gstreamer/docs/pwg/advanced-interfaces.xml

<chapter id="chapter-advanced-interfaces">
  <title>Interfaces</title>
  <para>
    Previously, in the chapter <xref linkend="chapter-building-args"/>, we have
    introduced the concept of GObject properties of controlling an element's
    behaviour. This is a very powerful, but has two big disadvantage: firstly,
    it is too generic, and secondly, it isn't dynamic.
  </para>
  <para>
    The first disadvantage has to do with customizability of the end-user
    interface that will be built to control the element. Some properties are
    more important than others. Some integer properties are better shown in a
    spin-button widget, whereas others would be better represented by a slider
    widget. Such things are not possible because the UI has no actual meaning
    in the application. A UI widget that stands for a bitrate property is the
    same as an UI widget that stands for the size of a video, as long as both
    are of the same <classname>GParamSpec</classname> type. Another problem,
    related to the one about parameter important, is that things like parameter
    grouping, function grouping or anything to make parameters coherent, is not
    really possible.
  </para>
  <para>
    The second argument against parameters are that they are not dynamic. In
    many cases, the allowed values for a property are not fixed, but depend
    on things that can only be detected at run-time. The names of inputs for
    a TV card in a video4linux source element, for example, can only be
    retrieved from the kernel driver when we've opened the device; this only
    happens when the element goes into the READY state. This means that we
    cannot create an enum property type to show this to the user.
  </para>
  <para>
    The solution to those problems is to create very specialized types of
    controls for certain often-used controls. We use the concept of interfaces
    to achieve this. The basis of this all is the glib
    <classname>GTypeInterface</classname> type. For each case where we think
    it's useful, we've created interfaces which can be implemented by elements
    at their own will. We've also created a small extension to
    <classname>GTypeInterface</classname> (which is static itself, too) which
    allows us to query for interface availability based on runtime properties.
    This extension is called <ulink type="http"
    url="../../gstreamer/html/GstImplementsInterface.html"><classname>
    GstImplementsInterface</classname></ulink>.
  </para>
  <para>
    One important note: interfaces do <emphasis>not</emphasis> replace
    properties. Rather, interfaces should be built <emphasis>next to</emphasis>
    properties. There are two important reasons for this. Firstly, properties
    can be saved in XML files. Secondly, properties can be specified on the
    commandline (<filename>gst-launch</filename>).
  </para>

  <sect1 id="section-iface-general" xreflabel="How to Implement Interfaces">
    <title>How to Implement Interfaces</title>
    <para>
      Implementing interfaces is intiated in the <function>_get_type ()</function>
      of your element. You can register one or more interfaces after having
      registered the type itself. Some interfaces have dependencies on other
      interfaces or can only be registered by certain types of elements. You
      will be notified of doing that wrongly when using the element: it will
      quit with failed assertions, which will explain what went wrong. In the
      case of GStreamer, the only dependency that <emphasis>some</emphasis>
      interfaces have is <ulink type="http"
      url="../../gstreamer/html/GstImplementsInterface.html"><classname>
      GstImplementsInterface</classname></ulink>. Per
      interface, we will indicate clearly when it depends on this extension.
      If it does, you need to register support for <emphasis>that</emphasis>
      interface before registering support for the interface that you're
      wanting to support. The example below explains how to add support for a
      simple interface with no further dependencies. For a small explanation
      on <ulink type="http" url="../../gstreamer/html/GstImplementsInterface.html">
      <classname>GstImplementsInterface</classname></ulink>, see the next section
      about the mixer interface: <xref linkend="section-iface-mixer"/>.
    </para>
    <programlisting>
static void	gst_my_filter_some_interface_init	(GstSomeInterface *iface);

