diff mbox series

[FFmpeg-devel,v3,3/5] avcodec/h264_mp4toannexb_bsf: fix missing PS before IDR frames

Message ID tencent_F4AFAA3B29C0D1B6DC69EE021168BE7F0A0A@qq.com
State New
Headers show
Series [FFmpeg-devel,v3,1/5] avcodec/h264_mp4toannexb_bsf: refactor start_code_size handling | expand

Checks

Context Check Description
yinshiyou/make_loongarch64 success Make finished
yinshiyou/make_fate_loongarch64 fail Make fate failed
andriy/make_x86 success Make finished
andriy/make_fate_x86 fail Make fate failed

Commit Message

Zhao Zhili Nov. 17, 2023, 5:24 p.m. UTC
From: Zhao Zhili <zhilizhao@tencent.com>

If there is a single group of SPS/PPS before an IDR frame, but no
SPS/PPS after that, we will miss the chance to reset
idr_sps_seen/idr_pps_seen. No SPS/PPS are inserted afterwards.

This patch saves in-band SPS/PPS and insert them before IDR frames
when necessary.
---
 libavcodec/h264_mp4toannexb_bsf.c | 83 ++++++++++++++++++++++++++++---
 tests/fate/h264.mak               |  8 ++-
 2 files changed, 84 insertions(+), 7 deletions(-)

Comments

Zhao Zhili Nov. 20, 2023, 3:09 a.m. UTC | #1
> On Nov 18, 2023, at 01:24, Zhao Zhili <quinkblack@foxmail.com> wrote:
> 
> From: Zhao Zhili <zhilizhao@tencent.com>
> 
> If there is a single group of SPS/PPS before an IDR frame, but no
> SPS/PPS after that, we will miss the chance to reset
> idr_sps_seen/idr_pps_seen. No SPS/PPS are inserted afterwards.
> 
> This patch saves in-band SPS/PPS and insert them before IDR frames
> when necessary.
> —


Please help upload the sample to fate h264/ directory. I have sent request
to samples-request too.

md5sum
09e56f41c2c804c3e1110bc2170aaa06  ps_prefix_first_idr.mp4



https://drive.google.com/file/d/1_jf8WmhEjiK2LKY8ZAtnCEubrHR0kJ1w/view?usp=sharing
diff mbox series

Patch

diff --git a/libavcodec/h264_mp4toannexb_bsf.c b/libavcodec/h264_mp4toannexb_bsf.c
index 846671abb6..4073c780c5 100644
--- a/libavcodec/h264_mp4toannexb_bsf.c
+++ b/libavcodec/h264_mp4toannexb_bsf.c
@@ -36,6 +36,8 @@  typedef struct H264BSFContext {
     uint8_t *pps;
     int      sps_size;
     int      pps_size;
+    unsigned sps_buf_size;
+    unsigned pps_buf_size;
     uint8_t  length_size;
     uint8_t  new_idr;
     uint8_t  idr_sps_seen;
@@ -130,16 +132,33 @@  pps:
         memset(out + total_size, 0, padding);
 
     if (pps_offset) {
-        s->sps      = out;
+        uint8_t *sps;
+
         s->sps_size = pps_offset;
+        sps = av_fast_realloc(s->sps, &s->sps_buf_size, s->sps_size);
+        if (!sps) {
+            av_free(out);
+            return AVERROR(ENOMEM);
+        }
+        s->sps = sps;
+        memcpy(s->sps, out, s->sps_size);
     } else {
         av_log(ctx, AV_LOG_WARNING,
                "Warning: SPS NALU missing or invalid. "
                "The resulting stream may not play.\n");
     }
     if (pps_offset < total_size) {
-        s->pps      = out + pps_offset;
+        uint8_t *pps;
+
         s->pps_size = total_size - pps_offset;
+        pps = av_fast_realloc(s->pps, &s->pps_buf_size, s->pps_size);
+        if (!pps) {
+            av_freep(&s->sps);
+            av_free(out);
+            return AVERROR(ENOMEM);
+        }
+        s->pps = pps;
+        memcpy(s->pps, out + pps_offset, s->pps_size);
     } else {
         av_log(ctx, AV_LOG_WARNING,
                "Warning: PPS NALU missing or invalid. "
@@ -153,6 +172,35 @@  pps:
     return length_size;
 }
 
