import logger sys.stdout = logger.Logger(sys.stdout, filename="check_extern.log") cont = True checked = 0 working = 0 nonworking = 0 totalchecked = 0 try: while cont: print i = 0 if len(todo) < 61 and do_all: for pl in wikipedia.allpages(start=start): todo.append(pl) i += 1 if i == 480: break start = todo[len(todo) - 1].title() + "_0" # todo is a list of pages to do, donow are the pages we will be doing in this run. if len(todo) > 60: # Take the first 60. donow = todo[0:60] todo = todo[60:] else: donow = todo # If there was more to do, the 'if len(todo)<61' part would have extended # todo beyond this size. cont = False
if globalvar.log: import logger sys.stdout = logger.Logger(sys.stdout, filename = 'treelang.log') unequal.read_exceptions() if skipfile: for pl in wikipedia.PageLinksFromFile(skipfile): globalvar.skip[pl] = None if start: if number: print "Treating %d pages starting at %s" % (number, start) i = 0 for pl in wikipedia.allpages(start = start): sa.add(pl) i += 1 if i >= number: break else: print "Treating pages starting at %s" % start sa.setGenerator(wikipedia.allpages(start = start)) inname = '_'.join(inname) if sa.isDone() and not inname: inname = raw_input('Which page to check:') if inname: inpl = wikipedia.PageLink(wikipedia.mylang, inname) sa.add(inpl, hints = hints)
if log: import logger sys.stdout = logger.Logger(sys.stdout, filename='check_extern.log') cont = True checked = 0 working = 0 nonworking = 0 totalchecked = 0 try: while cont: print i = 0 if len(todo) < 61 and do_all: for pl in wikipedia.allpages(start=start): todo.append(pl) i += 1 if i == 480: break start = todo[len(todo) - 1].title() + '_0' # todo is a list of pages to do, donow are the pages we will be doing in this run. if len(todo) > 60: # Take the first 60. donow = todo[0:60] todo = todo[60:] else: donow = todo # If there was more to do, the 'if len(todo)<61' part would have extended # todo beyond this size. cont = False
elif arg=='-file': mode=2 elif arg=='-start': mode=3 else: word.append(arg) files={} word=' '.join(word) if mode==1: pages=[word] elif mode==2: pages=open(word).readlines() elif mode==3: pages=wikipedia.allpages(start=word) else: raise "Please specify one of -one -file or -start" for pl in pages: print pl sys.stdout.flush() try: ll=pl.interwiki() except wikipedia.IsRedirectPage: continue except wikipedia.LockedPage: # Can't do anything with a locked page continue except wikipedia.NoPage: print "ERROR: Yikes, does not exist" continue