GType
gst_my_filter_get_type (void)
{
  static GType my_filter_type = 0;
                                                                                
  if (!my_filter_type) {
    static const GTypeInfo my_filter_info = {
      sizeof (GstMyFilterClass),
      (GBaseInitFunc) gst_my_filter_base_init,
      NULL,
      (GClassInitFunc) gst_my_filter_class_init,
      NULL,
      NULL,
      sizeof (GstMyFilter),
      0,
      (GInstanceInitFunc) gst_my_filter_init
    };
    static const GInterfaceInfo some_interface_info = {
      (GInterfaceInitFunc) gst_my_filter_some_interface_init,
      NULL,
      NULL
    };

    my_filter_type =
	g_type_register_static (GST_TYPE_MY_FILTER,
				"GstMyFilter",
				&amp;my_filter_info, 0);
    g_type_add_interface_static (my_filter_type,
				 GST_TYPE_SOME_INTERFACE,
                                 &amp;some_interface_info);
  }

  return my_filter_type;
}

static void
gst_my_filter_some_interface_init (GstSomeInterface *iface)
{
  /* here, you would set virtual function pointers in the interface */
}
    </programlisting>
  </sect1>

  <sect1 id="section-iface-mixer" xreflabel="Mixer Interface">
    <title>Mixer Interface</title>
    <para>
      The goal of the mixer interface is to provide a simple yet powerful API
      to applications for audio hardware mixer/volume control. Most soundcards
      have hardware mixers, where volume can be changed, they can be muted,
      inputs can be modified to mix their content into what will be read from
      the device by applications (in our case: audio source plugins). The
      mixer interface is the way to control those. The mixer interface can
      also be used for volume control in software (e.g. the <quote>volume</quote>
      element). The end goal of this interface is to allow development of
      hardware volume control applications and for the control of audio volume
      and input/output settings.
    </para>
    <para>
      The mixer interface requires the <ulink type="http"
      url="../../gstreamer/html/GstImplementsInterface.html"><classname>
      GstImplementsInterface</classname></ulink>
      interface to be implemented by the element. The example below will
      feature both, so it serves as an example for the <ulink type="http"
      url="../../gstreamer/html/GstImplementsInterface.html"><classname>
      GstImplementsInterface</classname></ulink>, too. In this
      interface, it is required to set a function pointer for the <function>
      supported ()</function> function.
      If you don't, this function will always return FALSE (default
      implementation) and the mixer interface implementation will not work. For
      the mixer interface, the only required function is
      <function>list_tracks ()</function>. All other function pointers in the
      mixer interface are optional, although it is strongly recommended to set
      function pointers for at least the <function>get_volume ()</function> and
      <function>set_volume ()</function> functions. The API reference for this
      interface documents the goal of each function, so we will limit ourselves
      to the implementation here.
    </para>
    <para>
      The following example shows a mixer implementation for a software N-to-1
      element. It does not show the actual process of stream mixing, that is
      far too complicated for this guide.
    </para>
    <programlisting>
#include &lt;gst/mixer/mixer.h&gt;

typedef struct _GstMyFilter {
[..]
  gint volume;
  GList *tracks;
} GstMyFilter;

static void	gst_my_filter_implements_interface_init	(GstImplementsInterfaceClass *iface);
static void	gst_my_filter_mixer_interface_init	(GstMixerClass *iface);

GType
gst_my_filter_get_type (void)
{
[..]
    static const GInterfaceInfo implements_interface_info = {
      (GInterfaceInitFunc) gst_my_filter_implements_interface_init,
      NULL,
      NULL
    };
    static const GInterfaceInfo mixer_interface_info = {
      (GInterfaceInitFunc) gst_my_filter_mixer_interface_init,
      NULL,
      NULL
    };
[..]
    g_type_add_interface_static (my_filter_type,
				 GST_TYPE_IMPLEMENTS_INTERFACE,
				 &amp;implements_interface_info);
    g_type_add_interface_static (my_filter_type,
				 GST_TYPE_MIXER,
				 &amp;mixer_interface_info);
[..]
}

static void
gst_my_filter_init (GstMyFilter *filter)
{
  GstMixerTrack *track = NULL;
[..]
  filter->volume = 100;
  filter->tracks = NULL;
  track = g_object_new (GST_TYPE_MIXER_TRACK, NULL);
  track->label = g_strdup ("MyTrack");
  track->num_channels = 1;
  track->min_volume = 0;
  track->max_volume = 100;
  track->flags = GST_MIXER_TRACK_SOFTWARE;
  filter->tracks = g_list_append (filter->tracks, track);
}

static gboolean
gst_my_filter_interface_supported (GstImplementsInterface *iface,
				   GType                   iface_type)
{
  g_return_val_if_fail (iface_type == GST_TYPE_MIXER, FALSE);

