Functions to get image dimensions quickly

[MysteriousJ]

Would SDL_image be open to a PR with functions to get the width and height of an image without returning pixel data? I imagine a function such as:

bool IMG_GetDimensions_IO(SDL_IOStream *src, bool closeio, int* w, int* h)

This would do as little processing on the input stream as possible to find the width and height of the image, and output them in the w and h parameters, respectively.

[slouken]

I'm not opposed to this, but what's the use case?

[Green-Sky]

This is very useful for anything that might take space off-screen. For example, I use this, if available, in the chatlog. Otherwise the scrollbar jumps around and other issues.

[MysteriousJ]

My use case is comic book pages. Usually 1 page is 1 image, but when an illustration spans across 2 pages, authors create it as a single image. To determine the actual number of pages in the book, the relative size of each image must be examined.

[Green-Sky]

Or to determine if the image is just too large to load (you can somewhat determine zip bombs that way).

[madebr]

There exist lots of metadata.

Wouldn't you also want to access more metadata, as it is currently possible for animations?

[MysteriousJ]

So like an IMG_Decoder API?

[LordOfXen]

Rather than an explicit IMG_GetDimensions_IO, I'd recommend a metadata getter and setter for non-animated decoders and encoders, similar to how we do in the animation decoder/encoder.

Examples

Non-standardized approach (Feature over existing API):

IMG_GetAnimationDecoderProperties -> IMG_GetPNGProperties, IMG_GetWEBPProperties OR IMG_GetProperties("png", stream)

Standardized approach (Revised API):

IMG_Decoder *decoder = IMG_CreateDecoder("PNG");
SDL_PropertiesID metadata = IMG_GetDecoderProperties(decoder);
// Access width, height, and other properties from metadata. This also brings the ability to check if it's animated and such.

The second one requires more work, but is a cleaner and more robust approach.

[slouken]

Metadata is probably the right way to go, but this is a bunch of work across a bunch of different image loading libraries that might have to load the image anyway to get the information.

I'm not opposed to someone creating a PR for this as long as they support all the existing image formats and loaders, but I don't know of anyone doing that work right now.