|
|
bugtest.webp (79,934 bytes) |
|
|
Another image with the same issue: > file bugtest2.webp bugtest2.webp: RIFF (little-endian) data, Web/P image, with alpha, 299+1x299+1 bugtest2.webp (5,042 bytes) |
|
|
'file' doesn't have a reasonable way to calculate WebP-style dimensions, so something's got to give, or it would need a major new feature. The technical problem is the need to truncate a 32-bit integer to a 24-bit integer, and add 1 to the result. 'file' can do one of those things, or the other, but not both. Another option would be to only print the dimensions if they fit into 16-bit integers. Like this: [magic/Magdir/riff, "VP8X" section] Change this: >>>12 lelong&0xffffff x \b, %d+1 >>>15 lelong&0xffffff x \bx%d+1 To this: # This only prints the dimensions if both are 65535 or less. >>>12 ulelong&0xffffff <65535 >>>>15 ulelong&0xffffff <65535 >>>>>12 uleshort+1 x \b, %u >>>>>15 uleshort+1 x \bx%u |
|
|
Thanks for the explanation and the patch, tested & works. IMO, for practical reasons it's better to restrict the dimension decoding to WebP images < 65535x65535 for now, than to introduce the "+1" syntax. There are probably very few applications needing to work with images exceeding these limits, and those will normally use other methods to determine the dimensions. |
|
|
Applied patch. |
- Anonymous
