summaryrefslogtreecommitdiffstats
path: root/libhb/reader.c
Commit message (Collapse)AuthorAgeFilesLines
* reader: fix live preview generation for some streamsJohn Stebbins2017-06-191-3/+23
| | | | | | | | | | Fixes https://github.com/HandBrake/HandBrake/issues/685 For some stream types, libav does not seek all streams within the file to the same position. So we get data from streams that is prior to the desired preview start position leaking through. The result, for example, is a preview that has audio that starts from the beginning of the file and video that starts much later at the desired start pos.
* decsrt: fix p-to-p start time after seekingJohn Stebbins2017-04-151-11/+5
| | | | | | | Reader can skip data at the beginning of the file. We were not informing decsrt how much was skipped when pts_to_start caused the skip. Fixes https://forum.handbrake.fr/viewtopic.php?f=11&t=36258
* reader: fix incorrect duration of UTF8 subtitlesJohn Stebbins2017-01-111-0/+4
| | | | The stop time for these was not getting adjusted by the scr_offset
* Update copyright dates to 2017.Bradley Sepos2017-01-011-1/+1
|
* reader: fix compiler warning "buf uninitialized"John Stebbins2016-12-151-0/+7
|
* libhb: do not free hb_work_private_t in reader_init() if the reader open ↵Damiano Galassi2016-11-121-1/+0
| | | | fail, it will be freed later in reader_close(). Fixes a crash that happens when hb_reader_open() fails because the source file was moved.
* reader: fix p-to-p start pos in avformat demuxed streamsJohn Stebbins2016-06-271-27/+27
|
* sync: correct timestamp discontinuities in sync instead of reader (#192)John Stebbins2016-05-241-387/+127
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | * sync: correct timestamp discontinuities in sync instead of reader This patch passes discontinuity information through the pipeline till it reaches sync.c. The timestamps are passed through the pipeline as read and unmodified to sync.c (instead of attempting to correct discontinuities in reader). In sync, when we see a discontinuity, we know where the next timestamp should be based on the timestamp and duration of the previous buffer (before the discontinuity). So we calculate an "SCR" offset based on the timestamp after the discontinuity and what we calculate it should be. The old discontinuity handling code was broken due to the following. The MPEG STD timing model relies heavily on the decoder having an STC that is phase lock looped to the PCRs in the stream. When decoding a broadcast stream, the decoder can count on the time measure between PCRs using the STC to match to a high degree of accuracy. I.e. STC - lastSTC == PCR - lastPCR. When a discontinuity occurs, the decoder calculates a new PCR offset = PCR - STC. I.e. the offset is the new PCR value minus what it would have been if there had been no discontinuity. The above does not work without a reliable STC, which we do not have. We have been attempting to approximate one by avereraging the duration of received packets and extrapolating an "STC" based on the last PTS and the average packet duration. But this is highly variable and unreliable. * decavcodec: fix data type of next_pts It needs to be double so that partial ticks are not lost * deccc608sub: clarify comment * sync: allow queueing more audio Audio is small, and there is often a significant amount of audio in the stream before the first video frame. * sync: improve handling of damaged streams When data is missing, the audio decoder was extrapolating timestamps from the last pts before the error caused by the missing data which caused sync issues. Also, missing data can cause the video decoder to output a frame out of order with the wrong scr sequence. Drop such frames.
* libhb: remove unused hb_buffer_t sequence numberJohn Stebbins2016-05-171-3/+0
|
* bd: change angles before seekingJohn Stebbins2016-04-041-4/+4
| | | | the angle can affect the seek position
* reader: fix memory leakJohn Stebbins2016-03-231-0/+6
|
* reader: fix p-to-p seconds for non-seekable streamsJohn Stebbins2016-03-111-1/+20
| | | | | This broke when sync was reworked. Sync now expects job->pts_to_start to be relative to the first frame that it sees.
* libhb: fix memory leaksJohn Stebbins2016-03-061-2/+3
|
* Update copyright dates to 2016.Bradley Sepos2016-01-011-1/+1
|
* reader: fix stream discontinuity splicing...John Stebbins2015-12-091-3/+9
| | | | | | | Splicing of buffers that got duplicated to multiple output fifos was broken. Caused https://forum.handbrake.fr/viewtopic.php?f=11&t=33666
* reader: remove track count limitationJohn Stebbins2015-11-131-7/+15
| | | | | | Don't use hard coded 100 fifo array, allocate what is needed. We probably just crashed if the number of tracks was > 99 since the limit of 100 fifos was not universally checked.
