ape: decode directly to the user-provided AVFrame

[ffmpeg] / libavcodec / mpegvideo.c

diff --git a/libavcodec/mpegvideo.c b/libavcodec/mpegvideo.c
index edd52532b258eb7352f854e25c75cf4b9b85e36f..4b68fd56a0515031445ab7b9a8e368897f76ea09 100644 (file)
--- a/libavcodec/mpegvideo.c
+++ b/libavcodec/mpegvideo.c
@@ -43,19 +43,19 @@
 //#include <assert.h>
 
 static void dct_unquantize_mpeg1_intra_c(MpegEncContext *s,
-                                   DCTELEM *block, int n, int qscale);
+                                   int16_t *block, int n, int qscale);
 static void dct_unquantize_mpeg1_inter_c(MpegEncContext *s,
-                                   DCTELEM *block, int n, int qscale);
+                                   int16_t *block, int n, int qscale);
 static void dct_unquantize_mpeg2_intra_c(MpegEncContext *s,
-                                   DCTELEM *block, int n, int qscale);
+                                   int16_t *block, int n, int qscale);
 static void dct_unquantize_mpeg2_intra_bitexact(MpegEncContext *s,
-                                   DCTELEM *block, int n, int qscale);
+                                   int16_t *block, int n, int qscale);
 static void dct_unquantize_mpeg2_inter_c(MpegEncContext *s,
-                                   DCTELEM *block, int n, int qscale);
+                                   int16_t *block, int n, int qscale);
 static void dct_unquantize_h263_intra_c(MpegEncContext *s,
-                                  DCTELEM *block, int n, int qscale);
+                                  int16_t *block, int n, int qscale);
 static void dct_unquantize_h263_inter_c(MpegEncContext *s,
-                                  DCTELEM *block, int n, int qscale);
+                                  int16_t *block, int n, int qscale);
 
 
 /* enable all paranoid tests for rounding, overflows, etc... */
@@ -131,9 +131,18 @@ const enum AVPixelFormat ff_pixfmt_list_420[] = {
 };
 
 const enum AVPixelFormat ff_hwaccel_pixfmt_list_420[] = {
+#if CONFIG_DXVA2
     AV_PIX_FMT_DXVA2_VLD,
+#endif
+#if CONFIG_VAAPI
     AV_PIX_FMT_VAAPI_VLD,
+#endif
+#if CONFIG_VDA
     AV_PIX_FMT_VDA_VLD,
+#endif
+#if CONFIG_VDPAU
+    AV_PIX_FMT_VDPAU,
+#endif
     AV_PIX_FMT_YUV420P,
     AV_PIX_FMT_NONE
 };
@@ -175,6 +184,7 @@ const uint8_t *avpriv_mpv_find_start_code(const uint8_t *restrict p,
 av_cold int ff_dct_common_init(MpegEncContext *s)
 {
     ff_dsputil_init(&s->dsp, s->avctx);
+    ff_videodsp_init(&s->vdsp, s->avctx->bits_per_raw_sample);
 
     s->dct_unquantize_h263_intra = dct_unquantize_h263_intra_c;
     s->dct_unquantize_h263_inter = dct_unquantize_h263_inter_c;
@@ -392,7 +402,7 @@ int ff_alloc_picture(MpegEncContext *s, Picture *pic, int shared)
         }
         if (s->avctx->debug&FF_DEBUG_DCT_COEFF) {
             FF_ALLOCZ_OR_GOTO(s->avctx, pic->f.dct_coeff,
-                              64 * mb_array_size * sizeof(DCTELEM) * 6, fail)
+                              64 * mb_array_size * sizeof(int16_t) * 6, fail)
         }
         pic->f.qstride = s->mb_stride;
         FF_ALLOCZ_OR_GOTO(s->avctx, pic->f.pan_scan,
@@ -445,7 +455,7 @@ static void free_picture(MpegEncContext *s, Picture *pic)
     }
 }
 
