hackbot/plugins/wpsearch.py

#!/usr/bin/python
# -*- coding: utf-8 -*-

# file: wpsearch.py
# date: 23.07.2020
# desc: generic plugin


import json
import common
import logging
import requests
import threading
from manager import Plugin
logging = logging.getLogger()

lock = threading.Lock()


class Plugin(Plugin):

    '''
    '''

    __module = __name__
    __command = 'wpsearch'
    __description = 'Seach wikipedia'

    @staticmethod
    def get_module():
        return Plugin.__module

    @staticmethod
    def get_command():
        return Plugin.__command

    @staticmethod
    def get_description():
        return Plugin.__description

    def __init__(self, callback=None):
        self.callback = callback

    def help(self):
        return ('!wp-search searches in wikipedia for a given search term. '
                'Arguments are language shortcut and search pattern. If no '
                'language shortcut is given, en is used as default. '
                '\nSyntax: !wp-search <lang> <pattern>')

    def run(self, stanza):
        '''
        Startfunktion des Plugins. Bekommt eine xmpp Stanza übergeben,
        extrahiert daraus den Nick des Absenders und aus dem Body den
        gewünschten Suchstring. Ist die Stringlänge ungleich Null oder
        False, wird beides dem Thread zur Wikipedia-Anfrage übergeben und
        eine Nachricht für den MUC zurück gegeben.
        param 1: xmpp stanza objekt
        returns: string
        '''
        lang = 'de'
        pattern = None
        call_msg = 'Call "!help {}"'.format(self.get_command())

        muc_nick = common.get_nick_from_stanza(stanza)
        arguments = common.get_arguments_from_body(stanza)
        logging.debug('Arguments: {}'.format(arguments))

        if arguments is False or len(arguments) == 0:
            logging.warning('No arguments for wp search. Abort.')
            msg = ' '.join(('No search pattern!', call_msg))
        else:
            pattern = ' '.join(arguments)
            api_thread = ApiThread(self.callback)
            api_thread.run(lang, pattern, muc_nick)
            msg = ' '.join(('Search started for:', pattern))
        self.callback(': '.join((muc_nick, msg)))
        return


class ApiThread(threading.Thread):
    '''
    The thread who fetched and returns the wp search. 
    '''
    def __init__(self, callback):
        threading.Thread.__init__(self)
        self.callback = callback

    def run(self, lang, pattern, muc_nick):
        '''
        Starts the thread.
        '''
        logging.debug('Start thread for wp search')
        data = self.get_file(lang, pattern)
        if data == False:
            self.callback('Error while connecting WP')
        else:
            # TODO: check if its a json needed !
            logging.debug(data)
            msg = self.string_factory(data)
            self.callback(': '.join((muc_nick, msg)))

    def get_file(self, lang, pattern):
        '''
        Grabs the API file, parse it and returns a json ... otherwise none.
        param 1: string
        param 2: string
        param 3: string
        returns: json or none
        '''
        api_url = 'https://{}.wikipedia.org/w/api.php'.format(lang)
        api_params = {
            'action': 'query',
            'prop': 'extracts|info',
            'explaintext': '',
            'redirects': '',
            'exchars': 200,
            'continue': '',
            'format': 'json',
            'titles': pattern,
            'inprop': 'url',
            'formatversion': 2
        }

        try:
            logging.debug('Try to fetch {}'.format(api_url))
            session = requests.Session()
            response = session.get(url = api_url, params = api_params)
            data = response.json()
            logging.debug('{}: successfull fetched'.format(api_url))
        except:
            logging.debug('{}: failed to fetch'.format(api_url))
            return False
        return data

    def string_factory(self, data):
        '''
        param 1: json
        returns: string
        '''
        msg = ''
        if 'redirects' in data['query'].keys():
            for i in data['query']['redirects']:
                rfr = i['from']
                rto = i['to']
                msg = '\n'.join((msg, 'Redirect from {} to {}'.format( \
                                 rfr, rto)))
                logging.debug('Message: {}'.format(msg))
        pages = data['query']['pages']
        for i in pages:
            if 'extract' in i.keys():
                msg = '\n'.join((msg, 'Summary:', i['extract']))
                logging.debug('Message: {}'.format(msg))
            else: msg = '\n'.join((msg, 'Nothing found'))
            msg = '\n'.join((msg, 'URL:', i['fullurl']))
            logging.debug('Message: {}'.format(msg))
        return msg