Use a heuristic for describing the RTP packets using sample data

Originally committed as revision 23165 to svn://svn.ffmpeg.org/ffmpeg/trunk

libavformat/movenc.h

@@ -51,6 +51,20 @@ typedef struct MOVIentry {
     uint32_t     flags;
 } MOVIentry;
 
+typedef struct HintSample {
+    uint8_t *data;
+    int size;
+    int sample_number;
+    int offset;
+    int own_data;
+} HintSample;
+
+typedef struct {
+    int size;
+    int len;
+    HintSample *samples;
+} HintSampleQueue;
+
 typedef struct MOVIndex {
     int         mode;
     int         entry;
@@ -82,6 +96,8 @@ typedef struct MOVIndex {
     uint32_t    prev_rtp_ts;
     int64_t     cur_rtp_ts_unwrapped;
     uint32_t    max_packet_size;
+
+    HintSampleQueue sample_queue;
 } MOVTrack;
 
 typedef struct MOVMuxContext {

libavformat/movenchint.c

@@ -98,6 +98,188 @@ fail:
     return ret;
 }
 
+/**
+ * Remove the first sample from the sample queue.
+ */
+static void sample_queue_pop(HintSampleQueue *queue)
+{
+    if (queue->len <= 0)
+        return;
+    if (queue->samples[0].own_data)
+        av_free(queue->samples[0].data);
+    queue->len--;
+    memmove(queue->samples, queue->samples + 1, sizeof(HintSample)*queue->len);
+}
+
+/**
+ * Empty the sample queue, releasing all memory.
+ */
+static void sample_queue_free(HintSampleQueue *queue)
+{
+    int i;
+    for (i = 0; i < queue->len; i++)
+        if (queue->samples[i].own_data)
+            av_free(queue->samples[i].data);
+    av_freep(&queue->samples);
+    queue->len = 0;
+    queue->size = 0;
+}
+
+/**
+ * Add a reference to the sample data to the sample queue. The data is
+ * not copied. sample_queue_retain should be called before pkt->data
+ * is reused/freed.
+ */
+static void sample_queue_push(HintSampleQueue *queue, AVPacket *pkt, int sample)
+{
+    /* No need to keep track of smaller samples, since describing them
+     * with immediates is more efficient. */
+    if (pkt->size <= 14)
+        return;
+    if (!queue->samples || queue->len >= queue->size) {
+        HintSample* samples;
+        queue->size += 10;
+        samples = av_realloc(queue->samples, sizeof(HintSample)*queue->size);
+        if (!samples)
+            return;
+        queue->samples = samples;
+    }
+    queue->samples[queue->len].data = pkt->data;
+    queue->samples[queue->len].size = pkt->size;
+    queue->samples[queue->len].sample_number = sample;
+    queue->samples[queue->len].offset = 0;
+    queue->samples[queue->len].own_data = 0;
+    queue->len++;
+}
+
+/**
+ * Make local copies of all referenced sample data in the queue.
+ */
+static void sample_queue_retain(HintSampleQueue *queue)
+{
+    int i;
+    for (i = 0; i < queue->len; ) {
+        HintSample *sample = &queue->samples[i];
+        if (!sample->own_data) {
+            uint8_t* ptr = av_malloc(sample->size);
+            if (!ptr) {
+                /* Unable to allocate memory for this one, remove it */
+                memmove(queue->samples + i, queue->samples + i + 1,
+                        sizeof(HintSample)*(queue->len - i - 1));
+                queue->len--;
+                continue;
+            }
+            memcpy(ptr, sample->data, sample->size);
+            sample->data = ptr;
+            sample->own_data = 1;
+        }
+        i++;
+    }
+}
+
+/**
+ * Find matches of needle[n_pos ->] within haystack. If a sufficiently
+ * large match is found, matching bytes before n_pos are included
+ * in the match, too (within the limits of the arrays).
+ *
+ * @param haystack buffer that may contain parts of needle
+ * @param h_len length of the haystack buffer
+ * @param needle buffer containing source data that have been used to
+ *               construct haystack
+ * @param n_pos start position in needle used for looking for matches
+ * @param n_len length of the needle buffer
+ * @param match_h_offset_ptr offset of the first matching byte within haystack
+ * @param match_n_offset_ptr offset of the first matching byte within needle
+ * @param match_len_ptr length of the matched segment
+ * @return 0 if a match was found, < 0 if no match was found
+ */
+static int match_segments(const uint8_t *haystack, int h_len,
+                          const uint8_t *needle, int n_pos, int n_len,
+                          int *match_h_offset_ptr, int *match_n_offset_ptr,
+                          int *match_len_ptr)
+{
+    int h_pos;
+    for (h_pos = 0; h_pos < h_len; h_pos++) {
+        int match_len = 0;
+        int match_h_pos, match_n_pos;
+
+        /* Check how many bytes match at needle[n_pos] and haystack[h_pos] */
+        while (h_pos + match_len < h_len && n_pos + match_len < n_len &&
+               needle[n_pos + match_len] == haystack[h_pos + match_len])
+            match_len++;
+        if (match_len <= 8)
+            continue;
+
+        /* If a sufficiently large match was found, try to expand
+         * the matched segment backwards. */
+        match_h_pos = h_pos;
+        match_n_pos = n_pos;
+        while (match_n_pos > 0 && match_h_pos > 0 &&
+               needle[match_n_pos - 1] == haystack[match_h_pos - 1]) {
+            match_n_pos--;
+            match_h_pos--;
+            match_len++;
+        }
+        if (match_len <= 14)
+            continue;
+        *match_h_offset_ptr = match_h_pos;
+        *match_n_offset_ptr = match_n_pos;
+        *match_len_ptr = match_len;
+        return 0;
+    }
+    return -1;
+}
+
+/**
+ * Look for segments in samples in the sample queue matching the data
+ * in ptr. Samples not matching are removed from the queue. If a match
+ * is found, the next time it will look for matches starting from the
+ * end of the previous matched segment.
+ *
+ * @param data data to find matches for in the sample queue
+ * @param len length of the data buffer
+ * @param queue samples used for looking for matching segments
+ * @param pos the offset in data of the matched segment
+ * @param match_sample the number of the sample that contained the match
+ * @param match_offset the offset of the matched segment within the sample
+ * @param match_len the length of the matched segment
+ * @return 0 if a match was found, < 0 if no match was found
+ */
+static int find_sample_match(const uint8_t *data, int len,
+                             HintSampleQueue *queue, int *pos,
+                             int *match_sample, int *match_offset,
+                             int *match_len)
+{
+    while (queue->len > 0) {
+        HintSample *sample = &queue->samples[0];
+        /* If looking for matches in a new sample, skip the first 5 bytes,
+         * since they often may be modified/removed in the output packet. */
+        if (sample->offset == 0 && sample->size > 5)
+            sample->offset = 5;
+
+        if (match_segments(data, len, sample->data, sample->offset,
+                           sample->size, pos, match_offset, match_len) == 0) {
+            *match_sample = sample->sample_number;
+            /* Next time, look for matches at this offset, with a little
+             * margin to this match. */
+            sample->offset = *match_offset + *match_len + 5;
+            if (sample->offset + 10 >= sample->size)
+                sample_queue_pop(queue); /* Not enough useful data left */
+            return 0;
+        }
+
+        if (sample->offset < 10 && sample->size > 20) {
+            /* No match found from the start of the sample,
+             * try from the middle of the sample instead. */
+            sample->offset = sample->size/2;
+        } else {
+            /* No match for this sample, remove it */
+            sample_queue_pop(queue);
+        }
+    }
+    return -1;
+}
+
 static void output_immediate(const uint8_t *data, int size,
                              ByteIOContext *out, int *entries)
 {
@@ -118,10 +300,36 @@ static void output_immediate(const uint8_t *data, int size,
     }
 }
 