-static int init_duplicate_context(MpegEncContext *s, MpegEncContext *base)
+static int init_duplicate_context(MpegEncContext *s)
 {
     int y_size = s->b8_stride * (2 * s->mb_height + 1);
     int c_size = s->mb_stride * (s->mb_height + 1);
@@ -469,7 +479,7 @@ static int init_duplicate_context(MpegEncContext *s, MpegEncContext *base)
                               2 * 64 * sizeof(int), fail)
         }
     }
-    FF_ALLOCZ_OR_GOTO(s->avctx, s->blocks, 64 * 12 * 2 * sizeof(DCTELEM), fail)
+    FF_ALLOCZ_OR_GOTO(s->avctx, s->blocks, 64 * 12 * 2 * sizeof(int16_t), fail)
     s->block = s->blocks[0];
 
     for (i = 0; i < 12; i++) {
@@ -979,7 +989,7 @@ av_cold int ff_MPV_common_init(MpegEncContext *s)
             }
 
             for (i = 0; i < nb_slices; i++) {
-                if (init_duplicate_context(s->thread_context[i], s) < 0)
+                if (init_duplicate_context(s->thread_context[i]) < 0)
                     goto fail;
                     s->thread_context[i]->start_mb_y =
                         (s->mb_height * (i) + nb_slices / 2) / nb_slices;
@@ -987,7 +997,7 @@ av_cold int ff_MPV_common_init(MpegEncContext *s)
                         (s->mb_height * (i + 1) + nb_slices / 2) / nb_slices;
             }
         } else {
-            if (init_duplicate_context(s, s) < 0)
+            if (init_duplicate_context(s) < 0)
                 goto fail;
             s->start_mb_y = 0;
             s->end_mb_y   = s->mb_height;
@@ -1108,7 +1118,7 @@ int ff_MPV_common_frame_size_change(MpegEncContext *s)
             }
 
             for (i = 0; i < nb_slices; i++) {
-                if (init_duplicate_context(s->thread_context[i], s) < 0)
+                if (init_duplicate_context(s->thread_context[i]) < 0)
                     goto fail;
                     s->thread_context[i]->start_mb_y =
                         (s->mb_height * (i) + nb_slices / 2) / nb_slices;
@@ -1116,7 +1126,7 @@ int ff_MPV_common_frame_size_change(MpegEncContext *s)
                         (s->mb_height * (i + 1) + nb_slices / 2) / nb_slices;
             }
         } else {
-            if (init_duplicate_context(s, s) < 0)
+            if (init_duplicate_context(s) < 0)
                 goto fail;
             s->start_mb_y = 0;
             s->end_mb_y   = s->mb_height;
@@ -1460,6 +1470,9 @@ int ff_MPV_frame_start(MpegEncContext *s, AVCodecContext *avctx)
              s->last_picture_ptr->f.data[0] == NULL) &&
             (s->pict_type != AV_PICTURE_TYPE_I ||
              s->picture_structure != PICT_FRAME)) {
+            int h_chroma_shift, v_chroma_shift;
+            av_pix_fmt_get_chroma_sub_sample(s->avctx->pix_fmt,
+                                             &h_chroma_shift, &v_chroma_shift);
             if (s->pict_type != AV_PICTURE_TYPE_I)
                 av_log(avctx, AV_LOG_ERROR,
                        "warning: first frame is no keyframe\n");
@@ -1478,6 +1491,16 @@ int ff_MPV_frame_start(MpegEncContext *s, AVCodecContext *avctx)
                 s->last_picture_ptr = NULL;
                 return -1;
             }
+
+            memset(s->last_picture_ptr->f.data[0], 0,
+                   avctx->height * s->last_picture_ptr->f.linesize[0]);
+            memset(s->last_picture_ptr->f.data[1], 0x80,
+                   (avctx->height >> v_chroma_shift) *
+                   s->last_picture_ptr->f.linesize[1]);
+            memset(s->last_picture_ptr->f.data[2], 0x80,
+                   (avctx->height >> v_chroma_shift) *
+                   s->last_picture_ptr->f.linesize[2]);
+
             ff_thread_report_progress(&s->last_picture_ptr->f, INT_MAX, 0);
             ff_thread_report_progress(&s->last_picture_ptr->f, INT_MAX, 1);
             s->last_picture_ptr->f.reference = 3;
@@ -1657,7 +1680,7 @@ static void draw_line(uint8_t *buf, int sx, int sy, int ex, int ey,
         buf += sx + sy * stride;
         ex  -= sx;
         f    = ((ey - sy) << 16) / ex;
-        for (x = 0; x = ex; x++) {
+        for (x = 0; x <= ex; x++) {
             y  = (x * f) >> 16;
             fr = (x * f) & 0xFFFF;
             buf[y * stride + x]       += (color * (0x10000 - fr)) >> 16;
@@ -2065,7 +2088,7 @@ unhandled:
 
