update

AudioTrackMerge.c

@@ -7,15 +7,12 @@
 #include <libavfilter/buffersink.h>
 #include <libavutil/opt.h>
 
-#define INPUT_FMT AV_SAMPLE_FMT_FLTP
-#define INPUT_SAMPLE_RATE 44100
-#define INPUT_TB_DEN 14112000
-#define INPUT_CHANNEL_LAYOUT AV_CH_LAYOUT_STEREO
-
-
-uint8_t errstr[1024];
+/*****************************************************************************/
+/*                                Declarations                               */
+/*****************************************************************************/
 
 typedef struct MixParameters {
+    unsigned int   index;
     AVCodecContext *cc;
     AVStream       *s;
     unsigned       offset;
@@ -32,7 +29,7 @@ typedef struct AudioStreamingContext {
 
 typedef struct FilteringContext {
     AVFilterGraph   *graph;
-    AVFilterContext *src;
+    AVFilterContext **src;
     AVFilterContext *sink;
     unsigned int     numOfSrcs;
 } FilteringContext;
@@ -59,6 +56,7 @@ FilterChain            build_filter_chain_mix_part(AVFilterGraph *graph, int num
 int                    mixing_audios(int numOfStream, AudioStreamingContext* inputs[],
                                      AudioStreamingContext *output,
                                      FilteringContext* filter);
+int                    write_out(AudioStreamingContext *encoder, AVFrame *frame);
 
 int prepare_audio_encoder(AudioStreamingContext *encoder,
                           AudioStreamingContext *decoder);
@@ -70,15 +68,42 @@ int add_to_src(AudioStreamingContext *decoder, AVFilterContext *src,
 
 int process_filtered_frames(AudioStreamingContext *encoder, AVFrame *frame);
 
-static int init_filter_graph(AVFilterGraph **graph, AVFilterContext **src,
-                             AVFilterContext **src_,
-                             AVFilterContext **sink,
-                             AudioStreamingContext *decoder_1,
-                             AudioStreamingContext *decoder_2);
 
 
+/*****************************************************************************/
+/*                                Definitions                                */
+/*****************************************************************************/
+void audioStreamCtx_Release(AudioStreamingContext **s_ctx) {
+    int err = 0;
+
+    AudioStreamingContext *ctx = *s_ctx;
+
+    // Format Cleanup
+    avformat_close_input(&ctx->fmt);
+
+    // CoDec context Cleanup
+    avcodec_free_context(&ctx->cc);
+    avcodec_close(ctx->cc);
+}
+
+void audioStreamCtx_Release_free(AudioStreamingContext **s_ctx) {
+    audioStreamCtx_Release(s_ctx);
+    free(*s_ctx);
+    *s_ctx = NULL;
+}
+
+void filterCtx_Release(FilteringContext **f_ctx) {
+    FilteringContext *ctx = *f_ctx;
+    avfilter_graph_free(&ctx->graph);
+
+    free(ctx->src);
+    free(*f_ctx);
+    *f_ctx = NULL;
+}
+
 int audioMixArbitary(int numOfAudios, AudioFile *files[], AudioFile *o_file) {
 
+    int err = 0;
     AudioStreamingContext *output_ctx;
     AudioStreamingContext **audio_ctx =
         (AudioStreamingContext**)calloc(1, numOfAudios*sizeof(AudioStreamingContext));
@@ -112,21 +137,150 @@ int audioMixArbitary(int numOfAudios, AudioFile *files[], AudioFile *o_file) {
         goto FAIL;
     }
 
-    return 0;
+    goto CLEANUP;
 
