fix: Correct condition for re-encoding avatar-images that include exif-data

[72374]

When exif-data is found in the avatar-image, it should be re-encoded to remove it.
When the avatar-image will be too large with a white background added, it should be re-encoded to fit within the file-size-limit.

The check for exif-data and for wether or not the image would be too large if encoded with a white background, currently depend on each other.

Both of these would set do_scale to false:

file-size is too large = false || is an avatar = true && (too high resolution = false || contains exif-data = true && is too large with a white background = false)

file-size is too large = false || is an avatar = true && (too high resolution = false || contains exif-data = false && is too large with a white background = true)

[iequidoo]

encoded_img_exceeds_bytes() already recodes the image, that's why it takes &mut encoded, this encoded is then passed to BlobObject::create_and_deduplicate_from_bytes(). Maybe rename it to just encode_img_ex() to avoid any wrong assumptions about what it does? Or, if i'm wrong, adding a test for the scenario being fixed will help, w/o tests we always get regressions.

[72374]

The function does add a file-size-check to encode_img, that returns false or true.

core/src/blob.rs

Lines 653 to 673 in f43b3b9

	fn encoded_img_exceeds_bytes(
	context: &Context,
	img: &DynamicImage,
	fmt: ImageOutputFormat,
	max_bytes: usize,
	encoded: &mut Vec<u8>,
	) -> anyhow::Result<bool> {
	encode_img(img, fmt, encoded)?;
	if encoded.len() > max_bytes {
	info!(
	context,
	"Image size {}B ({}x{}px) exceeds {}B, need to scale down.",
	encoded.len(),
	img.width(),
	img.height(),
	max_bytes,
	);
	return Ok(true);
	}
	Ok(false)
	}

So, i think the name makes sense as it is.

I added a comment that does explain that it actually changes the image (in the resizing-loop, where it is primarily used) to make that easier to notice, and added another comment explaining why the image will be encoded here.

[Hocuri]

Both of these would set do_scale to false:

file-size is too large = false || is an avatar = true && (too high resolution = false || contains exif-data = true && is too large with a white background = false)

It's good that do_scale is false, because there are no exif-data, so, nothing is recoded. Since there is no recoding step, it can't happen that the recoding makes the image too large.

file-size is too large = false || is an avatar = true && (too high resolution = false || contains exif-data = false && is too large with a white background = true)

It's good that do_scale is false, because we already called encoded_img_exceeds_bytes() there, so, we don't want to call it again.

Unless I misunderstood something, this PR here is not needed.

[72374]

It's good that do_scale is false, because there are no exif-data, so, nothing is recoded. Since there is no recoding step, it can't happen that the recoding makes the image too large.

"contains exif-data = true" means that there is exif-data.

It's good that do_scale is false, because we already called encoded_img_exceeds_bytes() there, so, we don't want to call it again.

This part is when a file is not too large in the beginning, but will be too large after a white background is added.
The image will be too large when this is reached.

[72374]

Ah, okay, the exif-data will also be removed further below, so it is not necessary to do that before the scaling-loop. I will close this PR.