Closed bradh closed 4 months ago
I have the DIS of the 8th Edition of ISOBMFF. It states in 8.11.1.1:
Box Type: 'meta'
Container: File, Segment, MovieBox, TrackBox, MovieFragmentBox or TrackFragmentBox
Mandatory: No
Quantity: Zero or one (in File, MovieBox, and TrackBox),
Zero or one (in Segment, MovieFragmentBox or TrackFragmentBox)
which is aligned the current code.
Can you send me your file separately for analysis?
I've tried guessing an email address (github user id + company domain), but it bounced. Can you ping me at bradh@frogmouth.net ?
The test was too restrictive by making a cross-check in a box full hierarchy instead of the parent only. This is obliviously wrong in your use-case because the meta
box can be located at several places in the same file.
I've pushed a fix. Let me know if it helps.
Works well for me. Happy for this to be closed.
@rbouqueau Thanks again.
Describe the issue
When running against a file I've cleaned up a bit (originally from a gopro camera), with the
isobmff
profile, I'm getting this:The second and third errors are probably valid - I need to remove some bogus headers and likely insert
nmhd
in place of the gopro (Quicktime) version of ISOBMFF.I think the first and last errors are false positive, or perhaps a difference between 6th Ed and 7th Ed. I understand that it is finding
meta
boxes at places other than a single entity at the root level (in particular, in themoov
). My file definitely does that - there are fourtrak
in themoov
, and each of them hasmeta
. The root also hasmeta
.From ISO/IEC 14496-12:2022 Section 8.11.1.1 and Table 1, I think that is valid.
To Reproduce Steps to reproduce the behavior:
Detailed analysis I think the arity needs to be refined.
Potential patch I may be able to address this given some direction.