This commit adds an optional transcript translation feature implemented as
request src Pads.
When requesting a src Pad, the user can specify the translation language code
using Pad properties 'language-code'.
The following properties are defined on the Element:
- 'transcribe-latency': formerly 'latency', defines the expected latency for
the Transcribe webservice.
- 'translate-latency': defines the expected latency for the Translate
webservice.
- 'transcript-lookahead': maximum transcript duration to send to translation
when a transcript is hitting its deadline and no punctuation was found.
When the input and output languages are the same, only the 'transcribe-latency'
is used for the Pad. Otherwise, the resulting latency is the addition of
'transcribe-latency' and 'translate-latency'.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-rs/-/merge_requests/1109>
This helps gather together the details related to the `TranscriberLoop`.
One difference with previous implementation is that the ws `Client` is
build each time the loop is started instead of being reused. With the new
approach, we don't keep the connection open after EOS and we should be
more resistant in case of a connection failure.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-rs/-/merge_requests/1104>
Instead of sending transcription events to the src pad loop, this commit
enqueues the transcribed buffers immediately in the ws loop, then notifies
the src pad loop. The src pad loop is only in charge of dequeuing the buffers.
This should help with upcoming evolutions.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-rs/-/merge_requests/1104>
If creating a playlist or fragment stream fails (disk is full, the
directory is removed, ...), we will currently crash because the signal
handler expects a non-None GIOStream. The actual callback is allowed to
return None values and we handle this in the caller, so let's not have
this restriction on the signal handler.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-rs/-/merge_requests/1093>
for uploaded object default content-type is set to binary/octet-stream,
which is correct.
metadata cannot be used to set content-type and content-disposition as
setting metadata add a prefix x-amz-meta to key
e.g. setting metadate "content-type=video/mp4" actually set value as
x-amz-meta-content-type. So these has to be seaprate property.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-rs/-/merge_requests/1085>
Commit ad3f1cf fixed the name of hlssink child element to be the same
for hlssink2 and hlssink3. However, we rely on element name to return
boolean in case of hlssink3 or None in case of hlssink2 as the return
value of the delete-fragment closure.
Fix this by using the factory name instead of the element name.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-rs/-/merge_requests/1076>
Simplifies state tracking and potentially reduces latency as it's not
necessary to wait until all fragments of an OBU are received.
The last OBU of a TU is marked with the marker flag to allow parsers to
detect this without first seeing the beginning of the next TU.
Also use a simple `Vec` for collecting complete OBUs instead of a
`gst_base::Adapter` as this reduces the number of allocations.
And also handle invalid packets a little bit more gracefully.
Fixes https://gitlab.freedesktop.org/gstreamer/gst-plugins-rs/-/issues/244
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-rs/-/merge_requests/1072>
We dispose of consumer pipelines asynchronously, potentially after the
session objects have been disposed of.
As session objects are the owner of the cc element, it is entirely
possible for the bwe-request signal to get emitted after cc has been
disposed of, as the closure only takes a weak reference to it.
Fix by simply checking if cc is None
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-rs/-/merge_requests/1044>
Without this auto-pluggers such as decodebin or parsebin will be unable to
process AV1 RTP payloads.
Tested with: `videotestsrc num-buffers=50 ! videoconvert ! av1enc ! av1parse ! rtpav1pay ! queue ! decodebin3 ! videoconvert ! queue ! autovideosink`
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-rs/-/merge_requests/1034>
Right now the code manually pieces together the components
in a String for efficiency. When credentials contain special
characters this can result in invalid URLs, so do it the proper
way (with Url::parse + format) to make sure components are escaped
as needed.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-rs/-/merge_requests/949>
WHIP endpoint providers like Cloudflare do not support Trickle ICE
and need candidates to be send along with the initial offer. Instead
of sending the offer in create-offer promise, send it once the ICE
candidates have been gathered.
While at it add properties to set STUN and TURN server along with the
ICE transport policy as at least when testing the Cloudflare WHIP
endpoint seems unreachable without it. This has also been observed
with Cloudflare provided demos.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-rs/-/merge_requests/949>
This implements WHEP specification based on
https://datatracker.ietf.org/doc/html/draft-murillo-whep-00
and has been tested with Cloudflare.
Server offers are likely to be removed from the WHEP specification
in upcoming revisions, to avoid compatibility issues. None of the
commercial services implementing WHEP support server initiated offers.
So we only support client side initiated offers.
