Large header extensions can result in failure to parse NIfTI headers

[effigies]

I manually decompressed and loaded the first kB of data from a failing file:

> readHeader(buf.buffer)
Uncaught RangeError: Offset is outside the bounds of the DataView
    at DataView.prototype.getInt32 (<anonymous>)
    at Function.getIntAt (file:///home/chris/.cache/deno/npm/registry.npmjs.org/@bids/nifti-reader-js/0.6.9/dist/src/utilities.js:28:21)
    at Function.getExtensionsAt (file:///home/chris/.cache/deno/npm/registry.npmjs.org/@bids/nifti-reader-js/0.6.9/dist/src/utilities.js:53:31)
    at NIFTI1.readHeader (file:///home/chris/.cache/deno/npm/registry.npmjs.org/@bids/nifti-reader-js/0.6.9/dist/src/nifti1.js:335:53)
    at readHeader (file:///home/chris/.cache/deno/npm/registry.npmjs.org/@bids/nifti-reader-js/0.6.9/dist/src/nifti.js:146:16)
    at <anonymous>:1:22

The relevant lines are here:

https://github.com/bids-standard/NIFTI-Reader-JS/blob/0e406b03775d5687911f9440dffa65d7b0886137/src/nifti1.ts#L386-L412

Because we pass in 540 bytes to allow for NIfTI2, the header parser attempts to parse extensions, and crashes if the extension extends beyond 540 bytes. If we passed 348 bytes in, we would safely ignore the extensions, but break on NIfTI2 images.

bids-validator/src/files/nifti.ts

Lines 32 to 37 in 9e8da14

	const buf = await file.readBytes(1024)
	const data = isCompressed(buf.buffer) ? await extract(buf, 540) : buf
	const header = readHeader(data.buffer)
	if (!header) {
	throw { key: 'NIFTI_HEADER_UNREADABLE' }
	}

I think probably the right way to fix this is to check if a file is NIfTI 1 or 2 (https://github.com/bids-standard/NIFTI-Reader-JS/blob/0e406b03775d5687911f9440dffa65d7b0886137/src/nifti.ts#L16-L74) and then load the header with the exact number of bytes.

The other approach I can see is to use a try/catch block and fall back to parsing 348 bytes.

Another approach could be to add a flag in NIFTI-Reader-JS to ignore extensions.