Tidied plugins/util/misc.py (hope I didn't break anything)
This commit is contained in:
parent
71661983df
commit
bb622efcd8
1 changed files with 0 additions and 13 deletions
|
@ -1,13 +1,7 @@
|
||||||
from htmlentitydefs import name2codepoint
|
from htmlentitydefs import name2codepoint
|
||||||
from time import time as unix_time
|
|
||||||
from HTMLParser import HTMLParser
|
from HTMLParser import HTMLParser
|
||||||
from datetime import datetime
|
|
||||||
import tempfile
|
|
||||||
import logging as log
|
|
||||||
import errno
|
import errno
|
||||||
import re
|
import re
|
||||||
import sys
|
|
||||||
import os
|
|
||||||
|
|
||||||
class HTMLStripper(HTMLParser):
|
class HTMLStripper(HTMLParser):
|
||||||
|
|
||||||
|
@ -28,13 +22,11 @@ class HTMLStripper(HTMLParser):
|
||||||
char = int(name)
|
char = int(name)
|
||||||
self._stripped.append(unichr(char))
|
self._stripped.append(unichr(char))
|
||||||
except Exception, error:
|
except Exception, error:
|
||||||
log.warn('invalid entity: %s' % error)
|
|
||||||
|
|
||||||
def handle_entityref(self, name):
|
def handle_entityref(self, name):
|
||||||
try:
|
try:
|
||||||
char = unichr(name2codepoint[name])
|
char = unichr(name2codepoint[name])
|
||||||
except Exception, error:
|
except Exception, error:
|
||||||
log.warn('unknown entity: %s' % error)
|
|
||||||
char = u'&%s;' % name
|
char = u'&%s;' % name
|
||||||
self._stripped.append(char)
|
self._stripped.append(char)
|
||||||
|
|
||||||
|
@ -45,10 +37,5 @@ class HTMLStripper(HTMLParser):
|
||||||
def stripped(self):
|
def stripped(self):
|
||||||
return ''.join(self._stripped)
|
return ''.join(self._stripped)
|
||||||
|
|
||||||
def superscript(text):
|
|
||||||
if isinstance(text, str):
|
|
||||||
text = decode(text, 'utf-8')
|
|
||||||
return text.translate(SUPER_MAP)
|
|
||||||
|
|
||||||
def strip_html(data):
|
def strip_html(data):
|
||||||
return HTMLStripper(data).stripped
|
return HTMLStripper(data).stripped
|
||||||
|
|
Reference in a new issue