0
0
mirror of https://github.com/ezyang/htmlpurifier.git synced 2024-11-08 14:58:42 +00:00
htmlpurifier/docs/proposal-new-directives.txt

48 lines
2.0 KiB
Plaintext

Configuration Ideas
Here are some theoretical configuration ideas that we could implement some
time. Note the naming convention: %Namespace.Directive
%Attr.IDPrefix - prefix all ids with this
%Attr.RewriteFragments - if there's %Attr.IDPrefix we may want to transparently
rewrite the URLs we parse too. However, we can only do it when it's a pure
anchor link, so it's not foolproof
%Attr.ClassBlacklist,
%Attr.ClassWhitelist,
%Attr.ClassPolicy - determines what classes are allowed. When
%Attr.ClassPolicy is set to Blacklist, only allow those not in
%Attr.ClassBlacklist. When it's Whitelist, only allow those in
%Attr.ClassWhitelist.
%Attr.MaxWidth,
%Attr.MaxHeight - caps for width and height related checks.
(the hack in Pixels for an image crashing attack could be replaced by this)
%URI.AddRelNofollow - will add rel="nofollow" to all links, preventing the
spread of ill-gotten pagerank
%URI.RelativeToAbsolute - transforms all relative URIs to absolute form
%URI.HostBlacklist - strings that if found in the host of a URI are disallowed
%URI.HostBlacklistRegex - regexes that if matching the host are disallowed
%URI.HostWhitelist - domain names that are excluded from the host blacklist
%URI.HostPolicy - determines whether or not its reject all and then whitelist
or allow all in then do specific blacklists with whitelist intervening.
'DenyAll' or 'AllowAll' (default)
%URI.DisableIPHosts - URIs that have IP addresses for hosts are disallowed.
Be sure to also grab unusual encodings (dword, hex and octal), which may
be currently be caught by regular DNS
%URI.DisableIDN - Disallow raw internationalized domain names. Punycode
will still be permitted.
%URI.ConvertUnusualIPHosts - transform dword/hex/octal IP addresses to the
regular form
%URI.ConvertAbsoluteDNS - Remove extra dots after host names that trigger
absolute DNS. While this is actually the preferred method according to
the RFC, most people opt to use a relative domain name relative to . (root).