gstvideodecoder.c 106 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
/* GStreamer
 * Copyright (C) 2008 David Schleef <ds@schleef.org>
 * Copyright (C) 2011 Mark Nauwelaerts <mark.nauwelaerts@collabora.co.uk>.
 * Copyright (C) 2011 Nokia Corporation. All rights reserved.
 *   Contact: Stefan Kost <stefan.kost@nokia.com>
 * Copyright (C) 2012 Collabora Ltd.
 *	Author : Edward Hervey <edward@collabora.com>
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Library General Public
 * License as published by the Free Software Foundation; either
 * version 2 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Library General Public License for more details.
 *
 * You should have received a copy of the GNU Library General Public
 * License along with this library; if not, write to the
Tim-Philipp Müller's avatar
Tim-Philipp Müller committed
21 22
 * Free Software Foundation, Inc., 51 Franklin St, Fifth Floor,
 * Boston, MA 02110-1301, USA.
23 24 25 26 27 28 29 30 31 32
 */

/**
 * SECTION:gstvideodecoder
 * @short_description: Base class for video decoders
 * @see_also: 
 *
 * This base class is for video decoders turning encoded data into raw video
 * frames.
 *
33
 * The GstVideoDecoder base class and derived subclasses should cooperate as follows:
34 35 36 37 38
 * <orderedlist>
 * <listitem>
 *   <itemizedlist><title>Configuration</title>
 *   <listitem><para>
 *     Initially, GstVideoDecoder calls @start when the decoder element
39
 *     is activated, which allows the subclass to perform any global setup.
40 41
 *   </para></listitem>
 *   <listitem><para>
42
 *     GstVideoDecoder calls @set_format to inform the subclass of caps
43 44 45 46 47 48
 *     describing input video data that it is about to receive, including
 *     possibly configuration data.
 *     While unlikely, it might be called more than once, if changing input
 *     parameters require reconfiguration.
 *   </para></listitem>
 *   <listitem><para>
49 50 51
 *     Incoming data buffers are processed as needed, described in Data Processing below.
 *   </para></listitem>
 *   <listitem><para>
52 53 54 55 56 57 58 59
 *     GstVideoDecoder calls @stop at end of all processing.
 *   </para></listitem>
 *   </itemizedlist>
 * </listitem>
 * <listitem>
 *   <itemizedlist>
 *   <title>Data processing</title>
 *     <listitem><para>
60
 *       The base class gathers input data, and optionally allows subclass
61 62 63 64
 *       to parse this into subsequently manageable chunks, typically
 *       corresponding to and referred to as 'frames'.
 *     </para></listitem>
 *     <listitem><para>
65 66
 *       Each input frame is provided in turn to the subclass' @handle_frame callback.
 *       The ownership of the frame is given to the @handle_frame callback.
67 68
 *     </para></listitem>
 *     <listitem><para>
69
 *       If codec processing results in decoded data, the subclass should call
70
 *       @gst_video_decoder_finish_frame to have decoded data pushed.
71 72 73
 *       downstream. Otherwise, the subclass must call @gst_video_decoder_drop_frame, to
 *       allow the base class to do timestamp and offset tracking, and possibly to
 *       requeue the frame for a later attempt in the case of reverse playback.
74 75 76 77 78 79
 *     </para></listitem>
 *   </itemizedlist>
 * </listitem>
 * <listitem>
 *   <itemizedlist><title>Shutdown phase</title>
 *   <listitem><para>
80
 *     The GstVideoDecoder class calls @stop to inform the subclass that data
81 82 83 84
 *     parsing will be stopped.
 *   </para></listitem>
 *   </itemizedlist>
 * </listitem>
85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108
 * <listitem>
 *   <itemizedlist><title>Additional Notes</title>
 *   <listitem>
 *     <itemizedlist><title>Seeking/Flushing</title>
 *     <listitem><para>
 *   When the pipeline is seeked or otherwise flushed, the subclass is informed via a call
 *   to its @reset callback, with the hard parameter set to true. This indicates the
 *   subclass should drop any internal data queues and timestamps and prepare for a fresh
 *   set of buffers to arrive for parsing and decoding.
 *     </para></listitem>
 *     </itemizedlist>
 *   </listitem>
 *   <listitem>
 *     <itemizedlist><title>End Of Stream</title>
 *     <listitem><para>
 *   At end-of-stream, the subclass @parse function may be called some final times with the 
 *   at_eos parameter set to true, indicating that the element should not expect any more data
 *   to be arriving, and it should parse and remaining frames and call
 *   gst_video_decoder_have_frame() if possible.
 *     </para></listitem>
 *     </itemizedlist>
 *   </listitem>
 *   </itemizedlist>
 * </listitem>
109 110
 * </orderedlist>
 *
111
 * The subclass is responsible for providing pad template caps for
112
 * source and sink pads. The pads need to be named "sink" and "src". It also
113 114 115 116 117
 * needs to provide information about the ouptput caps, when they are known.
 * This may be when the base class calls the subclass' @set_format function,
 * though it might be during decoding, before calling
 * @gst_video_decoder_finish_frame. This is done via
 * @gst_video_decoder_set_output_state
118
 *
119
 * The subclass is also responsible for providing (presentation) timestamps
120
 * (likely based on corresponding input ones).  If that is not applicable
121
 * or possible, the base class provides limited framerate based interpolation.
122
 *
123 124
 * Similarly, the base class provides some limited (legacy) seeking support
 * if specifically requested by the subclass, as full-fledged support
125 126
 * should rather be left to upstream demuxer, parser or alike.  This simple
 * approach caters for seeking and duration reporting using estimated input
127 128
 * bitrates. To enable it, a subclass should call
 * @gst_video_decoder_set_estimate_rate to enable handling of incoming byte-streams.
129
 *
130
 * The base class provides some support for reverse playback, in particular
131
 * in case incoming data is not packetized or upstream does not provide
132 133 134 135 136
 * fragments on keyframe boundaries.  However, the subclass should then be prepared
 * for the parsing and frame processing stage to occur separately (in normal
 * forward processing, the latter immediately follows the former),
 * The subclass also needs to ensure the parsing stage properly marks keyframes,
 * unless it knows the upstream elements will do so properly for incoming data.
137
 *
138
 * The bare minimum that a functional subclass needs to implement is:
139 140 141
 * <itemizedlist>
 *   <listitem><para>Provide pad templates</para></listitem>
 *   <listitem><para>
142
 *      Inform the base class of output caps via @gst_video_decoder_set_output_state
143 144
 *   </para></listitem>
 *   <listitem><para>
145
 *      Parse input data, if it is not considered packetized from upstream
146
 *      Data will be provided to @parse which should invoke @gst_video_decoder_add_to_frame and
147
 *      @gst_video_decoder_have_frame to separate the data belonging to each video frame.
148 149 150
 *   </para></listitem>
 *   <listitem><para>
 *      Accept data in @handle_frame and provide decoded results to
151
 *      @gst_video_decoder_finish_frame, or call @gst_video_decoder_drop_frame.
152 153 154 155 156 157 158 159 160 161 162 163 164 165 166
 *   </para></listitem>
 * </itemizedlist>
 */