* libhb: repair split packets in readerJohn Stebbins2015-11-101-1/+79
| | | | | | | | We split PES packets when there is a PCR change in the middle of the packet. This works fine for audio and video where the decoder parses the ES to find frame boundaries. But it does not work for some decoders such as PGS subtitles. So mark split buffers and reassemble them in reader after processing the PCR change.
* libhb: make muxer, sync, and reader behave like other work objectsJohn Stebbins2015-11-091-342/+315
| | | | | | | | | | | | | simplify job initialization sequence, clean up code, and document dependencies in the sequence better. Make hb_add set job->sequence_id. It is no longer necessary for the frontend to do this. If the frontend needs the sequence_id, it is returned by hb_add(). Clean up use of interjob. do_job() now uses sequence_id to detect when a new sequence of related jobs is running and automatically clears interjob.
* libhb: add hb_buffer_listJohn Stebbins2015-09-241-13/+12
| | | | | This brings together several independent implementations of a simple buffer list manager.
* libhb: detect read errors and propagatejstebbins2015-06-291-1/+1
| | | | git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@7332 b64f7644-9d1e-0410-96f1-a4d463321fa5
* libhb: Use a buffer flat to indicate EOFjstebbins2015-05-011-3/+3
| | | | | | | | | | | | ... instead of a 0 length buffer. This fixes this issue: https://forum.handbrake.fr/viewtopic.php?f=12&t=31959 Theora can create 0 length output. These 0 length frames indicate duplicate frames. So we can't use 0 length buffers to indicate the end of the stream. git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@7143 b64f7644-9d1e-0410-96f1-a4d463321fa5
* p-to-p: fix seek to start pts source initial PTS != 0jstebbins2015-04-241-17/+2
| | | | | | | When a file demuxed by libav does not start to time 0, our seek to the initial start pts tried to seek too far forward. git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@7124 b64f7644-9d1e-0410-96f1-a4d463321fa5
* demuxmpeg: Don't drop frames with bad timestampsjstebbins2015-04-221-4/+0
| | | | | | Just invalidate the timestamps and let the decoders interpolate. git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@7121 b64f7644-9d1e-0410-96f1-a4d463321fa5
* reader: fix subtitle start time not advancingjstebbins2015-04-071-2/+37
| | | | | | | | | We must set a start time for subtitles or the vobsub decoder (and probably other subtitle decoders) will assign duplicate timestamps. So when a discontinuity occurs, make the closest guess that we can. git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@7072 b64f7644-9d1e-0410-96f1-a4d463321fa5
* libhb: Eliminate global variable hb_gui_use_hwd_flagjstebbins2015-03-291-2/+4
| | | | | | | | | | | | This global was shared between the CLI and libhb and used as a back door to force scan and encode passes to use the same ffmpeg context for hardware decoding. Aside from the fact that this context sharing should not be necessary and needs fixing, this information belongs in the hb_handle_t that is shared between the scan and the encode. So put it there and make sure the hb_handle_t get propagated to where the flag is needed. git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@7028 b64f7644-9d1e-0410-96f1-a4d463321fa5
* json: automatically scan title when processing json jobsjstebbins2015-03-071-0/+1
| | | | | | | | | | | | | | | Simplifies the WinGui. This also changes how jobs are processed. Creating the sub-jobs for multiple passes is delayed until after scanning and immediately before running the job. Working status has also changed. Sub-job passes are identified in status with an ID that allows the frontend to definitively identify what pass is in progress. git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@6976 b64f7644-9d1e-0410-96f1-a4d463321fa5
* Update year to 2015.bradleys2015-02-011-1/+1
| | | | git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@6852 b64f7644-9d1e-0410-96f1-a4d463321fa5
* json: add json APIsjstebbins2014-12-161-3/+4
| | | | | | | | | | There are several changes to job and title structs that break current windows interop code. The interop code should be changed such that it only uses json APIs. So if there is any missing features (or bugs) in these APIs, please let me know. git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@6602 b64f7644-9d1e-0410-96f1-a4d463321fa5
* libhb: fix p-to-p start for some streamsjstebbins2014-11-051-18/+21
| | | | | | | | | The primary problem was in setting our "zero" time in reader based on a stream that is not decoded. Since this stream never reaches sync, there would appear to be a long initial frame from syncs perspective. git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@6506 b64f7644-9d1e-0410-96f1-a4d463321fa5
* libhb: fix stop time of subtitlesjstebbins2014-08-271-0/+4
