1 files changed, 117 insertions, 10 deletions

diff --git a/media/ffvpx/libavcodec/vp9_parser.c b/media/ffvpx/libavcodec/vp9_parser.c
index 9531f34a3..9900e7ab1 100644
--- a/media/ffvpx/libavcodec/vp9_parser.c
+++ b/media/ffvpx/libavcodec/vp9_parser.c
@@ -25,19 +25,21 @@
 #include "libavcodec/get_bits.h"
 #include "parser.h"
 
-static int parse(AVCodecParserContext *ctx,
-                 AVCodecContext *avctx,
-                 const uint8_t **out_data, int *out_size,
-                 const uint8_t *data, int size)
+typedef struct VP9ParseContext {
+    int n_frames; // 1-8
+    int size[8];
+    int marker_size;
+    int64_t pts;
+} VP9ParseContext;
+
+static int parse_frame(AVCodecParserContext *ctx, const uint8_t *buf, int size)
 {
+    VP9ParseContext *s = ctx->priv_data;
     GetBitContext gb;
-    int res, profile, keyframe;
+    int res, profile, keyframe, invisible;
 
-    *out_data = data;
-    *out_size = size;
-
-    if ((res = init_get_bits8(&gb, data, size)) < 0)
-        return size; // parsers can't return errors
+    if ((res = init_get_bits8(&gb, buf, size)) < 0)
+        return res;
     get_bits(&gb, 2); // frame marker
     profile  = get_bits1(&gb);
     profile |= get_bits1(&gb) << 1;
@@ -45,8 +47,10 @@ static int parse(AVCodecParserContext *ctx,
 
     if (get_bits1(&gb)) {
         keyframe = 0;
+        invisible = 0;
     } else {
         keyframe  = !get_bits1(&gb);
+        invisible = !get_bits1(&gb);
     }
 
     if (!keyframe) {
@@ -57,10 +61,113 @@ static int parse(AVCodecParserContext *ctx,
         ctx->key_frame = 1;
     }
 
+    if (!invisible) {
+        if (ctx->pts == AV_NOPTS_VALUE)
+            ctx->pts = s->pts;
+        s->pts = AV_NOPTS_VALUE;
+    } else if (ctx->pts != AV_NOPTS_VALUE) {
+        s->pts = ctx->pts;
+        ctx->pts = AV_NOPTS_VALUE;
+    }
+
+    return 0;
+}
+
+static int parse(AVCodecParserContext *ctx,
+                 AVCodecContext *avctx,
+                 const uint8_t **out_data, int *out_size,
+                 const uint8_t *data, int size)
+{
+    VP9ParseContext *s = ctx->priv_data;
+    int full_size = size;
+    int marker;
+
+    if (size <= 0) {
+        *out_size = 0;
+        *out_data = data;
+
+        return 0;
+    }
+
+    if (s->n_frames > 0) {
+        int i;
+        int size_sum = 0;
+
+        for (i = 0; i < s->n_frames ;i++)
+            size_sum += s->size[i];
+        size_sum += s->marker_size;
+
+        if (size_sum != size) {
+            av_log(avctx, AV_LOG_ERROR, "Inconsistent input frame sizes %d %d\n",
+                   size_sum, size);
+            s->n_frames = 0;
+        }
+    }
+
+    if (s->n_frames > 0) {
+        *out_data = data;
+        *out_size = s->size[--s->n_frames];
+        parse_frame(ctx, *out_data, *out_size);
+
+        return s->n_frames > 0 ? *out_size : size /* i.e. include idx tail */;
+    }
+
+    marker = data[size - 1];
+    if ((marker & 0xe0) == 0xc0) {
+        int nbytes = 1 + ((marker >> 3) & 0x3);
+        int n_frames = 1 + (marker & 0x7), idx_sz = 2 + n_frames * nbytes;
+
+        if (size >= idx_sz && data[size - idx_sz] == marker) {
+            const uint8_t *idx = data + size + 1 - idx_sz;
+            int first = 1;
+
+            switch (nbytes) {
+#define case_n(a, rd) \
+            case a: \
+                while (n_frames--) { \
+                    unsigned sz = rd; \
+                    idx += a; \
+                    if (sz == 0 || sz > size) { \
+                        s->n_frames = 0; \
+                        *out_size = size; \
+                        *out_data = data; \
+                        av_log(avctx, AV_LOG_ERROR, \
+                               "Invalid superframe packet size: %u frame size: %d\n", \
+                               sz, size); \
+                        return full_size; \
+                    } \
+                    if (first) { \
+                        first = 0; \
+                        *out_data = data; \
+                        *out_size = sz; \
+                        s->n_frames = n_frames; \
+                    } else { \
+                        s->size[n_frames] = sz; \
+                    } \
+                    data += sz; \
+                    size -= sz; \
+                } \
+                s->marker_size = size; \
+                parse_frame(ctx, *out_data, *out_size); \
+                return s->n_frames > 0 ? *out_size : full_size
+
+                case_n(1, *idx);
+                case_n(2, AV_RL16(idx));
+                case_n(3, AV_RL24(idx));
+                case_n(4, AV_RL32(idx));
+            }
+        }
+    }
+
+    *out_data = data;
+    *out_size = size;
+    parse_frame(ctx, data, size);
+
     return size;
 }
 
 AVCodecParser ff_vp9_parser = {
     .codec_ids      = { AV_CODEC_ID_VP9 },
+    .priv_data_size = sizeof(VP9ParseContext),
     .parser_parse   = parse,
 };