Python:Create redirect accent.py

# -*- coding: utf-8  -*-
"""
Bot to create redirects. This version create redirect to title containing accented letters and various special chars
 
do NOT uses this version it with -start:, as it would currently change any page name
containing a negative number (which we do not want to effect). Plus it'd be a huge waste of resources.
instead, feed it a file ... basic example of SQL query to create said file : 
echo "SELECT CONCAT('[[', page_title, ']]')  FROM page  WHERE page_namespace=0   AND page_is_redirect=0   AND page_title REGEXP \"[^a-zA-Z_ 0-9,:!%'`-/+=.\[\]\<\>]\";" | sql frwiki_p > e.txt
of course, this query is not perfect. but it's a good start
-- Darkoneko 28/12/2007
 
Command-line arguments:
 
    -file       Work on all pages listed in a text file.
                Argument can also be given as "-file:filename".
 
    -cat        Work on all pages which are in a specific category.
                Argument can also be given as "-cat:categoryname".
 
    -ref        Work on all pages that link to a certain page.
                Argument can also be given as "-ref:referredpagetitle".
 
    -links      Work on all pages that are linked from a certain page.
                Argument can also be given as "-link:linkingpagetitle".
 
    -start      Work on all pages on the home wiki, starting at the named
                page.
 
    -page       Work on a single page.
 
    -namespace  Run over especific namespace.
                Argument can also be given as "-namespace:100" or
                "-namespace:Image".
 
    -always     Don't prompt to make changes, just do them.
"""
#
# (C) Yrithinnd, 2007
# (C) Filnik, 2007
# (C) Darkoneko, 2007
# (C) Phe, 2007
# Class licensed under terms of the MIT license
#
__version__ = '$Id: capitalize_redirects.py 4251 2007-09-12 10:36:03Z wikipedian $'
 
import time, sys, string, re
import wikipedia, pagegenerators
 
msg = {
    'en': u'Robot: Create redirect to [[%s]]',
    'es': u'Bot: Creando redirección a [[%s]]',
    'it': u'Bot: Creo redirect a [[%s]]',
    'fr': u'Robot : Creation redirect vers [[%s]]',
    'pt': u'Bot: Criando redirect para [[%s]]',
    }
 
 
def getTranscriptionTable():
    #pre-creation of char change transcription table, for use with string.maketrans()
    origin=u'×.,;:-ẢĄÀÁÂÃÄÅǍȦǠẠẬẶāąảàáâãäåąǎȧǡạậặăĒĘĖẸỆÈÉÊËĚēęėẹệèéêëěĪỊİÌÍÎÏǏĮīɨịìíîïǐıŌỔȮȰȱƠÒÓÔÕÖØǑƏōổȯỌọỘộơòóôõöøǒőŪŲǙÙÚÛÜǓỤūųǚùúûüǔụŭḂḄßḃḅĊÇČ¢ćċçčḌḊĎÐĐḍḋđďðĠǤǦġǥǧğĹḶŁĽḸḸĺḷłľḹḹṄṆŇṅṇńňṘŘṚṜřṛṝṙŞṠṢṦṤṨṦŚşṡṣṧṥṩŠšṧşŦŤṪṬŧťṫṭţẎỴẏỵÿýŻẒŽżẓžżźḞḟĦħȞȟḢḣḤḥƗįǰḲḳǨǩṀṁṂṃṖṗƯưṾṿẆẇẈẉҲҳẊẋñÑ'
    dest=u'x     AAAAAAAAAAAAAAaaaaaaaaaaaaaaaaaEEEEEEEEEEeeeeeeeeeeIIIIIIIIIiiiiiiiiiOOOOOOOOOOOOOoooooooooooooooooUUUUUUUUUuuuuuuuuuuBBBbbCCCcccccDDDDDdddddGGGggggLLLLLLllllllNNNnnnnRRRRrrrrSSSSSSSSssssssssssTTTTtttttYYyyyyZZZzzzzzFfHhHhHhHhIijKkKkMmMmPpUuVvWWWwXxXxnN'
    table = {}
    for pos, ch in enumerate(origin):
        table[ord(ch)] = ord(dest[pos])
    return table
 
 
#create the redirect title from the transcription table (every 1 -> 1 char change) + some others (1 -> 2+ char change)
def getRedirectTitle(table, page):
   redirect =  page.translate(table)
   redirect = redirect.replace(u'æ', u'ae').replace(u'Æ', u'Ae')
   redirect = redirect.replace('\u0153', u'oe').replace('\u0152', 'Oe')
   redirect = redirect.replace('\u2026', '...')
   redirect = redirect.replace(u'þ', 'Th').replace(u'Þ', 'th')
   return redirect
 
 
 
def checkRedirect(page):
    try:
        text = page.get()
    # Double redirect!
    except wikipedia.IsRedirectPage:
        return True
    # Break Redirect!
    except wikipedia.NoPage:
        return True
    # Something-else! :-)
    except wikipedia.Error:
        return True
    text = text.lower()
    # Checking the text in the page.
    res = re.findall(r'([^>]#redirect\s*?\[\[)', text)
    if res == []:
        return False
    else:
        return True
 
class CapitalizeBot:
    def __init__(self, generator, acceptall, msg):
        self.generator = generator
        self.acceptall = acceptall
        self.msg = msg
 
    def run(self):
        #-- put before the for..in so its not generated at each loop
        table = getTranscriptionTable()
        stat_created = 0       #number of created redirects
        stat_existed = 0       #redirect already existed
        stat_total = 0
 
        for page in self.generator:
            page_t = page.title()
 
