#r'c:\program files\bozzy\bozzycompare\'
#File name: bozzycompare.py
#BozzyCompare, supplement to Bozzy, both by Mirabai Knight
#Final working version: January 15, 2007
import sets
import os
import time
import re
# puts empty line at end of list; otherwise last word of edic will appear in newlist whether matched or not
int_words = file(r'c:\program files\bozzy\bozzycompare\edic.txt','a')
hard_break = '\n'
int_words.write(hard_break)
int_words.close()
# edic = external dictionary; pdic = personal dictionary (only DigitalCAT rtf/cre currently supported)
int_words = file(r'c:\program files\bozzy\bozzycompare\edic.txt','r')
rtf_words = file(r'c:\program files\bozzy\bozzycompare\pdic.rtf','r')
intersec = file(r'c:\program files\bozzy\bozzycompare\newlist.txt','w')
# strips rtf/cre markup to leave only English, one word per line
rtf_words = rtf_words.read()
rtf_words = rtf_words.replace(r'{\rtf1\ansi{\*\cxrev100}\cxdict{\*\cxsystem','')
rtf_words = rtf_words.replace(r'''}\deff0{\fonttbl {\f0\fmodern Courier New;}}{\stylesheet{\s0 Normal;}{\s1 Question;
}{\s2 Answer;}{\s3 Colloquy;}}{\*\cxtranopt\numbar}''','')
rtf_words = rtf_words.replace(r'Digital CAT 00190 RTF by Stenovations,','')
rtf_words = rtf_words.replace(r'Inc.','')
rtf_words = re.sub('\*','',rtf_words)
rtf_words = re.sub(r'\\','',rtf_words)
rtf_words = re.sub('{.*?}','',rtf_words)
rtf_words = re.sub('}','',rtf_words)
rtf_words = rtf_words.lstrip()
# sets work from files but not strings
rtf_manip = file(r'c:\program files\bozzy\bozzycompare\rtfmanip.txt','w')
rtf_manip.write(str(rtf_words))
rtf_manip.close()
rtf_manip = file(r'c:\program files\bozzy\bozzycompare\rtfmanip.txt','r')
# compares edic to pdic; all words in edic that aren't in pdic are written to newlist
rtf_words = set(rtf_manip)
int_words = set(int_words)
intersec_set = int_words.difference(rtf_words)
intersec_set = str(intersec_set)
# formats list so it's Bozzy-compatible, one word per line
intersec_set = intersec_set.replace(r'set([','')
intersec_set = intersec_set.replace(r', ','\n')
intersec_set = intersec_set.replace(r'])','')
intersec_set = intersec_set.replace(r"'","")
intersec_set = intersec_set.replace(r'\n','')
intersec.write(intersec_set)
intersec.close()
rtf_manip.close()
# removes temporary file
os.remove(r'c:\program files\bozzy\bozzycompare\rtfmanip.txt')
# writes newlist to unique filename
os.rename(r'c:\program files\bozzy\bozzycompare\newlist.txt','bc' + time.strftime('%m%d%y%H%M%S') + '.txt')
print 'Done'