#ifdef HAVE_CONFIG_H
#include "config.h"
#endif

/* TODO
 *
 * * Add a flag/boolean for I-frame-only/image decoders so we can do extra
 *   features, like applying QoS on input (as opposed to after the frame is
 *   decoded).
 * * Add a flag/boolean for decoders that require keyframes, so the base
 *   class can automatically discard non-keyframes before one has arrived
167
 * * Detect reordered frame/timestamps and fix the pts/dts
168 169 170 171 172 173 174
 * * Support for GstIndex (or shall we not care ?)
 * * Calculate actual latency based on input/output timestamp/frame_number
 *   and if it exceeds the recorded one, save it and emit a GST_MESSAGE_LATENCY
 * * Emit latency message when it changes
 *
 */

175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295
/* Implementation notes:
 * The Video Decoder base class operates in 2 primary processing modes, depending
 * on whether forward or reverse playback is requested.
 *
 * Forward playback:
 *   * Incoming buffer -> @parse() -> add_to_frame()/have_frame() -> handle_frame() -> 
 *     push downstream
 *
 * Reverse playback is more complicated, since it involves gathering incoming data regions
 * as we loop backwards through the upstream data. The processing concept (using incoming
 * buffers as containing one frame each to simplify things) is:
 *
 * Upstream data we want to play:
 *  Buffer encoded order:  1  2  3  4  5  6  7  8  9  EOS
 *  Keyframe flag:            K        K        
 *  Groupings:             AAAAAAA  BBBBBBB  CCCCCCC
 *
 * Input:
 *  Buffer reception order:  7  8  9  4  5  6  1  2  3  EOS
 *  Keyframe flag:                       K        K
 *  Discont flag:            D        D        D
 *
 * - Each Discont marks a discont in the decoding order.
 * - The keyframes mark where we can start decoding.
 *
 * Initially, we prepend incoming buffers to the gather queue. Whenever the
 * discont flag is set on an incoming buffer, the gather queue is flushed out
 * before the new buffer is collected.
 *
 * The above data will be accumulated in the gather queue like this:
 *
 *   gather queue:  9  8  7
 *                        D
 *
 * Whe buffer 4 is received (with a DISCONT), we flush the gather queue like
 * this:
 *
 *   while (gather)
 *     take head of queue and prepend to parse queue (this reverses the sequence,
 *     so parse queue is 7 -> 8 -> 9)
 *
 *   Next, we process the parse queue, which now contains all un-parsed packets (including
 *   any leftover ones from the previous decode section)
 *
 *   for each buffer now in the parse queue:
 *     Call the subclass parse function, prepending each resulting frame to
 *     the parse_gather queue. Buffers which precede the first one that
 *     produces a parsed frame are retained in the parse queue for re-processing on
 *     the next cycle of parsing.
 *
 *   The parse_gather queue now contains frame objects ready for decoding, in reverse order.
 *   parse_gather: 9 -> 8 -> 7
 *
 *   while (parse_gather)
 *     Take the head of the queue and prepend it to the decode queue
 *     If the frame was a keyframe, process the decode queue
 *   decode is now 7-8-9
 *
 *  Processing the decode queue results in frames with attached output buffers
 *  stored in the 'output_queue' ready for outputting in reverse order.
 *
 * After we flushed the gather queue and parsed it, we add 4 to the (now empty) gather queue.
 * We get the following situation:
 *
 *  gather queue:    4
 *  decode queue:    7  8  9
 *
 * After we received 5 (Keyframe) and 6:
 *
 *  gather queue:    6  5  4
 *  decode queue:    7  8  9
 *
 * When we receive 1 (DISCONT) which triggers a flush of the gather queue:
 *
 *   Copy head of the gather queue (6) to decode queue:
 *
 *    gather queue:    5  4
 *    decode queue:    6  7  8  9
 *
 *   Copy head of the gather queue (5) to decode queue. This is a keyframe so we
 *   can start decoding.
 *
 *    gather queue:    4
 *    decode queue:    5  6  7  8  9
 *
 *   Decode frames in decode queue, store raw decoded data in output queue, we
 *   can take the head of the decode queue and prepend the decoded result in the
 *   output queue:
 *
 *    gather queue:    4
 *    decode queue:    
 *    output queue:    9  8  7  6  5
 *
 *   Now output all the frames in the output queue, picking a frame from the
 *   head of the queue.
 *
 *   Copy head of the gather queue (4) to decode queue, we flushed the gather
 *   queue and can now store input buffer in the gather queue:
 *
 *    gather queue:    1
 *    decode queue:    4
 *
 *  When we receive EOS, the queue looks like:
 *
 *    gather queue:    3  2  1
 *    decode queue:    4
 *
 *  Fill decode queue, first keyframe we copy is 2:
 *
 *    gather queue:    1
 *    decode queue:    2  3  4
 *
 *  Decoded output:
 *
 *    gather queue:    1
 *    decode queue:    
 *    output queue:    4  3  2
 *
 *  Leftover buffer 1 cannot be decoded and must be discarded.
 */

296 297 298
#include "gstvideodecoder.h"
#include "gstvideoutils.h"

299
#include <gst/video/video.h>
300
#include <gst/video/video-event.h>
301 302
#include <gst/video/gstvideopool.h>
#include <gst/video/gstvideometa.h>
303 304 305 306 307 308 309 310 311 312 313 314 315
#include <string.h>

GST_DEBUG_CATEGORY (videodecoder_debug);
#define GST_CAT_DEFAULT videodecoder_debug

#define GST_VIDEO_DECODER_GET_PRIVATE(obj)  \
    (G_TYPE_INSTANCE_GET_PRIVATE ((obj), GST_TYPE_VIDEO_DECODER, \
        GstVideoDecoderPrivate))

struct _GstVideoDecoderPrivate
{
  /* FIXME introduce a context ? */

316
  GstBufferPool *pool;
317 318
  GstAllocator *allocator;
  GstAllocationParams params;
319

320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336
  /* parse tracking */
  /* input data */
  GstAdapter *input_adapter;
  /* assembles current frame */
  GstAdapter *output_adapter;

  /* Whether we attempt to convert newsegment from bytes to
   * time using a bitrate estimation */
  gboolean do_estimate_rate;

  /* Whether input is considered packetized or not */
  gboolean packetized;

  /* Error handling */
  gint max_errors;
  gint error_count;

337 338
  gboolean do_caps;

339 340 341 342 343
  /* ... being tracked here;
   * only available during parsing */
  GstVideoCodecFrame *current_frame;
  /* events that should apply to the current frame */
  GList *current_frame_events;
344 345
  /* events that should be pushed before the next frame */
  GList *pending_events;
346 347 348 349 350 351 352 353

  /* relative offset of input data */
  guint64 input_offset;
  /* relative offset of frame */
  guint64 frame_offset;
  /* tracking ts and offsets */
  GList *timestamps;

354
  /* last outgoing ts */
355
  GstClockTime last_timestamp_out;
356 357
  /* incoming pts - dts */
  GstClockTime pts_delta;
358
  gboolean reordered_output;
359 360 361 362 363 364 365 366 367 368

