Ensure images with dimensions in name are supported #62
Conversation
Tachyon would previsouly assume that an image with a suffix like `-123x456.jpg` was from a thumbnail and attempt to infer the size from that. This update checks whether that suffix matches an image GUID, if it does then its an original file and should not have the dimensions inferred or have the suffix stripped off. Fixes #43
![hm-linter[bot]](https://avatars.githubusercontent.com/in/5455?s=60&v=4){kind=small}
|
I'm a little hesitant as to this exact solution due to the DB lookups but I can't think of another way to determine if it's a valid original image URL. |
| $src_path = wp_parse_url( $src, PHP_URL_PATH ); | ||
|
|
||
| // Try to fetch the post ID by GUID. | ||
| $id = $wpdb->get_var( $wpdb->prepare( "SELECT ID FROM {$wpdb->posts} WHERE post_type = 'attachment' AND guid LIKE '%%%s' LIMIT 1;", $src_path ) ); |
There was a problem hiding this comment.
Does this need to be a like query? AFAIK $src should be the full URL? can we just match against that?
I think a LIKE on GUID is going to be too poor for performance to take this route.
There was a problem hiding this comment.
Yeah we can do, I just opted for path matching instead as the URL in the content might not match whatever the original GUID was. Could potentially look for _wp_attached_file in post meta for a direct match but that will probably have a similar issue as it's a much bigger table.
There was a problem hiding this comment.
Hmm yeah good thinking / point. I think if we go GUID we should go with URL. I think _wp_attached_file is probably the closest thing we have to "correct" and we'll likely get a better hitrate on this case, but it's going to be a tablescan alteast on the amount of rows that is the number of attachments in the DB.
The alternative would be a file_exists on the original file, though I'm not sure if that's better or worse.
Do we know how common this case is? I'm thinking atleast any non-altis authored content will hit this condition, as g-berg or the Classic Editor will be inserting -300x200.jpg suffixed images, so I'm thinking this condition is actually hit quite a lot.
We could also add object cacing for a url -> ID map in to this function or somewhere else.
There was a problem hiding this comment.
Yeah you're right, it'll be hit continually. Seems there's not going to be a consistently performant way of handling this.
Closing in favour of writing a WP CLI migration command instead to update old upload file names.
Tachyon would previsouly assume that an image with a suffix like
-123x456.jpgwas from a thumbnail and attempt to infer the size from that. This update checks whether that suffix matches an image GUID, if it does then its an original file and should not have the dimensions inferred or have the suffix stripped off.Fixes #43