summaryrefslogtreecommitdiffstats
path: root/demux
<
Commit message (Collapse)AuthorAgeFilesLines
* demux_mkv: better seeking after video endwm42015-04-231-15/+7
| | | | | | | | | | | | | | | | | This change allows forward seeking even if there are no more video keyframes in forward direction. This helps with files that e.g. encode cover art as a single video frame (within a _real_ video stream - ffmpeg seems to like to produce such files). Seeking backwards will still jump to the nearest video frame, so this improvement has limited use. The old code didn't do this because of the logic the min_diff variable followed. Instead of somehow using the timestamp of the last packet read for min_diff, use the first index entry for it. This actually makes it fall back to the first/last index entry as the (removed) comment claims. Note that last_pts is basically random at this point (because the demuxer can be far ahead of playback position), so this didn't make sense in the first place.
* demux: report correct cache state close to EOFwm42015-04-201-5/+4
| | | | | | | | | | | | | | | | | | | | | On EOF, this stopped reporting the actual cache duration, and just signalled unknown duration. Fix this and keep reporting whatever is left in the packet queue. This reverts commit 5438a8b3. The commit doesn't give a good explanation as to why it is needed, but I guess it was because the reporting was imperfect (it switched between unknown or 0, and the correct duration). This also removes a line added in commit 848546f2. The line is ds->active = false; The "active" flag basically says that data from this stream is actively needed, and it's used to calculate the minimum data that can actually be played (approximately). If this were ignored, a sparse subtitle stream would set the cache duration to 0s. The commit message adding the line says "actually does nothing, but in theory it's cleaner". Well, screw it.
* player: change video-bitrate and audio-bitrate propertieswm42015-04-203-6/+0
| | | | | | | | | | | | | | Remove the old implementation for these properties. It was never very good, often returned very innaccurate values or just 0, and was static even if the source was variable bitrate. Replace it with the implementation of "packet-video-bitrate". Mark the "packet-..." properties as deprecated. (The effective difference is different formatting, and returning the raw value in bits instead of kilobits.) Also extend the documentation a little. It appears at least some decoders (sipr?) need the AVCodecContext.bit_rate field set, so this one is still passed through.
* player: allow playing directorieswm42015-04-171-1/+53
| | | | | | | | | | | If a directory is encountered, replace it with its contents in the internal playlist. This is messed into demux_playlist.c, because why not. STREAMTYPE_DIR could be avoided by unconditonally trying opendir() in demux_playlist.c, but it seems nicer not to do weird things like calling it on real files. This does not work on Windows, because msvcrt is retarded.
* demuxer_lavf: add an option for disabling hackswm42015-04-171-0/+5
|
* Update license headersMarcin Kurczewski2015-04-1315-77/+62
| | | | Signed-off-by: wm4 <wm4@nowhere>
* demux_lavf: use mimetype instead of codec type for embedded fontswm42015-04-051-9/+3
| | | | | | | We handle picking out font attachments by mime type ourselves in a higher level, so we really just want to use the mimetype. Also, Matroska is currently the only code in libavformat which uses the fonts at all, and we can drop use of the codec IDs completely.
* demux: fix rar support for files containing DTS audio trackswm42015-03-243-26/+64
| | | | | | | | | | | | | | | | | With a recent cleanup, rar support was stuffed into demux_playlist.c (because "opening" rar files pretty much just lists archive contents and adds them to a playlist using a special rar:// protocol, which will actually access the rar file contents). Since demux_playlist.c is probed _after_ demux_lavf.c (and should/must be), libavformat was given the chance to detect DTS streams embedded within the rar file. This is not really what we want, and a regression what happened before rar listing was moved to demux_playlist.c. Fix it by moving the rar listing into its own pseudo-demuxer, and let ir probe before demux_lavf.c. (Yes, this feature still has users.)
* demux_lavf: print seek failures in verbose modewm42015-03-241-6/+12
| | | | | Don't bother with making these visible by default, because often they are bogus and/or useless.