  /* reverse playback */
  /* collect input */
  GList *gather;
  /* to-be-parsed */
  GList *parse;
  /* collected parsed frames */
  GList *parse_gather;
  /* frames to be handled == decoded */
  GList *decode;
369 370
  /* collected output - of buffer objects, not frames */
  GList *output_queued;
371

372 373

  /* base_picture_number is the picture number of the reference picture */
374
  guint64 base_picture_number;
375 376 377
  /* combine with base_picture_number, framerate and calcs to yield (presentation) ts */
  GstClockTime base_timestamp;

378 379 380 381
  /* FIXME : reorder_depth is never set */
  int reorder_depth;
  int distance_from_sync;

382 383
  guint32 system_frame_number;
  guint32 decode_frame_number;
384 385 386

  GList *frames;                /* Protected with OBJECT_LOCK */
  GstVideoCodecState *input_state;
387
  GstVideoCodecState *output_state;     /* OBJECT_LOCK and STREAM_LOCK */
388 389 390
  gboolean output_state_changed;

  /* QoS properties */
391 392 393
  gdouble proportion;           /* OBJECT_LOCK */
  GstClockTime earliest_time;   /* OBJECT_LOCK */
  GstClockTime qos_frame_duration;      /* OBJECT_LOCK */
394 395 396 397 398 399 400 401 402 403 404
  gboolean discont;
  /* qos messages: frames dropped/processed */
  guint dropped;
  guint processed;

  /* Outgoing byte size ? */
  gint64 bytes_out;
  gint64 time;

  gint64 min_latency;
  gint64 max_latency;
405 406 407

  GstTagList *tags;
  gboolean tags_changed;
408 409
};

410 411 412 413 414
static GstElementClass *parent_class = NULL;
static void gst_video_decoder_class_init (GstVideoDecoderClass * klass);
static void gst_video_decoder_init (GstVideoDecoder * dec,
    GstVideoDecoderClass * klass);

415 416
static void gst_video_decoder_finalize (GObject * object);

417 418 419 420 421 422 423 424 425 426 427 428 429 430
static gboolean gst_video_decoder_setcaps (GstVideoDecoder * dec,
    GstCaps * caps);
static gboolean gst_video_decoder_sink_event (GstPad * pad, GstObject * parent,
    GstEvent * event);
static gboolean gst_video_decoder_src_event (GstPad * pad, GstObject * parent,
    GstEvent * event);
static GstFlowReturn gst_video_decoder_chain (GstPad * pad, GstObject * parent,
    GstBuffer * buf);
static gboolean gst_video_decoder_sink_query (GstPad * pad, GstObject * parent,
    GstQuery * query);
static GstStateChangeReturn gst_video_decoder_change_state (GstElement *
    element, GstStateChange transition);
static gboolean gst_video_decoder_src_query (GstPad * pad, GstObject * parent,
    GstQuery * query);
431 432
static void gst_video_decoder_reset (GstVideoDecoder * decoder, gboolean full);

433 434
static GstFlowReturn gst_video_decoder_decode_frame (GstVideoDecoder * decoder,
    GstVideoCodecFrame * frame);
435

436 437
static void gst_video_decoder_release_frame (GstVideoDecoder * dec,
    GstVideoCodecFrame * frame);
438 439
static GstClockTime gst_video_decoder_get_frame_duration (GstVideoDecoder *
    decoder, GstVideoCodecFrame * frame);
440 441
static GstVideoCodecFrame *gst_video_decoder_new_frame (GstVideoDecoder *
    decoder);
442 443
static GstFlowReturn gst_video_decoder_clip_and_push_buf (GstVideoDecoder *
    decoder, GstBuffer * buf);
444 445
static GstFlowReturn gst_video_decoder_flush_parse (GstVideoDecoder * dec,
    gboolean at_eos);
446

447 448
static void gst_video_decoder_clear_queues (GstVideoDecoder * dec);

449 450 451 452
static gboolean gst_video_decoder_sink_event_default (GstVideoDecoder * decoder,
    GstEvent * event);
static gboolean gst_video_decoder_src_event_default (GstVideoDecoder * decoder,
    GstEvent * event);
453 454
static gboolean gst_video_decoder_decide_allocation_default (GstVideoDecoder *
    decoder, GstQuery * query);
455 456
static gboolean gst_video_decoder_propose_allocation_default (GstVideoDecoder *
    decoder, GstQuery * query);
457
static gboolean gst_video_decoder_negotiate_default (GstVideoDecoder * decoder);
458
static GstFlowReturn gst_video_decoder_parse_available (GstVideoDecoder * dec,
459
    gboolean at_eos, gboolean new_buffer);
460

461 462 463 464
/* we can't use G_DEFINE_ABSTRACT_TYPE because we need the klass in the _init
 * method to get to the padtemplates */
GType
gst_video_decoder_get_type (void)
465
{
466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486
  static volatile gsize type = 0;

  if (g_once_init_enter (&type)) {
    GType _type;
    static const GTypeInfo info = {
      sizeof (GstVideoDecoderClass),
      NULL,
      NULL,
      (GClassInitFunc) gst_video_decoder_class_init,
      NULL,
      NULL,
      sizeof (GstVideoDecoder),
      0,
      (GInstanceInitFunc) gst_video_decoder_init,
    };

    _type = g_type_register_static (GST_TYPE_ELEMENT,
        "GstVideoDecoder", &info, G_TYPE_FLAG_ABSTRACT);
    g_once_init_leave (&type, _type);
  }
  return type;
487 488 489 490 491 492 493 494 495 496 497
}

static void
gst_video_decoder_class_init (GstVideoDecoderClass * klass)
{
  GObjectClass *gobject_class;
  GstElementClass *gstelement_class;

  gobject_class = G_OBJECT_CLASS (klass);
  gstelement_class = GST_ELEMENT_CLASS (klass);

498 499 500 501
  GST_DEBUG_CATEGORY_INIT (videodecoder_debug, "videodecoder", 0,
      "Base Video Decoder");

  parent_class = g_type_class_peek_parent (klass);
502 503 504 505 506 507
  g_type_class_add_private (klass, sizeof (GstVideoDecoderPrivate));

  gobject_class->finalize = gst_video_decoder_finalize;

  gstelement_class->change_state =
      GST_DEBUG_FUNCPTR (gst_video_decoder_change_state);
508 509 510

  klass->sink_event = gst_video_decoder_sink_event_default;
  klass->src_event = gst_video_decoder_src_event_default;
511
  klass->decide_allocation = gst_video_decoder_decide_allocation_default;
512
  klass->propose_allocation = gst_video_decoder_propose_allocation_default;
513
  klass->negotiate = gst_video_decoder_negotiate_default;
514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554
}

static void
gst_video_decoder_init (GstVideoDecoder * decoder, GstVideoDecoderClass * klass)
{
  GstPadTemplate *pad_template;
  GstPad *pad;

