diff mbox series

[FFmpeg-devel,RFC,GSoC,v2,3/6] avformat/hls: use abr to switch streams

Message ID 20200716125116.187031-3-sj.hc_Zhong@sjtu.edu.cn
State Superseded
Headers show
Series [FFmpeg-devel,RFC,GSoC,v2,1/6] avformat/abr: Adaptive Bitrate support
Related show

Checks

Context Check Description
andriy/default pending
andriy/make success Make finished
andriy/make_fate success Make fate finished

Commit Message

Hongcheng Zhong July 16, 2020, 12:51 p.m. UTC
From: spartazhc <spartazhc@gmail.com>

When abr is enable, it will take over the task to call http to
download segments, and will return a switch-request for hls to
switch streams.
For reason not to waste segments that have been downloaded,
switch will become effective after old segments is used out.
Abr cannot work with http_persistent option, and currently use
http_multiple.

v1 fixed:
1. fix memory leak

v2 fixed:
1. check malloc result and return AVERROR(ENOMEM)
2. define ABR_THROUGHPUT_FIFO_LEN instead of hardcode

Signed-off-by: spartazhc <spartazhc@gmail.com>
---
 doc/demuxers.texi |   3 +
 libavformat/hls.c | 237 ++++++++++++++++++++++++++++++++++++++++++++--
 2 files changed, 234 insertions(+), 6 deletions(-)
diff mbox series

Patch

diff --git a/doc/demuxers.texi b/doc/demuxers.texi
index 3c15ab9eee..4cdbd95962 100644
--- a/doc/demuxers.texi
+++ b/doc/demuxers.texi
@@ -321,6 +321,9 @@  available in a metadata key named "variant_bitrate".
 It accepts the following options:
 
 @table @option
+@item abr
+enable abr to switch streams.
+
 @item live_start_index
 segment index to start live streams at (negative values are from the end).
 
diff --git a/libavformat/hls.c b/libavformat/hls.c
index ba17c4ed96..0b297303d8 100644
--- a/libavformat/hls.c
+++ b/libavformat/hls.c
@@ -47,6 +47,7 @@ 
 #define MPEG_TIME_BASE 90000
 #define MPEG_TIME_BASE_Q (AVRational){1, MPEG_TIME_BASE}
 
+#define ABR_THROUGHPUT_FIFO_LEN 20
 /*
  * An apple http stream consists of a playlist with media segment files,
  * played sequentially. There may be several playlists with the same
@@ -189,6 +190,15 @@  struct variant {
     char subtitles_group[MAX_FIELD_LEN];
 };
 
+struct throughput {
+    int n_throughputs;
+
+    /* throughputs are in kbps */
+    float throughput_fifo[ABR_THROUGHPUT_FIFO_LEN];
+    int head;
+    int tail;
+};
+
 typedef struct HLSContext {
     AVClass *class;
     AVFormatContext *ctx;
@@ -213,8 +223,36 @@  typedef struct HLSContext {
     int http_multiple;
     int http_seekable;
     AVIOContext *playlist_pb;
+
+    int abr;
+    struct throughput *throughputs;
+    int can_switch;
+    int switch_request2;
+    int switch_delay;
+    int64_t switch_timestamp;
+    int64_t delta_timestamp;
+    int cur_pls;
 } HLSContext;
 
+static struct segment *next_segment(struct playlist *pls);
+static int open_input(HLSContext *c, struct playlist *pls, struct segment *seg, AVIOContext **in);
+
+static void sync_cur_seq(HLSContext *c) {
+    int i;
+    for (i = 0; i < c->n_playlists; i++) {
+        struct playlist *pls = c->playlists[i];
+        pls->cur_seq_no = c->cur_seq_no;
+    }
+}
+
+static struct segment *next2_segment(struct playlist *pls)
+{
+    int n = pls->cur_seq_no - pls->start_seq_no + 2;
+    if (n >= pls->n_segments)
+        return NULL;
+    return pls->segments[n];
+}
+
 static void free_segment_dynarray(struct segment **segments, int n_segments)
 {
     int i;
@@ -624,6 +662,31 @@  static int open_url_keepalive(AVFormatContext *s, AVIOContext **pb,
 #endif
 }
 
