_def.dim = len(dicts[i])
write_proto(out, header)
num_sequences = 0
sequence = []
for line in f:
line = line.strip()
if not line:
make_features(sequence)
write_sequence(out, sequence)
sequence = []
num_sequences += 1
continue
features = line.split(' ')
sequence.append(features)
f.close()
out.close()
logger.info("num_sequences=%s" % num_sequences)