   FAIL:
-    /* FIXME: Need to release all allocated resources */
-    return 1;
+    err = 1;
+
+  CLEANUP:
+    audioStreamCtx_Release_free(&output_ctx);
+    for (int i = 0; i < numOfAudios; ++i) {
+        audioStreamCtx_Release_free(&audio_ctx[i]);
+    }
+    free(audio_ctx);
+    filterCtx_Release(&filter_ctx);
+    return err;
+}
+
+int mixing_audios(int numOfStream, AudioStreamingContext* inputs[],
+                  AudioStreamingContext *output,
+                  FilteringContext* filter) {
+
+    int                    err    = 0;
+    AVPacket              *packet = av_packet_alloc();
+    AVFrame               *frame  = av_frame_alloc();
+    AudioStreamingContext *asctx  = NULL;
+
+    int finCount = 0;
+    _Bool *finished = (_Bool*)calloc(numOfStream, sizeof(_Bool));
+
+    while (1) {
+        // Read frame from AudioStreams
+        for (int i = 0; i < numOfStream; ++i) {
+            // All frame of this stream is readed.
+            if (finished[i] == 1)
+                continue;
+
+            asctx = inputs[i];
+
+            err = av_read_frame(asctx->fmt, packet);
+            if (err < 0) {
+                finished[i] = 1;
+                ++finCount;
+                err = 0;
+            } else {
+                err = avcodec_send_packet(asctx->cc, packet);
+                if (err < 0 && err != AVERROR(EAGAIN)) {
+                    printf("Failed to send packet to decoder: %s\n",
+                           av_err2str(err));
+                    goto Done;
+                }
+                av_packet_unref(packet);
+            }
+
+            // Send an empty frame to notify to filter that this stream
+            // has no more frames.
+            if (finished[i] == 1) {
+                err = av_buffersrc_add_frame(filter->src[i], NULL);
+                if (err < 0) {
+                    printf("Failed to add frame to filter\n");
+                    goto Done;
+                }
+            } else {
+                while (err >= 0) {
+                    err = avcodec_receive_frame(asctx->cc, frame);
+                    if (err == AVERROR(EAGAIN) || err == AVERROR_EOF) {
+                        break;
+                    } else if (err < 0) {
+                        printf("Failed to decode frames: %s\n", av_err2str(err));
+                        goto Done;
+                    }
+
+                    err = av_buffersrc_add_frame(filter->src[i], frame);
+                    if (err < 0) {
+                        printf("Failed to filter frame\n");
+                        goto Done;
+                    }
+                    av_frame_unref(frame);
+                }
+            }
+        }
+
+        if (finCount == numOfStream) {
+            err = 0;
+
+            write_out(output, NULL);
+            break;
+        }
+
+
+        // Get filtered frame
+        while ((err = av_buffersink_get_frame(filter->sink, frame)) >= 0) {
+            write_out(output, frame);
+            av_frame_unref(frame);
+        }
+    }
+
+    av_write_trailer(output->fmt);
+
+  Done:
+    free(finished);
+    av_packet_free(&packet);
+    av_frame_free(&frame);
+    return err;
+}
+
+int write_out(AudioStreamingContext *encoder, AVFrame *frame) {
+    int response = 0;
+    AVPacket *output_packet = av_packet_alloc();
+
+    response = avcodec_send_frame(encoder->cc, frame);
+
+    while (response >= 0) {
+        response = avcodec_receive_packet(encoder->cc, output_packet);
+        if (response == AVERROR(EAGAIN) || response == AVERROR_EOF) {
+            break;
+        } else if (response < 0) {
+            printf("Error while receiving packet from encoder: %s", av_err2str(response));
+            return 1;
+        }
+
+        output_packet->stream_index = 0;
+
+        response = av_interleaved_write_frame(encoder->fmt, output_packet);
+        if (response != 0) {
+            printf("Error %d while receiving packet from decoder: %s",
+                response, av_err2str(response));
+            return 1;
+        }
+
+    }
+
+    av_packet_unref(output_packet);
+    av_packet_free(&output_packet);
+
+    return 0;
 }
 
+
 /* This function create a filter chain for each of input audio medias.
  * Filter chain looks like:
- * abuffer -> asetpts -> amix -> abuffersink
- * abuffer -> asetpts -----\
+ * abuffer -> adelay -> amix -> abuffersink
+ * abuffer -> adelay -----\
  * ........................\
- * abuffer -> asetpts -----\
+ * abuffer -> adelay -----\
  *
- * "abuffer -> asetpts" is called shift part of filter graph
+ * "abuffer -> adelay" is called shift part of filter graph
  * "amix    -> abuffersink" is called mix part of filter graph */
 FilteringContext* create_filter_context(int numOfStream, AudioStreamingContext *s_ctx[]) {
     int            err   = 0;
@@ -147,12 +301,14 @@ FilteringContext* create_filter_context(int numOfStream, AudioStreamingContext *
     FilterChain *shift_chains = (FilterChain*)malloc(
         numOfStream * sizeof(FilterChain));
 
-    if (shift_chains->src->nb_inputs < numOfStream)
+    if (mix_chain.src->nb_inputs < (unsigned int)numOfStream)
         return NULL;
 
     for (int i = 0; i < numOfStream; ++i) {
         // Create filter chain for shifting purposes for
         // each of all input streams
+        s_ctx[i]->mixpar.cc = s_ctx[i]->cc;
+        s_ctx[i]->mixpar.s  = s_ctx[i]->s;
         chain = build_filter_chain_shift_part(graph, s_ctx[i]->mixpar);
         shift_chains[i] = chain;
 
@@ -162,16 +318,31 @@ FilteringContext* create_filter_context(int numOfStream, AudioStreamingContext *
         }
     }
 
