Original commit message from CVS:
* gst/id3demux/gstid3demux.c: (gst_id3demux_reset),
(gst_id3demux_send_new_segment), (gst_id3demux_chain),
(gst_id3demux_sink_event):
* gst/id3demux/gstid3demux.h:
* gst/apetag/gsttagdemux.c: (gst_tag_demux_reset),
(gst_tag_demux_chain), (gst_tag_demux_sink_event),
(gst_tag_demux_send_new_segment):
Handle and adjust new-segment events so that downstream really
sees a stream with the tag pieces stripped off the front and back.
Fixes strangeness in seeking when mp3 decoders use the new-segment
byte position to estimate their current playback position timestamp
and then the arriving buffers don't match up.
Original commit message from CVS:
* gst/apetag/gsttagdemux.c: (gst_tag_demux_get_type):
Rename registered type in preparation of GstTagDemux moving to
-base at some point in the future.
Original commit message from CVS:
* gst/apetag/gsttagdemux.c: (gst_tag_demux_srcpad_event):
Fix handling of -1 values for start and stop values when seeking,
and SEEK_CUR+SEEK_END here as well.
Original commit message from CVS:
* gst/apetag/gsttagdemux.c: (gst_tag_demux_read_range):
* gst/id3demux/gstid3demux.c: (gst_id3demux_read_range):
Set correct caps on outgoing pulled buffers, or things blow up
after recent core changes.
Original commit message from CVS:
* gst/apetag/gsttagdemux.c: (gst_tag_demux_chain_parse_tag),
(gst_tag_demux_chain):
* gst/id3demux/gstid3demux.c: (gst_id3demux_chain):
In streaming mode, if the first buffer we get doesn't have an
offset, fix it up to be 0, otherwise trimming won't work later on
and we'll be typefinding application/x-id3, which may result in
decodebin plugging an endless number of id3demux elements as a
consequence. Fixes#385031.
Original commit message from CVS:
Patch by: Sebastian Dröge <mail at slomosnail de>
* gst/apetag/gstapedemux.c: (ape_demux_parse_tags):
We need to be able to read and parse any possible floating point string
format ("1,234" or "1.234") irrespective of the current locale. g_strod()
will parse the former only in certain locales though, so we really need
to canonicalise the separator to '.' and then use g_ascii_strtod() to
make sure we can parse either version at all times.
Fixes#382982 for real.
Original commit message from CVS:
* gst/apetag/gstapedemux.c: (ape_demux_parse_tags):
Use g_strtod() instead of sscanf to parse doubles, so that it will
try parsing in the C locale if the current locale fails.
Fixes: #382982
Patch by: Sebastian Dröge <mail at slomosnail de >
Original commit message from CVS:
* gst/apetag/gsttagdemux.c: (gst_tag_demux_add_srcpad):
* gst/icydemux/gsticydemux.c: (gst_icydemux_add_srcpad):
* gst/id3demux/gstid3demux.c: (gst_id3demux_add_srcpad):
Activate pad before adding it to the already-running element.
* tests/check/elements/icydemux.c: (icydemux_found_pad):
Activate newly-created pad too.
Original commit message from CVS:
* gst/apetag/gstapedemux.c: (ape_demux_parse_tags):
Extract disc/album/medium number and count and try harder
to extract track number/count.
Original commit message from CVS:
* gst/apetag/gsttagdemux.c: (gst_tag_demux_chain_parse_tag):
* gst/id3demux/gstid3demux.c: (gst_id3demux_chain):
Don't interpret a first buffer with an offset of NONE as
'from the middle of the stream', but only a first buffer
that has a valid buffer offset that's non-zero (see #345449).
Original commit message from CVS:
* gst/apetag/gsttagdemux.c: (gst_tag_demux_trim_buffer),
(gst_tag_demux_read_range):
* gst/id3demux/gstid3demux.c: (gst_id3demux_trim_buffer),
(gst_id3demux_read_range):
Don't return FLOW_UNEXPECTED when a buffer is before
the start of the stream (which might happen with
large ID3v2 tags if the tag reading was done pullrange
based and we then switched to push mode later on).
