[wp-trac] [WordPress Trac] #5305: permalinks broken when article name is numeric

[WordPress Trac]

#5305: permalinks broken when article name is numeric
-------------------------------------------------+-------------------------
 Reporter:  thomask                              |       Owner:
     Type:  defect (bug)                         |  valendesigns
 Priority:  normal                               |      Status:  assigned
Component:  Permalinks                           |   Milestone:  4.2
 Severity:  major                                |     Version:  2.3.1
 Keywords:  has-patch needs-testing dev-         |  Resolution:
  feedback                                       |     Focuses:
-------------------------------------------------+-------------------------

Comment (by boonebgorges):

 valendesigns - Thanks for your work on this. Broadly speaking, I think the
 strategy here is good: when a date query is detected and the permalink
 structure contains `%postname%`, give precedence to a post matching the
 postname, rather than the corresponding date archive.

 I have some implementation suggestions, some large and some small. See
 [5305.7.diff].

 First, a 'pre_get_posts' callback doesn't seem like the right place to do
 this. For one thing, while you've given it a priority of 1, it's still
 possible for plugins to intervene before it runs, which could cause
 various problems. It's also not clear to me why we would want the proposed
 to be unhookable. It seems to me that the proper place is in
 `WP_Query::parse_query()`, right after the existing date checks.

 I've also simplified (and, I think, generalized) your `$token_index`
 trick. What we really care about is the location of `%postname%`, and
 whether it's preceded directly by `%year%` or `%month%` or whether it's in
 the 0th position. My patch tests this more directly.

 I also combined the two functions into a single method. By returning from
 the method whenever we know there's nothing to do, we prevent the nesting
 that I'm assuming you were trying to avoid.

 I simplified `<!--nextpage-->` support as well. Most of the necessary work
 - like the nextpage count - is done for us later on in `setup_postdata()`.
 No need to repeat it here. The only thing required is to translate the
 date query var into 'page'.

 I've left a `@todo` comment on the `zeroise()` block. There is still a bug
 here. Month and day URL chunks are parsed properly with or without the
 leading zero (`([0-9]{4})/([0-9]{1,2})/([0-9]{1,2})/page/?([0-9]{1,})/?$'`
 - note the {1,2}`). This means that `2015/2` is a potential clash when
 post_name=2, in addition to `2015/02` and post_name=02. Does this mean we
 have to do two calls to `get_page_by_path()`?

--
Ticket URL: <https://core.trac.wordpress.org/ticket/5305#comment:63>
WordPress Trac <https://core.trac.wordpress.org/>
WordPress publishing platform