Follows session setup and tear down as covered in Figure 1, Section 3
of the specification.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-rs/-/merge_requests/949>
error[E0425]: cannot find value `LIBRARY_NAME` in this scope
--> net/ndi/src/ndisys.rs:336:23
|
336 | path.push(LIBRARY_NAME);
| ^^^^^^^^^^^^ not found in this scope
error[E0425]: cannot find value `LIBRARY_NAME` in this scope
--> net/ndi/src/ndisys.rs:339:33
|
339 | path::PathBuf::from(LIBRARY_NAME)
| ^^^^^^^^^^^^ not found in this scope
Commit 24b7cfc8 applied changes related to nullability as declared
by gir. One consequence was that some functions signature ended up
requiring users to pass `Some(val)` when they could use `val`
before.
This commit applies changes on `gstreamer-rs` which, will honoring
the nullability stil allow users to pass `val` for the few affected
functions.
This commit also fixes the signature for `Element::request_new_pad`
which was updated upstream.
And use a `Vec` plus offset for consuming partial byte buffers.
Removing the first element from a `Vec` repeatedly is not very cheap.
Also simplify calculation of the current packet by removing a mostly
unused type and keeping track of the calculations always locally instead
of sometimes storing it in the element state.
- NDI HX Camera Android in the past used 1ns instead of 100ns as unit
for timecodes/timestamps.
- NDI HX Camera iOS uses 0 for all timecodes and the same non-zero
value for all audio timestamps
Detect such situations and try to compensate for them. Also add a new
"auto" timestamping mode that prefers to use timecodes and otherwise
falls back to timestamps or receive times.
Fixes https://github.com/teltek/gst-plugin-ndi/issues/79
Audio/video are in practice not always from the same clock and can have
different behaviours with regards to clock rate and jitter. Handling
them separately generally gives better results for the timestamps output
by the source element.
This is no longer available as this could lead to building a defined
value in Rust which could be interpreted as undefined in C due to
the sentinel `u64::MAX` for `None`.
Use the constants (e.g. `ONE`, `K`, `M`, ...) and operations to build
a value and deref (`*`) to get the quantity as an integer.
Always first try draining queued data in the loop and only start waiting
if there's nothing to drain right now. Otherwise data might have to be
drained right now but we still wait and nothing is ever waking up the
source pad task again.
Also make sure to not wait multiple times on the same gst::ClockId but
instead unset it after waiting on it and no new one was scheduled in the
meantime. Future waits on the same ClockId will immediately return and
instead we should wait on the condvar if no new ClockId is available.
When call_timeout is triggered, request will fail
irrespective of the retry setting. call_timeout define
max time request can take along with retry.
It can be solved by either setting call_timeout to
retry * call_attempt_timeout or not setting the call_timeout.
As per thread call_attempt and rety setting is enough.
https://github.com/awslabs/aws-sdk-rust/issues/558
Fixes https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/issues/1410
Created a new plugin 'webrtchttp' to implement all the
WebRTC HTTP protocols under /net/webrtc-http directory.
WhipSink wraps around 'webrtcbin' with HTTP capabilites
to exchange SDP offer/answer so an ICE/DTLS session can
be established between the encoder/media producer (WHIP client)
and the broadcasting ingestion endpoint (Media Server).
Once the ICE/DTLS session is set up, the media will
flow unidirectionally from the WHIP client to the
broadcasting ingestion endpoint (Media Server).
Spec:
https://www.ietf.org/archive/id/draft-ietf-wish-whip-04.html
The encoding of ONVIF metadata is always UTF-8. ONVIF metadata may
or may not be encoded with gzip, but we don't see a use case for
transporting compressed ONVIF metadata between elements for now.
The aggregator was consuming meta buffers too greedily, causing
potential interleaving starvation upstream. Refactor to consume
media and meta buffers synchronously
Also expect parsed=true metadata caps (requiring an upstream
onvifmetadataparse element).
warning: you are deriving `PartialEq` and can implement `Eq`
--> net/raptorq/src/fecscheme.rs:13:24
|
13 | #[derive(Clone, Debug, PartialEq)]
| ^^^^^^^^^ help: consider deriving `Eq` as well: `PartialEq, Eq`
|
= note: `#[warn(clippy::derive_partial_eq_without_eq)]` on by default
= help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#derive_partial_eq_without_eq
warning: you are deriving `PartialEq` and can implement `Eq`
--> net/raptorq/src/fecscheme.rs:38:24
|
38 | #[derive(Clone, Debug, PartialEq)]
| ^^^^^^^^^ help: consider deriving `Eq` as well: `PartialEq, Eq`
|
= help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#derive_partial_eq_without_eq
A regression was introduced during the migration to AWS SDK. One used
to be able to provide credentials in multiple ways with the earlier
Rusoto ChainProvider (config file / environment variables). Now one
has to explicitly set the properties.
