import getopt
import binarize
if __name__ == '__main__':
import sys
import config
try:
opts, args = getopt.getopt(sys.argv[1:], "nul:")
except getopt.GetoptError:
print "\nUsage: binarize.py [-nu] [-llogfile] rule_file config_file > output\n"
print "-n: not binarize\n"
sys.exit(1)
if len(args) < 2:
print "\nUsage: binarize.py [-nu] [-llogfile] rule_file config_file > output\n"
print "-n: not binarize\n"
print "-u: remove unary nodes\n"
sys.exit(1)
sLogs = None
bBinarize = True
bRemoveUnary = False
for opt in opts:
if opt[0] == '-n':
bBinarize = False
elif opt[0] == '-l':
sLogs = opt[1]
elif opt[0] == '-u':
bRemoveUnary = True
rule = binarize.CBinarizer(args[0], sLogs, bBinarize, bRemoveUnary)
cf = config.CConfig(args[1])
fiditer.sentence_iterator(rule.process, cf.directory, cf.range)
return node
else:
node.type = "token"
sToken, nIndex = __find_token(sSentence, nIndex)
node.token = sToken
assert sSentence[nIndex] == ")"
node.end_index = nIndex
return node
#================================================================
def fromString(sSentence):
if sSentence.strip() == "":
return "Empty sentence"
return __find_node(sSentence, 0)
#================================================================
if __name__ == '__main__':
import fiditer
import config
cf = config.CConfig("fid.config")
def printx(x):
print fromString(x)
fiditer.sentence_iterator(printx, cf.directory, cf.range)