On 8/16/2018 12:23 PM, Paul B Mahol wrote:
> On 8/16/18, James Almer <jamr...@gmail.com> wrote:
>> On 8/16/2018 8:00 AM, Paul B Mahol wrote:
>>> Hi,
>>>
>>> another patch attached, please review.
>>
>>> +static int decode_frame(AVCodecContext *avctx, void *data,
>>> + int *got_frame, AVPacket *avpkt)
>>> +{
>>> + IMM4Context *s = avctx->priv_data;
>>> + GetBitContext *gb = &s->gb;
>>> + AVFrame *frame = data;
>>> + unsigned type;
>>> + int ret, scaled;
>>> +
>>> + if (avpkt->size <= 32)
>>> + return AVERROR_INVALIDDATA;
>>> +
>>> + av_fast_padded_malloc(&s->bitstream, &s->bitstream_size,
>>> + FFALIGN(avpkt->size, 4));
>>> + if (!s->bitstream)
>>> + return AVERROR(ENOMEM);
>>> +
>>> + s->bdsp.bswap_buf((uint32_t *)s->bitstream,
>>> + (uint32_t *)avpkt->data,
>>> + (avpkt->size + 3) >> 2);
>>> +
>>> + if ((ret = init_get_bits8(gb, s->bitstream, FFALIGN(avpkt->size, 4)))
>>> < 0)
>>
>> Split this in two, just to be safe.
>
> Why?, not mandatory.
It's not, but it's a better practice. Changing old lines is pointless,
but lets try to not add new ones.
>
>>
>>> + return ret;
>>> +
>>> + avctx->pix_fmt = AV_PIX_FMT_YUV420P;
>>> + avctx->color_range = AVCOL_RANGE_JPEG;
>>> +
>>> + scaled = avpkt->data[8];
>>> + if (scaled < 2) {
>>> + int width, height;
>>> + int mode = avpkt->data[10];
>>> +
>>> + switch (mode) {
>>> + case 1:
>>> + width = 352;
>>> + height = 240;
>>> + break;
>>> + case 2:
>>> + width = 704;
>>> + height = 240;
>>> + break;
>>> + case 4:
>>> + width = 480;
>>> + height = 704;
>>> + break;
>>> + case 17:
>>> + width = 352;
>>> + height = 288;
>>> + break;
>>> + case 18:
>>> + width = 704;
>>> + height = 288;
>>> + break;
>>> + default:
>>> + width = 704;
>>> + height = 576;
>>> + break;
>>> + }
>>> +
>>> + if (s->changed_size == 1 &&
>>> + (avctx->width != width || avctx->height != height)) {
>>> + av_log(avctx, AV_LOG_ERROR, "Frame size change is
>>> unsupported.\n");
>>> + return AVERROR_INVALIDDATA;
>>> + }
>>> + avctx->width = width;
>>> + avctx->height = height;
>>
>> ff_set_dimensions()?
>
> These are hardcoded values, always sane.
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel@ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
