[FFmpeg-devel] [PATCH 2/5] h264_metadata: Add support for A/53 closed captions

[Mark Thompson]

On 26/03/18 14:33, Alex Giladi wrote:
> Is there any documentation on the side data file format?

The format for the side data is the existing one used in AV_PKT_DATA_A53_CC.

The documentation for that could probably be improved - basically it's just a sequence of the 3-byte CC data packets as defined by CEA-708.

> On Sun, Mar 25, 2018 at 6:18 PM, James Almer <jamrial at gmail.com> wrote:
>> On 3/25/2018 2:41 PM, Mark Thompson wrote:
>>> Allows insertion (from side data), extraction (to side data), and removal
>>> of closed captions in SEI messages.
>>> ---
>>>  libavcodec/Makefile            |   2 +-
>>>  libavcodec/h264_metadata_bsf.c | 138 ++++++++++++++++++++++++++++++
>> +++++++++++
>>>  2 files changed, 139 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/libavcodec/Makefile b/libavcodec/Makefile
>>> index aaef6c3ab8..cfde104055 100644
>>> --- a/libavcodec/Makefile
>>> +++ b/libavcodec/Makefile
>>> @@ -1044,7 +1044,7 @@ OBJS-$(CONFIG_DCA_CORE_BSF)               +=
>> dca_core_bsf.o
>>>  OBJS-$(CONFIG_EXTRACT_EXTRADATA_BSF)      += extract_extradata_bsf.o
>>   \
>>>                                               h2645_parse.o
>>>  OBJS-$(CONFIG_FILTER_UNITS_BSF)           += filter_units_bsf.o
>>> -OBJS-$(CONFIG_H264_METADATA_BSF)          += h264_metadata_bsf.o
>>> +OBJS-$(CONFIG_H264_METADATA_BSF)          += h264_metadata_bsf.o
>> cbs_misc.o
>>>  OBJS-$(CONFIG_H264_MP4TOANNEXB_BSF)       += h264_mp4toannexb_bsf.o
>>>  OBJS-$(CONFIG_H264_REDUNDANT_PPS_BSF)     += h264_redundant_pps_bsf.o
>>>  OBJS-$(CONFIG_HAPQA_EXTRACT_BSF)          += hapqa_extract_bsf.o hap.o
>>> diff --git a/libavcodec/h264_metadata_bsf.c b/libavcodec/h264_metadata_
>> bsf.c
>>> index 27053dbdcf..d79e6c0c87 100644
>>> --- a/libavcodec/h264_metadata_bsf.c
>>> +++ b/libavcodec/h264_metadata_bsf.c
>>> @@ -24,6 +24,7 @@
>>>  #include "bsf.h"
>>>  #include "cbs.h"
>>>  #include "cbs_h264.h"
>>> +#include "cbs_misc.h"
>>>  #include "h264.h"
>>>  #include "h264_sei.h"
>>>
>>> @@ -74,6 +75,8 @@ typedef struct H264MetadataContext {
>>>      int display_orientation;
>>>      double rotate;
>>>      int flip;
>>> +
>>> +    int a53_cc;
>>>  } H264MetadataContext;
>>>
>>>
>>> @@ -222,6 +225,8 @@ static int h264_metadata_filter(AVBSFContext *bsf,
>> AVPacket *out)
>>>      int err, i, j, has_sps;
>>>      uint8_t *displaymatrix_side_data = NULL;
>>>      size_t displaymatrix_side_data_size = 0;
>>> +    uint8_t *a53_side_data = NULL;
>>> +    size_t a53_side_data_size = 0;
>>>
>>>      err = ff_bsf_get_packet(bsf, &in);
>>>      if (err < 0)
>>> @@ -516,6 +521,116 @@ static int h264_metadata_filter(AVBSFContext
>> *bsf, AVPacket *out)
>>>          }
>>>      }
>>>
>>> +    if (ctx->a53_cc == INSERT) {
>>
>> This function is becoming pretty big. Could you split it, either before
>> or after this patch, for readability sake? One function per each
>> AVOption with pass/insert/remove, basically.