  GST_DEBUG_OBJECT (decoder, "gst_video_decoder_init");

  decoder->priv = GST_VIDEO_DECODER_GET_PRIVATE (decoder);

  pad_template =
      gst_element_class_get_pad_template (GST_ELEMENT_CLASS (klass), "sink");
  g_return_if_fail (pad_template != NULL);

  decoder->sinkpad = pad = gst_pad_new_from_template (pad_template, "sink");

  gst_pad_set_chain_function (pad, GST_DEBUG_FUNCPTR (gst_video_decoder_chain));
  gst_pad_set_event_function (pad,
      GST_DEBUG_FUNCPTR (gst_video_decoder_sink_event));
  gst_pad_set_query_function (pad,
      GST_DEBUG_FUNCPTR (gst_video_decoder_sink_query));
  gst_element_add_pad (GST_ELEMENT (decoder), decoder->sinkpad);

  pad_template =
      gst_element_class_get_pad_template (GST_ELEMENT_CLASS (klass), "src");
  g_return_if_fail (pad_template != NULL);

  decoder->srcpad = pad = gst_pad_new_from_template (pad_template, "src");

  gst_pad_set_event_function (pad,
      GST_DEBUG_FUNCPTR (gst_video_decoder_src_event));
  gst_pad_set_query_function (pad,
      GST_DEBUG_FUNCPTR (gst_video_decoder_src_query));
  gst_pad_use_fixed_caps (pad);
  gst_element_add_pad (GST_ELEMENT (decoder), decoder->srcpad);

  gst_segment_init (&decoder->input_segment, GST_FORMAT_TIME);
  gst_segment_init (&decoder->output_segment, GST_FORMAT_TIME);

555
  g_rec_mutex_init (&decoder->stream_lock);
556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692

  decoder->priv->input_adapter = gst_adapter_new ();
  decoder->priv->output_adapter = gst_adapter_new ();
  decoder->priv->packetized = TRUE;

  gst_video_decoder_reset (decoder, TRUE);
}

static gboolean
gst_video_rawvideo_convert (GstVideoCodecState * state,
    GstFormat src_format, gint64 src_value,
    GstFormat * dest_format, gint64 * dest_value)
{
  gboolean res = FALSE;
  guint vidsize;
  guint fps_n, fps_d;

  g_return_val_if_fail (dest_format != NULL, FALSE);
  g_return_val_if_fail (dest_value != NULL, FALSE);

  if (src_format == *dest_format || src_value == 0 || src_value == -1) {
    *dest_value = src_value;
    return TRUE;
  }

  vidsize = GST_VIDEO_INFO_SIZE (&state->info);
  fps_n = GST_VIDEO_INFO_FPS_N (&state->info);
  fps_d = GST_VIDEO_INFO_FPS_D (&state->info);

  if (src_format == GST_FORMAT_BYTES &&
      *dest_format == GST_FORMAT_DEFAULT && vidsize) {
    /* convert bytes to frames */
    *dest_value = gst_util_uint64_scale_int (src_value, 1, vidsize);
    res = TRUE;
  } else if (src_format == GST_FORMAT_DEFAULT &&
      *dest_format == GST_FORMAT_BYTES && vidsize) {
    /* convert bytes to frames */
    *dest_value = src_value * vidsize;
    res = TRUE;
  } else if (src_format == GST_FORMAT_DEFAULT &&
      *dest_format == GST_FORMAT_TIME && fps_n) {
    /* convert frames to time */
    *dest_value = gst_util_uint64_scale (src_value, GST_SECOND * fps_d, fps_n);
    res = TRUE;
  } else if (src_format == GST_FORMAT_TIME &&
      *dest_format == GST_FORMAT_DEFAULT && fps_d) {
    /* convert time to frames */
    *dest_value = gst_util_uint64_scale (src_value, fps_n, GST_SECOND * fps_d);
    res = TRUE;
  } else if (src_format == GST_FORMAT_TIME &&
      *dest_format == GST_FORMAT_BYTES && fps_d && vidsize) {
    /* convert time to frames */
    *dest_value = gst_util_uint64_scale (src_value,
        fps_n * vidsize, GST_SECOND * fps_d);
    res = TRUE;
  } else if (src_format == GST_FORMAT_BYTES &&
      *dest_format == GST_FORMAT_TIME && fps_n && vidsize) {
    /* convert frames to time */
    *dest_value = gst_util_uint64_scale (src_value,
        GST_SECOND * fps_d, fps_n * vidsize);
    res = TRUE;
  }

  return res;
}

static gboolean
gst_video_encoded_video_convert (gint64 bytes, gint64 time,
    GstFormat src_format, gint64 src_value, GstFormat * dest_format,
    gint64 * dest_value)
{
  gboolean res = FALSE;

  g_return_val_if_fail (dest_format != NULL, FALSE);
  g_return_val_if_fail (dest_value != NULL, FALSE);

  if (G_UNLIKELY (src_format == *dest_format || src_value == 0 ||
          src_value == -1)) {
    if (dest_value)
      *dest_value = src_value;
    return TRUE;
  }

  if (bytes <= 0 || time <= 0) {
    GST_DEBUG ("not enough metadata yet to convert");
    goto exit;
  }

  switch (src_format) {
    case GST_FORMAT_BYTES:
      switch (*dest_format) {
        case GST_FORMAT_TIME:
          *dest_value = gst_util_uint64_scale (src_value, time, bytes);
          res = TRUE;
          break;
        default:
          res = FALSE;
      }
      break;
    case GST_FORMAT_TIME:
      switch (*dest_format) {
        case GST_FORMAT_BYTES:
          *dest_value = gst_util_uint64_scale (src_value, bytes, time);
          res = TRUE;
          break;
        default:
          res = FALSE;
      }
      break;
    default:
      GST_DEBUG ("unhandled conversion from %d to %d", src_format,
          *dest_format);
      res = FALSE;
  }

exit:
  return res;
}

static GstVideoCodecState *
_new_input_state (GstCaps * caps)
{
  GstVideoCodecState *state;
  GstStructure *structure;
  const GValue *codec_data;

  state = g_slice_new0 (GstVideoCodecState);
  state->ref_count = 1;
  gst_video_info_init (&state->info);
  if (G_UNLIKELY (!gst_video_info_from_caps (&state->info, caps)))
    goto parse_fail;
  state->caps = gst_caps_ref (caps);

  structure = gst_caps_get_structure (caps, 0);

  codec_data = gst_structure_get_value (structure, "codec_data");
  if (codec_data && G_VALUE_TYPE (codec_data) == GST_TYPE_BUFFER)
693
    state->codec_data = GST_BUFFER (g_value_dup_boxed (codec_data));
694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724

  return state;

parse_fail:
  {
    g_slice_free (GstVideoCodecState, state);
    return NULL;
  }
}

static GstVideoCodecState *
_new_output_state (GstVideoFormat fmt, guint width, guint height,
    GstVideoCodecState * reference)
{
  GstVideoCodecState *state;

  state = g_slice_new0 (GstVideoCodecState);
  state->ref_count = 1;
  gst_video_info_init (&state->info);
  gst_video_info_set_format (&state->info, fmt, width, height);

  if (reference) {
    GstVideoInfo *tgt, *ref;

    tgt = &state->info;
    ref = &reference->info;