+static int h264_mp4toannexb_save_ps(uint8_t **dst, int *dst_size,
+                                    unsigned *dst_buf_size,
+                                    const uint8_t *nal, uint32_t nal_size,
+                                    int first)
+{
+    static const uint8_t nalu_header[4] = { 0, 0, 0, 1 };
+    const int start_code_size = sizeof(nalu_header);
+    uint8_t *ptr;
+    uint32_t size;
+
+    if (first)
+        size = 0;
+    else
+        size = *dst_size;
+
+    ptr = av_fast_realloc(*dst, dst_buf_size, size + nal_size + start_code_size);
+    if (!ptr)
+        return AVERROR(ENOMEM);
+
+    memcpy(ptr + size, nalu_header, start_code_size);
+    size += start_code_size;
+    memcpy(ptr + size, nal, nal_size);
+    size += nal_size;
+
+    *dst = ptr;
+    *dst_size = size;
+    return 0;
+}
+
 static int h264_mp4toannexb_init(AVBSFContext *ctx)
 {
     H264BSFContext *s = ctx->priv_data;
@@ -211,6 +259,9 @@  static int h264_mp4toannexb_filter(AVBSFContext *ctx, AVPacket *opkt)
     if (j) \
         av_log(__VA_ARGS__)
     for (int j = 0; j < 2; j++) {
+        int sps_count = 0;
+        int pps_count = 0;
+
         buf      = in->data;
         new_idr  = s->new_idr;
         sps_seen = s->idr_sps_seen;
@@ -241,8 +292,18 @@  static int h264_mp4toannexb_filter(AVBSFContext *ctx, AVPacket *opkt)
 
             if (unit_type == H264_NAL_SPS) {
                 sps_seen = new_idr = 1;
+                if (!j) {
+                    h264_mp4toannexb_save_ps(&s->sps, &s->sps_size, &s->sps_buf_size,
+                                             buf, nal_size, !sps_count);
+                    sps_count++;
+                }
             } else if (unit_type == H264_NAL_PPS) {
                 pps_seen = new_idr = 1;
+                if (!j) {
+                    h264_mp4toannexb_save_ps(&s->pps, &s->pps_size, &s->pps_buf_size,
+                                             buf, nal_size, !pps_count);
+                    pps_count++;
+                }
                 /* if SPS has not been seen yet, prepend the AVCC one to PPS */
                 if (!sps_seen) {
                     if (!s->sps_size) {
@@ -262,9 +323,10 @@  static int h264_mp4toannexb_filter(AVBSFContext *ctx, AVPacket *opkt)
 
             /* prepend only to the first type 5 NAL unit of an IDR picture, if no sps/pps are already present */
             if (new_idr && unit_type == H264_NAL_IDR_SLICE && !sps_seen && !pps_seen) {
-                if (ctx->par_out->extradata)
-                    count_or_copy(&out, &out_size, ctx->par_out->extradata,
-                                  ctx->par_out->extradata_size, PS_OUT_OF_BAND, j);
+                if (s->sps_size)
+                    count_or_copy(&out, &out_size, s->sps, s->sps_size, PS_OUT_OF_BAND, j);
+                if (s->pps_size)
+                    count_or_copy(&out, &out_size, s->pps, s->pps_size, PS_OUT_OF_BAND, j);
                 new_idr = 0;
             /* if only SPS has been seen, also insert PPS */
             } else if (new_idr && unit_type == H264_NAL_IDR_SLICE && sps_seen && !pps_seen) {
@@ -280,7 +342,7 @@  static int h264_mp4toannexb_filter(AVBSFContext *ctx, AVPacket *opkt)
             else
                 ps = PS_NONE;
             count_or_copy(&out, &out_size, buf, nal_size, ps, j);
-            if (!new_idr && unit_type == H264_NAL_SLICE) {
+            if (unit_type == H264_NAL_SLICE) {
                 new_idr  = 1;
                 sps_seen = 0;
                 pps_seen = 0;
@@ -320,6 +382,14 @@  fail:
     return ret;
 }
 
