[FFmpeg-devel] [PATCH] WIP: lavc/h264_mp4toannexb_bsf: Support HEVC

Rodger Combs rodger.combs at gmail.com
Fri Jun 26 01:46:48 CEST 2015


This is the result of me screwing around to see how complex of a task this
would be; the answer is evidently "not particularly complex". I seem to have
at least a few issues to work out, since the .ts files I'm generating don't
always seek correctly, but it's a start.
Would appreciate it if anyone who knows a bit more about how AVC and HEVC
differ here could provide some guidance on what's left to do.
---
 libavcodec/h264_mp4toannexb_bsf.c | 126 ++++++++++++++++++++++++--------------
 1 file changed, 81 insertions(+), 45 deletions(-)

diff --git a/libavcodec/h264_mp4toannexb_bsf.c b/libavcodec/h264_mp4toannexb_bsf.c
index ae96ee9..7cf8bc2 100644
--- a/libavcodec/h264_mp4toannexb_bsf.c
+++ b/libavcodec/h264_mp4toannexb_bsf.c
@@ -32,6 +32,7 @@ typedef struct H264BSFContext {
     uint8_t  new_idr;
     uint8_t  idr_sps_seen;
     uint8_t  idr_pps_seen;
+    uint8_t  idr_vps_seen;
     int      extradata_parsed;
 } H264BSFContext;
 
@@ -63,56 +64,83 @@ static int alloc_and_copy(uint8_t **poutbuf, int *poutbuf_size,
     return 0;
 }
 
+#define NAL_SPS 33
+#define NAL_PPS 34
+
 static int h264_extradata_to_annexb(H264BSFContext *ctx, AVCodecContext *avctx, const int padding)
 {
-    uint16_t unit_size;
+    int hevc = avctx->codec_id == AV_CODEC_ID_HEVC;
+    uint16_t unit_size, unit_nb;
     uint64_t total_size                 = 0;
-    uint8_t *out                        = NULL, unit_nb, sps_done = 0,
-             sps_seen                   = 0, pps_seen = 0;
-    const uint8_t *extradata            = avctx->extradata + 4;
+    uint8_t *out                        = NULL, sps_done = 0,
+             sps_seen                   = 0, pps_seen = 0, set_nb;
+    const uint8_t *extradata            = avctx->extradata + (hevc ? 21 : 4);
     static const uint8_t nalu_header[4] = { 0, 0, 0, 1 };
     int length_size = (*extradata++ & 0x3) + 1; // retrieve length coded size
 
     ctx->sps_offset = ctx->pps_offset = -1;
 
-    /* retrieve sps and pps unit(s) */
-    unit_nb = *extradata++ & 0x1f; /* number of sps unit(s) */
-    if (!unit_nb) {
-        goto pps;
-    } else {
-        ctx->sps_offset = 0;
-        sps_seen = 1;
-    }
-
-    while (unit_nb--) {
-        int err;
+    set_nb = hevc ? *extradata++ : 1;
 
-        unit_size   = AV_RB16(extradata);
-        total_size += unit_size + 4;
-        if (total_size > INT_MAX - padding) {
-            av_log(avctx, AV_LOG_ERROR,
-                   "Too big extradata size, corrupted stream or invalid MP4/AVCC bitstream\n");
-            av_free(out);
-            return AVERROR(EINVAL);
-        }
-        if (extradata + 2 + unit_size > avctx->extradata + avctx->extradata_size) {
-            av_log(avctx, AV_LOG_ERROR, "Packet header is not contained in global extradata, "
-                   "corrupted stream or invalid MP4/AVCC bitstream\n");
-            av_free(out);
-            return AVERROR(EINVAL);
-        }
-        if ((err = av_reallocp(&out, total_size + padding)) < 0)
-            return err;
-        memcpy(out + total_size - unit_size - 4, nalu_header, 4);
-        memcpy(out + total_size - unit_size, extradata + 2, unit_size);
-        extradata += 2 + unit_size;
-pps:
-        if (!unit_nb && !sps_done++) {
-            unit_nb = *extradata++; /* number of pps unit(s) */
-            if (unit_nb) {
-                ctx->pps_offset = (extradata - 1) - (avctx->extradata + 4);
+    while (set_nb--) {
+        int type = hevc ? *extradata++ & 0x3f : 0;
+        /* retrieve sps and pps unit(s) */
+        if (hevc) {
+            unit_nb = AV_RB16(extradata);
+            extradata += 2;
+            if (type == NAL_SPS) {
+                ctx->sps_offset = total_size;
+                sps_seen = 1;
+            } else if (type == NAL_PPS) {
+                ctx->pps_offset = total_size;
                 pps_seen = 1;
             }
+        } else {
+            unit_nb = *extradata++ & 0x1f; /* number of units */
+            if (!unit_nb) {
+                type = NAL_PPS;
+                goto pps;
+            } else {
+                type = NAL_SPS;
+                ctx->sps_offset = 0;
+                sps_seen = 1;
+            }
+        }
+
+        while (unit_nb--) {
+            int err;
+
+            unit_size   = AV_RB16(extradata);
+            if (type == NAL_SPS || type == NAL_PPS || !hevc) {
+                if (!hevc)
+                    ctx->sps_offset = total_size;
+                total_size += unit_size + 4;
+                if (total_size > INT_MAX - padding) {
+                    av_log(avctx, AV_LOG_ERROR,
+                           "Too big extradata size, corrupted stream or invalid MP4/AVCC bitstream\n");
+                    av_free(out);
+                    return AVERROR(EINVAL);
+                }
+                if (extradata + 2 + unit_size > avctx->extradata + avctx->extradata_size) {
+                    av_log(avctx, AV_LOG_ERROR, "Packet header is not contained in global extradata, "
+                           "corrupted stream or invalid MP4/AVCC bitstream\n");
+                    av_free(out);
+                    return AVERROR(EINVAL);
+                }
+                if ((err = av_reallocp(&out, total_size + padding)) < 0)
+                    return err;
+                memcpy(out + total_size - unit_size - 4, nalu_header, 4);
+                memcpy(out + total_size - unit_size, extradata + 2, unit_size);
+            }
+            extradata += 2 + unit_size;
+    pps:
+            if (!hevc && !unit_nb && !sps_done++) {
+                unit_nb = *extradata++; /* number of pps unit(s) */
+                if (unit_nb) {
+                    ctx->pps_offset = (extradata - 1) - (avctx->extradata + 4);
+                    pps_seen = 1;
+                }
+            }
         }
     }
 