    /* Copy over extra fields from reference state */
    tgt->interlace_mode = ref->interlace_mode;
    tgt->flags = ref->flags;
    tgt->chroma_site = ref->chroma_site;
725 726 727 728 729 730 731 732 733 734
    /* only copy values that are not unknown so that we don't override the
     * defaults. subclasses should really fill these in when they know. */
    if (ref->colorimetry.range)
      tgt->colorimetry.range = ref->colorimetry.range;
    if (ref->colorimetry.matrix)
      tgt->colorimetry.matrix = ref->colorimetry.matrix;
    if (ref->colorimetry.transfer)
      tgt->colorimetry.transfer = ref->colorimetry.transfer;
    if (ref->colorimetry.primaries)
      tgt->colorimetry.primaries = ref->colorimetry.primaries;
735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750
    GST_DEBUG ("reference par %d/%d fps %d/%d",
        ref->par_n, ref->par_d, ref->fps_n, ref->fps_d);
    tgt->par_n = ref->par_n;
    tgt->par_d = ref->par_d;
    tgt->fps_n = ref->fps_n;
    tgt->fps_d = ref->fps_d;
  }

  GST_DEBUG ("reference par %d/%d fps %d/%d",
      state->info.par_n, state->info.par_d,
      state->info.fps_n, state->info.fps_d);

  return state;
}

static gboolean
751
gst_video_decoder_setcaps (GstVideoDecoder * decoder, GstCaps * caps)
752 753 754 755 756 757 758 759 760
{
  GstVideoDecoderClass *decoder_class;
  GstVideoCodecState *state;
  gboolean ret = TRUE;

  decoder_class = GST_VIDEO_DECODER_GET_CLASS (decoder);

  GST_DEBUG_OBJECT (decoder, "setcaps %" GST_PTR_FORMAT, caps);

761 762 763 764 765 766 767 768 769 770
  GST_VIDEO_DECODER_STREAM_LOCK (decoder);

  if (decoder->priv->input_state) {
    GST_DEBUG_OBJECT (decoder,
        "Checking if caps changed old %" GST_PTR_FORMAT " new %" GST_PTR_FORMAT,
        decoder->priv->input_state->caps, caps);
    if (gst_caps_is_equal (decoder->priv->input_state->caps, caps))
      goto caps_not_changed;
  }

771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789
  state = _new_input_state (caps);

  if (G_UNLIKELY (state == NULL))
    goto parse_fail;

  if (decoder_class->set_format)
    ret = decoder_class->set_format (decoder, state);

  if (!ret)
    goto refused_format;

  if (decoder->priv->input_state)
    gst_video_codec_state_unref (decoder->priv->input_state);
  decoder->priv->input_state = state;

  GST_VIDEO_DECODER_STREAM_UNLOCK (decoder);

  return ret;

790 791 792 793 794 795
caps_not_changed:
  {
    GST_DEBUG_OBJECT (decoder, "Caps did not change - ignore");
    GST_VIDEO_DECODER_STREAM_UNLOCK (decoder);
    return TRUE;
  }
796

797
  /* ERRORS */
798 799 800
parse_fail:
  {
    GST_WARNING_OBJECT (decoder, "Failed to parse caps");
801
    GST_VIDEO_DECODER_STREAM_UNLOCK (decoder);
802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822
    return FALSE;
  }

refused_format:
  {
    GST_VIDEO_DECODER_STREAM_UNLOCK (decoder);
    GST_WARNING_OBJECT (decoder, "Subclass refused caps");
    gst_video_codec_state_unref (state);
    return FALSE;
  }
}

static void
gst_video_decoder_finalize (GObject * object)
{
  GstVideoDecoder *decoder;

  decoder = GST_VIDEO_DECODER (object);

  GST_DEBUG_OBJECT (object, "finalize");

823
  g_rec_mutex_clear (&decoder->stream_lock);
824 825 826 827 828 829 830 831 832 833 834 835 836 837 838

  if (decoder->priv->input_adapter) {
    g_object_unref (decoder->priv->input_adapter);
    decoder->priv->input_adapter = NULL;
  }
  if (decoder->priv->output_adapter) {
    g_object_unref (decoder->priv->output_adapter);
    decoder->priv->output_adapter = NULL;
  }

  if (decoder->priv->input_state)
    gst_video_codec_state_unref (decoder->priv->input_state);
  if (decoder->priv->output_state)
    gst_video_codec_state_unref (decoder->priv->output_state);

839
  if (decoder->priv->pool) {
840 841 842 843 844
    gst_object_unref (decoder->priv->pool);
    decoder->priv->pool = NULL;
  }

  if (decoder->priv->allocator) {
Wim Taymans's avatar
Wim Taymans committed
845
    gst_object_unref (decoder->priv->allocator);
846
    decoder->priv->allocator = NULL;
847 848
  }

849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870
  G_OBJECT_CLASS (parent_class)->finalize (object);
}

/* hard == FLUSH, otherwise discont */
static GstFlowReturn
gst_video_decoder_flush (GstVideoDecoder * dec, gboolean hard)
{
  GstFlowReturn ret = GST_FLOW_OK;

  GST_LOG_OBJECT (dec, "flush hard %d", hard);

  /* FIXME make some more distinction between hard and soft,
   * but subclass may not be prepared for that */
  /* FIXME perhaps also clear pending frames ?,
   * but again, subclass may still come up with one of those */
  if (!hard) {
    /* TODO ? finish/drain some stuff */
  } else {
    gst_segment_init (&dec->input_segment, GST_FORMAT_UNDEFINED);
    gst_segment_init (&dec->output_segment, GST_FORMAT_UNDEFINED);
    gst_video_decoder_clear_queues (dec);
  }
871

872 873 874 875 876 877
  /* and get (re)set for the sequel */
  gst_video_decoder_reset (dec, FALSE);

  return ret;
}

878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908
static gboolean
gst_video_decoder_push_event (GstVideoDecoder * decoder, GstEvent * event)
{
  switch (GST_EVENT_TYPE (event)) {
    case GST_EVENT_SEGMENT:
    {
      GstSegment segment;

      GST_VIDEO_DECODER_STREAM_LOCK (decoder);

      gst_event_copy_segment (event, &segment);

      GST_DEBUG_OBJECT (decoder, "segment %" GST_SEGMENT_FORMAT, &segment);

      if (segment.format != GST_FORMAT_TIME) {
        GST_DEBUG_OBJECT (decoder, "received non TIME newsegment");
        GST_VIDEO_DECODER_STREAM_UNLOCK (decoder);
        break;
      }

      decoder->output_segment = segment;
      GST_VIDEO_DECODER_STREAM_UNLOCK (decoder);
      break;
    }
    default:
      break;
  }

  return gst_pad_push_event (decoder->srcpad, event);
}

909
static GstFlowReturn
910 911
gst_video_decoder_parse_available (GstVideoDecoder * dec, gboolean at_eos,
    gboolean new_buffer)
912 913 914 915 916 917 918 919 920
{
  GstVideoDecoderClass *decoder_class = GST_VIDEO_DECODER_GET_CLASS (dec);
  GstVideoDecoderPrivate *priv = dec->priv;
  GstFlowReturn ret = GST_FLOW_OK;
  gsize start_size, available;

  available = gst_adapter_available (priv->input_adapter);
  start_size = 0;

