Was hoping to use ncredir to implement redirects towards status.wikimedia.org for many sources of the form like:
status.wikibooks.org
status.wikivoyage.com
status.wikipedia.net
etc etc
However if I'm understanding nc_redirects.dat correctly, it looks like that wildcards are supported only as prefixes, not suffixes? What's the best way to do this?
| Status | Subtype | Assigned | Task | ||
|---|---|---|---|---|---|
| Open | Feature | None | T22079 Provide a better means of status update delivery in WMF error message | ||
| Open | None | T202061 Implement an accurate and easy to understand status page for all wikis | |||
| Stalled | None | T318804 ncredir redirects for status.wiki* --> status.wikimedia.org |
Looks like interpret_wildcard() in compile_redirects.rb does not support suffixed wildcards:
def interpret_wildcard(wildcard)
case wildcard
when /^\*\.([^*]*)/
[{
:domain_regex => '^(.+)\.' << Regexp.quote(Regexp.last_match[1]) << '$',
:alias => wildcard,
}]
when /^\*([^*]*)/
[{
:domain_regex => '=' + Regexp.last_match[1],
:alias => Regexp.last_match[1],
}, {
:domain_regex => '^(.+)\.' << Regexp.quote(Regexp.last_match[1]) << '$',
:alias => '*.' << Regexp.last_match[1],
}]
when /\*/
error("invalid use of asterisk in domain pattern")
else
[{
:domain_regex => '=' + wildcard,
:alias => wildcard,
}]
end
endI imagine this could be extended but compile_redirects.rb is really gnarly and I'd be afraid of unintentional breakage. ._.
Looking into it further, it seems this is a very possible change! nginx mappings/site names support wildcards.
Pulling back a bit, does anything other than ncredir even use compile_redirects.rb? If not, why not just... maintain the maps file directly? This would be a non-issue if that were the case.
There are two main users of compile_redirects.rb:
A list of canonical domains can be found here: https://wikitech.wikimedia.org/wiki/HTTPS#Foundation_canonical_domain_names.
From my understanding of the task description, it impacts both non-canonical domain redirects and canonical domain redirects
Thank you for the clarification, @Vgutierrez. Do you have a suggestion on how to reconcile this? My instinct is to remove the abstraction entirely and maintain two separate mappings if they're going to diverge anyway.