Otherwise it is global and it retains a reference to the `parent`
WpSpaDevice object until this function is called again, which prevents
some camera nodes - in some cases - from being destroyed
Fixes#640
Enhance the parsing logic to consider more than one device number. libcamera
devices some times use up multiple V4L2 devices, in this case libcamera should
be given a chance to enumerate the device.
Fixes#623
This patch makes the check more robust when detecting main filter nodes. This is
because some filters might append '/Internal' to their main node media class.
The direction check has also been improved to work with Video filters.
event:set_data() coerces Lua tables to properties, so that keys and
values become strings. Take into account that profile.index is a string
due to this.
Fixes WP setting the profile, even though it is already the active one.
Using parseBool is actually more broken if the property is already parsed as a
boolean, and causes the node to remain unhidden even when hide-parent = true.
Revert 2a45842169 to fix this behaviour.
Signed-off-by: James Calligeros <jcalligeros99@gmail.com>
Using the new Conf() constructor, we can load and parse a filter graph
from a file on disk. This is useful when, for example, maintaining a
large database of filter graphs. It also keeps wireplumber.conf.d free
from clutter.
Signed-off-by: James Calligeros <jcalligeros99@gmail.com>
The cam_data structure stores a reference to the "parent" WpSpaDevice
and doesn't allow it to be destroyed when the monitor detects that
the device is no longer present. Clear it right after pushing the event
to make sure there's no dangling reference left around
Fixes: #627
We apparently store things in the <media-class>:<key>:<value>
format, while pipewire-pulse expects <media-class>.<key>:<value>
and this detail was missed in the latest refactoring.
It also seems that I forgot to ensure that restoring this metadata
key worked. The hook was there, but not registered.
Remove the hook and directly populate the metadata object when
it is activated, as it seems simpler.
Fixes: #604
The existence of props["hide-parent"] is truthy. Parse the boolean explicitly
to get the correct logic.
Signed-off-by: James Calligeros <jcalligeros99@gmail.com>
This allows DSP rules to be specified in a wireplumber.conf.d file under
the node.software-dsp.rules section. Properties are specified in the
software-dsp action.
Signed-off-by: James Calligeros <jcalligeros99@gmail.com>
As this DSP "policy" is intended to run on the creation/deletion
of a node, move it into the node subdir.
Signed-off-by: James Calligeros <jcalligeros99@gmail.com>
Prevously, we were only looking at node["global-properties"] for
properties to match on. Change this to instead run against
node.properties, so that we can use type-specific properties such as alsa.id.
Signed-off-by: James Calligeros <jcalligeros99@gmail.com>
The current code was not updated to use `clients_om`, thus never
found a client, ending up never sending client errors.
Just use the shared helper to avoid such issues in the future.
These were documented in src/scripts/lib initially because the settings-manager
scripts used to be also there and it made sense to keep them close together.
Autoswitch should make sure the saved headset profile always has an
input route. Check this on loading/saving it, otherwise reselect it.
Headset mode profile without mic used to make some sense when it wasn't
using the loopback node (ie. easy way to disable it), but now with the
loopback node this doesn't make sense any more and it's confusing if it
occurs, so we shouldn't allow it.
Profile autoswitch should support also A2DP profiles with input route,
so the loopback should not assume the source node is SCO one.
Also fix handling of AG SCO output stream nodes, which should not be
hidden as internal.
A smart filter should be considered "targetless" (i.e. interpose on
streams going to default target) only if filter.smart.target is not set.
Currently any smart filter with specified target not found is considered
such, which is wrong. This causes misbehavior, such as all recording
streams going to the bluetooth dummy source.
Fix this by doing it correctly.
And name the loopback source node the same as bluez source without 'internal'
prefix. This keeps consistency with input/output node names when switching
bluetooth profiles.
Like WirePlumber 0.4.17, we need to mark the current routes as 'active' if they
were previously not active as soon as we detect it. This avoids a possible
infinite loop that restores the routes and saves them constantly, which happens
when the device's Route param has changed more than once before the event
'select-routes' is triggered.
The boolean values of properties in rules are strings in JSON config files and
they will retain the same type when they are translated to Lua.
Use cutils.parseBool() function when they have to be interpreted as bools.
Fixes: #586
This is needed for some devices that expose both Headset and Speaker nodes, so
that the applications are automatically linked to the Headset node or Speakers
node automatically when plugging and unplugging a headset.
Cache the rules in a global variable in each script, as JSON,
and use JsonUtils directly to evaluate them. This will allow us to
close the WpConf in the future after loading the scripts.
Also change the order of the return values of the match_rules_apply_properties
function to be able to easily ignore the number of changed values,
which is useless in most cases.
In some cases we need to get a section as JSON, so that we can pass it
down to the rules parser, while in other cases we neeed to get it as a
table to use it natively, and in that case we even need to differentiate
between it being an object, an array or an object with WpProperties.
Make it also possible to optionally pass tables with default values to
the functions so that we can get rid of cutils.get_config_section()
as well.
Do not send an error to the client when the target is not defined
and the 'node.linger' property is set.
It is not absolutely necessary that every node has a defined target.
We can have a 'Stream/Output/*' node which can be linked to
multiple 'Stream/Input/*' nodes and only the 'Stream/Input/*'
nodes have a defined target.
This is because 'filter-forward-format.lua' only configures nodes, and therefore
does not have anything to do with the linking logic. The setting has also been
renamed to 'node.filter.forward-format'.
This is because the associated node does not exist anymore. This change also
logs the session item Id when handling it, so that it is easy to know if the
unhandled session item was handled or not.
Since the loopback bluetooth source node is meant to be always used by
applications, this change hides the actual bluez sco-source node by marking
them as internal. This avoids showing 2 input devices in pavucontrol per BT
device if HSP/HFP profile is enabled.
This hides the loopback stream node from applications, making desktop
environments to not show the 'recording from microphone' icon (Eg gnome-panel)
when the bluetooth device is not recording yet.
This patch improves the bluetooth profile autoswitch so that it works with any
application that wants to capture from a bluetooth device. To do so, a loopback
source filter is created per connected bluetooth device. If an application wants
to capture audio from such loopback source filter, the profile in the associated
bluetooth device is changed to HSP/HFP. If there isn't any application connected
to the loopback source filter, the profile switches back to A2DP.