921 922 923
  while (ret == GST_FLOW_OK && ((available && start_size != available)
          || new_buffer)) {
    new_buffer = FALSE;
924 925 926 927 928 929 930 931 932 933 934 935 936 937
    /* current frame may have been parsed and handled,
     * so we need to set up a new one when asking subclass to parse */
    if (priv->current_frame == NULL)
      priv->current_frame = gst_video_decoder_new_frame (dec);

    start_size = available;
    ret = decoder_class->parse (dec, priv->current_frame,
        priv->input_adapter, at_eos);
    available = gst_adapter_available (priv->input_adapter);
  }

  return ret;
}

938
static GstFlowReturn
939
gst_video_decoder_drain_out (GstVideoDecoder * dec, gboolean at_eos)
940 941 942 943 944 945 946 947 948 949 950
{
  GstVideoDecoderClass *decoder_class = GST_VIDEO_DECODER_GET_CLASS (dec);
  GstVideoDecoderPrivate *priv = dec->priv;
  GstFlowReturn ret = GST_FLOW_OK;

  GST_VIDEO_DECODER_STREAM_LOCK (dec);

  if (dec->input_segment.rate > 0.0) {
    /* Forward mode, if unpacketized, give the child class
     * a final chance to flush out packets */
    if (!priv->packetized) {
951
      ret = gst_video_decoder_parse_available (dec, TRUE, FALSE);
952 953 954 955 956 957
    }
  } else {
    /* Reverse playback mode */
    ret = gst_video_decoder_flush_parse (dec, TRUE);
  }

958 959 960 961
  if (at_eos) {
    if (decoder_class->finish)
      ret = decoder_class->finish (dec);
  }
962 963 964 965 966 967

  GST_VIDEO_DECODER_STREAM_UNLOCK (dec);

  return ret;
}

968
static gboolean
969 970
gst_video_decoder_sink_event_default (GstVideoDecoder * decoder,
    GstEvent * event)
971 972
{
  GstVideoDecoderPrivate *priv;
973
  gboolean ret = FALSE;
974
  gboolean forward_immediate = FALSE;
975 976 977 978

  priv = decoder->priv;

  switch (GST_EVENT_TYPE (event)) {
979 980 981 982 983 984 985 986 987 988 989 990 991 992
    case GST_EVENT_STREAM_START:
    {
      GstFlowReturn flow_ret = GST_FLOW_OK;

      flow_ret = gst_video_decoder_drain_out (decoder, FALSE);
      ret = (flow_ret == GST_FLOW_OK);

      /* Forward STREAM_START immediately. Everything is drained after
       * the STREAM_START event and we can forward this event immediately
       * now without having buffers out of order.
       */
      forward_immediate = TRUE;
      break;
    }
993 994 995 996 997
    case GST_EVENT_CAPS:
    {
      GstCaps *caps;

      gst_event_parse_caps (event, &caps);
998 999
      ret = TRUE;
      decoder->priv->do_caps = TRUE;
1000
      gst_event_unref (event);
1001
      event = NULL;
1002 1003
      break;
    }
1004 1005 1006 1007
    case GST_EVENT_EOS:
    {
      GstFlowReturn flow_ret = GST_FLOW_OK;

1008
      flow_ret = gst_video_decoder_drain_out (decoder, TRUE);
1009
      ret = (flow_ret == GST_FLOW_OK);
1010 1011 1012 1013 1014 1015 1016 1017
      /* Forward EOS immediately. This is required because no
       * buffer or serialized event will come after EOS and
       * nothing could trigger another _finish_frame() call.
       *
       * The subclass can override this behaviour by overriding
       * the ::sink_event() vfunc and not chaining up to the
       * parent class' ::sink_event() until a later time.
       */
1018
      forward_immediate = TRUE;
1019 1020
      break;
    }
Jan Schmidt's avatar
Jan Schmidt committed
1021 1022 1023 1024 1025 1026
    case GST_EVENT_GAP:
    {
      GstFlowReturn flow_ret = GST_FLOW_OK;

      flow_ret = gst_video_decoder_drain_out (decoder, FALSE);
      ret = (flow_ret == GST_FLOW_OK);
1027 1028 1029 1030 1031

      /* Forward GAP immediately. Everything is drained after
       * the GAP event and we can forward this event immediately
       * now without having buffers out of order.
       */
Jan Schmidt's avatar
Jan Schmidt committed
1032 1033 1034
      forward_immediate = TRUE;
      break;
    }
Jan Schmidt's avatar
Jan Schmidt committed
1035
    case GST_EVENT_CUSTOM_DOWNSTREAM:
1036 1037 1038 1039 1040 1041
    {
      gboolean in_still;
      GstFlowReturn flow_ret = GST_FLOW_OK;

      if (gst_video_event_parse_still_frame (event, &in_still)) {
        if (in_still) {
Jan Schmidt's avatar
Jan Schmidt committed
1042
          GST_DEBUG_OBJECT (decoder, "draining current data for still-frame");
1043 1044 1045
          flow_ret = gst_video_decoder_drain_out (decoder, FALSE);
          ret = (flow_ret == GST_FLOW_OK);
        }
1046 1047 1048 1049
        /* Forward STILL_FRAME immediately. Everything is drained after
         * the STILL_FRAME event and we can forward this event immediately
         * now without having buffers out of order.
         */
Jan Schmidt's avatar
Jan Schmidt committed
1050
        forward_immediate = TRUE;
1051
      }
Jan Schmidt's avatar
Jan Schmidt committed
1052
      break;
1053
    }
1054
    case GST_EVENT_SEGMENT:
1055
    {
1056
      GstSegment segment;
1057 1058 1059

      GST_VIDEO_DECODER_STREAM_LOCK (decoder);

1060 1061 1062
      gst_event_copy_segment (event, &segment);

      if (segment.format == GST_FORMAT_TIME) {
1063
        GST_DEBUG_OBJECT (decoder,
1064
            "received TIME SEGMENT %" GST_SEGMENT_FORMAT, &segment);
1065
      } else {
1066
        gint64 start;
1067 1068

        GST_DEBUG_OBJECT (decoder,
1069
            "received SEGMENT %" GST_SEGMENT_FORMAT, &segment);
1070 1071 1072 1073

        /* handle newsegment as a result from our legacy simple seeking */
        /* note that initial 0 should convert to 0 in any case */
        if (priv->do_estimate_rate &&
1074 1075
            gst_pad_query_convert (decoder->sinkpad, GST_FORMAT_BYTES,
                segment.start, GST_FORMAT_TIME, &start)) {
1076 1077 1078 1079 1080 1081
          /* best attempt convert */
          /* as these are only estimates, stop is kept open-ended to avoid
           * premature cutting */
          GST_DEBUG_OBJECT (decoder,
              "converted to TIME start %" GST_TIME_FORMAT,
              GST_TIME_ARGS (start));
1082 1083 1084
          segment.start = start;
          segment.stop = GST_CLOCK_TIME_NONE;
          segment.time = start;
1085 1086
          /* replace event */
          gst_event_unref (event);
1087
          event = gst_event_new_segment (&segment);
1088 1089 1090 1091 1092 1093
        } else {
          GST_VIDEO_DECODER_STREAM_UNLOCK (decoder);
          goto newseg_wrong_format;
        }
      }

