# creates a file of word pairs from a simple thesaurus
# works on a plain text file in which all non-blank lines
# must have exactly two fields (words) in them: a wordpair file
# the output is a thesaurus file in which synonyms are sorted
# note: the input file does not have to be sorted, therefore
# this is not exactly an inverse to the related program
# ths2prs.py since that program produces a sorted file
# also in ths2prs the synonyms on each line do not have
# to be sorted
#
# usage: python prs2ths.py wordpair-file thesaurus-file
#
# note: written to be read and maintained, not to be
# short and sweet -- as all good programs should be
#
# prs2ths.py is Copyright (C) 2010 Douglas Pardoe Wilson
# This program comes with ABSOLUTELY NO WARRANTY
# This is free software, and you are welcome to redistribute it
# subject to the conditions of the GNU General Public License (GPL)

import sys
import string

comlineargs = sys.argv
argcount = len(comlineargs)
# print "command line argument count was " + `argcount`
# print "command line: "
# print comlineargs
if len(comlineargs) != 3:
   print "usage: python ths2prs.py thesaurus-infile wordpair-outfile"
else:
   infile = open(comlineargs[1], "r")
   outfile = open(comlineargs[2], "w")
   wordpairlines = list()
   while 1:  # loop indefinitely, but will break out when no line found
      thisline = infile.readline()
      if thisline:
         pairline = string.strip(thisline)
         pair = string.split(pairline)
         if len(pair) != 2:
            print "error, not a pair ", thisline
            break
         wordpairlines.append(pairline)
      else:
         break
   infile.close()
   wordpairlines.sort()
   print len(wordpairlines), " lines read"
   outlinecount = 0
   wordpairs = list()
   if len(wordpairlines) > 0:
      for pairline in wordpairlines:
         pair = string.split(pairline)
         wordpairs.append(pair)
      lastlemma = ""
      outstring = ""
      for pair in wordpairs:         
         if len(pair) != 2:
            print "error, not a pair ", pairline
            break
         lemma = pair[0]
         if len(lemma) < 1:
            print "error, null lemma"
            break
         synonym = pair[1]
#         print lastlemma, lemma
         if lemma != lastlemma:
            if len(outstring) > 0:
               justchecking = string.split(outstring)
               if len(justchecking) < 2:
                  print "error, output less than a pair: ", outstring
                  break
               outfile.write(outstring + "\n") # note: writes syns for previous lemma
               outlinecount = outlinecount + 1
            lastlemma = lemma
            outstring = lemma + " " + synonym
         else:
            outstring = outstring + " " + synonym
      outfile.write(outstring) # must write syns for last lemma here
      outlinecount = outlinecount + 1
   outfile.close()
   print `outlinecount`, " lines written"