@@ -143,12 +171,18 @@ static int h264_mp4toannexb_filter(AVBitStreamFilterContext *bsfc,
                                    int keyframe)
 {
     H264BSFContext *ctx = bsfc->priv_data;
+    int hevc = avctx->codec_id == AV_CODEC_ID_HEVC;
     int i;
     uint8_t unit_type;
     int32_t nal_size;
     uint32_t cumul_size    = 0;
     const uint8_t *buf_end = buf + buf_size;
     int ret = 0;
+    uint8_t vps_type = hevc ? 32 : 0;
+    uint8_t sps_type = hevc ? 33 : 7;
+    uint8_t pps_type = hevc ? 34 : 8;
+    uint8_t idr_type = hevc ? 19 : 5;
+    uint8_t idr_type2 = hevc ? 20 : 5;
 
     /* nothing to filter */
     if (!avctx->extradata || avctx->extradata_size < 6) {
@@ -180,14 +214,16 @@ static int h264_mp4toannexb_filter(AVBitStreamFilterContext *bsfc,
             nal_size = (nal_size << 8) | buf[i];
 
         buf      += ctx->length_size;
-        unit_type = *buf & 0x1f;
+        unit_type = hevc ? (*buf >> 1) & 0x3f : *buf & 0x1f;
 
         if (buf + nal_size > buf_end || nal_size < 0)
             goto fail;
 
-        if (unit_type == 7)
+        if (unit_type == vps_type)
+            ctx->idr_vps_seen = ctx->new_idr = 1;
+        else if (unit_type == sps_type)
             ctx->idr_sps_seen = ctx->new_idr = 1;
-        else if (unit_type == 8) {
+        else if (unit_type == pps_type) {
             ctx->idr_pps_seen = ctx->new_idr = 1;
             /* if SPS has not been seen yet, prepend the AVCC one to PPS */
             if (!ctx->idr_sps_seen) {
@@ -208,18 +244,18 @@ static int h264_mp4toannexb_filter(AVBitStreamFilterContext *bsfc,
         /* if this is a new IDR picture following an IDR picture, reset the idr flag.
          * Just check first_mb_in_slice to be 0 as this is the simplest solution.
          * This could be checking idr_pic_id instead, but would complexify the parsing. */
-        if (!ctx->new_idr && unit_type == 5 && (buf[1] & 0x80))
+        if (!ctx->new_idr && (unit_type == idr_type || unit_type == idr_type2) && (buf[1] & 0x80))
             ctx->new_idr = 1;
 
         /* prepend only to the first type 5 NAL unit of an IDR picture, if no sps/pps are already present */
-        if (ctx->new_idr && unit_type == 5 && !ctx->idr_sps_seen && !ctx->idr_pps_seen) {
+        if (ctx->new_idr && (unit_type == idr_type || unit_type == idr_type2) && !ctx->idr_sps_seen && !ctx->idr_pps_seen) {
             if ((ret=alloc_and_copy(poutbuf, poutbuf_size,
                                avctx->extradata, avctx->extradata_size,
                                buf, nal_size)) < 0)
                 goto fail;
             ctx->new_idr = 0;
         /* if only SPS has been seen, also insert PPS */
-        } else if (ctx->new_idr && unit_type == 5 && ctx->idr_sps_seen && !ctx->idr_pps_seen) {
+        } else if (ctx->new_idr && (unit_type == idr_type || unit_type == idr_type2) && ctx->idr_sps_seen && !ctx->idr_pps_seen) {
             if (ctx->pps_offset == -1) {
                 av_log(avctx, AV_LOG_WARNING, "PPS not present in the stream, nor in AVCC, stream may be unreadable\n");
                 if ((ret = alloc_and_copy(poutbuf, poutbuf_size,
-- 
2.4.1



More information about the ffmpeg-devel mailing list