Document vs Oembed / Instagram blocked

[reflexxion]

Hi,

I'm using v4 for resolving our embeds.
Mainly (probably only) we are using embeds which does provide an oembed endpoint.
So we've a big overhead by first getting the main document of that "url".

https://github.com/oscarotero/Embed/blob/master/src/Embed.php#L23-L24

I think it'll be a big benefit to first check if that url does provide an oembed endpoint (detectEndpointFromProviders) and if so to only query that specific api endpoint.

Is there any build in way that I'm missing?

Regards

[oscarotero]

Hello.
This initial request checks the url status and resolve possible redirects. For example, a user can share a short link such https://t.co/4DK3Jc2BTt that must be resolved to the final url.

The way Oembed is detected is by priorizing the links found in the document and then the detectEndpointFromProviders (https://github.com/oscarotero/Embed/blob/master/src/OEmbed.php#L55) The main reason is that the urls in the document are more reliable because are provided directly by the site. The other values are hardcoded in a php array (https://github.com/oscarotero/Embed/blob/master/src/resources/oembed.php) and can become obsolete.

[reflexxion]

Ok, I understand.

Now we've an issue with instagram.
Instagram changed something (or blocked us) so that they are now redirecting a non matching URL.
The url was something like https://www.instagram.com/accounts/login?next=ORIGINAL_URL. (and seems to be some blocking)
This stopped oembed from working for instagram.

[oscarotero]

Mmm, I see.
FYI, instagram works fine in localhost, but not in my server (Example)

Perhaps it's because instagram detects too many requests from this ip?

[reflexxion]

Yeah, that's the same issue I had.
Any idea how we can solve this?
I did a workaround on my side by only crawling the oembed api for certain providers.
But would be great to have this in the library.

[oscarotero]

I made some changes to check the oembed endpoint not only with the final url, but also with the inital url and seems to work fine.

[nteath]

Any chance this could be implemented in version 3 as well ?

Update for anyone using version 3.
What I did as a temporary solution is I changed the Instagram provider (oembed provider) to check the response's url, if it contains 'accounts/login' I use the getStartingUrl() method instead of the getUrl().

@oscarotero , Is there any way to use a custom oembed Provider instead?

[oscarotero]

New versions released with fixes for v4 and v3

[zwan2]

1. Still not works for remote server (embed v3.4.7, PHP v7.3 remote server)
2. Shortend url can't works even in v3 demo site (https://oscarotero.com/embed3/demo/)
   (Example: http://instagr.am/p/CAkFeVZAgB1/)

[reflexxion]

Hey @zwan2,

I think this is due to a "not registered" URL pattern for "instagr.am" -> https://github.com/oscarotero/Embed/blob/v3.x/src/Providers/OEmbed/Instagram.php#L10

[davidwhthomas]

There is a similar case for instagram TV, it's not matching the pattern and returning the login page, Embed v3.4

e.g https://www.instagram.com/tv/CC8AXWhgm_c/?utm_source=ig_web_copy_link

P.S awesome library, thanks