The mov demuxer does not correctly read the `cprt` box

[Xiaolei Yu]

Summary of the bug:
The mov demuxer does not correctly read the cprt box specified in ISO/IEC 14496-12.
Metadata boxes specified in 3GPP use the same layout and are also handled incorrectly.

How to reproduce:
Use the following command to generate a 3gp file

% ffmpeg -f lavfi -f color -frames:v 1 -c:v libx264 -metadata copyright=abcdefg out.3gp

% ffmpeg -i out.3gp

produces:

ffmpeg version N-106627-g61d0b1d21f Copyright (c) 2000-2022 the FFmpeg developers
  built with gcc 11 (Debian 11.2.0-18)
  configuration: --cc=/usr/lib/ccache/gcc --enable-gpl --enable-libx264
  libavutil      57. 24.101 / 57. 24.101
  libavcodec     59. 26.100 / 59. 26.100
  libavformat    59. 22.100 / 59. 22.100
  libavdevice    59.  6.100 / 59.  6.100
  libavfilter     8. 33.100 /  8. 33.100
  libswscale      6.  6.100 /  6.  6.100
  libswresample   4.  6.100 /  4.  6.100
  libpostproc    56.  5.100 / 56.  5.100
Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'out.3gp':
  Metadata:
    major_brand     : 3gp6
    minor_version   : 256
    compatible_brands: 3gp6isomiso2avc1
    copyright       : 
    copyright-eng   : 
  Duration: 00:00:00.04, start: 0.000000, bitrate: 300 kb/s
  Stream #0:0[0x1](und): Video: h264 (High) (avc1 / 0x31637661), yuv420p(progressive), 320x240 [SAR 1:1 DAR 4:3], 150 kb/s, 25 fps, 25 tbr, 12800 tbn (default)
    Metadata:
      handler_name    : VideoHandler
      vendor_id       : [0][0][0][0]

[Balling]

Deleted

[IvanB]

As I understand, current implementation of 'cprt' box reading is broken indeed (or at least doesn't conform to ISO/IEC 14496-12).

According to the specification the box should be a Full Box that contain language field, followed by the copyright text in utfstring format (so either utf-8 or utf-16).

Current implementation reads this as a string length, followed by the actual string, which results in empty string. ("version" field, =0, goes to "string length").

I wonder if somebody have any samples (maybe from some custom mov/mp4 dialect) where 'cprt' box correspond to the current implementation. If there are some, probably we need to find a way to somehow distinguish the types of the boxes we read.

The same thing is also true for the most of 3gpp metadata boxes. For example, according to 3GPP specification, 'gnre' (genre) is written in a similar way to ISO/IEC 14496-12's 'cprt' (so as a FullBox + language + utfstring), but we read it as id3v1_genre index.