Add convenience bindings for javascript to check filetype and to get primary image handle

[EzraH442]

Currently if you want to check the filetype of some buffer in javascript you need to do something like:

const dummy_buffer = new Uint8Array([]) // pretend this has the image data

const ptr = Module._malloc(12);
if (!ptr) {
  //...
}
Module.HEAPU8.set(dummy_buffer.subarray(0, 12), ptr);
const res = Module._heif_check_filetype(ptr, 12);
Module._free(ptr);
if (res === 1) {
 // valid
}
// handle other cases

This is not only alot of manual memory management on the javascript side but also requires defining a copy of the heif_filetype_result enum on the javascript side to avoid using hardcoded numbers since the enum is not exported by the emscripten bindings.

With new bindings it would be simplified to:

const res = Module.heif_js_check_filetype(dummy_buffer)
if (res.value === Module.heif_filetype_result.heif_filetype_yes_supported.value) {
 // valid
}
// handle other cases

Additionally, (as far I know) the only way to get a proper javascript handle to the primary image is to first use heif_js_context_get_list_of_top_level_image_IDs to get all the image ids, then use heif_context_get_image_handle on each id, then finally use heif_image_handle_is_primary_image on each handle until the primary image handle is found. A special js binding like heif_js_context_get_primary_image_handle would be very convenient to avoid having to do this, especially since the function heif_context_get_primary_image_handle already exists.

[farindk]

Thank you. This is great.

I'm not an active user of the JS-binding myself and have only limited JS knowledge, thus I cannot really maintain this.
If you are interested in helping with the JS part, let me know.

[EzraH442]

Thank you for replying so quickly. I would definitely be interested in help maintain the JS part. Some of the first things I'd like to do are update the build-emscripten.sh script to be compatible with the latest version of the emscripten sdk, add an option to the build script to use the --emit-tsd flag to generate typescript types, and add more JS convenience bindings for thumbnail images. Is there anything else that I should look into doing or anything I should avoid?

[farindk]

Great, that all makes sense. In fact, the JS binding was mostly written in one of the first versions and has never been updated much. Thus, a lot of the API is missing. Many things that were recently added and cover more the low-level access to the files are probably not that important, but I'm soon going to add support for extracting single tiles of the image (for just decoding a ROI), and for access to lower-resolution images in a multi-resolution pyramid. That could be very helpful for interactively browsing of large images. Especially if we could add a way to browse large files without first downloading the whole file.