| | | | | | | scr_offset was not accounted for in stop time git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@6372 b64f7644-9d1e-0410-96f1-a4d463321fa5
* libhb: work around libav decoder error durint p-to-p encodingjstebbins2014-07-141-23/+40
| | | | | | | | | In some cases, initial data when in p-to-p mode causes libav decoder initialization to fail. This only happens when multi-threaded encoding is enabled. git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@6229 b64f7644-9d1e-0410-96f1-a4d463321fa5
* Update some copyright dates to 2014.Rodeo2014-02-181-1/+1
| | | | git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@6042 b64f7644-9d1e-0410-96f1-a4d463321fa5
* Replace invaled timestamp flag "-1" with AV_NOPTSjstebbins2014-01-261-8/+10
| | | | | | | | | -1 is not a good value as a flag for invalid timestamps. There are cases where small negative timestamps are useful. So this eliminates a potential ambiguity. git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@6001 b64f7644-9d1e-0410-96f1-a4d463321fa5
* libhb: fix pts p-to-p start time in files parsed by libavjstebbins2013-06-261-1/+1
| | | | | | | The adjustment made to the start time was made incorrectly. git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@5614 b64f7644-9d1e-0410-96f1-a4d463321fa5
* Updated all the copyright headers to 2013.sr552013-03-091-1/+1
| | | | git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@5318 b64f7644-9d1e-0410-96f1-a4d463321fa5
* libhb: fix audio sync regression with DVDsjstebbins2013-01-061-6/+25
| | | | | | | | A "fix" for another sync issue caused a regression in handling of DVD sync. So revert the change and make other improvements. git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@5153 b64f7644-9d1e-0410-96f1-a4d463321fa5
* libhb: fix memory consumption problemjstebbins2012-11-261-2/+2
| | | | | | | | | Issues with timestamps made cfr think it needed to duplicate a few thousand frames. this leads to an over-cunsumption of memory since all duplicates are placed in a list at once. git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@5082 b64f7644-9d1e-0410-96f1-a4d463321fa5
* Improve management of titles and jobsjstebbins2012-11-121-10/+10
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Cleans up several several unavoidable memory leaks caused by old api. Clearly separates titles from jobs. Titles are set during scan and never modified now. Since titles are immutable, this lead to API some changes. For example, We were setting chapter names in the title from the front ends. Now these get set in the job. These new APIs allow us to start moving away from our use of title->job. Eventually, I would like to eliminate title->job completely, but the mac ui is too tightly tied to using this field to allow removing it at this time. So there is temporarily a convenience function used only by the mac ui that allows it to continue using title->job and also use the new APIs. New APIs: typedef struct hb_title_set_s hb_title_set_t; struct hb_title_set_s { hb_list_t * list_title; int feature; // Detected DVD feature title }; hb_title_set_t * hb_get_title_set( hb_handle_t * ); This is just something I added to clean up how "feature title" info is passed. hb_job_t * hb_job_init( hb_title_t * title ); Initializes a new job with default settings from the title. hb_job_t * hb_job_init_by_index( hb_handle_t *h, int title_index ); Same as hb_job_init(). For use by win Interop lib. void hb_job_reset( hb_job_t * job ); Convenience function for the MacUi. Clears audio, subtitle, and filter lists. The macui still uses title->job because it is so intricately tied to it. So I created this convenience function that it can call after adding a job. void hb_job_close( hb_job_t ** job ); Releases the job an all resources it contains. void hb_job_set_advanced_opts( hb_job_t *job, const char *advanced_opts ); Makes a copy of "advanced_opts" and stores in job. Freed by hb_job_close(). void hb_job_set_file( hb_job_t *job, const char *file ); Makes a copy of "file" and stores in job. Freed by hb_job_close(). void hb_chapter_set_title(hb_chapter_t *chapter, const char *title); Makes a copy of "title" and stores in chapter. Freed by hb_chapter_close(). Recommended usage (cli and lingui are updated to do this): job = hb_job_init( title ); // set job settings ... hb_add(h, job); hb_job_close( &job ); I have also added new APIs for managing metadata. These are used to add metadata to a job. void hb_metadata_set_name( hb_metadata_t *metadata, const char *name ); void hb_metadata_set_artist( hb_metadata_t *metadata, const char *artist ); void hb_metadata_set_composer( hb_metadata_t *metadata, const char *composer ); void hb_metadata_set_release_date( hb_metadata_t *metadata, const char *release_date ); void hb_metadata_set_comment( hb_metadata_t *metadata, const char *comment ); void