avcodec/dca: avoid using bitstream reader in a non-standard way
[ffmpeg.git] / libavcodec / libtwolame.c
index b57fbc9..12d71e7 100644 (file)
 
 #include <twolame.h>
 
+#include "libavutil/common.h"
 #include "libavutil/opt.h"
+
 #include "avcodec.h"
 #include "internal.h"
 #include "mpegaudio.h"
 
 typedef struct TWOLAMEContext {
-    AVClass  *class;
-    int      mode;
-    int      psymodel;
-    int      energy;
-    int      error_protection;
-    int      copyright;
-    int      original;
+    AVClass *class;
+    int mode;
+    int psymodel;
+    int energy;
+    int error_protection;
+    int copyright;
+    int original;
+    int verbosity;
 
     twolame_options *glopts;
-    int64_t  next_pts;
+    int64_t next_pts;
 } TWOLAMEContext;
 
 static av_cold int twolame_encode_close(AVCodecContext *avctx)
@@ -57,35 +60,44 @@ static av_cold int twolame_encode_init(AVCodecContext *avctx)
     int ret;
 
     avctx->frame_size = TWOLAME_SAMPLES_PER_FRAME;
+    avctx->initial_padding = 512 - 32 + 1;
 
     s->glopts = twolame_init();
     if (!s->glopts)
         return AVERROR(ENOMEM);
 
-    twolame_set_verbosity(s->glopts, 0);
+    twolame_set_verbosity(s->glopts, s->verbosity);
     twolame_set_mode(s->glopts, s->mode);
     twolame_set_psymodel(s->glopts, s->psymodel);
     twolame_set_energy_levels(s->glopts, s->energy);
     twolame_set_error_protection(s->glopts, s->error_protection);
+    twolame_set_copyright(s->glopts, s->copyright);
+    twolame_set_original(s->glopts, s->original);
 
     twolame_set_num_channels(s->glopts, avctx->channels);
     twolame_set_in_samplerate(s->glopts, avctx->sample_rate);
     twolame_set_out_samplerate(s->glopts, avctx->sample_rate);
-    if (avctx->flags & CODEC_FLAG_QSCALE || !avctx->bit_rate) {
+
+    if (!avctx->bit_rate)
+        avctx->bit_rate = avctx->sample_rate < 28000 ? 160000 : 384000;
+
+    if (avctx->flags & AV_CODEC_FLAG_QSCALE || !avctx->bit_rate) {
         twolame_set_VBR(s->glopts, TRUE);
-        twolame_set_VBR_level(s->glopts, avctx->global_quality);
-        av_log(avctx, AV_LOG_WARNING, "VBR mode is experimental!\n");
+        twolame_set_VBR_level(s->glopts,
+                              avctx->global_quality / (float) FF_QP2LAMBDA);
+        av_log(avctx, AV_LOG_WARNING,
+               "VBR in MP2 is a hack, use another codec that supports it.\n");
     } else {
         twolame_set_bitrate(s->glopts, avctx->bit_rate / 1000);
     }
 
-    if ((ret = twolame_init_params(s->glopts)))
-        goto error;
+    ret = twolame_init_params(s->glopts);
+    if (ret) {
+        twolame_encode_close(avctx);
+        return AVERROR_UNKNOWN;
+    }
 
     return 0;
-error:
-    twolame_encode_close(avctx);
-    return ret;
 }
 
 static int twolame_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
@@ -94,103 +106,124 @@ static int twolame_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
     TWOLAMEContext *s = avctx->priv_data;
     int ret;
 
-    if ((ret = ff_alloc_packet2(avctx, avpkt, MPA_MAX_CODED_FRAME_SIZE)))
+    if ((ret = ff_alloc_packet2(avctx, avpkt, MPA_MAX_CODED_FRAME_SIZE, 0)) < 0)
         return ret;
 