  /* for the sake of this example, we'll always support it. However, normally,
   * you would check whether the device you've opened supports mixers. */
  return TRUE;
}

static void
gst_my_filter_implements_interface_init (GstImplementsInterfaceClass *iface)
{
  iface->supported = gst_my_filter_interface_supported;
}

/*
 * This function returns the list of support tracks (inputs, outputs)
 * on this element instance. Elements usually build this list during
 * _init () or when going from NULL to READY.
 */

static const GList *
gst_my_filter_mixer_list_tracks (GstMixer *mixer)
{
  GstMyFilter *filter = GST_MY_FILTER (mixer);

  return filter->tracks;
}

/*
 * Set volume. volumes is an array of size track->num_channels, and
 * each value in the array gives the wanted volume for one channel
 * on the track.
 */

static void
gst_my_filter_mixer_set_volume (GstMixer      *mixer,
				GstMixerTrack *track,
				gint          *volumes)
{
  GstMyFilter *filter = GST_MY_FILTER (mixer);

  filter->volume = volumes[0];

  g_print ("Volume set to %d\n", filter->volume);
}

static void
gst_my_filter_mixer_get_volume (GstMixer      *mixer,
				GstMixerTrack *track,
				gint          *volumes)
{
  GstMyFilter *filter = GST_MY_FILTER (mixer);

  volumes[0] = filter->volume;
}

static void
gst_my_filter_mixer_interface_init (GstMixerClass *iface)
{
  /* the mixer interface requires a definition of the mixer type:
   * hardware or software? */
  GST_MIXER_TYPE (iface) = GST_MIXER_SOFTWARE;

  /* virtual function pointers */
  iface->list_tracks = gst_my_filter_mixer_list_tracks;
  iface->set_volume  = gst_my_filter_mixer_set_volume;
  iface->get_volume  = gst_my_filter_mixer_get_volume;
}
    </programlisting>
    <para>
      The mixer interface is very audio-centric. However, with the software
      flag set, the mixer can be used to mix any kind of stream in a N-to-1
      element to join (not aggregate!) streams together into one output stream.
      Conceptually, that's called mixing too. You can always use the element
      factory's <quote>category</quote> to indicate type of your element. In
      a software element that mixes random streams, you would not be required
      to implement the <function>_get_volume ()</function> or
      <function>_set_volume ()</function> functions. Rather, you would only
      implement the <function>_set_record ()</function> to enable or disable
      tracks in the output stream. to make sure that a mixer-implementing
      element is of a certain type, check the element factory's category.
    </para>
  </sect1>

  <sect1 id="section-iface-tuner" xreflabel="Tuner Interface">
    <title>Tuner Interface</title>
    <para>
      As opposed to the mixer interface, that's used to join together N streams
      into one output stream by mixing all streams together, the tuner
      interface is used in N-to-1 elements too, but instead of mixing the input
      streams, it will select one stream and push the data of that stream to
      the output stream. It will discard the data of all other streams. There
      is a flag that indicates whether this is a software-tuner (in which case
      it is a pure software implementation, with N sink pads and 1 source pad)
      or a hardware-tuner, in which case it only has one source pad, and the
      whole stream selection process is done in hardware. The software case can
      be used in elements such as <emphasis>switch</emphasis>. The hardware
      case can be used in elements with channel selection, such as video source
      elements (v4lsrc, v4l2src, etc.). If you need a specific element type,
      use the element factory's <quote>category</quote> to make sure that the
      element is of the type that you need. Note that the interface itself is
      highly analog-video-centric.
    </para>
    <para>
      This interface requires the <ulink type="http"
      url="../../gstreamer/html/GstImplementsInterface.html"><classname>
      GstImplemensInterface</classname></ulink>
      interface to work correctly.
    </para>
    <para>
      The following example shows how to implement the tuner interface in an
      element. It does not show the actual process of stream selection, that
      is irrelevant for this section.
    </para>
    <programlisting>
#include &lt;gst/tuner/tuner.h&gt;