+static int update_throughputs(struct throughput *thr, float time, int pb_size)
+{
+    if (pb_size <= 0 || time <= 0)
+        return AVERROR(EINVAL);
+    if (thr->n_throughputs < ABR_THROUGHPUT_FIFO_LEN) {
+        ++thr->n_throughputs;
+    } else {
+        ++thr->head;
+    }
+    thr->throughput_fifo[thr->tail] = (float)(pb_size) / time;
+    thr->tail = (thr->tail + 1) % ABR_THROUGHPUT_FIFO_LEN;
+    return 0;
+}
+
+static int64_t get_switch_timestamp(HLSContext *c, struct playlist *pls)
+{
+    int64_t pos = c->first_timestamp == AV_NOPTS_VALUE ?
+                  0 : c->first_timestamp;
+
+    for (int i = 0; i < pls->cur_seq_no + 2; i++) {
+        pos += pls->segments[i]->duration;
+    }
+    return pos;
+}
+
 static int open_url(AVFormatContext *s, AVIOContext **pb, const char *url,
                     AVDictionary *opts, AVDictionary *opts2, int *is_http_out)
 {
@@ -639,6 +702,9 @@  static int open_url(AVFormatContext *s, AVIOContext **pb, const char *url,
     } else if (av_strstart(url, "data", NULL)) {
         if (url[4] == '+' || url[4] == ':')
             proto_name = avio_find_protocol_name(url + 5);
+    } else if (av_strstart(url, "ffabr", NULL)) {
+        if (url[5] == '+' || url[5] == ':')
+            proto_name = avio_find_protocol_name(url + 6);
     }
 
     if (!proto_name)
@@ -669,6 +735,8 @@  static int open_url(AVFormatContext *s, AVIOContext **pb, const char *url,
         ;
     else if (av_strstart(url, "data", NULL) && !strncmp(proto_name, url + 5, strlen(proto_name)) && url[5 + strlen(proto_name)] == ':')
         ;
+    else if (av_strstart(url, "ffabr", NULL) && !strncmp(proto_name, url + 6, strlen(proto_name)) && url[6 + strlen(proto_name)] == ':')
+        ;
     else if (strcmp(proto_name, "file") || !strncmp(url, "file,", 5))
         return AVERROR_INVALIDDATA;
 
@@ -690,6 +758,43 @@  static int open_url(AVFormatContext *s, AVIOContext **pb, const char *url,
     } else {
         ret = s->io_open(s, pb, url, AVIO_FLAG_READ, &tmp);
     }
+    if (c->abr && ret >= 0) {
+        AVDictionary *abr_ret = NULL;
+        AVDictionaryEntry *en = NULL;
+        struct segment *seg;
+        int pb_size, switch_request;
+        av_opt_get_dict_val(*pb, "abr-metadata", AV_OPT_SEARCH_CHILDREN, &abr_ret);
+        if (abr_ret) {
+            en = av_dict_get(abr_ret, "download_time", NULL, 0);
+            if (en) {
+                pb_size = avio_size(*pb);
+                update_throughputs(c->throughputs, atoi(en->value) / 1000.0, pb_size);
+                av_log(s, AV_LOG_VERBOSE, "[abr] time=%.2fms, size=%.2fkb\n", atoi(en->value) / 1000.0, pb_size / 1000.0);
+            }
+            en = av_dict_get(abr_ret, "switch_request", NULL, 0);
+            if (en) {
+                switch_request = atoi(en->value);
+                av_log(s, AV_LOG_VERBOSE, "[abr] switch request: %s\n", en->value);
+            }
+            if (switch_request != -1) {
+                c->switch_request2 = switch_request;
+                c->switch_delay = 2;
+                c->can_switch = 0;
+                sync_cur_seq(c);
+                seg = next2_segment(c->playlists[c->switch_request2]);
+                if (!seg) {
+                    c->switch_request2 = -1;
+                    av_log(s, AV_LOG_INFO, "[abr] no more segment need to switch\n");
+                } else {
+                    c->switch_timestamp = get_switch_timestamp(c,  c->playlists[c->switch_request2]);
+                    av_log(s, AV_LOG_VERBOSE, "[abr] switch timestamp: %ld\n", c->switch_timestamp);
+                    c->playlists[c->switch_request2]->input_next_requested = 1;
+                    ret = open_input(c, c->playlists[c->switch_request2], seg, &c->playlists[c->switch_request2]->input_next);
+                }
+            }
+            av_dict_free(&abr_ret);
+        }
+    }
     if (ret >= 0) {
         // update cookies on http response with setcookies.
         char *new_cookies = NULL;
@@ -1219,6 +1324,35 @@  static void intercept_id3(struct playlist *pls, uint8_t *buf,
         pls->is_id3_timestamped = (pls->id3_mpegts_timestamp != AV_NOPTS_VALUE);
 }
 
