Data is appended in fill_buffer() when there is sufficient space left
and the data pointer only reset when needed.
Previously the data pointer was more often reset, loosing more seekback
space than otherwise needed.
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
The sample from ticket #2691 currently does not trigger "Consider increasing
the value for analzeduration and probesize" because the audio streams are
only added after calling estimate_timings(). Attached patch moves the message
below this function call.
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
Allow applications to request reading streamcast metadata. This uses
AVOptions as API, and requires the application to explicitly request
and read metadata. Metadata can be updated mid-stream; if an
application is interested in that, it has to poll for the data by
reading the "icy_metadata_packet" option in regular intervals.
There doesn't seem to be a nice way to transfer the metadata in a nicer
way. Converting the metadata to ID3v2 tags might be a nice idea, but
the libavformat mp3 demuxer doesn't seem to read these tags mid-stream,
and even then we couldn't guarantee that tags are not inserted in the
middle of mp3 packet data.
This commit provides the minimum to enable applications to retrieve
this information at all.
Signed-off-by: Stefano Sabatini <stefasab@gmail.com>
The code that copies any extradata from the MXFDescriptor to the codec does
not set the size, which it should otherwise the copied data is useless.
Attached it a patch to correct this.
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
This is part of normal operation with some formats. A warning should
indicate that something is wrong, and the documentation for AV_LOG_WARNING
says: "Something somehow does not look correct."
Since this message is most likely useful for debugging only, raise the
log level accordingly.
Plus-one-by: Stefano Sabatini <stefasab@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
WebVTT subtitle tracks have four kinds. Certain downstream muxers
(such as for WebM) need to know which WebVTT kind this is, in order to
specify the codec id of the output track.
A new private input option, "-kind", has been added to the WebVTT
demuxer. It accepts as a value any of "subtitles" (the default),
"captions", "descriptions", and "metadata". The kind option value is
used to assign a value to the stream disposition flag, to which four
new values have been added, corresponding the four WebVTT kinds.
Signed-off-by: Matthew Heaney <matthewjheaney@google.com>
Signed-off-by: Clément Bœsch <ubitux@gmail.com>
There are 4 separate WebVTT text track kinds: subtitles (the default
if not otherwise specified), captions, descriptions, and metadata.
The WebM muxer needs to know which WebVTT text track kind this is, in
order to synthesize the correct track type and codec id.
To allow a demuxer to indicate the text track kind of the input, a new
set of AV_DISPOSITION flag values has been added, corresponding to
each of the non-default text track kind values.
Michael Niedermayer
Stefano Sabatini
Matthieu Bouron
Zhang Rui
Paul B Mahol
Tudor Suciu
wm4
Carl Eugen Hoyos
Anton Khirnov
Reuben Martin
Timothy Gu
Nigel Touati-Evans
Luca Barbato
Ronald S. Bultje
Matthew Heaney