+static void output_match(ByteIOContext *out, int match_sample,
+                         int match_offset, int match_len, int *entries)
+{
+    put_byte(out, 2); /* sample constructor */
+    put_byte(out, 0); /* track reference */
+    put_be16(out, match_len);
+    put_be32(out, match_sample);
+    put_be32(out, match_offset);
+    put_be16(out, 1); /* bytes per block */
+    put_be16(out, 1); /* samples per block */
+    (*entries)++;
+}
+
 static void describe_payload(const uint8_t *data, int size,
-                             ByteIOContext *out, int *entries)
+                             ByteIOContext *out, int *entries,
+                             HintSampleQueue *queue)
 {
     /* Describe the payload using different constructors */
+    while (size > 0) {
+        int match_sample, match_offset, match_len, pos;
+        if (find_sample_match(data, size, queue, &pos, &match_sample,
+                              &match_offset, &match_len) < 0)
+            break;
+        output_immediate(data, pos, out, entries);
+        data += pos;
+        size -= pos;
+        output_match(out, match_sample, match_offset, match_len, entries);
+        data += match_len;
+        size -= match_len;
+    }
     output_immediate(data, size, out, entries);
 }
 
@@ -195,7 +403,7 @@ static int write_hint_packets(ByteIOContext *out, const uint8_t *data,
 
         entries = 0;
         /* Write one or more constructors describing the payload data */
-        describe_payload(data, packet_len, out, &entries);
+        describe_payload(data, packet_len, out, &entries, &trk->sample_queue);
         data += packet_len;
         size -= packet_len;
 
@@ -230,6 +438,8 @@ int ff_mov_add_hinted_packet(AVFormatContext *s, AVPacket *pkt,
     if (!rtp_ctx->pb)
         return AVERROR(ENOMEM);
 
+    sample_queue_push(&trk->sample_queue, pkt, sample);
+
     /* Feed the packet to the RTP muxer */
     local_pkt = *pkt;
     local_pkt.stream_index = 0;
@@ -269,6 +479,7 @@ int ff_mov_add_hinted_packet(AVFormatContext *s, AVPacket *pkt,
         ff_mov_write_packet(s, &hint_pkt);
 done:
     av_free(buf);
+    sample_queue_retain(&trk->sample_queue);
     return ret;
 }
 
@@ -277,6 +488,7 @@ void ff_mov_close_hinting(MOVTrack *track) {
     uint8_t *ptr;
 
     av_freep(&track->enc);
+    sample_queue_free(&track->sample_queue);
     if (!rtp_ctx)
         return;
     if (rtp_ctx->pb) {