+static int abrinfo_to_dict(HLSContext *c, char **abr_info)
+{
+    struct throughput *thr = c->throughputs;
+    char buffer[MAX_URL_SIZE];
+    int size, i;
+    size = snprintf(buffer, sizeof(buffer), "format=hls:");
+    size += snprintf(buffer + size, sizeof(buffer) - size, "cur_pls=%d:", c->cur_pls);
+    size += snprintf(buffer + size, sizeof(buffer) - size, "can_switch=%d:", c->can_switch);
+    size += snprintf(buffer + size, sizeof(buffer) - size, "n_variants=%d:", c->n_variants);
+    for (i = 0; i < c->n_variants; i++) {
+        struct variant *v = c->variants[i];
+        size += snprintf(buffer + size, sizeof(buffer) - size, "variant_bitrate%d=%d:", i, v->bandwidth);
+    }
+    size += snprintf(buffer + size, sizeof(buffer) - size, "n_throughputs=%d:", thr->n_throughputs);
+    if (thr->n_throughputs > 0) {
+        i = thr->head;
+        do {
+            size += snprintf(buffer + size, sizeof(buffer) - size, "throughputs%d=%.2f:", i, thr->throughput_fifo[i]);
+            i = (i + 1) % ABR_THROUGHPUT_FIFO_LEN;
+        } while (i != thr->tail);
+    }
+    *abr_info = av_malloc(size);
+    if (!abr_info)
+        return AVERROR(ENOMEM);
+    snprintf(*abr_info, size, "%s", buffer);
+    av_log(c, AV_LOG_DEBUG, "[abr] abr_info: %s\n", *abr_info);
+    return 0;
+}
+
 static int open_input(HLSContext *c, struct playlist *pls, struct segment *seg, AVIOContext **in)
 {
     AVDictionary *opts = NULL;
@@ -1238,8 +1372,20 @@  static int open_input(HLSContext *c, struct playlist *pls, struct segment *seg,
     av_log(pls->parent, AV_LOG_VERBOSE, "HLS request for url '%s', offset %"PRId64", playlist %d\n",
            seg->url, seg->url_offset, pls->index);
 
+    if (c->abr) {
+        char *abr_opts;
+        abrinfo_to_dict(c, &abr_opts);
+        av_dict_set(&opts, "abr-params", abr_opts, 0); // how to setup flag?
+        av_free(abr_opts);
+    }
     if (seg->key_type == KEY_NONE) {
-        ret = open_url(pls->parent, in, seg->url, c->avio_opts, opts, &is_http);
+        char url_abr[MAX_URL_SIZE];
+        if (c->abr) {
+            snprintf(url_abr, sizeof(url_abr), "ffabr:%s", seg->url); // + or : tbd
+            ret = open_url(pls->parent, in, url_abr, c->avio_opts, opts, &is_http);
+        } else {
+            ret = open_url(pls->parent, in, seg->url, c->avio_opts, opts, &is_http);
+        }
     } else if (seg->key_type == KEY_AES_128) {
         char iv[33], key[33], url[MAX_URL_SIZE];
         if (strcmp(seg->key, pls->key_url)) {
@@ -1273,6 +1419,7 @@  static int open_input(HLSContext *c, struct playlist *pls, struct segment *seg,
             goto cleanup;
         }
         ret = 0;
+        // TODO: Add abr prefix for crypto
     } else if (seg->key_type == KEY_SAMPLE_AES) {
         av_log(pls->parent, AV_LOG_ERROR,
                "SAMPLE-AES encryption is not supported yet\n");
@@ -1435,7 +1582,17 @@  restart:
         /* Check that the playlist is still needed before opening a new
          * segment. */
         v->needed = playlist_needed(v);
-
+        if (c->abr) {
+            if (c->switch_request2 == -1)
+                ;
+            else if (v->needed && c->switch_request2 != v->index && c->switch_delay <= 0) {
+                av_log(v->parent, AV_LOG_VERBOSE, "read_data: needed but not download playlist %d ('%s')\n",
+                       v->index, v->url);
+                return AVERROR_EOF;
+            }
+            if (!c->can_switch)
+                c->can_switch = 1;
+        }
         if (!v->needed) {
             av_log(v->parent, AV_LOG_INFO, "No longer receiving playlist %d ('%s')\n",
                    v->index, v->url);
@@ -1530,7 +1687,12 @@  reload:
     }
 
