#!/usr/bin/python # dictionary2.py by Nadeem Abdul Hamid, August 2006 # based on code by Saketh Bhamidipati, 2005 # This version uses regular expressions to search and remove # all links from the output file. # Demonstrates Python regular expressions, doc strings """ This program will read in a word list from a file and construct as output an HTML file (web page) containing the definitions of all the words, as found at dictionary.reference.com. """ import re import sys from urllib2 import * def buildDictURL( word ): return 'http://dictionary.reference.com/search?q=%s' % word; def lookupWord( word, outFileHandle ): """ Compile a list of definitions for the given word and write them out, in HTML format, to the output file handle. """ inDef = False # flag to indicate that we are reading in the # definitions portion of the web page and it should # be printed to the output file dict_url = buildDictURL( word ) lines = [] # lines to be output for line in urlopen( dict_url ): if '' in line: inDef = True elif '' in line: inDef = False if inDef: lines.append( line ) output = ''.join( lines ) # concatenates together all the lines, # separated by '' (i.e. nothing) outFileHandle.write( cleanLinks( output ) ); def lookupWords( inFileHandle, outFileHandle ): """ Read word list from input file and write a compilation of definition in HTML format to the output file. """ outFileHandle.write('\n') for word in inFileHandle.readlines(): lookupWord( word, outFileHandle ) outFileHandle.write('\n'); def cleanLinks( string ): """ Remove all hyperlinks from HTML source code and delete 'pron.jpg' images. """ p = re.compile( r'(.+?)', re.DOTALL | re.IGNORECASE ); string = p.sub( r'\1', string ); p = re.compile( r'' ); string = p.sub( '', string ); return string; def main(argv=sys.argv): if len(argv) == 3: try: inFileH = file( argv[1], 'r' ); outFileH = file( argv[2], 'w' ); lookupWords( inFileH, outFileH ); inFileH.close(); outFileH.close(); except IOError, e: print "Sorry, an exception occurred:", e else: # print usage message print \ """ dictionary2.py, 2006 Nadeem Abdul Hamid Usage: dictionary2.py inFileName outFileName """ if __name__ == '__main__': main();