do_enclose() can ping the same URL many times, can't filter URLs to ping

[mdawaffe]

If the same URL is included in a post several times, do_enclose() can check that URL for inclusion as a possible enclosure several times.

Additionally, there's no way to filter what URLs do_enclose() should ping for possible inclusion as enclosures.

Attached adds an array_unique() call to do_enclose().

Attached adds a do_action_ref_array( 'pre_enclose', array( &$post_links, &$pung ) ) hook to allow filtration of URLs to ping. (Matches the pre_ping hook.)

do_enclose() has several more problems this patch addresses.

1. Correct non-functional DELETE FROM postmeta query (bad use of wpdb::prepare()).
2. Use like_escape() in several LIKE queries.
3. Add an apply_filters( 'enclosure_mime_types', array( 'video', 'audio' ) ) hook.
4. Currently do_enclose() doesn't allow root URLs as enclosures (e.g. ​http://example.com/), only URLs with a non-trivial path or query. Move that functionality to the new pre_enclose filter.
5. Efficiency improvements in conditional logic.
6. Clean up code by reducing control structure nesting depth.

[mdawaffe]

Suggested test post content:

http://wordpress.org/

http://wordpress.org

http://wordpress.org/news/2010/07/wordpress-3-0-1/

http://wordpress.org/?howdy

http://upload.wikimedia.org/wikipedia/commons/5/50/Sine_wave_440.ogg
http://upload.wikimedia.org/wikipedia/commons/5/50/Sine_wave_440.ogg
http://upload.wikimedia.org/wikipedia/commons/5/50/Sine_wave_440.ogg

[nacin]

Let's do this early 3.2. Probably too late for 3.1.

[solarissmoke]

Was going to open a new ticket but I think it can be addressed here:

do_enclose() does not check that the response to wp_get_http_headers() was good (i.e., HTTP 200) before trying to process it. That means that if you happen to have a link that is either broken or redirected (301), it doesn't realise this and still tries to parse the result, resulting in PHP notices (e.g., missing content-length header).

Two things:

(1) We should at least be checking for a 200 response

(2) Maybe we could be clever and follow 301 redirects. Seeing as there are so many shortlinks floating about everywhere, I think this would greatly increase the number of hits for do_enclose()

[nacin]

I just did a lot of similar cleanup in #20417.

[nacin]

Said cleanup is in ​http://core.trac.wordpress.org/attachment/ticket/20417/20417.3.diff.

[SergeyBiryukov]

Related: #19890, #26530

[wonderboymusic]

Half of this has improved greatly since this ticket was opened