            #stat show one time out of 50 pages : at the 10th, 60th, etc.
            if stat_total % 50 == 10:
                wikipedia.output(u'  Status : working on \03{lightblue}%s\03{default}. \n  \03{lightyellow}%s\03{default} pages created, \03{lightyellow}%s\03{default} already existed, \03{lightyellow}%s\03{default} pages parsed' % (page_t, stat_created, stat_existed, stat_total) )
 
            stat_total = stat_total + 1
            #is the page we're working on a redirect ?
            if checkRedirect(page):
                wikipedia.output(u'\03{lightblue}%s\03{default} is a redirect! Skip...' % page_t)
                continue
            site = wikipedia.getSite()
            redirectpage = wikipedia.Page(site, getRedirectTitle(table, page_t) )
 
            #is the redirect title the same as the original title ?
            if page.title() == redirectpage.title():
                continue #nothing to do, no output
 
            #does the redirect we want to create already exist ?            
            if redirectpage.exists():
                stat_existed = stat_existed + 1                
               # wikipedia.output(u'\03{lightyellow}%s\03{default} already exists, skip!' % redirectpage.title())
                continue
 
            # Show the title of the page we're working on.
            wikipedia.output(u"\n>>> \03{lightpurple}%s\03{default} <<<" % page_t)
            wikipedia.output(u'Creating \03{lightyellow}%s\03{default}...' % redirectpage.title())
            created = True
            if not self.acceptall:
                choice = wikipedia.inputChoice(
                        u'Do you want to create the redirect?',
                        ['Yes', 'No', 'All'], ['y', 'N', 'a'], 'N')
                if choice in ['a', 'A']:
                    self.acceptall = True
            if self.acceptall or choice in ['y', 'Y']:
                try:
                    comment = wikipedia.translate(wikipedia.getSite(), self.msg) % page_t
                    #wikipedia.setAction(comment)
                    redirectpage._putPage(u"#REDIRECT [[%s]]" % page_t, comment, newPage = True)
                except wikipedia.EditConflict:
                    wikipedia.output(u"An edit conflict has occurred. Skip!...")
                    continue
                except wikipedia.LockedPage:
                    wikipedia.output(u"page was protected against creation. Skip!...")
                    continue
            stat_created = stat_created + 1
 
def main():
    gen = None
    source = None
    textfilename = None
    categoryname = None
    pageNames = []
    referredPageName = None
    acceptall = False
    namespaces = []
    startpage = None
 
    for arg in wikipedia.handleArgs():
        if arg.startswith('-file'):
            if len(arg) == 5:
                textfilename = wikipedia.input(u'Please enter the filename:')
            else:
                textfilename = arg[6:]
            source = 'textfile'
        elif arg.startswith('-cat'):
            if len(arg) == 4:
                categoryname = wikipedia.input(
                               u'Please enter the category name:')
            else:
                categoryname = arg[5:]
            source = 'category'
        elif arg.startswith('-page'):
            if len(arg) == 5:
                pageNames.append(wikipedia.input(
                                 u'Which page do you want to change?'))
            else:
                pageNames.append(arg[6:])
            source = 'singlepage'
        elif arg.startswith('-ref'):
            if len(arg) == 4:
                referredPageName = wikipedia.input(
                                   u'Links to which page should be processed?')
            else:
                referredPageName = arg[5:]
            source = 'ref'
        elif arg.startswith('-start'):
            if len(arg) == 6:
                firstPageTitle = wikipedia.input(
                                 u'Which page do you want to change?')
            else:
                firstPageTitle = arg[7:]
            source = 'allpages'
        elif arg == '-always':
            acceptall = True
        elif arg.startswith('-namespace:'):
            try:
                namespaces.append(int(arg[11:]))
            except ValueError:
                namespaces.append(arg[11:])
        else:
            commandline_replacements.append(arg)
 
    if source == 'textfile':
        gen = pagegenerators.TextfilePageGenerator(textfilename)
    elif source == 'category':
        cat = catlib.Category(wikipedia.getSite(), categoryname)
        gen = pagegenerators.CategorizedPageGenerator(cat)
    elif source == 'singlepage':
        pages = [wikipedia.Page(wikipedia.getSite(), pageName)
                 for pageName in pageNames]
        gen = iter(pages)
    elif source == 'allpages':
        namespace = wikipedia.Page(wikipedia.getSite(),
                                   firstPageTitle).namespace()
        gen = pagegenerators.AllpagesPageGenerator(firstPageTitle, namespace)
    elif source == 'ref':
        referredPage = wikipedia.Page(wikipedia.getSite(), referredPageName)
        gen = pagegenerators.ReferringPageGenerator(referredPage)
    elif source == None or len(commandline_replacements) not in [0, 2]:
        wikipedia.stopme()
        wikipedia.showHelp(u'capitalize_redirects')
        sys.exit()
    if namespaces != []:
        gen =  pagegenerators.NamespaceFilterPageGenerator(gen, namespaces)
    preloadingGen = pagegenerators.PreloadingGenerator(gen, pageNumber = 25)
    bot = CapitalizeBot(preloadingGen, acceptall, msg)
    bot.run()
 
if __name__ == "__main__":
    try:
        main()
    finally:
        wikipedia.stopme()