merge read and and in add_paeth_prediction

[ffmpeg] / libavcodec / aacdec.c

diff --git a/libavcodec/aacdec.c b/libavcodec/aacdec.c
index a362d6ad4e30b6c7eaee711309de485ddf3ac812..96b1323c191d61164e11481e2da01738246458b3 100644 (file)
--- a/libavcodec/aacdec.c
+++ b/libavcodec/aacdec.c
@@ -87,6 +87,8 @@
 #include "fft.h"
 #include "fmtconvert.h"
 #include "lpc.h"
+#include "kbdwin.h"
+#include "sinewin.h"
 
 #include "aac.h"
 #include "aactab.h"
@@ -178,9 +180,8 @@ static ChannelElement *get_che(AACContext *ac, int type, int elem_id)
  * @return  Returns error status. 0 - OK, !0 - error
  */
 static av_cold int che_configure(AACContext *ac,
-                         enum ChannelPosition che_pos[4][MAX_ELEM_ID],
-                         int type, int id,
-                         int *channels)
+                                 enum ChannelPosition che_pos[4][MAX_ELEM_ID],
+                                 int type, int id, int *channels)
 {
     if (che_pos[type][id]) {
         if (!ac->che[type][id] && !(ac->che[type][id] = av_mallocz(sizeof(ChannelElement))))
@@ -210,9 +211,9 @@ static av_cold int che_configure(AACContext *ac,
  * @return  Returns error status. 0 - OK, !0 - error
  */
 static av_cold int output_configure(AACContext *ac,
-                            enum ChannelPosition che_pos[4][MAX_ELEM_ID],
-                            enum ChannelPosition new_che_pos[4][MAX_ELEM_ID],
-                            int channel_config, enum OCStatus oc_type)
+                                    enum ChannelPosition che_pos[4][MAX_ELEM_ID],
+                                    enum ChannelPosition new_che_pos[4][MAX_ELEM_ID],
+                                    int channel_config, enum OCStatus oc_type)
 {
     AVCodecContext *avctx = ac->avctx;
     int i, type, channels = 0, ret;
@@ -229,7 +230,7 @@ static av_cold int output_configure(AACContext *ac,
                 return ret;
         }
 
-        memset(ac->tag_che_map, 0,       4 * MAX_ELEM_ID * sizeof(ac->che[0][0]));
+        memset(ac->tag_che_map, 0, 4 * MAX_ELEM_ID * sizeof(ac->che[0][0]));
 
         avctx->channel_layout = aac_channel_layout[channel_config - 1];
     } else {
@@ -344,8 +345,8 @@ static int decode_pce(AVCodecContext *avctx, MPEG4AudioConfig *m4ac,
  * @return  Returns error status. 0 - OK, !0 - error
  */
 static av_cold int set_default_channel_config(AVCodecContext *avctx,
-                                      enum ChannelPosition new_che_pos[4][MAX_ELEM_ID],
-                                      int channel_config)
+                                              enum ChannelPosition new_che_pos[4][MAX_ELEM_ID],
+                                              int channel_config)
 {
     if (channel_config < 1 || channel_config > 7) {
         av_log(avctx, AV_LOG_ERROR, "invalid default channel configuration (%d)\n",
@@ -462,6 +463,11 @@ static int decode_audio_specific_config(AACContext *ac,
     GetBitContext gb;
     int i;
 
+    av_dlog(avctx, "extradata size %d\n", avctx->extradata_size);
+    for (i = 0; i < avctx->extradata_size; i++)
+         av_dlog(avctx, "%02x ", avctx->extradata[i]);
+    av_dlog(avctx, "\n");
+
     init_get_bits(&gb, data, data_size * 8);
 
     if ((i = ff_mpeg4audio_get_config(m4ac, data, data_size)) < 0)
@@ -488,6 +494,10 @@ static int decode_audio_specific_config(AACContext *ac,
         return -1;
     }
 
+    av_dlog(avctx, "AOT %d chan config %d sampling index %d (%d) SBR %d PS %d\n",
+            m4ac->object_type, m4ac->chan_config, m4ac->sampling_index,
+            m4ac->sample_rate, m4ac->sbr, m4ac->ps);
+
     return get_bits_count(&gb);
 }
 
@@ -547,7 +557,8 @@ static av_cold int aac_decode_init(AVCodecContext *avctx)
             return -1;
     }
 
-    avctx->sample_fmt = AV_SAMPLE_FMT_S16;
+    avctx->sample_fmt = avctx->request_sample_fmt == AV_SAMPLE_FMT_FLT ?
+                        AV_SAMPLE_FMT_FLT : AV_SAMPLE_FMT_S16;
 
