[FFmpeg-devel] [PATCH] libavformat/aac: Parse ID3 tags between ADTS frames.

[Richard Shaffer]

On Fri, Feb 9, 2018 at 1:21 PM, wm4 <nfxjfg at googlemail.com> wrote:
> On Fri, 9 Feb 2018 12:54:29 -0800
> Richard Shaffer <rshaffer at tunein.com> wrote:
>
>> I just wanted to send a final ping about this patch. While most
>> streams will not encounter the case that it addresses, those that do
>> will have a pretty bad experience. The demuxer's read_packet function
>> currently only reads 7 bytes (ADTS_HEADER_SIZE) of input. If the first
>> 12 bits aren't the sync word, it immediately gives up and returns
>> AVERROR_INVALIDDATA. This means that if it encounters an ID3 tag, it
>> won't re-sync unless the tag length just happens to be a multiple of 7
>> bytes.
>>
>> I'm assuming that the demuxer should be enhanced to search input for
>> the next sync word. (That might be some of the additional work that
>> James mentioned, and is something I'd be interested in working on.) In
>> the mean time, I think this patch addresses a likely case of lost
>> synchronization, while also providing some utility to API users.
>>
>> I know that people are probably busy, so I don't want to be annoying
>> or overly persistent. However, if anyone has a little time to review
>> this and provide feedback or accept the changes, I'd appreciate it.
>
> I think I forgot about this patch. I can apply it on Monday or so,
> unless someone else has comments. James Almer wanted a test - what's
> the status on this and does anyone demand that this is pushed only with
> such a test added?

I was also able to provide a patch and sample file to add a fate test.

Patch: https://patchwork.ffmpeg.org/patch/7504/
Sample: http://ffmpeg.org/pipermail/ffmpeg-devel/2018-February/224906.html

I think I sent that out late on a Saturday, so it probably got
overlooked in the shuffle.

-Richard


> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel at ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel