scripts/parse/map-berk-parses.py

#! /usr/bin/python

## This script changes the syntactic tags in the input syntax trees to 
## tags from another tag set according to the specified mapping.
##
## Mappings are defined in berekelyparses.py.
##
## Current version: 1.0

import sys, optparse, berkeleyparses


##======================================================================
## main
def main(argv=None):
 if argv is None:
  argv = sys.argv
 
 parser = optparse.OptionParser(usage="%prog <INPUT PARSE FILE>" + 
                                      "\nChanges the syntactic tags in the input syntax trees to tags from another tag set according to the specified mapping.", version="%prog 1.0")
 
 parser.add_option("-m", "--mapping", help="mapping to be used (1: FTB-trained Berkeley to a common tag set over itself and frenchFactored-trained Stanford)", metavar="MAPPING", dest="mapping", default='1', action="store")
 parser.add_option("-t", "--toptag", help="top node tag to be used in output (unchanged if not specified)", metavar="OUTPUT TOP NODE TAG", dest="outToptag", action="store")

 # processing input arguments

 (opts, posArgs) = parser.parse_args()
 
 if len(posArgs) < 1:
  parser.error("At least 1 arguments are required")
 
 # opening input parse file
 try:
  vfInput  = open(posArgs[0])
 except IOError:
  sys.exit('Can\'t open input parse file: ' + posArgs[0])


 # processing options
 if opts.outToptag == None:
  vChangeTopNode = False
 else:
  vChangeTopNode = True
 
 
 for vTree in vfInput:
  if vTree.strip('\n ') == '(())':      # treating Berkely unparsed sentences
   print vTree.strip('\n')
  elif len(vTree.strip()) != 0:
   berkeleyparses.mapToCommonTagSet(vTree, opts.mapping, vChangeTopNode, opts.outToptag, True)

 
 vfInput.close()

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