* demux_mkv_timeline: don't continue if reopening file failedwm42015-03-231-1/+1
| | | | | Could theoretically dereference "d" later in the loop. It's on an error codepath, so just give up.
* demux_lavf: assume h264/hevc raw streams have no timestampswm42015-03-201-4/+12
| | | | | There are obscure methods to add timestamps to such streams, but assume they're unused.
* player: better handling of video with no timestampswm42015-03-202-10/+5
| | | | | | | | | | | Trying to handle such video is almost worthless, but it was requested by at least 2 users. If there are no timestamps, enable byte seeking by setting ts_resets_possible. Use the video FPS (wherever it comes from) and the audio samplerate for timing. The latter was already done by making the first packet emit DTS=0; remove this again and do it "properly" in a higher level.
* demux_mkv: never warn when trying to resync on EOFwm42015-03-091-2/+4
| | | | | | If this happens, the file is actually broken, but the assumption is simply that the file was truncated, and printing a warning would be strange.
* demux_mkv: check for playback abortswm42015-03-093-0/+9
| | | | | | | | | Check async abort notification. libavformat already do something equivalent. Before this commit, the demuxer could enter resync mode (and print silly warning messages) when the stream stopped returning data because of an abort.
* demux_mkv: actually skip elements out of reachwm42015-03-061-1/+2
| | | | | This is missing from the previous commit. Not that harmful, but also slightly un-nice since even a failed seek will reset the cache.
* demux_mkv: fix issues with unseekable streamswm42015-03-061-14/+16
| | | | | | | | | | | | | | | A user reported a webm stream that couldn't be played. The issue was that this stream 1. was on an unseekable HTTP connection, and 2. had a SeekHead element (wtf?). The code reading the SeekHead marked the element as unreadable too early: although you can't seek in the stream, reading the header elements after the SeekHead read them anyway. Marking them as unreadable only after the normal header reading fixes this. (The way the failing stream was setup was pretty retarded: inserting these SeekHead elements makes absolutely no sense for a stream that cannot be seeked.) Fixes #1656.
* sub: this isn't needed eitherwm42015-03-032-3/+0
|
* Revert "demux_lavf: workaround for broken libavformat subtitle seeking"wm42015-03-031-23/+4
| | | | | | | This reverts commit c8f49be919ffaf983bde77b63d75f96a593ec7a8. Not needed anymore; fixed in all supported FFmpeg releases. Though I could not test again, because all sample files are gone (oops).
* Remove some FFmpeg/Libav compatibility hackswm42015-03-032-35/+3
| | | | | | All of these are now in the supported FFmpeg and Libav versions. The 3 remaining API checks are for FFmpeg-only things.
* demux_lavf: use this for image formatswm42015-03-021-6/+13
| | | | | | | | | | | | | | | | | | Use the (relatively new) libavformat image format probing functionality, instead of letting demux_mf guess by file extension and MIME type. The libavformat support is weird, though. Traditionally, it uses an absolutely terrible hack to detect images by extension, _and_ (which is the horrible part) will randomly interpret parts of the filename as specifiers for matching by number. So something like '%03d' will be interpreted as placeholder for a frame number. The worst part is that such character sequences can be perfectly valid and common in http URLs. This is known as "image2" demuxer. The newer support, which probes by examining the file header, is split into several format-specific demuxers with names ending in "_pipe". So we check for such a name suffix. (At this point we're doing fine-grained hacking around ffmpeg weirdness, so a clean solution is impossible anyway until upstream changes.)
* player: refine rar:// playlist-safety handlingwm42015-03-021-0/+1
| | | | | | | | | | | | It was possible to make the player play local files by putting rar:// links into remote playlists, and some other potentially unsafe things. Redo the handling of it. Now the rar-redirector (the thing in demux_playlist.c) sets disable_safety, which makes the player open any playlist entries returned. This is fine, because it redirects to the same file anyway (just with different selection/interpretation of the contents). On the other hand, rar:// itself is now considered fully unsafe, which means that it is ignored if found in normal playlists.
