daemon.shit/plugins/wakaba_thread.py

   1 import dbus.service
   2
   3 from my_thread import Thread
   4 from post import Post
   5
   6 import urllib
   7 import re
   8
   9 INTERFACE = 'py.chans.dbus'
  10
  11 class WakabaThread(Thread):
  12
  13     def __init__(self, bus_name, obj_path, board, id):
  14         Thread.__init__(self, bus_name, obj_path, board, id)
  15
  16     #TODO: _normal_ parsing needs
  17     @dbus.service.method(dbus_interface=INTERFACE, in_signature='', out_signature='')
  18     def refresh(self):
  19         def get_posts_dicts():
  20             def parse_msg_table(post_data):
  21                 post = {}
  22                 post['id'] = re.findall('<a name="([0-9]+)">', post_data)[0]
  23                 post['date'] = ''
  24                 post['name'] = ''
  25                 post['title'] = ''
  26                 post['is_sage'] = ''
  27                 img_src = re.findall('<a target="_blank" href="(.*?)">', post_data)
  28                 if img_src:
  29                     post['img_src'] = img_src[0]
  30                 else:
  31                     post['img_src'] = ''
  32                 post['text'] = re.findall('<blockquote>(.*?)</blockquote>', post_data)[0]
  33                 return post
  34
  35             def get_msg_tables(data):
  36                 return re.findall('<table><tbody><tr><td class="doubledash">(.*?)</td></tr></tbody></table>', data)
  37
  38             data = urllib.urlopen('%s/%s/res/%s.html' %(self.board.base_uri, self.board.name, self.id)).read()
  39
  40             first_post = re.findall('(<div id="thread-.*?</blockquote>)', data)[0]
  41             posts = [parse_msg_table(first_post)]
  42             for msg_table in get_msg_tables(data):
  43                 posts.append(parse_msg_table(msg_table))
  44             return posts
  45
  46         self.posts = []
  47         posts_dicts = get_posts_dicts()
  48         for post_dict in posts_dicts:
  49             self.posts.append( Post(bus_name=self.bus_name, obj_path=self.obj_path +'/'+ post_dict['id'], post=post_dict) )