Use the DefaultCredentialsChain from AWS SDK to restore the previous
functionality.
See
https://docs.rs/aws-config/0.46.0/aws_config/default_provider/credentials/struct.DefaultCredentialsChain.html.
Allow specifying an endpoint to be used for S3 requests. This makes
it possible to use integrations providing object storage based on S3
API like MinIO.
When the endpoint-uri property is specified, the endpoint resolver to
use will be overridden when making S3 requests.
Ideally, when player encounter PLAYLIST-TYPE is VOD, player should
not reload the playlist. For playlist-type=vod, initially we put
PLAYLIST-TYPE=EVENT, and later change it to VOD, which confuse some
players so we shold put ENDLIST here.
In any case putting ENDLIST is right thing to do to indicate no new
segment will be added to playlist.
in current implementation EXT-X-ENDLIST is never set for any playlist-type.
After calling playlist.stop(), during write_final_playlist() is called.
in final playlist write, segment is not added but update_playlist is called.
and update_playlist reset end_list again, so plugin never put ENDLIST.
STS provide temporary credentials to access AWS resource. Temporary
credentials include, AccessKeyId, SecretAccessKey and SessionToken.
With session-token property, element will be able to use temporary
credentials. When session-token is not set, element can use long
term credentials.
Keeping the upload-part-retry-duration & complete-upload-retry-duration
properties changes the semantics in comparison to their usage in
rusotos3sink. Deprecate these two properties and add a warning while
making them noop.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-rs/-/merge_requests/759>
`PadTemplate::caps()` returns a reference to the caps now instead of a
new strong reference, so keeping the template in scope as long as the
caps reference is required.
warning: this expression borrows a value the compiler would automatically borrow
--> net/reqwest/tests/reqwesthttpsrc.rs:126:56
|
126 | async move { Ok::<_, hyper::Error>((&mut *http_func.lock().unwrap())(req)) }
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ help: change this to: `(*http_func.lock().unwrap())`
|
= note: `#[warn(clippy::needless_borrow)]` on by default
= help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#needless_borrow
This implements a default timeout and retry duration for the remaining
S3 requests that were still able to be blocked indefinitely. There are 3
classes of operations: multipart upload creation/abort (should not take
too long), uploads (duration depends on part size), multipart upload
completion (can take several minutes according to documentation).
We currently only expose the part upload times as configurable, and hard
code the rest. If it seems sensible, we can expose the other two sets of
parameters as well.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-rs/-/merge_requests/690>
Previously, the actual reading from the streaming body of a GetObject
request was not within the same timeout/retry path as the dispatch of
the HTTP request itself. We consolidate these two into a single async
block and create a sum type to encapsulate the rusoto and std library
error paths within that future.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-rs/-/merge_requests/690>
Have seen a few times where machines that are in perfect time sync with a good source the requests fail with `RequestExpired` errors.
https://docs.aws.amazon.com/transcribe/latest/dg/CommonErrors.html
While not perfect, bumping to five minutes gives more a chance that the signed requests to start streaming won't be expired.
Rusoto does not implement timeouts or retries for any of its HTTP
requests. This is particularly problematic for the part upload stage of
multipart uploads, as a blip in the network could cause part uploads to
freeze for a long duration and eventually bail.
To avoid this, for part uploads, we add (a) (configurable) timeouts for
each request, and (b) retries with exponential backoff, upto a
configurable duration.
It is not clear if/how we want to do this for other types of requests.
The creation of a multipart upload should be relatively quick, but the
completion of an upload might take several minutes, so there is no
one-size-fits-all configuration, necessarily.
It would likely make more sense to implement some sensible hard-coded
defaults for these other sorts of requests.
A multipart upload should either be completed or aborted on error. In
the current state of things, a multipart upload would neither be
completed nor aborted, putting the onus on an external entity to take
care of finishing incomplete uploads or relying on a sane bucket
life cycle policy configured to abort incomplete multipart uploads.
An incomplete multipart upload still contributes to the storage costs as
long as it exists.
We introduce a property here to allow the user to select either aborting
or completing multipart uploads on error. Aborting the upload causes
whole of data to be discarded and the same upload ID is not usable for
uploading more parts to the same.