* buid: readd -Wparentheseswm42015-03-021-2/+2
| | | | | | | | | This warning wasn't overly helpful in the past, and warned against perfectly fine code. But at least with recent gcc versions, this is the warning that complains about assignments in if expressions (why???), so we want to enable it. Also change all the code this warning complains about for no reason.
* stream: remove stream filter conceptwm42015-02-271-1/+1
| | | | Unused since the previous commit. (Apparently it was a stupid idea.)
* stream_rar: treat rar files as playlistswm42015-02-271-0/+25
| | | | | | | | | | | | Refactors an older hack, which for some reason used a more complicated way. This generates the playlist representing the contents of the rar file in demux_playlist.c. The pseudo-demuxer could easily be separate from the the playlist parsers (and in fact there's almost no shared code), but I don't think this obscure feature deserves a separate file. Sample files created with: rar a -v20000k -m0 files.rar file1.mkv file1.mkv
* demux_mkv_timeline: fix potential issue when enabling cachewm42015-02-251-0/+1
| | | | | | | | | | | | If the cache is enabled, the demuxer is closed and opened again (because currently, the cache can not be enabled atfer data was already read). The call for opening a new demuxer uses the same params struct, which references the ctx->uids array. But there is a MP_TARRAY_GROW() invocation somewhere on the way, which can reallocate the ctx->uids array, making params.uids a dangling pointer. This issue probably existed for a longer time, probably since 5cd33853 (slightly more obvious since f50b105d).
* demux_mkv_timeline: move uids array to context toowm42015-02-251-31/+26
| | | | | | | Again removes some indirections and extra arguments. Also replace some memcpy/memmoves with assignments. (Assignments became possible only later, when reference UIDs were turned into a struct.)
* demux_mkv_timeline: pass sources array as part of contextwm42015-02-251-45/+39
| | | | Removes tripple pointer indirections and such.
* demux_mkv: minor refactor for deferred cue readingwm42015-02-251-45/+41
| | | | | | Should behave about the same, but reduces code some duplication with seeking and reading a header element pointed to by a SeekHead. It also makes behavior with incomplete files slightly better.
* demux_mkv: don't spam too many warnings with partial fileswm42015-02-251-14/+15
| | | | | Limit it to a single message. It often printed more than that, and in some cases (old files with "cluster" index), spammed a lot.
* video: remove redundant codec parameterswm42015-02-242-3/+0
| | | | | | | Remove coded_width and coded_height. This was originally added in commit fd7dde40, when BITMAPINFOHEADER was killed. The separate fields became redundant in commit e68f4be1. Remove them (nothing passed to the decoders actually changes with _this_ commit).
* demux: timeline: honor quit requestswm42015-02-205-4/+12
|
* demux: add a demux_open_url() functionwm42015-02-204-58/+43
| | | | | Often stream and a demuxer are opened at the same time. Provide a function for this and replace most of its uses.
* demux: change demux_open() signaturewm42015-02-206-14/+17
| | | | Fold the relatively obscure force_format parameter into demuxer_params.
* demux: add free_demuxer_and_stream() functionwm42015-02-204-9/+14
| | | | | Although their lifetimes are conceptually different, it happens often that a demuxer is destroyed together with its stream.
* demux_lavf: apply hacks even if format is forcedwm42015-02-201-27/+28
| | | | | | | | | | | | | Some of the hacks were not applied if the file format was forced. Commit 37a0c914 moved them to a table, which is checked with normal probing only. Fixes #1612 (DVD forces mpeg, which in turn has to export native stream IDs specifically). Do some code restructuring on the way. For example, the probescore can simply be set to the correct initial value, instead of checking whether it was set at all.
* demux_lavf: blacklist bintext fileswm42015-02-181-0/+3
| | | | | | Whatever the hell that is. FFmpeg tries to open any files with .bin file extension with this demuxer (unless it finds a better demuxer), and then reads the whole damn file, along with spamming dumb crap.