     if (frame) {
         switch (avctx->sample_fmt) {
         case AV_SAMPLE_FMT_FLT:
             ret = twolame_encode_buffer_float32_interleaved(s->glopts,
-                                        frame->data[0],
-                                        frame->nb_samples,
-                                        avpkt->data, avpkt->size);
+                                                            (const float *)frame->data[0],
+                                                            frame->nb_samples,
+                                                            avpkt->data,
+                                                            avpkt->size);
             break;
         case AV_SAMPLE_FMT_FLTP:
             ret = twolame_encode_buffer_float32(s->glopts,
-                                        frame->data[0], frame->data[1],
-                                        frame->nb_samples,
-                                        avpkt->data, avpkt->size);
+                                                (const float *)frame->data[0],
+                                                (const float *)frame->data[1],
+                                                frame->nb_samples,
+                                                avpkt->data, avpkt->size);
             break;
         case AV_SAMPLE_FMT_S16:
             ret = twolame_encode_buffer_interleaved(s->glopts,
-                                        frame->data[0],
-                                        frame->nb_samples,
-                                        avpkt->data, avpkt->size);
+                                                    (const short int *)frame->data[0],
+                                                    frame->nb_samples,
+                                                    avpkt->data, avpkt->size);
             break;
         case AV_SAMPLE_FMT_S16P:
             ret = twolame_encode_buffer(s->glopts,
-                                        frame->data[0], frame->data[1],
+                                        (const short int *)frame->data[0],
+                                        (const short int *)frame->data[1],
                                         frame->nb_samples,
                                         avpkt->data, avpkt->size);
             break;
         default:
+            av_log(avctx, AV_LOG_ERROR,
+                   "Unsupported sample format %d.\n", avctx->sample_fmt);
             return AVERROR_BUG;
         }
     } else {
         ret = twolame_encode_flush(s->glopts, avpkt->data, avpkt->size);
     }
 
-    if (ret > 0) {
-        avpkt->duration = ff_samples_to_time_base(avctx, avctx->frame_size);
-        if (frame) {
-            if (frame->pts != AV_NOPTS_VALUE)
-                avpkt->pts  = frame->pts;
-        } else {
-            avpkt->pts = s->next_pts;
-        }
-        if (avpkt->pts != AV_NOPTS_VALUE)
-            s->next_pts = avpkt->pts + avpkt->duration;
-
-        avpkt->size     = ret;
-        *got_packet_ptr = 1;
+    if (!ret)     // no bytes written
         return 0;
+    if (ret < 0)  // twolame error
+        return AVERROR_UNKNOWN;
+
+    if (frame) {
+        avpkt->duration = ff_samples_to_time_base(avctx, frame->nb_samples);
+        if (frame->pts != AV_NOPTS_VALUE)
+            avpkt->pts = frame->pts - ff_samples_to_time_base(avctx, avctx->initial_padding);
+    } else {
+        avpkt->pts = s->next_pts;
     }
+    // this is for setting pts for flushed packet(s).
+    if (avpkt->pts != AV_NOPTS_VALUE)
+        s->next_pts = avpkt->pts + avpkt->duration;
 
-    return ret;
+    av_shrink_packet(avpkt, ret);
+    *got_packet_ptr = 1;
+    return 0;
 }
 
 #define OFFSET(x) offsetof(TWOLAMEContext, x)
 #define AE AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
 static const AVOption options[] = {
-    { "mode",         "Mpeg Mode", OFFSET(mode), AV_OPT_TYPE_INT, { .i64 = TWOLAME_AUTO_MODE }, TWOLAME_AUTO_MODE, TWOLAME_MONO, AE, "mode"},
-    { "auto",         NULL,         0, AV_OPT_TYPE_CONST, { .i64 = TWOLAME_AUTO_MODE },          0, 0, AE, "mode" },
-    { "stereo",       NULL,         0, AV_OPT_TYPE_CONST, { .i64 = TWOLAME_STEREO },             0, 0, AE, "mode" },
-    { "joint_stereo", NULL,         0, AV_OPT_TYPE_CONST, { .i64 = TWOLAME_JOINT_STEREO },       0, 0, AE, "mode" },
-    { "dual_channel", NULL,         0, AV_OPT_TYPE_CONST, { .i64 = TWOLAME_DUAL_CHANNEL },       0, 0, AE, "mode" },
-    { "mono",         NULL,         0, AV_OPT_TYPE_CONST, { .i64 = TWOLAME_MONO },               0, 0, AE, "mode" },
-    { "psymodel",    "Psychoacoustic Model",  OFFSET(psymodel), AV_OPT_TYPE_INT, { .i64 = 3 }, -1, 4, AE},
-    { "energy_levels","enable energy levels", OFFSET(energy),   AV_OPT_TYPE_INT, { .i64 = 0 },  0, 1, AE},
+    { "mode", "Mpeg Mode", OFFSET(mode), AV_OPT_TYPE_INT, { .i64 = TWOLAME_AUTO_MODE }, TWOLAME_AUTO_MODE, TWOLAME_MONO, AE, "mode"},
+        { "auto", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = TWOLAME_AUTO_MODE }, 0, 0, AE, "mode" },
+        { "stereo", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = TWOLAME_STEREO }, 0, 0, AE, "mode" },
+        { "joint_stereo", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = TWOLAME_JOINT_STEREO }, 0, 0, AE, "mode" },
+        { "dual_channel", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = TWOLAME_DUAL_CHANNEL }, 0, 0, AE, "mode" },
+        { "mono", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = TWOLAME_MONO }, 0, 0, AE, "mode" },
+    { "psymodel", "Psychoacoustic Model", OFFSET(psymodel), AV_OPT_TYPE_INT, { .i64 = 3 }, -1, 4, AE},
+    { "energy_levels","enable energy levels", OFFSET(energy), AV_OPT_TYPE_INT, { .i64 = 0 },  0, 1, AE},
     { "error_protection","enable CRC error protection", OFFSET(error_protection), AV_OPT_TYPE_INT, { .i64 = 0 },  0, 1, AE},
-    { "copyright",    "set MPEG Audio Copyright flag",  OFFSET(copyright), AV_OPT_TYPE_INT, { .i64 = 0 },  0, 1, AE},
-    { "original",     "set MPEG Audio Original flag",   OFFSET(original),  AV_OPT_TYPE_INT, { .i64 = 0 },  0, 1, AE},
+    { "copyright", "set MPEG Audio Copyright flag", OFFSET(copyright), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, AE},
+    { "original", "set MPEG Audio Original flag", OFFSET(original), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, AE},
+    { "verbosity", "set library optput level (0-10)", OFFSET(verbosity), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 10, AE},
     { NULL },
 };
 
