.. _fbf.plugs.extra.wikipedia:

wikipedia
~~~~~~~~~

.. automodule:: fbf.plugs.extra.wikipedia
        :show-inheritance:
        :members:
        :undoc-members:

CODE
----

    ::

        # fbf/plugs/common/wikipedia.py
        #
        #
        
        """ query wikipedia .. use countrycode to select a country specific wikipedia. """
        
.. _fbf.plugs.extra.wikipedia_fbf_imports:

fbf imports
--------------

    ::

        
        from fbf.utils.url import geturl, striphtml
        from fbf.utils.generic import splittxt, handle_exception, fromenc
        from fbf.lib.commands import cmnds
        from fbf.lib.examples import examples
        from fbf.utils.rsslist import rsslist
        
.. _fbf.plugs.extra.wikipedia_generic_imports:

generic imports
------------------

    ::

        
        from urllib.parse import quote
        import re
        import logging
        
.. _fbf.plugs.extra.wikipedia_defines_:

defines
----------

    ::

        
        wikire = re.compile('start content(.*?)end content', re.M)
        
.. _fbf.plugs.extra.wikipedia_searchwiki_function:

searchwiki function
----------------------

    ::

        
        def searchwiki(txt, lang='en'):
            """ parse wiki data. """
            input = []     
            for i in txt.split():
                if i.startswith('-'):
                    if len(i) != 3: continue
                    else: lang = i[1:]
                    continue
                input.append(i.strip().capitalize())
            what = "_".join(input)
            url = 'http://%s.wikipedia.org/wiki/Special:Export/%s' % (lang, quote(what.encode('utf-8')))
            url2 = 'http://%s.wikipedia.org/wiki/%s' % (lang, quote(what.encode('utf-8')))
            txt = getwikidata(url)
            if not txt: return ("", url2)
            if 'from other capitalisation' in txt:
                what = what.title()
                url = 'http://%s.wikipedia.org/wiki/Special:Export/%s' % (lang, quote(what.encode('utf-8')))
                url2 = 'http://%s.wikipedia.org/wiki/%s' % (lang, quote(what.encode('utf-8')))
                txt = getwikidata(url)
            if '#REDIRECT' in txt or '#redirect' in txt:
                redir = ' '.join(txt.split()[1:])
                url = 'http://%s.wikipedia.org/wiki/Special:Export/%s' % (lang, quote(redir.encode('utf-8')))
                url2 = 'http://%s.wikipedia.org/wiki/%s' % (lang, quote(redir.encode('utf-8')))
                txt = getwikidata(url)
            return (txt, url2)
        
.. _fbf.plugs.extra.wikipedia_getwikidata_function:

getwikidata function
-----------------------

    ::

        
        def getwikidata(url):
            """ fetch wiki data """
            try:
                result = geturl(url)
            except IOError as ex: logging.error("error fetching %s: %s" % (url, str(ex))) ; return 
            if not result: return
            res = rsslist(result)
            txt = ""
            for i in res:
                try:
                    logging.debug(str(i))
                    txt = i['text']
                    break
                except: pass
            txt = re.sub('\[\[(.*?)\]\]', '<b>\g<1></b>', txt)
            txt = re.sub('{{(.*?)}}', '<i>\g<1></i>', txt)
            txt = re.sub('==(.*?)==', '<h3>\g<1></h3>', txt)
            txt = re.sub('=(.*?)=', '<h2>\g<1></h2>', txt)
            txt = re.sub('\*(.*?)\n', '<li>\g<1></li>', txt)
            txt = re.sub('\n\n', '<br><br>', txt)
            txt = re.sub('\s+', ' ', txt)
            txt = txt.replace('|', ' - ')
            return txt
        
.. _fbf.plugs.extra.wikipedia_wikipedia_command:

wikipedia command
--------------------

    ::

        
        resultre1 = re.compile("(<li>.*?</li>)")
        resultre2 = re.compile("(<h2>.*?</h2>)")
        
        def handle_wikipedia(bot, ievent):
            """ arguments: <searchtxt> ["-"<countrycode>] -  search wikipedia, you can provide an optional country code.  """
            if not ievent.rest: ievent.missing('<searchtxt>') ; return
            showall = False
            res = searchwiki(ievent.rest)
            if not res[0]: ievent.reply('no result found') ; return
            prefix = '%s ===> ' % res[1]
            result = resultre1.findall(res[0])
            if result:
                if bot.type == "sxmpp" and not ievent.groupchat: showall = True
                ievent.reply(prefix, result, dot="<br>", showall=showall)
                return
            result2 = resultre2.findall(res[0])
            if result2:
                if bot.type == "sxmpp" and not ievent.groupchat: showall = True
                ievent.reply(prefix, result2, dot="<br>", showall=showall)
                return
            else: ievent.reply("no data found on %s" % event.rest)
        
        cmnds.add('wikipedia', handle_wikipedia, ['USER', 'GUEST'])
        examples.add('wikipedia', 'search wikipedia for <what>','1) wikipedia bot 2) wikipedia -nl bot')