XMP containing invalid bytes leading fail to decode image

[eeyrw]

Describe the bug

Fail to decode image just because of some invalid bytes in XMP.

Steps/Code to Reproduce

from PIL import Image
from pillow_heif import register_heif_opener

register_heif_opener()

im = Image.open("images/input.heic")  # do whatever need with a Pillow image

input.heic: https://easyupload.io/2p5p0d

Expected Results

No error is thrown or warning prompt.

Actual Results

Versions

3.7.7 (tags/v3.7.7:d7c567b08f, Mar 10 2020, 10:41:24) [MSC v.1900 64 bit (AMD64)]
Windows-10-10.0.19041-SP0
0.9.2
{'version': {'libheif': '1.14.2', 'x265': 'x265 HEVC encoder (3.4+31-6722fce1f)', 'aom': 'AOMedia Project AV1 Encoder v3.5.0'}, 'decoders': {'HEVC': 1, 'AV1': 1, 'AVC': 0}, 'encoders': {'HEVC': 1, 'AV1': 1, 'AVC': 0}}

[bigcat88]

Hm, this is a valid XMP but me just forgot that maybe will be case when last byte can be zero.
I will add test & fix for this and will publish a hotfix release within 36 hours.

Thanks for replying this!

Added:
After closer look at sample, I see two bugs.

1. Zero byte at the end of XMP
2. XMP encoding is latin1

I will wrap decode in try handler and in case of exception during decoding will use latin1 to decode.
As defusedxml do not support data in latin1 encoding, XMP data in this case will be converted to utf-8.
After resaving file with such XMP data, it will be a bit differ in raw comparison(as string it should be equal, but not when comparing as bytes)

[eeyrw]

Hm, this is a valid XMP but me just forgot that maybe will be case when last byte can be zero. I will add test & fix for this and will publish a hotfix release within 36 hours.

Thanks for replying this!

Added: After closer look at sample, I see two bugs.

1. Zero byte at the end of XMP
2. XMP encoding is latin1

I will wrap decode in try handler and in case of exception during decoding will use latin1 to decode. As defusedxml do not support data in latin1 encoding, XMP data in this case will be converted to utf-8. After resaving file with such XMP data, it will be a bit differ in raw comparison(as string it should be equal, but not when comparing as bytes)

String encoding is always the trouble... There are so many local encodings and hard to identify.

[bigcat88]

Yeah, but not so many encodings can be used to encode XMP and latin1 encoding has no invalid bytes, so now it is fine.

[bigcat88]

@eeyrw can we close this?