avformat/flvenc: add add_keyframe_index option

Add keyframe index metadata
Used to facilitate seeking; particularly for HTTP pseudo streaming.
 1. read live streaming or file by sequence
 2. if use add_keyframe_index option, add a mark flag at the position,
    use to insert new context at the last step.
 3. add the keyframes *offset* and *timestamp* into a list
 4. if use add_keyframe_index option, shift the metadata data from
    mark flag offset
 5. insert the keyframes *offset* and *timestamp* from the list by
    sequence
 6. free the list
 7. end.

Add FATE test case;

Reviewed-by: Lou Logan <lou@lrcd.com>
Signed-off-by: Steven Liu <liuqi@gosun.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

6 files changed, 352 insertions, 10 deletions

diff --git a/doc/muxers.texi b/doc/muxers.texi
index 488ed431ab..806182a148 100644
--- a/doc/muxers.texi
+++ b/doc/muxers.texi
@@ -147,6 +147,9 @@ Place AAC sequence header based on audio stream data.
 
 @item no_sequence_end
 Disable sequence end tag.
+
+@item add_keyframe_index
+Used to facilitate seeking; particularly for HTTP pseudo streaming.
 @end table
 @end table
 
diff --git a/libavformat/flvenc.c b/libavformat/flvenc.c
index 94f95ac4ba..4ca6aea9f7 100644
--- a/libavformat/flvenc.c
+++ b/libavformat/flvenc.c
@@ -24,6 +24,8 @@
 #include "libavutil/intfloat.h"
 #include "libavutil/avassert.h"
 #include "libavutil/mathematics.h"
+#include "avio_internal.h"
+#include "avio.h"
 #include "avc.h"
 #include "avformat.h"
 #include "flv.h"
@@ -64,8 +66,15 @@ static const AVCodecTag flv_audio_codec_ids[] = {
 typedef enum {
     FLV_AAC_SEQ_HEADER_DETECT = (1 << 0),
     FLV_NO_SEQUENCE_END = (1 << 1),
+    FLV_ADD_KEYFRAME_INDEX = (1 << 2),
 } FLVFlags;
 
+typedef struct FLVFileposition {
+    int64_t keyframe_position;
+    double keyframe_timestamp;
+    struct FLVFileposition *next;
+} FLVFileposition;
+
 typedef struct FLVContext {
     AVClass *av_class;
     int     reserved;
@@ -74,6 +83,33 @@ typedef struct FLVContext {
     int64_t duration;
     int64_t delay;      ///< first dts delay (needed for AVC & Speex)
 
+    int64_t datastart_offset;
+    int64_t datasize_offset;
+    int64_t datasize;
+    int64_t videosize_offset;
+    int64_t videosize;
+    int64_t audiosize_offset;
+    int64_t audiosize;
+
+    int64_t metadata_size_pos;
+    int64_t metadata_totalsize_pos;
+    int64_t metadata_totalsize;
+    int64_t keyframe_index_size;
+
+    int64_t lasttimestamp_offset;
+    double lasttimestamp;
+    int64_t lastkeyframetimestamp_offset;
+    double lastkeyframetimestamp;
+    int64_t lastkeyframelocation_offset;
+    int64_t lastkeyframelocation;
+
+    int acurframeindex;
+    int64_t keyframes_info_offset;
+
+    int64_t filepositions_count;
+    FLVFileposition *filepositions;
+    FLVFileposition *head_filepositions;
+
     AVCodecParameters *audio_par;
     AVCodecParameters *video_par;
     double framerate;
@@ -211,6 +247,17 @@ static void put_amf_double(AVIOContext *pb, double d)
     avio_wb64(pb, av_double2int(d));
 }
 