1094
      gst_video_decoder_flush (decoder, FALSE);
1095

1096 1097
      priv->base_timestamp = GST_CLOCK_TIME_NONE;
      priv->base_picture_number = 0;
1098

1099
      decoder->input_segment = segment;
1100 1101 1102 1103 1104 1105 1106 1107 1108 1109

      GST_VIDEO_DECODER_STREAM_UNLOCK (decoder);
      break;
    }
    case GST_EVENT_FLUSH_STOP:
    {
      GST_VIDEO_DECODER_STREAM_LOCK (decoder);
      /* well, this is kind of worse than a DISCONT */
      gst_video_decoder_flush (decoder, TRUE);
      GST_VIDEO_DECODER_STREAM_UNLOCK (decoder);
1110 1111 1112 1113
      /* Forward FLUSH_STOP immediately. This is required because it is
       * expected to be forwarded immediately and no buffers are queued
       * anyway.
       */
1114
      forward_immediate = TRUE;
1115
      break;
1116
    }
1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130
    case GST_EVENT_TAG:
    {
      GstTagList *tags;

      gst_event_parse_tag (event, &tags);

      if (gst_tag_list_get_scope (tags) == GST_TAG_SCOPE_STREAM) {
        gst_video_decoder_merge_tags (decoder, tags, GST_TAG_MERGE_REPLACE);
        gst_event_unref (event);
        event = NULL;
        ret = TRUE;
      }
      break;
    }
1131 1132 1133 1134
    default:
      break;
  }

1135 1136 1137 1138
  /* Forward non-serialized events immediately, and all other
   * events which can be forwarded immediately without potentially
   * causing the event to go out of order with other events and
   * buffers as decided above.
1139 1140
   */
  if (event) {
1141
    if (!GST_EVENT_IS_SERIALIZED (event) || forward_immediate) {
1142 1143 1144 1145 1146 1147
      ret = gst_video_decoder_push_event (decoder, event);
    } else {
      GST_VIDEO_DECODER_STREAM_LOCK (decoder);
      decoder->priv->current_frame_events =
          g_list_prepend (decoder->priv->current_frame_events, event);
      GST_VIDEO_DECODER_STREAM_UNLOCK (decoder);
1148
      ret = TRUE;
1149 1150 1151 1152
    }
  }

  return ret;
1153 1154 1155 1156 1157 1158

newseg_wrong_format:
  {
    GST_DEBUG_OBJECT (decoder, "received non TIME newsegment");
    gst_event_unref (event);
    /* SWALLOW EVENT */
1159
    return TRUE;
1160 1161 1162 1163
  }
}

static gboolean
1164 1165
gst_video_decoder_sink_event (GstPad * pad, GstObject * parent,
    GstEvent * event)
1166 1167 1168 1169 1170
{
  GstVideoDecoder *decoder;
  GstVideoDecoderClass *decoder_class;
  gboolean ret = FALSE;

1171
  decoder = GST_VIDEO_DECODER (parent);
1172 1173 1174 1175 1176 1177
  decoder_class = GST_VIDEO_DECODER_GET_CLASS (decoder);

  GST_DEBUG_OBJECT (decoder, "received event %d, %s", GST_EVENT_TYPE (event),
      GST_EVENT_TYPE_NAME (event));

  if (decoder_class->sink_event)
1178
    ret = decoder_class->sink_event (decoder, event);
1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194

  return ret;
}

/* perform upstream byte <-> time conversion (duration, seeking)
 * if subclass allows and if enough data for moderately decent conversion */
static inline gboolean
gst_video_decoder_do_byte (GstVideoDecoder * dec)
{
  return dec->priv->do_estimate_rate && (dec->priv->bytes_out > 0)
      && (dec->priv->time > GST_SECOND);
}

static gboolean
gst_video_decoder_do_seek (GstVideoDecoder * dec, GstEvent * event)
{
1195
  GstFormat format;
1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228
  GstSeekFlags flags;
  GstSeekType start_type, end_type;
  gdouble rate;
  gint64 start, start_time, end_time;
  GstSegment seek_segment;
  guint32 seqnum;

  gst_event_parse_seek (event, &rate, &format, &flags, &start_type,
      &start_time, &end_type, &end_time);

  /* we'll handle plain open-ended flushing seeks with the simple approach */
  if (rate != 1.0) {
    GST_DEBUG_OBJECT (dec, "unsupported seek: rate");
    return FALSE;
  }

  if (start_type != GST_SEEK_TYPE_SET) {
    GST_DEBUG_OBJECT (dec, "unsupported seek: start time");
    return FALSE;
  }

  if (end_type != GST_SEEK_TYPE_NONE ||
      (end_type == GST_SEEK_TYPE_SET && end_time != GST_CLOCK_TIME_NONE)) {
    GST_DEBUG_OBJECT (dec, "unsupported seek: end time");
    return FALSE;
  }

  if (!(flags & GST_SEEK_FLAG_FLUSH)) {
    GST_DEBUG_OBJECT (dec, "unsupported seek: not flushing");
    return FALSE;
  }

  memcpy (&seek_segment, &dec->output_segment, sizeof (seek_segment));
1229
  gst_segment_do_seek (&seek_segment, rate, format, flags, start_type,
1230
      start_time, end_type, end_time, NULL);
1231
  start_time = seek_segment.position;
1232 1233

  if (!gst_pad_query_convert (dec->sinkpad, GST_FORMAT_TIME, start_time,
1234
          GST_FORMAT_BYTES, &start)) {
1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250
    GST_DEBUG_OBJECT (dec, "conversion failed");
    return FALSE;
  }

  seqnum = gst_event_get_seqnum (event);
  event = gst_event_new_seek (1.0, GST_FORMAT_BYTES, flags,
      GST_SEEK_TYPE_SET, start, GST_SEEK_TYPE_NONE, -1);
  gst_event_set_seqnum (event, seqnum);

  GST_DEBUG_OBJECT (dec, "seeking to %" GST_TIME_FORMAT " at byte offset %"
      G_GINT64_FORMAT, GST_TIME_ARGS (start_time), start);

  return gst_pad_push_event (dec->sinkpad, event);
}

static gboolean
1251 1252
gst_video_decoder_src_event_default (GstVideoDecoder * decoder,
    GstEvent * event)
1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265
{
  GstVideoDecoderPrivate *priv;
  gboolean res = FALSE;

  priv = decoder->priv;

  GST_DEBUG_OBJECT (decoder,
      "received event %d, %s", GST_EVENT_TYPE (event),
      GST_EVENT_TYPE_NAME (event));

  switch (GST_EVENT_TYPE (event)) {
    case GST_EVENT_SEEK:
    {
1266
      GstFormat format;
1267 1268
      gdouble rate;
      GstSeekFlags flags;
Wim Taymans's avatar
Wim Taymans committed
1269 1270 1271
      GstSeekType start_type, stop_type;
      gint64 start, stop;
      gint64 tstart, tstop;
1272 1273
      guint32 seqnum;