typedef struct _GstMyFilter {
[..]
  gint active_input;
  GList *channels;
} GstMyFilter;

static void	gst_my_filter_implements_interface_init	(GstImplementsInterfaceClass *iface);
static void	gst_my_filter_tuner_interface_init	(GstTunerClass *iface);

GType
gst_my_filter_get_type (void)
{
[..]
    static const GInterfaceInfo implements_interface_info = {
      (GInterfaceInitFunc) gst_my_filter_implements_interface_init,
      NULL,
      NULL
    };
    static const GInterfaceInfo tuner_interface_info = {
      (GInterfaceInitFunc) gst_my_filter_tuner_interface_init,
      NULL,
      NULL
    };
[..]
    g_type_add_interface_static (my_filter_type,
				 GST_TYPE_IMPLEMENTS_INTERFACE,
				 &amp;implements_interface_info);
    g_type_add_interface_static (my_filter_type,
				 GST_TYPE_TUNER,
				 &amp;tunerr_interface_info);
[..]
}

static void
gst_my_filter_init (GstMyFilter *filter)
{
  GstTunerChannel *channel = NULL;
[..]
  filter->active_input = 0;
  filter->channels = NULL;
  channel = g_object_new (GST_TYPE_TUNER_CHANNEL, NULL);
  channel->label = g_strdup ("MyChannel");
  channel->flags = GST_TUNER_CHANNEL_INPUT;
  filter->channels = g_list_append (filter->channels, channel);
}

static gboolean
gst_my_filter_interface_supported (GstImplementsInterface *iface,
				   GType                   iface_type)
{
  g_return_val_if_fail (iface_type == GST_TYPE_TUNER, FALSE);

  /* for the sake of this example, we'll always support it. However, normally,
   * you would check whether the device you've opened supports tuning. */
  return TRUE;
}

static void
gst_my_filter_implements_interface_init (GstImplementsInterfaceClass *iface)
{
  iface->supported = gst_my_filter_interface_supported;
}

static const GList *
gst_my_filter_tuner_list_channels (GstTuner *tuner)
{
  GstMyFilter *filter = GST_MY_FILTER (tuner);

  return filter->channels;
}

static GstTunerChannel *
gst_my_filter_tuner_get_channel (GstTuner *tuner)
{
  GstMyFilter *filter = GST_MY_FILTER (tuner);

  return g_list_nth_data (filter->channels,
			  filter->active_input);
}

static void
gst_my_filter_tuner_set_channel (GstTuner        *tuner,
				 GstTunerChannel *channel)
{
  GstMyFilter *filter = GST_MY_FILTER (tuner);

  filter->active_input = g_list_index (filter->channels, channel);
  g_assert (filter->active_input >= 0);
}

static void
gst_my_filter_tuner_interface_init (GstTunerClass *iface)
{
  iface->list_channels = gst_my_filter_tuner_list_channels;
  iface->get_channel   = gst_my_filter_tuner_get_channel;
  iface->set_channel   = gst_my_filter_tuner_set_channel;
}
    </programlisting>
    <para>
      As said, the tuner interface is very analog video-centric. It features
      functions for selecting an input or output, and on inputs, it features
      selection of a tuning frequency if the channel supports frequency-tuning
      on that input. Likewise, it allows signal-strength-acquiring if the input
      supports that. Frequency tuning can be used for radio or cable-TV tuning.
      Signal-strength is an indication of the signal and can be used for
      visual feedback to the user or for autodetection. Next to that, it also
      features norm selection, which is only useful for analog video elements.
    </para>
  </sect1>

