def retieve_raw_content (self):
fh = urllib.urlopen (self.thread_url)
self.raw_content = fh.read().decode("utf-8");
self.thread_title = common.get_title_from_content(self.raw_content);
self.thread_author = common.get_author_from_content(self.raw_content)
#set entry point for list parsing
entry_point = self.raw_content.find ('<span class="active_post">');
if entry_point == -1:
self.raw_content = None
return False;
entry_point += len ('<span class="active_post">');
stop = self.raw_content.find ('<ul class="forum_navi">', entry_point)
if stop == -1:
self.raw_content = None;
return False;
self.raw_content = self.raw_content[entry_point:stop];
return True;
def retieve_raw_content(self):
fh = urllib.urlopen(self.thread_url)
self.raw_content = fh.read().decode("utf-8")
self.thread_title = common.get_title_from_content(self.raw_content)
self.thread_author = common.get_author_from_content(self.raw_content)
#set entry point for list parsing
entry_point = self.raw_content.find('<span class="active_post">')
if entry_point == -1:
self.raw_content = None
return False
entry_point += len('<span class="active_post">')
stop = self.raw_content.find('<ul class="forum_navi">', entry_point)
if stop == -1:
self.raw_content = None
return False
self.raw_content = self.raw_content[entry_point:stop]
return True
def retrieve_post_author (self): self.post_author = common.get_author_from_content(self.raw_content) return True
def retrieve_post_author(self):
self.post_author = common.get_author_from_content(self.raw_content)
return True