* player: enable cache and demuxer thread for subtitles toowm42015-02-188-0/+19
| | | | | | | | | Includes some logic for not starting the demuxer thread for fully read subtitles. (Well, the cache will still waste _lots_ of resources, and the cache always has to be created, because we don't know whether it'll be needed _before_ opening the file.) See #1597.
* demux_lavf: reorganize hackswm42015-02-181-67/+70
| | | | | | | | | | | An attempt to make format-specifics more declarative. (In my opinion, all of this should be either provided by libavformat, or should not be needed.) I'm still leaving many checks with matches_avinputformat_name(), because they're so specific. Also useful for the following commit.
* demux_lavf: set interrupt_callbackwm42015-02-181-0/+11
| | | | | | Helps with terminating the stream if e.g. HLS streams are stuck. (For other demuxers, the stream's interrupt callback already takes care of this.)
* demux, matroska: remove demuxer type fieldwm42015-02-174-39/+29
| | | | | | | | | | | | | The Matroska timeline code was the only thing which still used the demuxer.type field. This field explicitly identifies a demuxer implementation. The purpose of the Matroska timeline code was to reject files that are not Matroska. But it already forces the Matroska format, meaning loading will explicitly only use the Matroska demuxer. If the demuxer can't open the file, no other demuxer will be tried, and thus checking the field is redundant. The change in demux_mkv_timeline.c removes the if condition, and unindents the if body.
* demux: remove file_contents fieldwm42015-02-174-11/+25
| | | | | | | Only demux_cue and demux_edl used it. It's a weird field and doesn't help with anything anymore - by now, it only saves a priv context in the mentioned demuxers. Reducing the number of confusing things the demuxer struct has is more important than minimizing the code.
* demux_edl: make independent from MPContextwm42015-02-172-42/+36
|
* demux_edl: move implementationwm42015-02-171-2/+280
| | | | Same deal as with demux_cue, and a separate commit for the same reasons.
* demux_cue: make independent from MPContextwm42015-02-172-35/+36
| | | | Also see previous commit(s).
* demux_cue: move implementationwm42015-02-171-4/+388
| | | | | | | | Move the implementation, of which most was in tl_cue.c, to demux_cue.c. Currently, this is illogical, because tl_cue.c still accesses MPContext. This is going to change, and then it will be better if everything is in demux_cue.c. This is only a separate commit to distinguish code movement and actual work; the next commit will do the actual work.
* demux: copy priv pointer toowm42015-02-171-0/+1
| | | | | Weird, but helps with the case a demuxer gets handed its own instance from outside.
* matroska: move timeline code to demux/wm42015-02-171-0/+654
| | | | | Separate from previous commit, because git is bad at tracking file renames when the file contents are also changed.
* matroska: make timeline code independent of MPContextwm42015-02-172-1/+5
|
* player: use a separate context for timeline loader stuffwm42015-02-173-0/+82
| | | | | | | | | | | | | | | | | | | | | | | | Instead of accessing MPContext in player/timeline/*, create a separate context struct, which the timeline loaders fill out. It turns out that there's not much in the way too big MPContext that these need to access. One major PITA is managing (and closing) the set of open demuxers. The problem is that we need a list of all demuxers to make sure no unneeded streams are enabled. This adds a callback to the demuxer_desc struct, with the intention of leaving to to the demuxer to call the right loader, instead of explicitly checking the demuxer type and dispatching manually in common code. I also considered making the timeline part of the demuxer state, but decided against: it's too much of a mess wrt. memory management and threading, and also doesn't make it clear who owns the child demuxers. With the struct timeline decoupled from the demuxer state, it's at least somewhat clear that the child demuxers are independent from the "main" demuxer. The actual changes to player/timeline/* are separated in the following commits, because they're quite verbose. Some artifacts will be removed later as soon as there's only 1 timeline loading mechanism.
* demux: chapters without metadata are allowedwm42015-02-171-2/+3
| | | | Makes some of the following commits slightly simpler. Also fix a typo.
