See also the very old T32961 and T231393, which described some technical problems preventing running that script. No idea if they are still relevant now.

We don't have as many webp files (26,729) as we have jpgs/pngs (92 million total), so I don't think it is a problem in this case. English Wikipedia has about 2000, which might be a better test ground than most other domains, as most of the wikis hardly have any webp files.

Does remind me that we should add an example to the ticket, for the person executing it to check the result.

In T364680#9795173, @TheDJ wrote:

We don't have as many webp files (26,729) as we have jpgs/pngs (92 million total), so I don't think it is a problem in this case. English Wikipedia has about 2000, which might be a better test ground than most other domains, as most of the wikis hardly have any webp files.

Does remind me that we should add an example to the ticket, for the person executing it to check the result.

If you add the example, I can run the script :)

Thank you @Zabe, I wasn't able to find a good example outside of Commons, but:

https://commons.wikimedia.org/wiki/File:Universiade_(大运)_Exit_A_F1_(2022-08-16).webp
After running should look like:
https://test.wikipedia.org/wiki/File:Universiade_webp_test.webp

I don't expect any issues. However it is image metadata, and metadata isn't generally the cleanest of data, so it might be that as the script runs, it encounters broken files halting the run prematurely but I doubt that will be very common as we would have many errors for webp files in Logstash in that case, and we don't.

Mentioned in SAL (#wikimedia-operations) [2024-06-12T08:42:14Z] <zabe> zabe@mwmaint1002:~$ mwscript refreshImageMetadata.php commonswiki --mime image/webp # T364680

Mentioned in SAL (#wikimedia-operations) [2024-06-12T10:07:44Z] <zabe> zabe@mwmaint1002:~$ foreachwikiindblist 'all - s4' refreshImageMetadata.php --mime image/webp # T364680

Thank you @Zabe