 /* put block[] to dest[] */
 static inline void put_dct(MpegEncContext *s,
-                           DCTELEM *block, int i, uint8_t *dest, int line_size, int qscale)
+                           int16_t *block, int i, uint8_t *dest, int line_size, int qscale)
 {
     s->dct_unquantize_intra(s, block, i, qscale);
     s->dsp.idct_put (dest, line_size, block);
@@ -2073,7 +2096,7 @@ static inline void put_dct(MpegEncContext *s,
 
 /* add block[] to dest[] */
 static inline void add_dct(MpegEncContext *s,
-                           DCTELEM *block, int i, uint8_t *dest, int line_size)
+                           int16_t *block, int i, uint8_t *dest, int line_size)
 {
     if (s->block_last_index[i] >= 0) {
         s->dsp.idct_add (dest, line_size, block);
@@ -2081,7 +2104,7 @@ static inline void add_dct(MpegEncContext *s,
 }
 
 static inline void add_dequant_dct(MpegEncContext *s,
-                           DCTELEM *block, int i, uint8_t *dest, int line_size, int qscale)
+                           int16_t *block, int i, uint8_t *dest, int line_size, int qscale)
 {
     if (s->block_last_index[i] >= 0) {
         s->dct_unquantize_inter(s, block, i, qscale);
@@ -2134,7 +2157,7 @@ void ff_clean_intra_table_entries(MpegEncContext *s)
    s->interlaced_dct : true if interlaced dct used (mpeg2)
  */
 static av_always_inline
-void MPV_decode_mb_internal(MpegEncContext *s, DCTELEM block[12][64],
+void MPV_decode_mb_internal(MpegEncContext *s, int16_t block[12][64],
                             int is_mpeg12)
 {
     const int mb_xy = s->mb_y * s->mb_stride + s->mb_x;
@@ -2146,7 +2169,7 @@ void MPV_decode_mb_internal(MpegEncContext *s, DCTELEM block[12][64],
     if(s->avctx->debug&FF_DEBUG_DCT_COEFF) {
        /* save DCT coefficients */
        int i,j;
-       DCTELEM *dct = &s->current_picture.f.dct_coeff[mb_xy * 64 * 6];
+       int16_t *dct = &s->current_picture.f.dct_coeff[mb_xy * 64 * 6];
        av_log(s->avctx, AV_LOG_DEBUG, "DCT coeffs of MB at %dx%d:\n", s->mb_x, s->mb_y);
        for(i=0; i<6; i++){
            for(j=0; j<64; j++){
@@ -2365,7 +2388,7 @@ skip_idct:
     }
 }
 
-void ff_MPV_decode_mb(MpegEncContext *s, DCTELEM block[12][64]){
+void ff_MPV_decode_mb(MpegEncContext *s, int16_t block[12][64]){
 #if !CONFIG_SMALL
     if(s->out_format == FMT_MPEG1) {
         MPV_decode_mb_internal(s, block, 1);
@@ -2476,6 +2499,35 @@ void ff_init_block_index(MpegEncContext *s){ //FIXME maybe rename
     }
 }
 