Fixes regression introduced by commit from June 29th.
Original commit message from CVS:
* gst/apetag/gsttagdemux.c: (gst_tag_demux_chain):
* gst/id3demux/gstid3demux.c: (gst_id3demux_chain):
Return FLOW_UNEXPECTED when at the end of the file, not
FLOW_ERROR. Fixes 'internal stream error' errors that
would sometimes occur in totem when scrubbing to the
end of an ID3v1 tagged mp3 file.
Original commit message from CVS:
* gst/id3demux/gstid3demux.c: (gst_id3demux_trim_buffer),
(gst_id3demux_read_range):
Make sure we don't return GST_FLOW_OK with a NULL buffer in
certain cases where a read beyond the end of the file is
requested. Fixes#345930.
* gst/apetag/gsttagdemux.c: (gst_tag_demux_trim_buffer),
(gst_tag_demux_read_range):
Fix same issue here as well.
Original commit message from CVS:
* gst/apetag/gsttagdemux.c: (gst_tag_demux_get_upstream_size):
* gst/id3demux/gstid3demux.c: (id3demux_get_upstream_size):
Use gst_pad_query_peer_duration() utility function here.
Original commit message from CVS:
Patch by: Sebastian Dröge <slomo at ubuntu dot com>
* gst/apetag/gstapedemux.c: (ape_demux_get_gst_tag_from_tag),
(ape_demux_parse_tags):
Some clean-ups and additions: map APE 'file' tag to
GST_TAG_LOCATION (#343123); add support for extracting
the track count and clean up parsing a bit (#343127).
Original commit message from CVS:
* gst/apetag/gsttagdemux.c: (gst_tag_demux_init),
(gst_tag_demux_sink_event):
* gst/id3demux/gstid3demux.c: (gst_id3demux_init),
(gst_id3demux_sink_event):
Post an error message when we get an EOS event and were not
able to find out the type of stream.
* tests/check/elements/id3v2mux.c: (fill_mp3_buffer), (got_buffer),
(test_taglib_id3mux_with_tags):
Decrease num-buffers to 16 per iteration again, otherwise the
many memcpy()s and reallocations in the test will hammer slow
CPUs completely and make the test timeout.
Original commit message from CVS:
* gst/apetag/gsttagdemux.c: (gst_tag_demux_chain):
* gst/id3demux/gstid3demux.c: (gst_id3demux_chain):
Don't attempt typefinding on too-short buffers that have been
completely trimmed away.
* gst/id3demux/id3tags.c: (id3demux_read_id3v2_tag):
Improve the debug output
Original commit message from CVS:
* gst/apetag/gsttagdemux.c: (gst_tag_demux_chain):
Ensure that we set caps on the buffers we pass.
* gst/id3demux/gstid3demux.c: (gst_id3demux_chain),
(gst_id3demux_sink_activate):
Ensure that we set caps on the buffers we pass.
Use STREAM, TYPE_NOT_FOUND as the error class when
typefinding fails.
Original commit message from CVS:
* configure.ac:
Bump requirements to current core and -base CVS
(core for new typefind helper API, and -base for the
WAVFORMATEX support that was added to libgstriff and
is needed by wavparse).
* gst/apetag/Makefile.am:
* gst/apetag/gsttagdemux.c: (gst_tag_demux_chain),
(gst_tag_demux_sink_activate):
Use new typefind helpers for typefinding instead of our
home-grown stuff; also, do typefinding in pull-mode if
upstream supports that.
Original commit message from CVS:
* gst/apetag/gsttagdemux.c: (gst_tag_demux_get_upstream_size),
(gst_tag_demux_do_typefind):
... and fix the very same leaks in GstTagDemux.