+static void h264_mp4toannexb_close(AVBSFContext *ctx)
+{
+    H264BSFContext *s = ctx->priv_data;
+
+    av_freep(&s->sps);
+    av_freep(&s->pps);
+}
+
 static void h264_mp4toannexb_flush(AVBSFContext *ctx)
 {
     H264BSFContext *s = ctx->priv_data;
@@ -339,5 +409,6 @@  const FFBitStreamFilter ff_h264_mp4toannexb_bsf = {
     .priv_data_size = sizeof(H264BSFContext),
     .init           = h264_mp4toannexb_init,
     .filter         = h264_mp4toannexb_filter,
+    .close          = h264_mp4toannexb_close,
     .flush          = h264_mp4toannexb_flush,
 };
diff --git a/tests/fate/h264.mak b/tests/fate/h264.mak
index 0ff3d6df55..c7e0d0a84e 100644
--- a/tests/fate/h264.mak
+++ b/tests/fate/h264.mak
@@ -224,7 +224,9 @@  FATE_H264-$(call FRAMECRC, MOV, H264) += fate-h264-unescaped-extradata
 # this sample contains field-coded frames, with both fields in a single packet
 FATE_H264-$(call FRAMECRC, MOV, H264) += fate-h264-twofields-packet
 
-FATE_H264-$(call DEMMUX, MOV, H264, H264_MP4TOANNEXB_BSF) += fate-h264-bsf-mp4toannexb
+FATE_H264-$(call DEMMUX, MOV, H264, H264_MP4TOANNEXB_BSF) += fate-h264-bsf-mp4toannexb \
+                                                             fate-h264-bsf-mp4toannexb-2
+
 FATE_H264-$(call FRAMECRC, MATROSKA, H264) += fate-h264-direct-bff
 FATE_H264-$(call FRAMECRC, FLV, H264, SCALE_FILTER) += fate-h264-brokensps-2580
 FATE_H264-$(call FRAMECRC, MXF, H264, PCM_S24LE_DECODER SCALE_FILTER ARESAMPLE_FILTER) += fate-h264-xavc-4389
@@ -426,6 +428,10 @@  fate-h264-conformance-sva_nl1_b:                  CMD = framecrc -i $(TARGET_SAM
 fate-h264-conformance-sva_nl2_e:                  CMD = framecrc -i $(TARGET_SAMPLES)/h264-conformance/SVA_NL2_E.264
 
 fate-h264-bsf-mp4toannexb:                        CMD = md5 -i $(TARGET_SAMPLES)/h264/interlaced_crop.mp4 -c:v copy -f h264
+# First IDR is prefixed by SPS/PPS
+fate-h264-bsf-mp4toannexb-2:                      CMD = md5 -i $(TARGET_SAMPLES)/h264/ps_prefix_first_idr.mp4 -c:v copy -f h264
+fate-h264-bsf-mp4toannexb-2:                      CMP = oneline
+fate-h264-bsf-mp4toannexb-2:                      REF = cffcfa6a2d0b58c9de1f5785f099f41d
 
 fate-h264-crop-to-container:                      CMD = framemd5 -i $(TARGET_SAMPLES)/h264/crop-to-container-dims-canon.mov
 fate-h264-direct-bff:                             CMD = framecrc -i $(TARGET_SAMPLES)/h264/direct-bff.mkv