+/**
+ * Permute an 8x8 block.
+ * @param block the block which will be permuted according to the given permutation vector
+ * @param permutation the permutation vector
+ * @param last the last non zero coefficient in scantable order, used to speed the permutation up
+ * @param scantable the used scantable, this is only used to speed the permutation up, the block is not
+ *                  (inverse) permutated to scantable order!
+ */
+void ff_block_permute(int16_t *block, uint8_t *permutation, const uint8_t *scantable, int last)
+{
+    int i;
+    int16_t temp[64];
+
+    if(last<=0) return;
+    //if(permutation[1]==1) return; //FIXME it is ok but not clean and might fail for some permutations
+
+    for(i=0; i<=last; i++){
+        const int j= scantable[i];
+        temp[j]= block[j];
+        block[j]=0;
+    }
+
+    for(i=0; i<=last; i++){
+        const int j= scantable[i];
+        const int perm_j= permutation[j];
+        block[perm_j]= temp[j];
+    }
+}
+
 void ff_mpeg_flush(AVCodecContext *avctx){
     int i;
     MpegEncContext *s = avctx->priv_data;
@@ -2504,7 +2556,7 @@ void ff_mpeg_flush(AVCodecContext *avctx){
 }
 
 static void dct_unquantize_mpeg1_intra_c(MpegEncContext *s,
-                                   DCTELEM *block, int n, int qscale)
+                                   int16_t *block, int n, int qscale)
 {
     int i, level, nCoeffs;
     const uint16_t *quant_matrix;
@@ -2536,7 +2588,7 @@ static void dct_unquantize_mpeg1_intra_c(MpegEncContext *s,
 }
 
 static void dct_unquantize_mpeg1_inter_c(MpegEncContext *s,
-                                   DCTELEM *block, int n, int qscale)
+                                   int16_t *block, int n, int qscale)
 {
     int i, level, nCoeffs;
     const uint16_t *quant_matrix;
@@ -2565,7 +2617,7 @@ static void dct_unquantize_mpeg1_inter_c(MpegEncContext *s,
 }
 
 static void dct_unquantize_mpeg2_intra_c(MpegEncContext *s,
-                                   DCTELEM *block, int n, int qscale)
+                                   int16_t *block, int n, int qscale)
 {
     int i, level, nCoeffs;
     const uint16_t *quant_matrix;
@@ -2595,7 +2647,7 @@ static void dct_unquantize_mpeg2_intra_c(MpegEncContext *s,
 }
 
 static void dct_unquantize_mpeg2_intra_bitexact(MpegEncContext *s,
-                                   DCTELEM *block, int n, int qscale)
+                                   int16_t *block, int n, int qscale)
 {
     int i, level, nCoeffs;
     const uint16_t *quant_matrix;
@@ -2628,7 +2680,7 @@ static void dct_unquantize_mpeg2_intra_bitexact(MpegEncContext *s,
 }
 
 static void dct_unquantize_mpeg2_inter_c(MpegEncContext *s,
-                                   DCTELEM *block, int n, int qscale)
+                                   int16_t *block, int n, int qscale)
 {
     int i, level, nCoeffs;
     const uint16_t *quant_matrix;
@@ -2659,7 +2711,7 @@ static void dct_unquantize_mpeg2_inter_c(MpegEncContext *s,
 }
 
 static void dct_unquantize_h263_intra_c(MpegEncContext *s,
-                                  DCTELEM *block, int n, int qscale)
+                                  int16_t *block, int n, int qscale)
 {
     int i, level, qmul, qadd;
     int nCoeffs;
@@ -2696,7 +2748,7 @@ static void dct_unquantize_h263_intra_c(MpegEncContext *s,
 }
 
 static void dct_unquantize_h263_inter_c(MpegEncContext *s,
-                                  DCTELEM *block, int n, int qscale)
+                                  int16_t *block, int n, int qscale)
 {
     int i, level, qmul, qadd;
     int nCoeffs;