     seg = next_segment(v);
-    if (c->http_multiple == 1 && !v->input_next_requested &&
+
+    // when get switch_request, old stream should stop downloading next seg
+    if (c->abr && c->switch_request2 != v->index && c->switch_timestamp != AV_NOPTS_VALUE
+         && ((c->cur_timestamp + seg->duration * 2) >= c->switch_timestamp) )
+        ;
+    else if (c->http_multiple && !v->input_next_requested &&
         seg && seg->key_type == KEY_NONE && av_strstart(seg->url, "http", NULL)) {
         ret = open_input(c, v, seg, &v->input_next);
         if (ret < 0) {
@@ -1563,11 +1725,17 @@  reload:
 
         return ret;
     }
-    if (c->http_persistent &&
+    if (c->http_persistent && !c->abr &&
         seg->key_type == KEY_NONE && av_strstart(seg->url, "http", NULL)) {
         v->input_read_done = 1;
     } else {
         ff_format_io_close(v->parent, &v->input);
+        if (c->abr) {
+            if (c->switch_delay > 0)
+                c->switch_delay--;
+            av_log(v->parent, AV_LOG_VERBOSE, "read_data: close pls[%d]->input, v->cur_seq_no=%d, c->switch_delay= %d\n",
+                    v->index, v->cur_seq_no, c->switch_delay);
+        }
     }
     v->cur_seq_no++;
 
@@ -1860,6 +2028,21 @@  static int hls_read_header(AVFormatContext *s)
     c->first_packet = 1;
     c->first_timestamp = AV_NOPTS_VALUE;
     c->cur_timestamp = AV_NOPTS_VALUE;
+    c->switch_request2 = -1;
+    c->switch_delay = 0;
+    c->switch_timestamp = AV_NOPTS_VALUE;
+    c->delta_timestamp = AV_NOPTS_VALUE;
+    c->can_switch = -1;
+
+    if (c->abr) {
+        c->http_persistent = 0;
+        c->http_multiple = 1;
+        c->throughputs = av_mallocz(sizeof(struct throughput));
+        if (!c->throughputs) {
+            ret = AVERROR(ENOMEM);
+            goto fail;
+        }
+    }
 
     if ((ret = save_avio_options(s)) < 0)
         goto fail;
@@ -2049,6 +2232,9 @@  static int hls_read_header(AVFormatContext *s)
         add_metadata_from_renditions(s, pls, AVMEDIA_TYPE_SUBTITLE);
     }
 
+    if (c->abr && c->can_switch == -1)
+        c->can_switch = 1;
+
     update_noheader_flag(s);
 
     return 0;
@@ -2057,6 +2243,13 @@  fail:
     return ret;
 }
 
+static void change_discard_flags(struct playlist *pls, int flag)
+{
+    for (int i = 0; i < pls->n_main_streams; i++) {
+        pls->main_streams[i]->discard = flag;
+    }
+}
+
 static int recheck_discard_flags(AVFormatContext *s, int first)
 {
     HLSContext *c = s->priv_data;
@@ -2067,6 +2260,22 @@  static int recheck_discard_flags(AVFormatContext *s, int first)
     for (i = 0; i < c->n_playlists; i++) {
         struct playlist *pls = c->playlists[i];
 
