Moved fact.py over to BS4 (I need to rewrite fact.py soon, it's a bit outdated and somewhat broken), removed old htmlstipper library
This commit is contained in:
parent
b536bbf754
commit
16573eb482
2 changed files with 5 additions and 49 deletions
|
@ -1,8 +1,6 @@
|
|||
import re
|
||||
from util import hook
|
||||
from util import http
|
||||
from util import misc
|
||||
from BeautifulSoup import BeautifulSoup
|
||||
from util import hook, http
|
||||
from bs4 import BeautifulSoup
|
||||
|
||||
|
||||
@hook.command(autohelp=False)
|
||||
|
@ -22,10 +20,10 @@ def fact(inp, say=False, nick=False):
|
|||
def get_fact():
|
||||
page = http.get('http://www.omg-facts.com/random')
|
||||
soup = BeautifulSoup(page)
|
||||
container = soup.find('a', {'class': 'surprise'})
|
||||
link = container['href']
|
||||
response = soup.find('a', {'class': 'surprise'})
|
||||
link = response['href']
|
||||
|
||||
fact = misc.strip_html(container.renderContents())
|
||||
fact = response.contents[0]
|
||||
|
||||
if fact:
|
||||
return (fact, link)
|
||||
|
|
|
@ -1,42 +0,0 @@
|
|||
from HTMLParser import HTMLParser
|
||||
import htmlentitydefs
|
||||
import errno
|
||||
import re
|
||||
|
||||
class HTMLStripper(HTMLParser):
|
||||
|
||||
def __init__(self, data):
|
||||
HTMLParser.__init__(self)
|
||||
self._stripped = []
|
||||
self.feed(data)
|
||||
|
||||
def handle_starttag(self, tag, attrs):
|
||||
if tag.lower() == 'br':
|
||||
self._stripped.append('\n')
|
||||
|
||||
def handle_charref(self, name):
|
||||
try:
|
||||
if name.lower().startswith('x'):
|
||||
char = int(name[1:], 16)
|
||||
else:
|
||||
char = int(name)
|
||||
self._stripped.append(unichr(char))
|
||||
except Exception, error:
|
||||
return
|
||||
|
||||
def handle_entityref(self, name):
|
||||
try:
|
||||
char = unichr(htmlentitydefs.name2codepoint[name])
|
||||
except Exception, error:
|
||||
char = u'&%s;' % name
|
||||
self._stripped.append(char)
|
||||
|
||||
def handle_data(self, data):
|
||||
self._stripped.append(data)
|
||||
|
||||
@property
|
||||
def stripped(self):
|
||||
return ''.join(self._stripped)
|
||||
|
||||
def strip_html(data):
|
||||
return HTMLStripper(data).stripped
|
Reference in a new issue