Before this commit, SRT demuxing and display actually happened to work
on Libav. But it was using the libavcodec srt converter (which is
essentially unmaintained in Libav), and timing postprocessing didn't
work. For some background explanations see sd_lavf_srt.c.
Until now, timing and charset recoding postprocessing was applied on
packets as they were output by the demuxer, and then passed to the
decoders. Make it so that postprocessing can happen after some decoders
in special situations.
Old code used to use libass' recoding feature, which is a copy of the
old MPlayer code. We dropped that a few commits ago. Unfortunately,
this made it impossible to load some subtitle files, like UTF-16 files.
Make .ass loading respect -subcp again. We do this by recoding the
probe buffer to UTF-8, and then trying to load it normally. (Yep.)
Since UTF-16 in particular will effectively half the probe buffer size,
double the probe size.
This code was once part of subreader.c, then traveled to libass, and now
made its way back to the fork of the fork of the original code, MPlayer.
It works pretty much the same as subreader.c, except that we have to
concatenate some packets to do auto-detection. This is rather annoying,
but for all we know the actual source file could be a binary format.
Unlike subreader.c, the iconv context is reopened on each packet. This
is simpler, and with respect to multibyte encodings, more robust.
Reopening is probably not a very fast, but I suspect subtitle charset
conversion is not an operation that happens often or has to be fast.
Also, this auto-detection is disabled for microdvd - this is the only
format we know that has binary data in its packets, but is actually
decoded to text. FFmpeg doesn't really allow us to solve this properly,
because a) the input packets can be binary, and b) the output will be
checked whether it's UTF-8, and if it's not, the output is thrown away
and an error message is printed. We could just recode the decoded
subtitles before sd_ass if it weren't for that.
demux_libass.c allows us to make subtitle format detection part of the
normal file loading process. libass has no probe function, but trying to
load the start of a file (the first 4 KB) is good enough. Hope that
libass can even handle random binary input gracefully without printing
stupid log messages, and that the libass parser doesn't accept too many
non-ASS files as input.
This doesn't handle the -subcp option correctly yet. This will be fixed
later.
The default correct-pts mode depended on which demuxer was opened last.
Often this is the subtitle demuxer. The correct-pts mode should be
decided on the demuxer for video instead.
subreader.c (before this commit renamed to demux_subreader.c) was
special cased to the -sub option. The plan is using the normal demuxer
codepath for all subtitle formats (so we can prefer libavformat demuxers
for most formats).
There are some subtle changes. The probe size is restricted to 32 KB
(instead of unlimitted + giving up after 100 lines of input). For
formats like MicroDVD, the video FPS isn't used anymore, because it's
not available on the subtitle demuxer level. Instead, hardcode it to
23.976 FPS (libavformat seems to do the same). The user can probably
still use -sub-fps to fix the timing. Checking the file extension for
".utf"/".utf8"/".utf-8" is simply removed (seems worthless, was in the
way, and I've never seen this anywhere).
Simpler, reduces the amount of copying.
We still have to malloc+memcpy the probe buffer though, because padding
with FF_INPUT_BUFFER_PADDING_SIZE is required by libavformat.
stream_read_unbuffered() can sometimes return negative values on error.
Change that to return 0 - the negative values are nowhere used anyway.
If distinguishing errors and EOF is really needed, a flag could be added
instead.
This also fixes the stream_read_partial() call in cache.c, which assumes
the return values is always >= 0.
Actually check the newly added text for whitespace, and not the
uninitialized buffer after it. Also, if an even is only whitespace,
don't add it at all.
sd_ass contains some code that treats subtitle events with duration 0
specially, and adjust their duration so that they will disappear with
the next event.
This is most likely not needed anymore. Some subtitle formats allow
omitting the duration so that the event is visible until the next one,
but both subreader.c as well as libavformat subtitle demuxers already
handle this.