+        if (c->abr) {
+            if (c->switch_request2 == -1)
+                ;
+            else if (c->switch_request2 == i && c->switch_delay <= 0
+                    && c->cur_timestamp + c->delta_timestamp >= c->switch_timestamp
+                    && c->switch_timestamp != AV_NOPTS_VALUE) {
+                av_log(s, AV_LOG_VERBOSE, "[switch point] cur_timestamp:%ld\n", c->cur_timestamp);
+                change_discard_flags(pls, AVDISCARD_DEFAULT);
+                c->switch_timestamp = AV_NOPTS_VALUE;
+            } else if (c->switch_request2 != -1 && c->switch_request2 != i && c->switch_delay <= 0
+                    && c->cur_timestamp + c->delta_timestamp >= c->switch_timestamp
+                    && c->switch_timestamp != AV_NOPTS_VALUE) {
+                change_discard_flags(pls, AVDISCARD_ALL);
+            }
+        }
+
         cur_needed = playlist_needed(c->playlists[i]);
 
         if (pls->broken) {
@@ -2077,6 +2286,11 @@  static int recheck_discard_flags(AVFormatContext *s, int first)
             changed = 1;
             pls->cur_seq_no = select_cur_seq_no(c, pls);
             pls->pb.eof_reached = 0;
+            if (c->abr) {
+                pls->cur_seq_no = select_cur_seq_no(c, pls) + 1;
+                avio_flush(&pls->pb);
+                avformat_flush(pls->ctx);
+            }
             if (c->cur_timestamp != AV_NOPTS_VALUE) {
                 /* catch up */
                 pls->seek_timestamp = c->cur_timestamp;
@@ -2084,7 +2298,7 @@  static int recheck_discard_flags(AVFormatContext *s, int first)
                 pls->seek_stream_index = -1;
             }
             av_log(s, AV_LOG_INFO, "Now receiving playlist %d, segment %d\n", i, pls->cur_seq_no);
-        } else if (first && !cur_needed && pls->needed) {
+        } else if ((first || c->abr) && !cur_needed && pls->needed) {
             ff_format_io_close(pls->parent, &pls->input);
             pls->input_read_done = 0;
             ff_format_io_close(pls->parent, &pls->input_next);
@@ -2093,6 +2307,9 @@  static int recheck_discard_flags(AVFormatContext *s, int first)
             changed = 1;
             av_log(s, AV_LOG_INFO, "No longer receiving playlist %d\n", i);
         }
+
+        if (c->abr && changed && cur_needed)
+            c->cur_pls = i;
     }
     return changed;
 }
@@ -2167,7 +2384,13 @@  static int hls_read_packet(AVFormatContext *s, AVPacket *pkt)
                         /* audio elementary streams are id3 timestamped */
                         fill_timing_for_id3_timestamped_stream(pls);
                     }
-
+                    if (c->abr && c->first_timestamp != AV_NOPTS_VALUE && c->delta_timestamp == AV_NOPTS_VALUE && !i) {
+                        c->delta_timestamp = av_rescale_q(pls->pkt.dts,
+                            get_timebase(pls), AV_TIME_BASE_Q) - c->first_timestamp + 1;
+                        av_log(s, AV_LOG_VERBOSE, "[abr] delta_timestamp=%ld, %ld, %ld\n",
+                             c->delta_timestamp, c->first_timestamp,av_rescale_q(pls->pkt.dts,
+                            get_timebase(pls), AV_TIME_BASE_Q));
+                    }
                     if (c->first_timestamp == AV_NOPTS_VALUE &&
                         pls->pkt.dts       != AV_NOPTS_VALUE)
                         c->first_timestamp = av_rescale_q(pls->pkt.dts,
@@ -2375,6 +2598,8 @@  static int hls_probe(const AVProbeData *p)
 #define OFFSET(x) offsetof(HLSContext, x)
 #define FLAGS AV_OPT_FLAG_DECODING_PARAM
 static const AVOption hls_options[] = {
+    {"abr", "enable abr to switch streams",
+        OFFSET(abr), AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, FLAGS },
     {"live_start_index", "segment index to start live streams at (negative values are from the end)",
         OFFSET(live_start_index), AV_OPT_TYPE_INT, {.i64 = -3}, INT_MIN, INT_MAX, FLAGS},
     {"allowed_extensions", "List of file extensions that hls is allowed to access",