Modernize Youtube embeds and exempt from Bleach cleaning, abandon Imgur embeds

KerbalStuff/common.py

@@ -23,12 +23,22 @@
 from .database import db, Base
 from .objects import Game, Mod, Featured, ModVersion, ReferralEvent, DownloadEvent, FollowEvent
 from .search import search_mods
+from .kerbdown import EmbedInlineProcessor
 
 TRUE_STR = ('true', 'yes', 'on')
 PARAGRAPH_PATTERN = re.compile('\n\n|\r\n\r\n')
 
-cleaner = bleach.Cleaner(tags=bleach_allowlist.markdown_tags,
- attributes=bleach_allowlist.markdown_attrs,
+def allow_iframe_attr(tagname: str, attrib: str, val: str) -> bool:
+ return (any(val.startswith(prefix) for prefix in EmbedInlineProcessor.IFRAME_SRC_PREFIXES)
+ if attrib == 'src' else
+ attrib in EmbedInlineProcessor.IFRAME_ATTRIBS)
+
+
+cleaner = bleach.Cleaner(tags=bleach_allowlist.markdown_tags + ['iframe'],
+ attributes={ # type: ignore[arg-type]
+ **bleach_allowlist.markdown_attrs,
+ 'iframe': allow_iframe_attr
+ },
  filters=[bleach.linkifier.LinkifyFilter])
 
 

KerbalStuff/kerbdown.py

@@ -1,69 +1,75 @@
 import urllib.parse
 from urllib.parse import parse_qs, urlparse
-from typing import Dict, Any, Match
+from typing import Dict, Any, Match, Tuple
 
 from markdown import Markdown
 from markdown.extensions import Extension
-from markdown.inlinepatterns import Pattern
+from markdown.inlinepatterns import InlineProcessor
 from markdown.util import etree
 
-EMBED_RE = r'\[\[(?P<url>.+?)\]\]'
 
+class EmbedInlineProcessor(InlineProcessor):
+ # Don't worry about re.compiling this, markdown.inlinepatterns.Pattern.__init__ does that for us
+ EMBED_RE = r'\[\[(?P<url>.+?)\]\]'
 
-def embed_youtube(link: urllib.parse.ParseResult) -> etree.Element:
- q = parse_qs(link.query)
- v = q['v'][0]
- el = etree.Element('iframe')
- el.set('width', '100%')
- el.set('height', '600')
- el.set('frameborder', '0')
- el.set('allowfullscreen', '')
- el.set('src', '//www.youtube-nocookie.com/embed/' + v + '?rel=0')
- return el
+ # Prefixes of the iframe src attributes we generate
+ YOUTUBE_SRC_PREFIX = '//www.youtube-nocookie.com/embed/'
+ IMGUR_SRC_PREFIX = '//imgur.com/a/'
+ IFRAME_SRC_PREFIXES = [YOUTUBE_SRC_PREFIX, IMGUR_SRC_PREFIX]
 
+ # Other iframe attributes we generate
+ IFRAME_ATTRIBS = ['width', 'height', 'frameborder', 'allowfullscreen']
 
-def embed_imgur(link: urllib.parse.ParseResult) -> etree.Element:
- a = link.path.split('/')[2]
- el = etree.Element('iframe')
- el.set('width', '100%')
- el.set('height', '550')
- el.set('frameborder', '0')
- el.set('allowfullscreen', '')
- el.set('src', '//imgur.com/a/' + a + '/embed')
- return el
-
-
-class EmbedPattern(Pattern):
- def __init__(self, pattern: str, m: Markdown, configs: Dict[str, Any]) -> None:
- super(EmbedPattern, self).__init__(pattern, m)
+ def __init__(self, md: Markdown, configs: Dict[str, Any]) -> None:
+ super().__init__(self.EMBED_RE, md)
  self.config = configs
 
- def handleMatch(self, m: Match[str]) -> etree.Element:
+ def handleMatch(self, m: Match[str], data: str) -> Tuple[etree.Element, int, int]: # type: ignore[override]
  d = m.groupdict()
  url = d.get('url')
  if not url:
  el = etree.Element('span')
  el.text = "[[]]"
- return el
+ return el, m.start(0), m.end(0)
  try:
  link = urlparse(url)
  host = link.hostname
  except:
  el = etree.Element('span')
  el.text = "[[" + url + "]]"
- return el
+ return el, m.start(0), m.end(0)
  el = None
  try:
  if host == 'youtube.com' or host == 'www.youtube.com':
- el = embed_youtube(link)
+ el = self._embed_youtube(link)
  if host == 'imgur.com' or host == 'www.imgur.com':
- el = embed_imgur(link)
+ el = self._embed_imgur(link)
  except:
  pass
  if el is None:
  el = etree.Element('span')
  el.text = "[[" + url + "]]"
- return el
+ return el, m.start(0), m.end(0)
+
+ def _embed_youtube(self, link: urllib.parse.ParseResult) -> etree.Element:
+ q = parse_qs(link.query)
+ v = q['v'][0]
+ el = etree.Element('iframe')
+ el.set('width', '100%')
+ el.set('height', '600')
+ el.set('frameborder', '0')
+ el.set('allowfullscreen', '')
+ el.set('src', self.YOUTUBE_SRC_PREFIX + v + '?rel=0')
+ return el
+
+ def _embed_imgur(self, link: urllib.parse.ParseResult) -> etree.Element:
+ a = link.path.split('/')[2]
+ el = etree.Element('iframe')
+ el.set('width', '100%')
+ el.set('height', '550')
+ el.set('frameborder', '0')
+ el.set('allowfullscreen', '')
+ el.set('src', self.IMGUR_SRC_PREFIX + a + '/embed')
  return el
 
 
@@ -75,5 +81,5 @@ def __init__(self, **kwargs: str) -> None:
  # noinspection PyMethodOverriding
  def extendMarkdown(self, md: Markdown) -> None:
  # BUG: the base method signature is INVALID, it's a bug in flask-markdown
- md.inlinePatterns['embeds'] = EmbedPattern(EMBED_RE, md, self.config) # type: ignore[attr-defined]
+ md.inlinePatterns.register(EmbedInlineProcessor(md, self.config), 'embed', 200)
  md.registerExtension(self)