+static void put_amf_byte(AVIOContext *pb, unsigned char abyte)
+{
+    avio_w8(pb, abyte);
+}
+
+static void put_amf_dword_array(AVIOContext *pb, uint32_t dw)
+{
+    avio_w8(pb, AMF_DATA_TYPE_ARRAY);
+    avio_wb32(pb, dw);
+}
+
 static void put_amf_bool(AVIOContext *pb, int b)
 {
     avio_w8(pb, AMF_DATA_TYPE_BOOL);
@@ -222,12 +269,12 @@ static void write_metadata(AVFormatContext *s, unsigned int ts)
     AVIOContext *pb = s->pb;
     FLVContext *flv = s->priv_data;
     int metadata_count = 0;
-    int64_t metadata_size_pos, data_size, metadata_count_pos;
+    int64_t metadata_count_pos;
     AVDictionaryEntry *tag = NULL;
 
     /* write meta_tag */
-    avio_w8(pb, 18);            // tag type META
-    metadata_size_pos = avio_tell(pb);
+    avio_w8(pb, FLV_TAG_TYPE_META);            // tag type META
+    flv->metadata_size_pos = avio_tell(pb);
     avio_wb24(pb, 0);           // size of data part (sum of all parts below)
     avio_wb24(pb, ts);          // timestamp
     avio_wb32(pb, 0);           // reserved
@@ -336,19 +383,87 @@ static void write_metadata(AVFormatContext *s, unsigned int ts)
         put_amf_double(pb, 0); // delayed write
     }
 
+    if (flv->flags & FLV_ADD_KEYFRAME_INDEX) {
+        flv->acurframeindex = 0;
+        flv->keyframe_index_size = 0;
+
+        put_amf_string(pb, "hasVideo");
+        put_amf_bool(pb, !!flv->video_par);
+        metadata_count++;
+
+        put_amf_string(pb, "hasKeyframes");
+        put_amf_bool(pb, 1);
+        metadata_count++;
+
+        put_amf_string(pb, "hasAudio");
+        put_amf_bool(pb, !!flv->audio_par);
+        metadata_count++;
+
+        put_amf_string(pb, "hasMetadata");
+        put_amf_bool(pb, 1);
+        metadata_count++;
+
+        put_amf_string(pb, "canSeekToEnd");
+        put_amf_bool(pb, 1);
+        metadata_count++;
+
+        put_amf_string(pb, "datasize");
+        flv->datasize_offset = avio_tell(pb);
+        flv->datasize = 0;
+        put_amf_double(pb, flv->datasize);
+        metadata_count++;
+
+        put_amf_string(pb, "videosize");
+        flv->videosize_offset = avio_tell(pb);
+        flv->videosize = 0;
+        put_amf_double(pb, flv->videosize);
+        metadata_count++;
+
+        put_amf_string(pb, "audiosize");
+        flv->audiosize_offset = avio_tell(pb);
+        flv->audiosize = 0;
+        put_amf_double(pb, flv->audiosize);
+        metadata_count++;
+
+        put_amf_string(pb, "lasttimestamp");
+        flv->lasttimestamp_offset = avio_tell(pb);
+        flv->lasttimestamp = 0;
+        put_amf_double(pb, 0);
+        metadata_count++;
+
+        put_amf_string(pb, "lastkeyframetimestamp");
+        flv->lastkeyframetimestamp_offset = avio_tell(pb);
+        flv->lastkeyframetimestamp = 0;
+        put_amf_double(pb, 0);
+        metadata_count++;
+
+        put_amf_string(pb, "lastkeyframelocation");
+        flv->lastkeyframelocation_offset = avio_tell(pb);
+        flv->lastkeyframelocation = 0;
+        put_amf_double(pb, 0);
+        metadata_count++;
+
+        put_amf_string(pb, "keyframes");
+        put_amf_byte(pb, AMF_DATA_TYPE_OBJECT);
+        metadata_count++;
+
+        flv->keyframes_info_offset = avio_tell(pb);
+    }
+
     put_amf_string(pb, "");
     avio_w8(pb, AMF_END_OF_OBJECT);
 