+    // Configure AVFilterGraph
+    err = avfilter_graph_config(graph, NULL);
+    if (err < 0) {
+        goto FAIL;
+    }
+
+    char *dump = avfilter_graph_dump(graph, NULL);
+    av_log(NULL, AV_LOG_ERROR, "Graph:\n%s\n", dump);
+    free(dump);
+
     FilteringContext *f_ctx = (FilteringContext*)malloc(
         sizeof(FilteringContext));
     f_ctx->graph = graph;
     f_ctx->sink = mix_chain.sink;
+    f_ctx->src = (AVFilterContext**)malloc(numOfStream*sizeof(AVFilterContext*));
 
     for (int i = 0; i < numOfStream; ++i) {
-        f_ctx->src = shift_chains[i].src;
+        f_ctx->src[i] = shift_chains[i].src;
     }
 
+    free(shift_chains);
+    return f_ctx;
+
   FAIL:
+    free(shift_chains);
     if (graph)
         avfilter_graph_free(&graph);
     return NULL;
@@ -179,10 +350,12 @@ FilteringContext* create_filter_context(int numOfStream, AudioStreamingContext *
 
 FilterChain build_filter_chain_mix_part(AVFilterGraph *graph, int numOfStreams) {
     int err = 0;
-    AVFilterContext *amix_ctx        = NULL;
-    AVFilterContext *abuffersink_ctx = NULL;
-    const AVFilter  *amix            = NULL;
-    const AVFilter  *abuffersink     = NULL;
+    AVFilterContext *amix_ctx         = NULL;
+    AVFilterContext *abuffersink_ctx  = NULL;
+    AVFilterContext *asetnsamples_ctx = NULL;
+    const AVFilter  *amix             = NULL;
+    const AVFilter  *asetnsamples     = NULL;
+    const AVFilter  *abuffersink      = NULL;
 
     /* Create amix filter */
     amix = avfilter_get_by_name("amix");
@@ -193,7 +366,8 @@ FilterChain build_filter_chain_mix_part(AVFilterGraph *graph, int numOfStreams)
     if (amix_ctx == NULL) {
         goto FAIL;
     }
-    av_opt_set_int(amix_ctx, "input", numOfStreams, AV_OPT_SEARCH_CHILDREN);
+    av_opt_set_int(amix_ctx, "inputs", numOfStreams, AV_OPT_SEARCH_CHILDREN);
+    av_opt_set(amix_ctx, "duration", "longest", AV_OPT_SEARCH_CHILDREN);
 
     err = avfilter_init_str(amix_ctx, NULL);
     if (err < 0) {
@@ -215,11 +389,30 @@ FilterChain build_filter_chain_mix_part(AVFilterGraph *graph, int numOfStreams)
         goto FAIL;
     }
 
+    /* Create asetnsamples filter */
+    asetnsamples = avfilter_get_by_name("asetnsamples");
+    if (asetnsamples == NULL) {
+        goto FAIL;
+    }
+    asetnsamples_ctx = avfilter_graph_alloc_filter(
+        graph, asetnsamples, "asetnsamples_out");
+    if (asetnsamples_ctx == NULL) {
+        goto FAIL;
+    }
+
+    av_opt_set_int(asetnsamples_ctx, "n", 1152, AV_OPT_SEARCH_CHILDREN);
+
+    err = avfilter_init_str(asetnsamples_ctx, NULL);
+    if (err < 0) {
+        goto FAIL;
+    }
+
     // Link amix with abuffersink
-    err = avfilter_link(amix_ctx, 0, abuffersink_ctx, 0);
+    err = avfilter_link(amix_ctx, 0, asetnsamples_ctx, 0);
     if (err != 0) {
         goto FAIL;
     }
+    err = avfilter_link(asetnsamples_ctx, 0, abuffersink_ctx, 0);
 
     return (FilterChain){amix_ctx, abuffersink_ctx};
 
@@ -234,12 +427,13 @@ FilterChain build_filter_chain_mix_part(AVFilterGraph *graph, int numOfStreams)
 FilterChain build_filter_chain_shift_part(AVFilterGraph *graph, MixParameters param) {
     int err = 0;
     char ch_layout[64] = {0};
+    char filter_name[64] = {0};
 
     const AVFilter  *abuffer     = NULL;
     AVFilterContext *abuffer_ctx = NULL;
 
-    const AVFilter  *asetpts     = NULL;
-    AVFilterContext *asetpts_ctx = NULL;
+    const AVFilter  *adelay     = NULL;
+    AVFilterContext *adelay_ctx = NULL;
 
     AVRational time_base = param.cc->time_base;
 
