Fix incorrectly constructed Dirac parse units that caused A/V sync loss.

Fixes issue 694.
patch by Anuradha Suraparaju, anuradha rd.bbc.co uk

Originally committed as revision 16261 to svn://svn.ffmpeg.org/ffmpeg/trunk

libavcodec/dirac_parser.c

@@ -1,7 +1,8 @@
 /*
  * Dirac parser
  *
- * Copyright (c) 2007 Marco Gerards <marco@gnu.org>
+ * Copyright (c) 2007-2008 Marco Gerards <marco@gnu.org>
+ * Copyright (c) 2008 BBC, Anuradha Suraparaju <asuraparaju@gmail.com>
  *
  * This file is part of FFmpeg.
  *
@@ -34,42 +35,200 @@
  * Finds the end of the current frame in the bitstream.
  * @return the position of the first byte of the next frame or -1
  */
-static int find_frame_end(ParseContext *pc, const uint8_t *buf, int buf_size)
+typedef struct DiracParseContext {
+    int state;
+    int is_synced;
+    int sync_offset;
+    int header_bytes_needed;
+    int overread_index;
+    int buffer_size;
+    int index;
+    uint8_t *buffer;
+    int dirac_unit_size;
+    uint8_t *dirac_unit;
+} DiracParseContext;
+
+static int find_frame_end(DiracParseContext *pc,
+                          const uint8_t *buf, int buf_size)
 {
     uint32_t state = pc->state;
-    int i;
+    int i = 0;
 
-    for (i = 0; i < buf_size; i++) {
+    if (!pc->is_synced) {
-        state = (state << 8) | buf[i];
+        for (i = 0; i < buf_size; i++) {
-        if (state == DIRAC_PARSE_INFO_PREFIX) {
+            state = (state << 8) | buf[i];
-            pc->frame_start_found ^= 1;
+            if (state == DIRAC_PARSE_INFO_PREFIX) {
-            if (!pc->frame_start_found) {
+                state                   = -1;
-                pc->state = -1;
+                pc->is_synced           = 1;
-                return i - 3;
+                pc->header_bytes_needed = 9;
+                pc->sync_offset         = i;
+                break;
             }
         }
     }
 
+    if (pc->is_synced) {
+        pc->sync_offset = 0;
+        for (; i < buf_size; i++) {
+            if (state == DIRAC_PARSE_INFO_PREFIX) {
+                if ((buf_size-i) >= pc->header_bytes_needed) {
+                    pc->state = -1;
+                    return i + pc->header_bytes_needed;
+                } else {
+                    pc->header_bytes_needed = 9-(buf_size-i);
+                    break;
+                }
+            } else
+              state = (state << 8) | buf[i];
+        }
+    }
     pc->state = state;
+    return -1;
+}
 
-    return END_NOT_FOUND;
+typedef struct DiracParseUnit
+{
+    int next_pu_offset;
+    int prev_pu_offset;
+    uint8_t pu_type;
+} DiracParseUnit;
+
+static int unpack_parse_unit(DiracParseUnit *pu, DiracParseContext *pc,
+                             int offset)
+{
+    uint8_t *start = pc->buffer + offset;
+    uint8_t *end   = pc->buffer + pc->index;
+    if (start < pc->buffer || (start+13 > end))
+        return 0;
+    pu->pu_type = start[4];
+
+    pu->next_pu_offset = AV_RB32(start+5);
+    pu->prev_pu_offset = AV_RB32(start+9);
+
+    if (pu->pu_type == 0x10 && pu->next_pu_offset == 0)
+        pu->next_pu_offset = 13;
+
+    return 1;
+}
+
+static int dirac_combine_frame(AVCodecParserContext *s, AVCodecContext *avctx,
+                               int next, const uint8_t **buf, int *buf_size)
+{
+    int parse_timing_info = (s->pts == AV_NOPTS_VALUE &&
+                             s->dts == AV_NOPTS_VALUE);
+    DiracParseContext *pc = s->priv_data;
+
+    if (pc->overread_index) {
+        memcpy(pc->buffer, pc->buffer + pc->overread_index,
+               pc->index - pc->overread_index);
+        pc->index -= pc->overread_index;
+        pc->overread_index = 0;
+        if (*buf_size == 0 && pc->buffer[4] == 0x10) {
+            *buf      = pc->buffer;
+            *buf_size = pc->index;
+            return 0;
+        }
+    }
+
+    if ( next == -1) {
+        /* Found a possible frame start but not a frame end */
+        void *new_buffer = av_fast_realloc(pc->buffer, &pc->buffer_size,
+                                           pc->index + (*buf_size -
+                                                        pc->sync_offset));
+        pc->buffer = new_buffer;
+        memcpy(pc->buffer+pc->index, (*buf + pc->sync_offset),
+               *buf_size - pc->sync_offset);
+        pc->index += *buf_size - pc->sync_offset;
+        return -1;
+    } else {
+        /* Found a possible frame start and a  possible frame end */
+        DiracParseUnit pu1, pu;
+        void *new_buffer = av_fast_realloc(pc->buffer, &pc->buffer_size,
+                                           pc->index + next);
+        pc->buffer = new_buffer;
+        memcpy(pc->buffer + pc->index, *buf, next);
+        pc->index += next;
+
+        /* Need to check if we have a valid Parse Unit. We can't go by the
+         * sync pattern 'BBCD' alone because arithmetic coding of the residual
+         * and motion data can cause the pattern triggering a false start of
+         * frame. So check if the previous parse offset of the next parse unit
+         * is equal to the next parse offset of the current parse unit then
+         * we can be pretty sure that we have a valid parse unit */
+        if (!unpack_parse_unit(&pu1, pc, pc->index - 13)                     ||
+            !unpack_parse_unit(&pu, pc, pc->index - 13 - pu1.prev_pu_offset) ||
+            pu.next_pu_offset != pu1.prev_pu_offset) {
+            pc->index -= 9;
+            *buf_size = next-9;
+            pc->header_bytes_needed = 9;
+            return -1;
+        }
+
+        /* All non-frame data must be accompanied by frame data. This is to
+         * ensure that pts is set correctly. So if the current parse unit is
+         * not frame data, wait for frame data to come along */
+
+        pc->dirac_unit = pc->buffer + pc->index - 13 -
+                         pu1.prev_pu_offset - pc->dirac_unit_size;
+
+        pc->dirac_unit_size += pu.next_pu_offset;
+
+        if ((pu.pu_type&0x08) != 0x08) {
+            pc->header_bytes_needed = 9;
+            *buf_size = next;
+            return -1;
+        }
+
+        /* Get the picture number to set the pts and dts*/
+        if (parse_timing_info) {
+            uint8_t *cur_pu = pc->buffer +
+                              pc->index - 13 - pu1.prev_pu_offset;
+            int pts =  AV_RB32(cur_pu + 13);
+            if (s->last_pts == 0 && s->last_dts == 0)
+                s->dts = pts - 1;
+            else
+                s->dts = s->last_dts+1;
+            s->pts = pts;
+            if (!avctx->has_b_frames && (cur_pu[4] & 0x03))
+                avctx->has_b_frames = 1;
+        }
+        if (avctx->has_b_frames && s->pts == s->dts)
+             s->pict_type = FF_B_TYPE;
+
+        /* Finally have a complete Dirac data unit */
+        *buf      = pc->dirac_unit;
+        *buf_size = pc->dirac_unit_size;
+
+        pc->dirac_unit_size     = 0;
+        pc->overread_index      = pc->index-13;
+        pc->header_bytes_needed = 9;
+    }
+    return next;
 }
 