     /* write total size of tag */
-    data_size = avio_tell(pb) - metadata_size_pos - 10;
+    flv->metadata_totalsize = avio_tell(pb) - flv->metadata_size_pos - 10;
 
     avio_seek(pb, metadata_count_pos, SEEK_SET);
     avio_wb32(pb, metadata_count);
 
-    avio_seek(pb, metadata_size_pos, SEEK_SET);
-    avio_wb24(pb, data_size);
-    avio_skip(pb, data_size + 10 - 3);
-    avio_wb32(pb, data_size + 11);
+    avio_seek(pb, flv->metadata_size_pos, SEEK_SET);
+    avio_wb24(pb, flv->metadata_totalsize);
+    avio_skip(pb, flv->metadata_totalsize + 10 - 3);
+    flv->metadata_totalsize_pos = avio_tell(pb);
+    avio_wb32(pb, flv->metadata_totalsize + 11);
 }
 
 static int unsupported_codec(AVFormatContext *s,
@@ -425,6 +540,111 @@ static void flv_write_codec_header(AVFormatContext* s, AVCodecParameters* par) {
     }
 }
 
+static int flv_append_keyframe_info(AVFormatContext *s, FLVContext *flv, double ts, int64_t pos)
+{
+    FLVFileposition *position = av_malloc(sizeof(FLVFileposition));
+
+    if (!position) {
+        av_log(s, AV_LOG_WARNING, "no mem for add keyframe index!\n");
+        return AVERROR(ENOMEM);
+    }
+
+    position->keyframe_timestamp = ts;
+    position->keyframe_position = pos;
+
+    if (!flv->filepositions_count) {
+        flv->filepositions = position;
+        flv->head_filepositions = flv->filepositions;
+        position->next = NULL;
+    } else {
+        flv->filepositions->next = position;
+        position->next = NULL;
+        flv->filepositions = flv->filepositions->next;
+    }
+
+    flv->filepositions_count++;
+
+    return 0;
+}
+
+static int shift_data(AVFormatContext *s)
+{
+    int ret = 0;
+    int n = 0;
+    int64_t metadata_size = 0;
+    FLVContext *flv = s->priv_data;
+    int64_t pos, pos_end = avio_tell(s->pb);
+    uint8_t *buf, *read_buf[2];
+    int read_buf_id = 0;
+    int read_size[2];
+    AVIOContext *read_pb;
+
+    metadata_size = flv->filepositions_count * 9 * 2 + 10; /* filepositions and times value */
+    metadata_size += 2 + 13; /* filepositions String */
+    metadata_size += 2 + 5; /* times String */
+    metadata_size += 3; /* Object end */
+
+    flv->keyframe_index_size = metadata_size;
+
+    if (metadata_size < 0)
+        return metadata_size;
+
+    buf = av_malloc_array(metadata_size, 2);
+    if (!buf) {
+        return AVERROR(ENOMEM);
+    }
+    read_buf[0] = buf;
+    read_buf[1] = buf + metadata_size;
+
+    avio_seek(s->pb, flv->metadata_size_pos, SEEK_SET);
+    avio_wb24(s->pb, flv->metadata_totalsize + metadata_size);
+
+    avio_seek(s->pb, flv->metadata_totalsize_pos, SEEK_SET);
+    avio_wb32(s->pb, flv->metadata_totalsize + 11 + metadata_size);
+    avio_seek(s->pb, pos_end, SEEK_SET);
+
+    /* Shift the data: the AVIO context of the output can only be used for
+     * writing, so we re-open the same output, but for reading. It also avoids
+     * a read/seek/write/seek back and forth. */
+    avio_flush(s->pb);
+    ret = s->io_open(s, &read_pb, s->filename, AVIO_FLAG_READ, NULL);
+    if (ret < 0) {
+        av_log(s, AV_LOG_ERROR, "Unable to re-open %s output file for "
+               "the second pass (add_keyframe_index)\n", s->filename);
+        goto end;
+    }
+
+    /* mark the end of the shift to up to the last data we wrote, and get ready
+     * for writing */
+    pos_end = avio_tell(s->pb);
+    avio_seek(s->pb, flv->keyframes_info_offset + metadata_size, SEEK_SET);
+
+    /* start reading at where the keyframe index information will be placed */
+    avio_seek(read_pb, flv->keyframes_info_offset, SEEK_SET);
+    pos = avio_tell(read_pb);
+
+    /* shift data by chunk of at most keyframe *filepositions* and *times* size */
+    read_size[read_buf_id] = avio_read(read_pb, read_buf[read_buf_id], metadata_size);  \
+    read_buf_id ^= 1;
+    do {
+
+        read_size[read_buf_id] = avio_read(read_pb, read_buf[read_buf_id], metadata_size);  \
+        read_buf_id ^= 1;
+        n = read_size[read_buf_id];
+        if (n < 0)
+            break;
+        avio_write(s->pb, read_buf[read_buf_id], n);
+        pos += n;
+    } while (pos <= pos_end);
+
+    ff_format_io_close(s, &read_pb);
+
+end:
+    av_free(buf);
+    return ret;
+}
+
+
 static int flv_write_header(AVFormatContext *s)
 {
     int i;
@@ -530,17 +750,75 @@ static int flv_write_header(AVFormatContext *s)
         flv_write_codec_header(s, s->streams[i]->codecpar);
     }
 
