url_to_postid() returning post ID for external URL

[jkhongusc]

One our site has a post that has a SoundCloud embed in it. The WP embed is embedding the post instead of the SoundCloud content. See bug#37359 as another person reported but flagged the wrong function.

Unfortunately I dont have a fix. This is what I see, the post (​https://bedrosian.usc.edu/podcast/drown/) embed eventually makes this call:

url_to_postid ('​https://soundcloud.com/uscbedrosian/drown')

This is obviously (to human eyes) an external url but url_to_postid() returns the postid of the post itself. This site uses rewrites, I see url_to_postid() make this database call:

SELECT wp_334_posts.* FROM wp_334_posts WHERE 1=1 AND wp_334_posts.post_name = 'drown' AND wp_334_posts.post_type = 'post' ORDER BY wp_334_posts.post_date DESC " posts

Due to rewrites, all it has to match is 'drown' to be flagged as an internal url... and returns the original post. Happens to all SoundCloud embeds on the same site.

[troutacular]

Sample iframe returned for Soundcloud oEmbed

[troutacular]

As jkhongusc described, the iframe being returned was being altered to being the post itself. Due to this site being a production site, we have applied a filter patch as noted in 37359:

remove_filter( 'pre_oembed_result', 'wp_filter_pre_oembed_result' );

The iframe source being returned before this patch was ​https://bedrosian.usc.edu/podcast/drown/embed/#?secret=z5fm4h1xpp

<iframe sandbox="allow-scripts" security="restricted" src="https://bedrosian.usc.edu/podcast/drown/embed/#?secret=z5fm4h1xpp" width="200" height="339" title="“Drown” — Bedrosian Center" frameborder="0" marginwidth="0" marginheight="0" scrolling="no" class="wp-embedded-content" data-secret="z5fm4h1xpp"></iframe>

[swissspidy]

Hey there,

Thanks for your report. This seems like a duplicate of #37359, but yet I haven't been able to reproduce this in a unit test.

However, I can see that url_to_postid() doesn't really check if the domain matches, so there is a potential for false positives. Such a check might need to be added to wp_filter_pre_oembed_result() or url_to_postid() directly.

[swissspidy]

As per my previous comment, I'm not sure yet what's the best approach.

[swissspidy]

#40368 was marked as a duplicate.

[swissspidy]

#40372 was marked as a duplicate.

[jbpaul17]

Per today's bug scrub and resulting discussion in a thread in #core, we're punting this to 4.9.

[ivankristianto]

Check if the $url is from current domain

[ivankristianto]

I agree to do direct check in url_to_postid so it won't cause any future false positive anymore.

I attached my patch above.

[swissspidy]

Simplified the patch a bit in 39373.diff​. Would be great to have this in trunk as soon as possible for testing.

[SergeyBiryukov]

In 41786:

Rewrite: In url_to_postid(), bail early if the URL does not belong to the site.

Props ivankristianto, swissspidy, jkhongusc, SergeyBiryukov.
Fixes #39373.