 static int dirac_parse(AVCodecParserContext *s, AVCodecContext *avctx,
                        const uint8_t **poutbuf, int *poutbuf_size,
                        const uint8_t *buf, int buf_size)
 {
-    ParseContext *pc = s->priv_data;
+    DiracParseContext *pc = s->priv_data;
     int next;
 
+    *poutbuf = NULL;
+    *poutbuf_size = 0;
+
     if (s->flags & PARSER_FLAG_COMPLETE_FRAMES) {
         next = buf_size;
-    }else{
+        *poutbuf = buf;
+        *poutbuf_size = buf_size;
+        /* Assume that data has been packetized into an encapsulation unit. */
+    } else {
         next = find_frame_end(pc, buf, buf_size);
+        if (!pc->is_synced && next == -1) {
+            /* No frame start found yet. So throw away the entire buffer. */
+            return buf_size;
+        }
 
-        if (ff_combine_frame(pc, next, &buf, &buf_size) < 0) {
+        if (dirac_combine_frame(s, avctx, next, &buf, &buf_size) < 0) {
-            *poutbuf = NULL;
-            *poutbuf_size = 0;
             return buf_size;
         }
     }
@@ -79,10 +238,18 @@ static int dirac_parse(AVCodecParserContext *s, AVCodecContext *avctx,
     return next;
 }
 
+static void dirac_parse_close(AVCodecParserContext *s)
+{
+    DiracParseContext *pc = s->priv_data;
+
+    if (pc->buffer_size > 0)
+        av_free(pc->buffer);
+}
+
 AVCodecParser dirac_parser = {
     { CODEC_ID_DIRAC },
-    sizeof(ParseContext),
+    sizeof(DiracParseContext),
     NULL,
     dirac_parse,
-    ff_parse_close,
+    dirac_parse_close,
 };

libavcodec/libdiracdec.c

@@ -88,10 +88,12 @@ static int libdirac_decode_frame(AVCodecContext *avccontext,
 
     *data_size = 0;
 
-    if (buf_size>0)
+    if (buf_size>0) {
         /* set data to decode into buffer */
         dirac_buffer (p_dirac_params->p_decoder, buf, buf+buf_size);
-
+        if ((buf[4] &0x08) == 0x08 && (buf[4] & 0x03))
+            avccontext->has_b_frames = 1;
+    }
     while (1) {
          /* parse data and process result */
         DecoderState state = dirac_parse (p_dirac_params->p_decoder);

libavcodec/libschroedingerdec.c

@@ -235,6 +235,9 @@ static int libschroedinger_decode_frame(AVCodecContext *avccontext,
     do {
         if ((enc_buf = FfmpegFindNextSchroParseUnit(&parse_ctx))) {
             /* Push buffer into decoder. */
+            if (SCHRO_PARSE_CODE_IS_PICTURE(enc_buf->data[4]) &&
+                SCHRO_PARSE_CODE_NUM_REFS(enc_buf->data[4]) > 0)
+                avccontext->has_b_frames = 1;
             state = schro_decoder_push (decoder, enc_buf);
             if (state == SCHRO_DECODER_FIRST_ACCESS_UNIT)
                   libschroedinger_handle_first_access_unit(avccontext);