cleaned up code and removed debugging statements

[backwash] / backwash.py

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

import sys
import re
from simplemediawiki import MediaWiki
import interwiki

# TODO/bugs:

# - the bot should loginto the system. the simplemediawiki library
#   should help us with that
# - we need to de-interwikify any interwiki links given to Wiki
# - the interwiki stuff isn't working for a few reason. i think we should turn it off
#   we should only add interwiki links to the api
#   we should only do it when we create an account on the wiki for the bot
# - normalize messages for non-unix linebreaks
# - get_revid_for_date is not being called at the moment a date w/o a
#   a revision is not showing up

class Wiki:
    def __init__(self, url):
        self.url = url
        self.wiki = MediaWiki(self.url + 'api.php')

    def get_revid_for_date(self, page, date):
        wc = self.wiki.call({
            'action': 'query',
            'prop': 'revisions',
            'titles': page,
            'rvlimit': 1,
            'redirects': 'true',
            'rvstart': date,
            'rvdir' : 'newer',
            'rvprop' : 'timestamp|ids'})
        pages = wc['query']['pages']
        return pages[pages.keys()[0]]['revisions'][0]['revid']

    def get_edit_token(self, page):
        wc = self.wiki.call({
            'action' : 'query',
            'prop': 'info|revisions',
            'titles': page,
            'intoken' : 'edit'})
        pages = wc['query']['pages']
        page = pages[pages.keys()[0]]

        return page['edittoken']

    def normalize_title(self, page):
        wc = self.wiki.call({
            'action': 'query',
            'prop': 'info',
            'titles': page,
            'redirects': 'true'})
        pages = wc['query']['pages']
        page = pages[pages.keys()[0]]

        return page['title'].replace(" ", "_")
        
    def get_talk_page(self, page):
       normalized = self.normalize_title(page)
       
       if ':' in normalized:
          (namespace, rest) = normalized.split(':', 1)
          return '%s_talk:%s' % (namespace, rest)
       else:
          return 'Talk:%s' % page

    def get_page_text(self, page):
        wc = self.wiki.call({
            'action': 'query',
            'prop': 'revisions',
            'titles': page,
            'redirects': 'true',
            'rvprop' : 'content'})
        pages = wc['query']['pages']
        # something here:LP
        page = pages[pages.keys()[0]]
        return page['revisions'][0]['*']
        
    def append_to_talk_page(self, page, text):
        talk_page = self.get_talk_page(page)
        edit_token = self.get_edit_token(talk_page)

        wc = self.wiki.call({
            'action' : 'edit',
            'title': talk_page,
            'bot' : 'true',
            'token' : edit_token,
            'notminor': 'true',
            'section' : 'new',
            'summary': 'Suggestion from an offline user (via [[User:BackWashBot|BackWashBot]])',
            'text': text })

def make_link(wiki, page, revision):
    return '%sindex.php?title=%s&oldid=%s' % (wiki, page.replace(' ', '_'), revision)

class WikiEdit:
    fields = ['page', 'date', 'e-mail', 'export-date', 'redirect', 'page-revision', 'user-agent', 'username', 'wiki']
    datefields = ['date', 'export-date']

    def __init__(self, headers={}, body='', normalize=True):
        self.headers = headers
        self.body = body

        if normalize:
            self.normalize_headers()

    def normalize_headers(self):
        self.headers.setdefault('wiki', 'http://en.wikipedia.org/wiki/')
        if self.headers['wiki'] in interwiki.mapping.keys():
            self.headers['wiki'] = interwiki.mapping[self.headers['wiki']]

    def to_comment(self):
        assembly = [('date', ' on %s,'),
                    ('username', ' [[User:%s]]', ' an anonymous user'),
                    ('e-mail', ' (e-mail: %s)'),
                    (' made the following comment',),
                    ('page-revision', 
        lambda x: ' on [' + make_link(self.headers['wiki'], self.headers['page'], self.headers['page-revision']) + ' this revision]'),
                    ('user-agent', ' (using %s)')
                    ]

        comment = ""
        for stage in assembly:
            if len(stage) == 1:
                comment += stage[0]
            elif self.headers.has_key(stage[0]):
                if isinstance(stage[1], str):
                    comment += stage[1] % (self.headers[stage[0]])
                else:
                    comment += stage[1](stage[0])
            elif len(stage) == 3:
                comment += stage[2]

        comment = comment.strip()
        return ":''" + comment[0].capitalize() + comment[1:] + \
               " using an offline wikireader:''\n\n" + self.body.strip() + \
               "\n\n—via [[User:BackWashBot|BackWashBot]] ~~~~~"

    @classmethod
    def from_string(cls, s):
        body_index = s.index('\n\n')
        # headers = dict([(X[:X.index(':')], X[X.index(':')+1:].strip()) for X in s[:body_index].split('\n')])
        kvs = [re.split(': *', key_value, 1) for key_value in s[:body_index].splitlines()]
        headers = dict([(key.lower(), value) for (key, value) in kvs])
        body = s[body_index+2:]
        return cls(headers, body)

def parse_edits(lines):
    edits = []
    acc = ""
    for line in lines:
        if '\n\n' in acc and line.startswith('Page:'):
            we = WikiEdit.from_string(acc)
            edits.append(we)
            #print we, we.headers, we.body
            acc = ""
        acc += line
    
    if acc != '':
        we = WikiEdit.from_string(acc)
        edits.append(we)
        #print (we, we.headers, we.body)

    return edits

if __name__ =='__main__':
    # parse the .wpe file on standard
    edits = parse_edits(sys.stdin)

    for edit in edits:
        wiki = Wiki(edit.headers['wiki'])
        wiki.append_to_talk_page(edit.headers['page'], edit.to_comment())