v0.56.0

Added

• LablBox for the Label box labl (ISO/IEC 14496-12:2026 Section 8.10.5) with
  the LablIsGroupLabelFlag flag mask and an IsGroupLabel() accessor, as used
  for track labels in DASH-IF Ingest
• UdtaBox.Labls with the labl children of a user data box, and
  UdtaBox.GroupLabl() to find the group label of a label group
• TrakBox.Udta reference to the user data box of a track
• QuickTime sound sample description versions 1 and 2 are parsed into
  AudioSampleEntryBox.QuickTimeVersion, QuickTimeV1, and QuickTimeV2,
  preserved on encode (along with the QuickTime revision level, vendor, and
  CompressionID), and the child boxes are read at the version-dependent
  offset. Info() reports the effective channel count, sample size, and
  sample rate from the version 2 struct
• WaveBox for the QuickTime siDecompressionParam atom, so an esds wrapped
  in a wave child of an audio sample entry becomes reachable as Wave.Esds.
  Children that are not well-formed boxes (such as the spec-mandated
  terminator atom when written with size zero) are preserved verbatim
• WaveBox.GetChildren, so a wave box satisfies the ContainerBox interface
• DecoderConfigDescriptor.StreamTypeValue and UpStream decompose the
  packed streamType byte, and named constants cover the common stream types
  (ISO/IEC 14496-1 Table 6) and object type indications (mp4ra.org), so
  callers no longer need magic numbers for the esds vocabulary
• aac.DecodeAudioSpecificConfigHeader parses the leading
  AudioSpecificConfig fields (object type including the 31-escape, base
  sampling frequency, channel configuration) for every audio object type,
  not just the AAC-LC and HE-AAC types the full decoder supports
• ElstEntry.MediaRateFixed32 and SetMediaRateFixed32 combine and split
  the media rate halves as one signed 16.16 fixed-point number, so callers
  no longer need to know the bit layout
• QuickTimeFormatFlagIsFloat and the other CoreAudio/QTFF linear-PCM
  format flags name the bits of
  QuickTimeV2SoundDescription.FormatSpecificFlags, and the IsFloat,
  IsBigEndian, and IsSignedInteger accessors read the common ones, so
  lpcm entries can be interpreted without magic numbers

Changed

• Info() escapes control characters in box types as \xNN, so a type such
  as the four zero bytes of the QuickTime wave terminator atom stays visible
• audio sample entries whose reserved bytes claim a QuickTime version whose
  layout does not parse fall back to the plain ISO interpretation, so
  previously decodable files keep decoding; the overlaid QuickTime bytes
  (version, revision level, vendor) are now preserved on encode instead of
  being zeroed
• DecodeAudioSampleEntry now errors on bodies too short for the fixed
  fields, and encoding errors on inconsistent
  QuickTimeVersion/QuickTimeV1/QuickTimeV2 combinations
• The QuickTime audio sample entries .mp3, lpcm, twos, and sowt
  decode as AudioSampleEntryBox (reachable as StsdBox.Mp3 and
  StsdBox.QtPcm) instead of falling through to UnknownBox. A body that
  does not parse as a sound sample description still becomes an
  UnknownBox, so previously decodable files keep decoding; a body that
  does parse now re-encodes with the same fidelity as mp4a (zeroed
  packet-size bytes, integer sample rate) instead of byte-verbatim. Note
  that a .mp3 entry carries no esds; its codec facts live in the sound
  description fields themselves

Fixed

• hevc.ParseSliceHeader applies the Section 7.4.7.1 inferences for slice
  header fields that are absent from the bitstream, instead of leaving them at
  the Go zero value:
  • slice_deblocking_filter_disabled_flag from
    pps_deblocking_filter_disabled_flag. This one also caused a parse failure:
    a stream with deblocking disabled in the PPS and no slice header override
    read a slice_loop_filter_across_slices_enabled_flag bit that is not there
    and then failed byte_alignment (x265 --no-deblock output)
  • slice_loop_filter_across_slices_enabled_flag from
    pps_loop_filter_across_slices_enabled_flag
  • slice_beta_offset_div2 and slice_tc_offset_div2 from
    pps_beta_offset_div2 and pps_tc_offset_div2 when the slice header does
    not override the deblocking parameters
  • pic_output_flag as 1 when output_flag_present_flag is 0
• avc.ParsePPSNALUnit rejects an out-of-range num_slice_groups_minus1
  (max 7 for any profile) instead of appending one entry per signalled slice
  group, which let a 7-byte malformed PPS NAL unit allocate over 10 GiB
• hevc.ParseVPSNALUnit rejects an out-of-range vps_num_hrd_parameters
  (max vps_num_layer_sets_minus1 + 1) and vps_num_layer_sets_minus1
  (valid range 0-1023), which let a 33-byte malformed VPS NAL unit churn
  30 GiB of HRD parameter allocations
• the alst sample group entry no longer underflows its count of optional
  entries when roll_count implies more bytes than the signalled
  description_length, which let a 32-byte sgpd box allocate over 4 GiB
• avc.ParseSPSNALUnit rejects an out-of-range
  num_ref_frames_in_pic_order_cnt_cycle (valid range 0-255) instead of
  allocating a slice sized from the raw field, which let a tiny malformed SPS
  NAL unit trigger a multi-gigabyte allocation (found by fuzzing)
• MdatBox.HeaderSize accounts for large lazy payloads, so trun data
  offsets are no longer 8 bytes short for fragments above 4 GiB