Subtitles embedded in mp4 files (movtext) used to trigger this code. But
these files appear to export subtitle duration correctly (at least
libavcodec's movtext decoder is using this assumption). Since commit
6dbedd2 changed demux_lavf to actually copy the packet duration field,
the code removed with this commit isn't needed anymore for correct
display of movtext subtitles. (The change in sd_movtext is for dropping
empty subtitle events, which would now be "displayed" - libavcodec does
the same.)
On the other hand, this code incorrectly displayed hidden events in .srt
subtitles. See for example the first event in SubRip_capability_tester.srt
(part of FFmpeg's FATE). These intentionally have a duration of 0, and
should not be displayed. (As of with this commit, they are still
displayed in external .srt subs because of subreader.c hacks.)
However, we can't be 100% sure that this code is really unneeded, so
just comment the code. Hopefully it can be removed if there are no
regressions after some weeks or months.
Currently, we are filtering libavformat style ASS packets by checking
whether they are prefixed "Dialogue: ". Unfortunately, comment packets
are demuxed too. These start with "Comment: ", so they are not caught.
Change the filtering, and use the codec ID instead. libavformat uses
"ssa" as codec ID for ASS subtitles, while mpv uses "ass". Also, at
least FFmpeg will change the ASS packet format to the same format mpv
and Matroska use, and identify these with "ass" as codec ID, so this is
works out nicely.
Some of this (fixing timing) is now done in dec_sub.c (although it's
not active for subreader.c code yet - this will be fixed when
subreader.c subs are read through a demuxer wrapper).
Another reason to remove this is that this code doesn't do much good
anymore. libass does handle overlap, and trying to fold overlapping
lines into single subtitle events will prevent libass from handling
this properly.
This fixes the -subfps option (which unfortunately is still useful),
and fixes minor annoying timing errors (which unfortunately still
happen).
Note that none of these affect ASS or image subtitles. ASS is specially
handled: libass loads subtitles as ASS_Track. There are no actual
packets passed around, and sd_ass just uses the ASS_Track.
Disable the --sub-no-text-pp option. It's misleading now and always was
completely useless.
If a subtitle is external, read it completely and add all subtitle
events in advance when the subtitle track is selected. This is done
for text subtitles only. (Note that subreader.c and subtitles loaded
with libass are different and don't have anything to do with this
commit.)
Seems like a completely unnecessary complication. Instead, always add a
1 byte padding (could be extended if a caller needs it), and clear it.
Also add some documentation. There was some, but it was outdated and
incomplete.
This is useless on the cache side. The sector is needed only to deal
with stream implementations which are not byte addressable, and the
cache is always byte addressable.
Also set a default read_chunk value. (This value is never used unless
you chain multiple caches, but it's cleaner.)
DVD and bluray packet streams carry (essentially) random timestamps,
which don't start at 0, can wrap, etc. libdvdread and libbluray provide
a linear timestamp additionally. This timestamp can be retrieved with
STREAM_CTRL_GET_CURRENT_TIME.
The problem is that this timestamp is bound to the current raw file
position, and the stream cache can be ahead of playback by an arbitrary
amount. This is a big problem for the user, because the displayed
playback time and actual time don't match (depending on cache size),
and relative seeking is broken completely.
Attempt to fix this by saving the linear timestamp all N bytes (where
N = BYTE_META_CHUNK_SIZE = 16 KB). This is a rather crappy hack, but
also very effective.
A proper solution would probably try to offset the playback time with
the packet PTS, but that would require at least knowing how the PTS can
wrap (e.g. how many bits is the PTS comprised of, and what are the
maximum and reset values). Another solution would be putting the cache
between libdvdread and the filesystem/DVD device, but that can't be done
currently. (Also isn't that the operating system's responsibility?)
This was probably done this way to ensure that after a successful seek,
the reported stream position is the same as the requested seek position.
But it doesn't make too much sense, since both stream->pos and the
stream implementation's internal position will go out of sync.