@@ -248,13 +442,15 @@ FilterChain build_filter_chain_shift_part(AVFilterGraph *graph, MixParameters pa
     if (abuffer == NULL) {
         goto FAIL;
     }
-    abuffer_ctx = avfilter_graph_alloc_filter(graph, abuffer, "abuffer");
+
+    sprintf(filter_name, "abuffer_%d", param.index);
+    abuffer_ctx = avfilter_graph_alloc_filter(graph, abuffer, filter_name);
     if (abuffer_ctx == NULL) {
         goto FAIL;
     }
 
     av_get_channel_layout_string(
-        (char*)ch_layout, sizeof(ch_layout), 0, AV_OPT_SEARCH_CHILDREN);
+        (char*)ch_layout, sizeof(ch_layout), 0, param.cc->channel_layout);
     av_opt_set(abuffer_ctx, "channel_layout", (char*)ch_layout, AV_OPT_SEARCH_CHILDREN);
     av_opt_set(abuffer_ctx, "sample_fmt",
                av_get_sample_fmt_name(param.cc->sample_fmt),
@@ -265,43 +461,43 @@ FilterChain build_filter_chain_shift_part(AVFilterGraph *graph, MixParameters pa
     av_opt_set_int(abuffer_ctx, "sample_rate", param.cc->sample_rate, AV_OPT_SEARCH_CHILDREN);
 
     err = avfilter_init_str(abuffer_ctx, NULL);
-    if (err < 0) {
+    if (err < 0) {;
         goto FAIL;
     }
 
-    // Create asetpts filter
-    asetpts = avfilter_get_by_name("asetpts");
-    if (asetpts == NULL) {
+    // Create adelay filter
+    adelay = avfilter_get_by_name("adelay");
+    if (adelay == NULL) {
         goto FAIL;
     }
-    asetpts_ctx = avfilter_graph_alloc_filter(graph, asetpts, "asetpts");
-    if (asetpts_ctx == NULL) {
+    sprintf(filter_name, "adelay_%d", param.index);
+    adelay_ctx = avfilter_graph_alloc_filter(graph, adelay, filter_name);
+    if (adelay_ctx == NULL) {
         goto FAIL;
     }
 
-    // Construct expr argument of asetpts
-    char expr[128] = { 0 };
-    sprintf(expr, "PTS+%u/TB", param.offset);
-
-    av_opt_set(asetpts_ctx, "expr", expr, AV_OPT_SEARCH_CHILDREN);
+    char delays[32] = {0};
+    unsigned int offset = param.offset*1000;
+    sprintf(delays, "%d|%d", offset, offset);
+    av_opt_set(adelay_ctx, "delays", delays, AV_OPT_SEARCH_CHILDREN);
 
-    err = avfilter_init_str(asetpts_ctx, NULL);
+    err = avfilter_init_str(adelay_ctx, NULL);
     if (err < 0) {
         goto FAIL;
     }
 
-    err = avfilter_link(abuffer_ctx, 0, asetpts_ctx, 0);
+    err = avfilter_link(abuffer_ctx, 0, adelay_ctx, 0);
     if (err < 0) {
         goto FAIL;
     }
 
-    return (FilterChain){ abuffer_ctx, asetpts_ctx };
+    return (FilterChain){ abuffer_ctx, adelay_ctx };
 
   FAIL:
     if (abuffer_ctx)
         avfilter_free(abuffer_ctx);
-    if (asetpts_ctx)
-        avfilter_free(asetpts_ctx);
+    if (adelay_ctx)
+        avfilter_free(adelay_ctx);
     return (FilterChain){NULL, NULL};
 }
 
@@ -356,6 +552,14 @@ AudioStreamingContext* prepare_output_streaming_context(
    }
    avcodec_parameters_from_context(stream->codecpar, codec_ctx);
 
+   if (avio_open(&fmt_ctx->pb, file->filepath, AVIO_FLAG_WRITE) < 0) {
+       goto FAIL;
+   }
+   AVDictionary *opts = NULL;
+   if (avformat_write_header(fmt_ctx, NULL) < 0) {
+       goto FAIL;
+   }
+
    AudioStreamingContext *o_streaming_ctx = (AudioStreamingContext*)malloc(
        sizeof(AudioStreamingContext));
    o_streaming_ctx->c        = codec;
@@ -440,6 +644,39 @@ AudioStreamingContext* open_audio_file(AudioFile *files) {
 
 
 int main(int argc, char *argv[]) {
+    AudioFile i_file_1 = {
+        argv[1],
+        (MixParameters){
+            0, NULL, NULL, 10
+        }
+    };
+    AudioFile i_file_2 = {
+        argv[2],
+        (MixParameters){
+            1, NULL, NULL, 10
+        }
+    };
+   AudioFile i_file_3 = {
+        argv[3],
+        (MixParameters){
+            2, NULL, NULL, 5
+        }
+    };
+
+    AudioFile o_file = {
+        argv[4],
+        (MixParameters){
+            0, NULL, NULL, 0
+        }
+    };
+
+    AudioFile *i_files[3] = {
+        &i_file_1,
+        &i_file_2,
+        &i_file_3
+    };
+
+    audioMixArbitary(3, i_files, &o_file);
 
     return 0;
 }