-static const AVClass libtwolame_class = {
+static const AVClass twolame_class = {
     .class_name = "libtwolame encoder",
     .item_name  = av_default_item_name,
     .option     = options,
     .version    = LIBAVUTIL_VERSION_INT,
 };
 
+static const AVCodecDefault twolame_defaults[] = {
+    { "b", "0" },
+    { NULL },
+};
+
+static const int twolame_samplerates[] = {
+    16000, 22050, 24000, 32000, 44100, 48000, 0
+};
+
 AVCodec ff_libtwolame_encoder = {
-    .name                  = "libtwolame",
-    .type                  = AVMEDIA_TYPE_AUDIO,
-    .id                    = AV_CODEC_ID_MP2,
-    .priv_data_size        = sizeof(TWOLAMEContext),
-    .init                  = twolame_encode_init,
-    .encode2               = twolame_encode_frame,
-    .close                 = twolame_encode_close,
-    .capabilities          = CODEC_CAP_DELAY,
-    .sample_fmts           = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_FLT,
-                                                             AV_SAMPLE_FMT_FLTP,
-                                                             AV_SAMPLE_FMT_S16,
-                                                             AV_SAMPLE_FMT_S16P,
-                                                             AV_SAMPLE_FMT_NONE },
-    .channel_layouts       = (const uint64_t[]) { AV_CH_LAYOUT_MONO,
-                                                  AV_CH_LAYOUT_STEREO,
-                                                  0 },
-    .supported_samplerates = (const int[]){ 16000, 22050, 24000, 32000, 44100, 48000, 0 },
-    .long_name             = NULL_IF_CONFIG_SMALL("libtwolame MP2 (MPEG audio layer 2)"),
-    .priv_class            = &libtwolame_class,
+    .name           = "libtwolame",
+    .long_name      = NULL_IF_CONFIG_SMALL("libtwolame MP2 (MPEG audio layer 2)"),
+    .type           = AVMEDIA_TYPE_AUDIO,
+    .id             = AV_CODEC_ID_MP2,
+    .priv_data_size = sizeof(TWOLAMEContext),
+    .init           = twolame_encode_init,
+    .encode2        = twolame_encode_frame,
+    .close          = twolame_encode_close,
+    .capabilities   = AV_CODEC_CAP_DELAY,
+    .defaults       = twolame_defaults,
+    .priv_class     = &twolame_class,
+    .sample_fmts    = (const enum AVSampleFormat[]) {
+        AV_SAMPLE_FMT_FLT,
+        AV_SAMPLE_FMT_FLTP,
+        AV_SAMPLE_FMT_S16,
+        AV_SAMPLE_FMT_S16P,
+        AV_SAMPLE_FMT_NONE
+    },
+    .channel_layouts = (const uint64_t[]) {
+        AV_CH_LAYOUT_MONO,
+        AV_CH_LAYOUT_STEREO,
+        0 },
+    .supported_samplerates = twolame_samplerates,
 };