The stream EOF flag should only be set when trying to read past the end
of the file (relatively similar to unix files). Always clear the EOF
flag on seeking. Trying to set it "properly" (depending whether data is
available at seek destination or not) might be an ok idea, but would
require attention to too many special cases. I suspect before this
commit (and in MPlayer etc. too), the EOF flag wasn't handled
consistently when the stream position was at the end of the file.
Fix one special case in ebml.c and stream_skip(): this function couldn't
distinguish between at-EOF and past-EOF either.
EOF should be set when reading more data fails. The stream
implementations have nothing to say here and should behave correctly
when trying to read when EOF was actually read.
Even when seeking, a correct EOF flag should be guaranteed. stream_seek()
(or actually stream_seek_long()) calls stream_fill_buffer() at least
once, which also updates the EOF flag.
This function was called in various places. Most time, it was used
before a seek. In other cases, the purpose was apparently resetting
the EOF flag. As far as I can see, this makes no sense anymore. At
least the stream_reset() calls paired with stream_seek() are completely
pointless. A seek will either seek inside the buffer (and reset the
EOF flag), or do an actual seek and reset all state.
This happens with something like "mpv https://www.youtube.com/watch".
The URL is obviously not valid, but the stream layer tries to reconnect.
This commit at least allows to use the terminal to abort gracefully.
(Other than killing the process.)
Basically rewrite all the code supporting the cache (i.e. anything other
than the ringbuffer logic). The underlying design is untouched.
Note that the old cache2.c (on which this code is based) already had a
threading implementation. This was mostly unused on Linux, and had some
problems, such as using shared volatile variables for communication and
uninterruptible timeouts, instead of using locks for synchronization.
This commit does use proper locking, while still retaining the way the
old cache worked. It's basically a big refactor.
Simplify the code too. Since we don't need to copy stream ctrl args
anymore (we're always guaranteed a shared address space now), lots of
annoying code just goes away. Likewise, we don't need to care about
sector sizes. The cache uses the high-level stream API to read from
other streams, and sector sizes are handled transparently.
demux_lavf probes up to 2 MB of data in the worst case. When the ffmpeg
demuxer is actually opened, the stream is seeked back to 0, and the
previously read data is thrown away.
This wasn't a problem for playback of local files, but it's less than
ideal for playing from slow media (like web streams), and breaks
completely if the media is not seekable (pipes, some web streams).
This new function is intended to allow fixing this. demux_lavf will use
it to put the read probe data back into the buffer.
The simplest way of implementing this function is by making it
transparently extend the normal stream buffer. This makes sure no
existing code is broken by new weird special cases. For simplicity
and to avoid possible performance loss due to extra dereferencing
when accessing the buffer, we just extend the static buffer from
8 KB to 2 MB. Normally, most of these 2 MB will stay uncommitted, so
there's no associated waste of memory. If demux_lavf really reads all
2 MB, the memory will be committed and stay unused, though.
Before this commit, the cache was franken-hacked on top of the stream
API. You had to use special functions (like cache_stream_fill_buffer()
instead of stream_fill_buffer()), which would access the stream in a
cached manner.
The whole idea about the previous design was that the cache runs in a
thread or in a forked process, while the cache awa functions made sure
the stream instance looked consistent to the user. If you used the
normal functions instead of the special ones while the cache was
running, you were out of luck.
Make it a bit more reasonable by turning the cache into a stream on its
own. This makes it behave exactly like a normal stream. The stream
callbacks call into the original (uncached) stream to do work. No
special cache functions or redirections are needed. The only different
thing about cache streams is that they are created by special functions,
instead of being part of the auto_open_streams[] array.
To make things simpler, remove the threading implementation, which was
messed into the code. The threading code could perhaps be kept, but I
don't really want to have to worry about this special case. A proper
threaded implementation will be added later.
Remove the cache enabling code from stream_radio.c. Since enabling the
cache involves replacing the old stream with a new one, the code as-is
can't be kept. It would be easily possible to enable the cache by
requesting a cache size (which is also much simpler). But nobody uses
stream_radio.c and I can't even test this thing, and the cache is
probably not really important for it either.