Wim Taymans's avatar
Wim Taymans committed
1274
      gst_event_parse_seek (event, &rate, &format, &flags, &start_type, &start,
1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290
          &stop_type, &stop);
      seqnum = gst_event_get_seqnum (event);

      /* upstream gets a chance first */
      if ((res = gst_pad_push_event (decoder->sinkpad, event)))
        break;

      /* if upstream fails for a time seek, maybe we can help if allowed */
      if (format == GST_FORMAT_TIME) {
        if (gst_video_decoder_do_byte (decoder))
          res = gst_video_decoder_do_seek (decoder, event);
        break;
      }

      /* ... though a non-time seek can be aided as well */
      /* First bring the requested format to time */
1291
      if (!(res =
Wim Taymans's avatar
Wim Taymans committed
1292 1293
              gst_pad_query_convert (decoder->srcpad, format, start,
                  GST_FORMAT_TIME, &tstart)))
1294
        goto convert_error;
1295
      if (!(res =
1296 1297
              gst_pad_query_convert (decoder->srcpad, format, stop,
                  GST_FORMAT_TIME, &tstop)))
1298 1299 1300 1301
        goto convert_error;

      /* then seek with time on the peer */
      event = gst_event_new_seek (rate, GST_FORMAT_TIME,
Wim Taymans's avatar
Wim Taymans committed
1302
          flags, start_type, tstart, stop_type, tstop);
1303 1304 1305 1306 1307 1308 1309
      gst_event_set_seqnum (event, seqnum);

      res = gst_pad_push_event (decoder->sinkpad, event);
      break;
    }
    case GST_EVENT_QOS:
    {
1310
      GstQOSType type;
1311 1312 1313 1314
      gdouble proportion;
      GstClockTimeDiff diff;
      GstClockTime timestamp;

1315
      gst_event_parse_qos (event, &type, &proportion, &diff, &timestamp);
1316 1317 1318 1319 1320

      GST_OBJECT_LOCK (decoder);
      priv->proportion = proportion;
      if (G_LIKELY (GST_CLOCK_TIME_IS_VALID (timestamp))) {
        if (G_UNLIKELY (diff > 0)) {
1321
          priv->earliest_time = timestamp + 2 * diff + priv->qos_frame_duration;
1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348
        } else {
          priv->earliest_time = timestamp + diff;
        }
      } else {
        priv->earliest_time = GST_CLOCK_TIME_NONE;
      }
      GST_OBJECT_UNLOCK (decoder);

      GST_DEBUG_OBJECT (decoder,
          "got QoS %" GST_TIME_FORMAT ", %" G_GINT64_FORMAT ", %g",
          GST_TIME_ARGS (timestamp), diff, proportion);

      res = gst_pad_push_event (decoder->sinkpad, event);
      break;
    }
    default:
      res = gst_pad_push_event (decoder->sinkpad, event);
      break;
  }
done:
  return res;

convert_error:
  GST_DEBUG_OBJECT (decoder, "could not convert format");
  goto done;
}

1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367
static gboolean
gst_video_decoder_src_event (GstPad * pad, GstObject * parent, GstEvent * event)
{
  GstVideoDecoder *decoder;
  GstVideoDecoderClass *decoder_class;
  gboolean ret = FALSE;

  decoder = GST_VIDEO_DECODER (parent);
  decoder_class = GST_VIDEO_DECODER_GET_CLASS (decoder);

  GST_DEBUG_OBJECT (decoder, "received event %d, %s", GST_EVENT_TYPE (event),
      GST_EVENT_TYPE_NAME (event));

  if (decoder_class->src_event)
    ret = decoder_class->src_event (decoder, event);

  return ret;
}

1368
static gboolean
1369
gst_video_decoder_src_query (GstPad * pad, GstObject * parent, GstQuery * query)
1370 1371 1372 1373
{
  GstVideoDecoder *dec;
  gboolean res = TRUE;

1374
  dec = GST_VIDEO_DECODER (parent);
1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390

  GST_LOG_OBJECT (dec, "handling query: %" GST_PTR_FORMAT, query);

  switch (GST_QUERY_TYPE (query)) {
    case GST_QUERY_POSITION:
    {
      GstFormat format;
      gint64 time, value;

      /* upstream gets a chance first */
      if ((res = gst_pad_peer_query (dec->sinkpad, query))) {
        GST_LOG_OBJECT (dec, "returning peer response");
        break;
      }

      /* we start from the last seen time */
1391
      time = dec->priv->last_timestamp_out;
1392 1393 1394 1395 1396 1397 1398 1399 1400 1401
      /* correct for the segment values */
      time = gst_segment_to_stream_time (&dec->output_segment,
          GST_FORMAT_TIME, time);

      GST_LOG_OBJECT (dec,
          "query %p: our time: %" GST_TIME_FORMAT, query, GST_TIME_ARGS (time));

      /* and convert to the final format */
      gst_query_parse_position (query, &format, NULL);
      if (!(res = gst_pad_query_convert (pad, GST_FORMAT_TIME, time,
1402
                  format, &value)))
1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416
        break;

      gst_query_set_position (query, format, value);

      GST_LOG_OBJECT (dec,
          "query %p: we return %" G_GINT64_FORMAT " (format %u)", query, value,
          format);
      break;
    }
    case GST_QUERY_DURATION:
    {
      GstFormat format;

      /* upstream in any case */
1417
      if ((res = gst_pad_query_default (pad, parent, query)))
1418 1419 1420 1421 1422 1423 1424
        break;

      gst_query_parse_duration (query, &format, NULL);
      /* try answering TIME by converting from BYTE if subclass allows  */
      if (format == GST_FORMAT_TIME && gst_video_decoder_do_byte (dec)) {
        gint64 value;

1425 1426
        if (gst_pad_peer_query_duration (dec->sinkpad, GST_FORMAT_BYTES,
                &value)) {
1427 1428
          GST_LOG_OBJECT (dec, "upstream size %" G_GINT64_FORMAT, value);
          if (gst_pad_query_convert (dec->sinkpad,
1429
                  GST_FORMAT_BYTES, value, GST_FORMAT_TIME, &value)) {
1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444
            gst_query_set_duration (query, GST_FORMAT_TIME, value);
            res = TRUE;
          }
        }
      }
      break;
    }
    case GST_QUERY_CONVERT:
    {
      GstFormat src_fmt, dest_fmt;
      gint64 src_val, dest_val;

      GST_DEBUG_OBJECT (dec, "convert query");

      gst_query_parse_convert (query, &src_fmt, &src_val, &dest_fmt, &dest_val);
1445
      GST_OBJECT_LOCK (dec);
1446 1447 1448 1449 1450
      if (dec->priv->output_state != NULL)
        res = gst_video_rawvideo_convert (dec->priv->output_state,
            src_fmt, src_val, &dest_fmt, &dest_val);
      else
        res = FALSE;
1451
      GST_OBJECT_UNLOCK (dec);
1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 <