+    flv->datastart_offset = avio_tell(pb);
     return 0;
 }
 
 static int flv_write_trailer(AVFormatContext *s)
 {
     int64_t file_size;
-
     AVIOContext *pb = s->pb;
     FLVContext *flv = s->priv_data;
-    int i;
+    int build_keyframes_idx = flv->flags & FLV_ADD_KEYFRAME_INDEX;
+    int i, res;
+    int64_t cur_pos = avio_tell(s->pb);
+
+    if (build_keyframes_idx) {
+        FLVFileposition *newflv_posinfo, *p;
+
+        avio_seek(pb, flv->videosize_offset, SEEK_SET);
+        put_amf_double(pb, flv->videosize);
+
+        avio_seek(pb, flv->audiosize_offset, SEEK_SET);
+        put_amf_double(pb, flv->audiosize);
+
+        avio_seek(pb, flv->lasttimestamp_offset, SEEK_SET);
+        put_amf_double(pb, flv->lasttimestamp);
+
+        avio_seek(pb, flv->lastkeyframetimestamp_offset, SEEK_SET);
+        put_amf_double(pb, flv->lastkeyframetimestamp);
 
+        avio_seek(pb, flv->lastkeyframelocation_offset, SEEK_SET);
+        put_amf_double(pb, flv->lastkeyframelocation + flv->keyframe_index_size);
+        avio_seek(pb, cur_pos, SEEK_SET);
+
+        res = shift_data(s);
+        if (res < 0) {
+             goto end;
+        }
+        avio_seek(pb, flv->keyframes_info_offset, SEEK_SET);
+        put_amf_string(pb, "filepositions");
+        put_amf_dword_array(pb, flv->filepositions_count);
+        for (newflv_posinfo = flv->head_filepositions; newflv_posinfo; newflv_posinfo = newflv_posinfo->next) {
+            put_amf_double(pb, newflv_posinfo->keyframe_position + flv->keyframe_index_size);
+        }
+
+        put_amf_string(pb, "times");
+        put_amf_dword_array(pb, flv->filepositions_count);
+        for (newflv_posinfo = flv->head_filepositions; newflv_posinfo; newflv_posinfo = newflv_posinfo->next) {
+            put_amf_double(pb, newflv_posinfo->keyframe_timestamp);
+        }
+
+        newflv_posinfo = flv->head_filepositions;
+        while (newflv_posinfo) {
+            p = newflv_posinfo->next;
+            if (p) {
+                newflv_posinfo->next = p->next;
+                av_free(p);
+                p = NULL;
+            } else {
+                av_free(newflv_posinfo);
+                newflv_posinfo = NULL;
+            }
+        }
+
+        put_amf_string(pb, "");
+        avio_w8(pb, AMF_END_OF_OBJECT);
+
+        avio_seek(pb, cur_pos + flv->keyframe_index_size, SEEK_SET);
+    }
+
+end:
     if (flv->flags & FLV_NO_SEQUENCE_END) {
         av_log(s, AV_LOG_DEBUG, "FLV no sequence end mode open\n");
     } else {
@@ -556,6 +834,11 @@ static int flv_write_trailer(AVFormatContext *s)
 