Completing an incomplete multipart upload can be useful in situations
like having a streamable MP4 where one might want to complete the upload
and have part of the data which was uploaded be preserved.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-rs/-/merge_requests/618>
The design of the element is based on the assumption that when
receiving a partial result, the following result will contain
at least as many items as there were stable items in the previous
result.
This patch adds a sanity check to make sure our "partial index"
isn't larger than the new received result, and errors out otherwise.
partial_index will eventually be reset to 0 once we receive a
new non-partial result.
For the region property this would be provided as
`region-name+https://region.end/point`
while for the URI this unfortunately has to be base32 encoded to allow
usage as the host part of the URI.
The default behavior for the transcriber is to output text buffers
synchronized with the input stream, introducing a configurable
latency.
For use cases where synchronization is not crucial, but latency
is, the lateness property can be used instead of or in combination
with the latency property, in order to introduce a configurable
offset with the input stream.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-rs/-/merge_requests/534>
As awstranscriber might in theory push out gap events without
any flow of input data, it needs to send its mandatory events
(stream-start, caps, segment) independently.
In addition, track a start time and use it to offset the 0-based
timestamps returned by AWS in order to output buffers timestamped
in the running-time domain, and perform item timing adjustment
only when dequeuing, instead of when queuing.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-rs/-/merge_requests/525>
<https://aws.amazon.com/blogs/machine-learning/amazon-transcribe-now-supports-partial-results-stabilization-for-streaming-audio/>
Amazon seem to have realized the previous iteration of their API
made it difficult to identify items from one result to the next,
which made the element much more complicated than it should have
been. With that new "stability" option, we can enqueue items as
soon as they stabilize, and simply rely on the current index in
the transcript to output them exactly once.
This also means the "use_partial_results" is now useless, as there
will be no difference in accuracy between a non-partial result and
and of its stable items that might have been pushed from previous
partial versions of the result.
The property is removed, instead a new option is exposed to let
users control how fast results should stabilize.
This greatly simplifies the code, and also improves the output as
punctuation doesn't need to be randomly discarded anymore.
Implemented analogously to souphttpsrc for compatibility. Proxy
prevents sharing the client between element instances.
Change-Id: I50d676fd55f0e1d7051d8cd7d5922b7be4f0c6e8
With the URI handler interface implemented, we can drop the old method
of specifying bucket, key and region. This also brings it in line with
how it is for s3src.
AWS offers the option of creating "vocabularies", lists of words
that are likely to be encountered. Those can be created through
the AWS console, and are given a name. That name can then be
specified when starting a transcription job.
cargo-c will produce a pkg-config file making it easier to statically
link plugins.
Also add 'static' features for plugins depending on < 1.14 as this is the
minimal required version to use static linking because of ABI changes in
core.
This mutex is actually only ever used from a single thread, so use
AtomicRefCell instead. It provides the guarantees of a mutex but panics
instead of blocking.
There is no way to dynamically ask Cargo to build static or dynamic lib
so we have to build both and pick the one we care when doing the meson
processing.
Fix#88
Only two uses of unsafely setting the pad functions is left:
- fallbacksrc for overriding the chain function of the proxy pad of a
ghost pad
- threadshare for overriding the pad functions after creationg, which
probably needs some fixing at some point
Since those are using the clock for sync, they need to also
provide a clock for good measure. The reason is that even if
downstream elements provide a clock, we don't want to have
that clock selected because it might not be running yet.
This moves to Rusoto 0.43, which has moved from futures to async/.await.
As a result, we implement a utility function to convert the
async/streaming bits to blocking operations backed by a tokio runtime.
In the process, we also need to restructure s3sink a little, so that the
client is now part of the started state (like it is for s3src). This is
a better model than a separate client, as it reflects the condition that
the client is only available in the started state.
This should start making navigating the tree a little easier to start
with, and we can then move to allowing building specific groups of
plugins as well.
The plugins are moved into the following hierarchy:
audio
/ gst-plugin-audiofx
/ gst-plugin-claxon
/ gst-plugin-csound
/ gst-plugin-lewton
generic
/ gst-plugin-file
/ gst-plugin-sodium
/ gst-plugin-threadshare
net
/ gst-plugin-reqwest
/ gst-plugin-rusoto
utils
/ gst-plugin-fallbackswitch
/ gst-plugin-togglerecord
video
/ gst-plugin-cdg
/ gst-plugin-closedcaption
/ gst-plugin-dav1d
/ gst-plugin-flv
/ gst-plugin-gif
/ gst-plugin-rav1e
gst-plugin-tutorial
gst-plugin-version-helper