     AAC_INIT_VLC_STATIC( 0, 304);
     AAC_INIT_VLC_STATIC( 1, 270);
@@ -962,19 +973,19 @@ static inline float *VMUL4S(float *dst, const float *v, unsigned idx,
     union float754 s = { .f = *scale };
     union float754 t;
 
-    t.i = s.i ^ (sign & 1<<31);
+    t.i = s.i ^ (sign & 1U<<31);
     *dst++ = v[idx    & 3] * t.f;
 
     sign <<= nz & 1; nz >>= 1;
-    t.i = s.i ^ (sign & 1<<31);
+    t.i = s.i ^ (sign & 1U<<31);
     *dst++ = v[idx>>2 & 3] * t.f;
 
     sign <<= nz & 1; nz >>= 1;
-    t.i = s.i ^ (sign & 1<<31);
+    t.i = s.i ^ (sign & 1U<<31);
     *dst++ = v[idx>>4 & 3] * t.f;
 
     sign <<= nz & 1; nz >>= 1;
-    t.i = s.i ^ (sign & 1<<31);
+    t.i = s.i ^ (sign & 1U<<31);
     *dst++ = v[idx>>6 & 3] * t.f;
 
     return dst;
@@ -1167,11 +1178,11 @@ static int decode_spectrum_and_dequant(AACContext *ac, float coef[1024],
                                     b += 4;
                                     n = (1 << b) + SHOW_UBITS(re, gb, b);
                                     LAST_SKIP_BITS(re, gb, b);
-                                    *icf++ = cbrt_tab[n] | (bits & 1<<31);
+                                    *icf++ = cbrt_tab[n] | (bits & 1U<<31);
                                     bits <<= 1;
                                 } else {
                                     unsigned v = ((const uint32_t*)vq)[cb_idx & 15];
-                                    *icf++ = (bits & 1<<31) | v;
+                                    *icf++ = (bits & 1U<<31) | v;
                                     bits <<= !!v;
                                 }
                                 cb_idx >>= 4;
@@ -1233,7 +1244,7 @@ static av_always_inline float flt16_trunc(float pf)
 
 static av_always_inline void predict(PredictorState *ps, float *coef,
                                      float sf_scale, float inv_sf_scale,
-                    int output_enable)
+                                     int output_enable)
 {
     const float a     = 0.953125; // 61.0 / 64
     const float alpha = 0.90625;  // 29.0 / 32
@@ -1750,7 +1761,7 @@ static void windowing_and_mdct_ltp(AACContext *ac, float *out,
         ac->dsp.vector_fmul_reverse(in + 1024 + 448, in + 1024 + 448, swindow, 128);
         memset(in + 1024 + 576, 0, 448 * sizeof(float));
     }
-    ff_mdct_calc(&ac->mdct_ltp, out, in);
+    ac->mdct_ltp.mdct_calc(&ac->mdct_ltp, out, in);
 }
 
 /**
@@ -1763,8 +1774,8 @@ static void apply_ltp(AACContext *ac, SingleChannelElement *sce)
     int i, sfb;
 
     if (sce->ics.window_sequence[0] != EIGHT_SHORT_SEQUENCE) {
-        float *predTime = ac->buf_mdct;
-        float *predFreq = sce->ret;
+        float *predTime = sce->ret;
+        float *predFreq = ac->buf_mdct;
         int16_t num_samples = 2048;
 
         if (ltp->lag < 1024)
@@ -1797,19 +1808,22 @@ static void update_ltp(AACContext *ac, SingleChannelElement *sce)
     const float *swindow = ics->use_kb_window[0] ? ff_aac_kbd_short_128 : ff_sine_128;
     int i;
 