hb_metadata_set_genre( hb_metadata_t *metadata, const char *genre ); void hb_metadata_set_album( hb_metadata_t *metadata, const char *album ); void hb_metadata_set_coverart( hb_metadata_t *metadata, const uint8_t *coverart, int size ); Example: job = hb_job_init( &job ); // set job settings ... hb_metadata_set_artist( job->metadata, "Danny Elfman" ); hb_add(h, job); hb_job_close( &job ); Some APIs have changed in order to avoid using title incorrectly and use the new hb_title_set_t. -void hb_autopassthru_apply_settings( hb_job_t * job, hb_title_t * title ); +void hb_autopassthru_apply_settings( hb_job_t * job ); -void hb_get_preview( hb_handle_t *, hb_title_t *, int, uint8_t * ); +void hb_get_preview( hb_handle_t *, hb_job_t *, int, uint8_t * ); hb_thread_t * hb_scan_init( hb_handle_t *, volatile int * die, const char * path, int title_index, - hb_list_t * list_title, int preview_count, + hb_title_set_t * title_set, int preview_count, int store_previews, uint64_t min_duration ); These APIs have been removed. Win Interop will need some changes. I think what I've provided will be suffecient, but let me know if it's not. -void hb_get_preview_by_index( hb_handle_t *, int, int, uint8_t * ); -void hb_set_anamorphic_size_by_index( hb_handle_t *, int, - int *output_width, int *output_height, - int *output_par_width, int *output_par_height ); git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@5058 b64f7644-9d1e-0410-96f1-a4d463321fa5
* More fine grained progress statistics during subtitle scanjstebbins2012-07-181-30/+66
| | | | | | | | Make reader compute subtitle scan progress based on timestamps seen and duration instead of chapter marks. git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@4864 b64f7644-9d1e-0410-96f1-a4d463321fa5
* libhb: fix live preview crash when generating preview while encodingjstebbins2012-07-031-11/+12
| | | | | | | GetFifoForId() was not re-entrant (it used a static array). git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@4809 b64f7644-9d1e-0410-96f1-a4d463321fa5
* Updated file copyright / GPL headerssr552012-06-151-4/+6
| | | | git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@4737 b64f7644-9d1e-0410-96f1-a4d463321fa5
* Simplify Foreign Audio Search code:Rodeo2012-05-011-15/+7
| | | | | | | | | | | | | | | | | | | | - move all subtitle hit counting to the decoders instead of reader ---> allows us to count actual subtitles rather than just packets - always count subtitles, even when not doing a scan (may be useful in the future) Miscellaneous improvements: - always insert select_subtitle at the head of the output subtitle list, to make it less likely to be dropped - when multiple subtitle tracks have forced hits, pick the track with the fewest forced hits ---> Foreign Audio Search should now work with Star Wars on Blu-ray - logging improvements (more readable, and log job->select_subtitle configuration - Forced Only vs. All, Render vs. Passthrough) git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@4622 b64f7644-9d1e-0410-96f1-a4d463321fa5
* Rework filter pipelinejstebbins2012-03-271-28/+28
| | | | | | | | | | | | | | | | | | | | This patch enhances the filter objects. The 2 key improvements are: 1. A filter can change the image dimensions as frames pass through it. 2. A filter can output more than one frame. In addition, I have: Moved cropping & scalling into a filter object Added 90 degree rotation to the rotate filter Moved subtitle burn-in rendering to a filter object. Moved VFR/CFR handling into a framerate shaping filter object. Removed render.c since all it's responsibilities got moved to filters. Improves VOBSUB and SSA subtitle handling. Allows subtitle animations. SSA karaoke support. My apologies in advance if anything breaks ;) git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@4546 b64f7644-9d1e-0410-96f1-a4d463321fa5
* Fix p-to-p start time detectionjstebbins2012-03-041-2/+0
| | | | | | | Brainfart caused start time detection in TS files to break. git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@4488 b64f7644-9d1e-0410-96f1-a4d463321fa5
* handle TS streams that have no PCRs betterjstebbins2011-09-261-11/+29
| | | | | | | | | | | | | | | | For TS streams that don't have PCRs, we substitute the DTS timestamp from the video track (or PTS if we don't see DTS). But these can bounce around or be wider spaced in the stream that PCRs are meant to be. So I have added a test to see if the timestamp looks like a discontinuity. Then I only pass the timestamp as a PCR if there appears to be a discontinuity. This prevents a lot of scr_offset thrashing. I have also fixed an error in our scr_offset processing. It is rarely triggered and it's effects are so minor with well behaved streams that it would be completely unnoticed. But with the test stream I was using, it caused a factor of 10 times more "audio went backwards" errors. git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@4254 b64f7644-9d1e-0410-96f1-a4d463321fa5