     file_size = avio_tell(pb);
 
+    if (build_keyframes_idx) {
+        flv->datasize = file_size - flv->datastart_offset;
+        avio_seek(pb, flv->datasize_offset, SEEK_SET);
+        put_amf_double(pb, flv->datasize);
+    }
     if (pb->seekable) {
         /* update information */
         if (avio_seek(pb, flv->duration_offset, SEEK_SET) < 0) {
@@ -583,6 +866,7 @@ static int flv_write_packet(AVFormatContext *s, AVPacket *pkt)
     int size = pkt->size;
     uint8_t *data = NULL;
     int flags = -1, flags_size, ret;
+    int64_t cur_offset = avio_tell(pb);
 
     if (par->codec_id == AV_CODEC_ID_VP6F || par->codec_id == AV_CODEC_ID_VP6A ||
         par->codec_id == AV_CODEC_ID_VP6  || par->codec_id == AV_CODEC_ID_AAC)
@@ -736,6 +1020,32 @@ static int flv_write_packet(AVFormatContext *s, AVPacket *pkt)
                               pkt->pts + flv->delay + pkt->duration);
     }
 
+    if (flv->flags & FLV_ADD_KEYFRAME_INDEX) {
+        switch (par->codec_type) {
+            case AVMEDIA_TYPE_VIDEO:
+                flv->videosize += (avio_tell(pb) - cur_offset);
+                flv->lasttimestamp = flv->acurframeindex / flv->framerate;
+                if (pkt->flags & AV_PKT_FLAG_KEY) {
+                    double ts = flv->acurframeindex / flv->framerate;
+                    int64_t pos = cur_offset;
+
+                    flv->lastkeyframetimestamp = flv->acurframeindex / flv->framerate;
+                    flv->lastkeyframelocation = pos;
+                    flv_append_keyframe_info(s, flv, ts, pos);
+                }
+                flv->acurframeindex++;
+                break;
+
+            case AVMEDIA_TYPE_AUDIO:
+                flv->audiosize += (avio_tell(pb) - cur_offset);
+                break;
+
+            default:
+                av_log(s, AV_LOG_WARNING, "par->codec_type is type = [%d]\n", par->codec_type);
+                break;
+        }
+    }
+
     av_free(data);
 
     return pb->error;
@@ -745,6 +1055,7 @@ static const AVOption options[] = {
     { "flvflags", "FLV muxer flags", offsetof(FLVContext, flags), AV_OPT_TYPE_FLAGS, {.i64 = 0}, INT_MIN, INT_MAX, AV_OPT_FLAG_ENCODING_PARAM, "flvflags" },
     { "aac_seq_header_detect", "Put AAC sequence header based on stream data", 0, AV_OPT_TYPE_CONST, {.i64 = FLV_AAC_SEQ_HEADER_DETECT}, INT_MIN, INT_MAX, AV_OPT_FLAG_ENCODING_PARAM, "flvflags" },
     { "no_sequence_end", "disable sequence end for FLV", 0, AV_OPT_TYPE_CONST, {.i64 = FLV_NO_SEQUENCE_END}, INT_MIN, INT_MAX, AV_OPT_FLAG_ENCODING_PARAM, "flvflags" },
+    { "add_keyframe_index", "Add keyframe index metadata", 0, AV_OPT_TYPE_CONST, {.i64 = FLV_ADD_KEYFRAME_INDEX}, INT_MIN, INT_MAX, AV_OPT_FLAG_ENCODING_PARAM, "flvflags" },
     { NULL },
 };
 