-    for (i = 0; i < 512; i++)
-        ac->buf_mdct[1535 - i] = ac->buf_mdct[512 + i];
-
     if (ics->window_sequence[0] == EIGHT_SHORT_SEQUENCE) {
         memcpy(saved_ltp,       saved, 512 * sizeof(float));
         memset(saved_ltp + 576, 0,     448 * sizeof(float));
-        ac->dsp.vector_fmul_reverse(saved_ltp + 448, ac->buf_mdct + 960,     swindow,     128);
+        ac->dsp.vector_fmul_reverse(saved_ltp + 448, ac->buf_mdct + 960,     &swindow[64],      64);
+        for (i = 0; i < 64; i++)
+            saved_ltp[i + 512] = ac->buf_mdct[1023 - i] * swindow[63 - i];
     } else if (ics->window_sequence[0] == LONG_START_SEQUENCE) {
         memcpy(saved_ltp,       ac->buf_mdct + 512, 448 * sizeof(float));
         memset(saved_ltp + 576, 0,                  448 * sizeof(float));
-        ac->dsp.vector_fmul_reverse(saved_ltp + 448, ac->buf_mdct + 960,     swindow,     128);
+        ac->dsp.vector_fmul_reverse(saved_ltp + 448, ac->buf_mdct + 960,     &swindow[64],      64);
+        for (i = 0; i < 64; i++)
+            saved_ltp[i + 512] = ac->buf_mdct[1023 - i] * swindow[63 - i];
     } else { // LONG_STOP or ONLY_LONG
-        ac->dsp.vector_fmul_reverse(saved_ltp,       ac->buf_mdct + 512,     lwindow,     1024);
+        ac->dsp.vector_fmul_reverse(saved_ltp,       ac->buf_mdct + 512,     &lwindow[512],     512);
+        for (i = 0; i < 512; i++)
+            saved_ltp[i + 512] = ac->buf_mdct[1023 - i] * lwindow[511 - i];
     }
 
     memcpy(sce->ltp_state, &sce->ltp_state[1024], 1024 * sizeof(int16_t));
@@ -1836,9 +1850,9 @@ static void imdct_and_windowing(AACContext *ac, SingleChannelElement *sce)
     // imdct
     if (ics->window_sequence[0] == EIGHT_SHORT_SEQUENCE) {
         for (i = 0; i < 1024; i += 128)
-            ff_imdct_half(&ac->mdct_small, buf + i, in + i);
+            ac->mdct_small.imdct_half(&ac->mdct_small, buf + i, in + i);
     } else
-        ff_imdct_half(&ac->mdct, buf, in);
+        ac->mdct.imdct_half(&ac->mdct, buf, in);
 
     /* window overlapping
      * NOTE: To simplify the overlapping code, all 'meaningless' short to long
@@ -2161,7 +2175,8 @@ static int aac_decode_frame_int(AVCodecContext *avctx, void *data,
         avctx->frame_size = samples;
     }
 
-    data_size_tmp = samples * avctx->channels * sizeof(int16_t);
+    data_size_tmp = samples * avctx->channels;
+    data_size_tmp *= avctx->sample_fmt == AV_SAMPLE_FMT_FLT ? sizeof(float) : sizeof(int16_t);
     if (*data_size < data_size_tmp) {
         av_log(avctx, AV_LOG_ERROR,
                "Output buffer too small (%d) or trying to output too many samples (%d) for this frame.\n",
@@ -2170,8 +2185,12 @@ static int aac_decode_frame_int(AVCodecContext *avctx, void *data,
     }
     *data_size = data_size_tmp;
 
-    if (samples)
-        ac->fmt_conv.float_to_int16_interleave(data, (const float **)ac->output_data, samples, avctx->channels);
+    if (samples) {
+        if (avctx->sample_fmt == AV_SAMPLE_FMT_FLT) {
+            float_interleave(data, (const float **)ac->output_data, samples, avctx->channels);
+        } else
+            ac->fmt_conv.float_to_int16_interleave(data, (const float **)ac->output_data, samples, avctx->channels);
+    }
 
     if (ac->output_configured)
         ac->output_configured = OC_LOCKED;
@@ -2489,7 +2508,7 @@ AVCodec ff_aac_decoder = {
     aac_decode_frame,
     .long_name = NULL_IF_CONFIG_SMALL("Advanced Audio Coding"),
     .sample_fmts = (const enum AVSampleFormat[]) {
-        AV_SAMPLE_FMT_S16,AV_SAMPLE_FMT_NONE
+        AV_SAMPLE_FMT_S16,AV_SAMPLE_FMT_FLT,AV_SAMPLE_FMT_NONE
     },
     .channel_layouts = aac_channel_layout,
 };
@@ -2509,7 +2528,7 @@ AVCodec ff_aac_latm_decoder = {
     .decode = latm_decode_frame,
     .long_name = NULL_IF_CONFIG_SMALL("AAC LATM (Advanced Audio Codec LATM syntax)"),
     .sample_fmts = (const enum AVSampleFormat[]) {
-        AV_SAMPLE_FMT_S16,AV_SAMPLE_FMT_NONE
+        AV_SAMPLE_FMT_S16,AV_SAMPLE_FMT_FLT,AV_SAMPLE_FMT_NONE
     },
     .channel_layouts = aac_channel_layout,
 };