* demux_lavf: hack against hls showing "100%" positionwm42015-02-171-0/+6
| | | | | | | | | | | | | The HLs protocol consists of a "playlist" main file, which mpv downloads and passes to the HLS demuxer. The HLS demuxer actually requests segment files containing media data on its own. The packets read from the demuxer have a source file position set, but it's not from the main file. This leads to a strange effect: as a last fallback, the player will calculate the approximate playback position from the file position/size ratio, and since the main file is tiny, this will always show 100%. Fix this by resetting the packet file position. This doesn't affect the case when HLS actually reports a duration.
* demux: fix dropped subtitle packets with the new stream switchingwm42015-02-142-3/+4
| | | | | | | | | | | | | | If the previous subtitle packet is too far back, and the refresh seek won't pick it up, and the packet never comes again. As a consequence, the refresh mode was never stopped on the subtitle stream, which caused all packets to be discarded. Fix by assuming the file position is monotonically increasing; then it will resume even if a packet _after_ the intended resume point is returned. This introduces a new requirement on how the demuxer behaves. (I'm not sure if mp4 actually satisfies this requirement in all cases.) Fixes a regression introduced by commit f9f2e1cc.
* demux: hack for instant stream switchingwm42015-02-134-4/+87
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This removes the delay when switching audio tracks in mkv or mp4 files. Other formats are not enabled, because it's not clear whether the demuxers fulfill the requirements listed in demux.h. (Many formats definitely do not with libavformat.) Background: The demuxer packet cache buffers a certain amount of packets. This includes only packets from selected streams. We discard packets from other streams for various reasons. This introduces a problem: switching to a different audio track introduces a delay. The delay is as big as the demuxer packet cache buffer, because while the file was read ahead to fill the packet buffer, the process of reading packets also discarded all packets from the previously not selected audio stream. Once the remaining packet buffer has been played, new audio packets are available and you hear audio again. We could probably just not discard packets from unselected streams. But this would require additional memory and CPU resources, and also it's hard to tell when packets from unused streams should be discarded (we don't want to keep them forever; it'd be a memory leak). We could also issue a player hr-seek to the current playback position, which would solve the problem in 1 line of code or so. But this can be rather slow. So what we do in this commit instead is: we just seek back to the position where our current packet buffer starts, and start demuxing from this position again. This way we can get the "past" packets for the newly selected stream. For streams which were already selected the packets are simply discarded until the previous position is reached again. That latter part is the hard part. We really want to skip packets exactly until the position where we left off previously, or we will skip packets or feed packets to the decoder twice. If we assume that the demuxer is deterministic (returns exactly the same packets after a seek to a previous position), then we can try to check whether it's the same packet as the one at the end of the packet buffer. If it is, we know that the packet after it is where we left off last time. Unfortunately, this is not very robust, and maybe it can't be made robust. Currently we use the demux_packet.pos field as unique packet ID - which works fine in some scenarios, but will break in arbitrary ways if the basic requirement to the demuxer (as listed in the demux.h additions) are broken. Thus, this is enabled only for the internal mkv demuxer and the libavformat mp4 demuxer. (libavformat mkv does not work, because the packet positions are not unique. Probably could be fixed upstream, but it's not clear whether it's a bug or a feature.)
* demux_mkv: return unique file positions for all packetswm42015-02-131-4/+5
| | | | | | | | | | | | | Until now, some packets could return the same file position if they were split off from a Matroska-level packet. This was perfectly fine, because the file position isn't used for anything overly important (it uses it to estimate playback position if no other information is available). The following commit will use the demux_packet.pos field as unique ID (as a simplification), so make the demuxer export more finegrained information. Also, the last_filepos field didn't have to be global, at least not anymore.
* demux_mkv: remove the realmedia fieldwm42015-02-051-25/+21
| | | | | | Granted, this doesn't help much with anything, other than the hate- driven desire to remove or at least reduce anything that has to do with RealMedia.
* demux_mkv: reindent, cosmeticswm42015-02-051-97/+79