* Improve mpeg PS supportjstebbins2011-09-141-13/+2
| | | | | | | | | | | | | | | | | | | | | | Adds support for MPEG-1 PS, HDDVD EVOB, and video codecs other than mpeg1/2 in PS Improves probing of unknown stream types by using Libav's probing utilities Use Libav to probe for dts profile in TS and PS files when profile is unknown Improves framerate detection (improved telecine detection) Fixes preview generation for mpeg video that has only a single sequence header Patches Libav to handle VC-1 pulldown flags properly Improve PS and TS stream log information git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@4220 b64f7644-9d1e-0410-96f1-a4d463321fa5
* libhb: fix or simplify several hacks involved with Libav supportjstebbins2011-07-271-62/+95
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | For files that are demuxed by Libav, we must share the format context with the decoder iso that it can obtain the codec context for each stream. The code that did this was very convoluted and difficult to understand. It is simplified by simply passing the context in hb_title_t. Reader was closing stream files before the decoder was finished with the context. This created the need to delay the actual close and cache the context. Changed reader so it behaves more like the rest of handbrake's work objects which lets us explicitly close after the decoders are finished. Libav does some probing of the file when av_find_stream_info is called. This probing leaves the format context in a bad state for some files and causes subsequent reads or seeks to misbehave. So open 2 contexts in ffmpeg_open. One is used only for probing, and the other only for reading. decavcodec.c had 2 separate decoders for files demuxed by hb and files demuxed by Libav. They have been combined and simplified. Previously, it was not possible to decode one source audio track multiple times in order to fan it out to multiple output tracks if the file is demuxed by Libav. We were using the codec context from the format context. Since there is only one of these for each stream, we could only do one decode for each stream. Use avcodec_copy_context to make copies of the codec context and allow multiple decodes. This allows removal of a lot of special case code for Libav streams that was necessary to duplicate the output of the decoder. Patch Libav's mkv demux to fix a seek problem. This has been pushed upstreams, so the next time we update Libav, we must remove this patch. git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@4141 b64f7644-9d1e-0410-96f1-a4d463321fa5
* Add support for TrueHD and DTS-HD from BD sourcesjstebbins2011-04-221-17/+13
| | | | | | | | | | | | | | TrueHD and DTS-HD now show up in the audio list along side their AC-3 and DTS counterparts. Note that currently the DTS-HD decoder we are using (ffmpeg) discards the HD portion of the stream and onle decodes the DTS core portion. So there is no advantage yet to using the DTS-HD stream. In the future I would like to add DTS-HD passthru support and hopefully ffmpeg will improve their DTS-HD decoder. git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@3950 b64f7644-9d1e-0410-96f1-a4d463321fa5
* Fix a deadlock that can happen with p-to-p encodingjstebbins2011-04-101-1/+15
| | | | | | | | | | | | | | | | | | With p-to-p, the audio sync thread waits for the video sync thread to reach the designated start point. There is a possibility that the video decoder will drop so many frames that the audio sync fifo fills before any frames reach the video sync thread. When this happens, drop some audio to unplug the pipeline. Also, to make this less likely to happen, start sending data to the video decoder 2 seconds before the actual desired start point. This will allow the decoder to find an initial i-frame before the audio stalls since the audio sync thread drops any audio that is before the designated start point. A side effect of this is our start time now more accurate since the decoder is only dropping frames before the start point instead of after. git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@3917 b64f7644-9d1e-0410-96f1-a4d463321fa5
* Fix garbled previews from BD h.264 sourcesjstebbins2011-04-021-1/+2
| | | | | | | | | | | Forome reason, frames that are tagged as recovery points in many BD h.264 streams do not result in complete frames when decoded. Pushing 2 extra frames through the decoder seems to always fix this. This patch extends something I was already doing when generating previews from a BD structure. This just applies the same logic to ffmpeg streams that have h.264 video. git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@3895 b64f7644-9d1e-0410-96f1-a4d463321fa5
generated by cgit v1.2.3 (git 2.39.1) at 2024-12-22 12:39:13 +0000