scripts/parse/parser-mismatch-id.py

#! /usr/bin/python

## This script checks for mismatched brackets in input parse trees.
## The following types of mismatches are supported:
## - total opening/closing bracket number conflict
## - early exceeding closing brackets
##
## This script is built upon no well-known bracketing mishmatch in 
## parsing context.
##
## To do: upgarde to support them
##
## Version 1.0

import sys, optparse


##======================================================================
## main
def main(argv=None):
 if argv is None:
  argv = sys.argv
 
 parser = optparse.OptionParser(usage="%prog <PARSE FILE>" + 
                                      "\nChecks the input parse trees and outputs the line numbers of trees containing mismatched brackets.", version="%prog 1.0")

 (opts, posArgs) = parser.parse_args()
 
 if len(posArgs) < 1:
  parser.error("At least 1 argument is needed!")

 ## opening input parse file
 vParseFileName = posArgs[0]
 if vParseFileName == None:
  print 'Weka pasre file is not provided!'
  sys.exit(2)
 else:
  try:
   vParseFile  = open(vParseFileName, 'r')
  except IOError:
   print 'Can\'t open parse file: ' + vParseFileName
   sys.exit(2)

 
 vTreeCntr = 1
 for vTree in vParseFile:
  vBracketCntr = 0
  for vChr in vTree:
   if vChr == '(':
    vBracketCntr += 1
   elif vChr == ')':
    vBracketCntr -= 1
   ## cheking for exceeding number of closing brackets throughout the tree
   ## It may happen that at some points in a tree number of closing brackets 
   ## exceed that of opening brackets but finally the total number of opening
   ## and closing brackets in tree be equal (e.g. ())(() which is wrong)
   if vBracketCntr < 0:
    print vTreeCntr
    continue
 
  # final check for the equality of numbers of openning and closing brackets
  if vBracketCntr != 0:
   print vTreeCntr
  
  vTreeCntr += 1


##======================================================================
## calling main
if __name__ == "__main__":
 sys.exit(main())