[5.10.x]: Uploading images with a © symbol in their EXIF/IPTC metadata fails with SQL error 1366 since 5.10

[johnvanhulsen]

What happened?

Description

Since updating to Craft 5.10, we can no longer upload certain JPEGs. The upload fails with this error:

SQLSTATE[HY000]: General error: 1366 Incorrect string value: '\xA9 Cor ...' for column 'alt' at row 1
The SQL being executed was: INSERT INTO craft_assets (..., alt, ...) VALUES (..., '... - Photo Vincent Kalut /PN /� Cor Vos 2026', ...)

These are press agency photos (cycling photos from Cor Vos / Photo News) that have a caption embedded in their metadata, ending with something like "© Cor Vos 2026". We've been uploading these photos for years without any problems. It only started failing after updating to 5.10.

After some digging I think I understand what's happening:

• Craft 5.10 introduced a new feature that automatically fills the asset's alt text from the image's embedded metadata. That explains why this never happened before.
• The caption in these photos is apparently not saved as UTF-8 by the photo agency's software, but in an older encoding (Latin-1, I believe). In that encoding the © symbol is the byte \xA9, which matches the byte MySQL is complaining about in the error.
• So Craft reads the caption from the file and tries to save it into the database as-is, MySQL refuses it because it's not valid UTF-8, and the whole upload fails.

At first I thought it was our database's fault, because our tables were still on utf8mb3. So I ran php craft db/convert-charset and converted everything to utf8mb4, but the error stays exactly the same. From what I understand that makes sense: \xA9 on its own simply isn't valid UTF-8, no matter which of the two charsets the table uses (in UTF-8 the © symbol would be \xC2\xA9).

So it looks like Craft needs to either convert the metadata text to proper UTF-8 before saving it, or just skip filling in the alt text when the text isn't valid UTF-8 instead of failing the entire upload.

This is a real problem for us, because basically every photo we get from this agency has "©" in the caption, so right now we can't upload any of them.

Steps to reproduce

1. Craft 5.10.0 or later, with MySQL.
2. Get a JPEG with a caption in its IPTC metadata that contains a © symbol and is saved in Latin-1 encoding instead of UTF-8. Press agency photos often have this. I've upload one original file here: https://oym-general.b-cdn.net/production/CORVOS_00039036-072.jpg

Expected behavior

The image uploads normally, like it did before 5.10. Ideally the alt text gets filled with the caption (with the © symbol intact), or stays empty if Craft can't read the text properly. But either way the upload itself shouldn't fail.

Actual behavior

The upload fails completely with:

SQLSTATE[HY000]: General error: 1366 Incorrect string value: '\xA9 Cor ...' for column 'alt' at row 1

Converting the database from utf8mb3 to utf8mb4 (via php craft db/convert-charset) makes no difference.

Craft CMS version

5.10.5

PHP version

8.5.1

Operating system and version

No response

Database type and version

No response

Image driver and version

No response

Installed plugins and versions

[linear-code]

CMS-2191

[i-just]

Hi, thanks for reporting and for all the details! I raised a PR for this.