diff --git a/tests/Makefile b/tests/Makefile
index 8e810ff195..07e839e316 100644
--- a/tests/Makefile
+++ b/tests/Makefile
@@ -132,6 +132,7 @@ include $(SRC_PATH)/tests/fate/fifo-muxer.mak
 include $(SRC_PATH)/tests/fate/filter-audio.mak
 include $(SRC_PATH)/tests/fate/filter-video.mak
 include $(SRC_PATH)/tests/fate/flac.mak
+include $(SRC_PATH)/tests/fate/flvenc.mak
 include $(SRC_PATH)/tests/fate/gapless.mak
 include $(SRC_PATH)/tests/fate/gif.mak
 include $(SRC_PATH)/tests/fate/h264.mak
diff --git a/tests/fate-run.sh b/tests/fate-run.sh
index c640cc5794..29a36cea3a 100755
--- a/tests/fate-run.sh
+++ b/tests/fate-run.sh
@@ -129,6 +129,10 @@ framecrc(){
     ffmpeg "$@" -flags +bitexact -fflags +bitexact -f framecrc -
 }
 
+ffmetadata(){
+    ffmpeg "$@" -flags +bitexact -fflags +bitexact -f ffmetadata -
+}
+
 framemd5(){
     ffmpeg "$@" -flags +bitexact -fflags +bitexact -f framemd5 -
 }
diff --git a/tests/fate/flvenc.mak b/tests/fate/flvenc.mak
new file mode 100644
index 0000000000..3881bcd105
--- /dev/null
+++ b/tests/fate/flvenc.mak
@@ -0,0 +1,11 @@
+tests/data/add_keyframe_index.flv: TAG = GEN
+tests/data/add_keyframe_index.flv: ffmpeg$(PROGSSUF)$(EXESUF) | tests/data
+	$(M)$(TARGET_EXEC) $(TARGET_PATH)/$< \
+		-f lavfi -i "sws_flags=+accurate_rnd+bitexact;testsrc=r=7:n=2:d=20" -sws_flags '+accurate_rnd+bitexact' -metadata "encoder=Lavf" -pix_fmt yuv420p -c:v flv1 -g 7 -f flv -flags +bitexact \
+		-flvflags add_keyframe_index -idct simple -dct int -y $(TARGET_PATH)/tests/data/add_keyframe_index.flv 2> /dev/null;
+
+FATE_AFILTER-$(call ALLYES, FLV_MUXER FLV_DEMUXER AVDEVICE TESTSRC_FILTER LAVFI_INDEV FLV_ENCODER) += fate-flv-add_keyframe_index
+fate-flv-add_keyframe_index: tests/data/add_keyframe_index.flv
+fate-flv-add_keyframe_index: CMD = ffmetadata -flags +bitexact -i $(TARGET_PATH)/tests/data/add_keyframe_index.flv
+
+
diff --git a/tests/ref/fate/flv-add_keyframe_index b/tests/ref/fate/flv-add_keyframe_index
new file mode 100644
index 0000000000..1c4da65d90
--- /dev/null
+++ b/tests/ref/fate/flv-add_keyframe_index
@@ -0,0 +1,12 @@
+;FFMETADATA1
+hasVideo=true
+hasKeyframes=true
+hasAudio=false
+hasMetadata=true
+canSeekToEnd=true
+datasize=629776
+videosize=629381
+audiosize=0
+lasttimestamp=20
+lastkeyframetimestamp=19
+lastkeyframelocation=597963