  <sect1 id="section-iface-colorbalance" xreflabel="Color Balance Interface">
    <title>Color Balance Interface</title>
    <para>
      WRITEME
    </para>
  </sect1>

  <sect1 id="section-iface-propprobe" xreflabel="Property Probe Interface">
    <title>Property Probe Interface</title>
    <para>
      Property probing is a generic solution to the problem that properties'
      value lists in an enumeration are static. We've shown enumerations in
      <xref linkend="chapter-building-args"/>. Property probing tries to accomplish
      a goal similar to enumeration lists: to have a limited, explicit list of
      allowed values for a property. There are two differences between
      enumeration lists and probing. Firstly, enumerations only allow strings
      as values; property probing works for any value type. Secondly, the
      contents of a probed list of allowed values may change during the life
      of an element. The contents of a enumeraiton list are static. Crrently,
      property probing is being used for detection of devices (e.g. for OSS
      elements, Video4linux elements, etc.). It could - in theory - be used
      for any property, though.
    </para>
    <para>
      Property probing stores the list of allowed (or recommended) values in a
      <classname>GValueArray</classname> and returns that to the user.
      <symbol>NULL</symbol> is a valid return value, too. The process of
      property probing is separated over two virtual functions: one for probing
      the property to create a <classname>GValueArray</classname>, and one to
      retrieve the current <classname>GValueArray</classname>. Those two are
      separated because probing might take a long time (several seconds). Also,
      this simpliies interface implementation in elements. For the application,
      there are functions that wrap those two. For more information on this,
      have a look at the API reference for the
      <!-- FIXME: add link, but this is in ./gst-plugins/gst-libs/gst/propertyprobe/propertyprobe.c-->
      <classname>GstPropertyProbe</classname> interface.
    </para>
    <para>
      Below is a example of property probing for the audio filter element; it
      will probe for allowed values for the <quote>silent</quote> property.
      Indeed, this value is a <type>gboolean</type> so it doesn't
      make much sense. Then again, it's only an example.
    </para>
    <programlisting>
#include &lt;gst/propertyprobe/propertyprobe.h&gt;

static void	gst_my_filter_probe_interface_init	(GstPropertyProbeInterface *iface);

GType
gst_my_filter_get_type (void)
{
[..]
    static const GInterfaceInfo probe_interface_info = {
      (GInterfaceInitFunc) gst_my_filter_probe_interface_init,
      NULL,
      NULL
    };
[..]
    g_type_add_interface_static (my_filter_type,
				 GST_TYPE_PROPERTY_PROBE,
				 &amp;probe_interface_info);
[..]
}

static const GList *
gst_my_filter_probe_get_properties (GstPropertyProbe *probe)
{
  GObjectClass *klass = G_OBJECT_GET_CLASS (probe);
  static GList *props = NULL;

  if (!props) {
    GParamSpec *pspec;

    pspec = g_object_class_find_property (klass, "silent");
    props = g_list_append (props, pspec);
  }

  return props;
}

static gboolean
gst_my_filter_probe_needs_probe (GstPropertyProbe *probe,
				 guint             prop_id,
				 const GParamSpec *pspec)
{
  gboolean res = FALSE;

  switch (prop_id) {
    case ARG_SILENT:
      res = FALSE;
      break;
    default:
      G_OBJECT_WARN_INVALID_PROPERTY_ID (probe, prop_id, pspec);
      break;
  }

  return res;
}

static void
gst_my_filter_probe_probe_property (GstPropertyProbe *probe,
				    guint             prop_id,
				    const GParamSpec *pspec)
{
  switch (prop_id) {
    case ARG_SILENT:
      /* don't need to do much here... */
      break;
    default:
      G_OBJECT_WARN_INVALID_PROPERTY_ID (probe, prop_id, pspec);
      break;
  }
}

static GValueArray *
gst_my_filter_get_silent_values (GstMyFilter *filter)
{
  GValueArray *array = g_value_array_new (2);
  GValue value = { 0 };

  g_value_init (&amp;value, G_TYPE_BOOLEAN);

  /* add TRUE */
  g_value_set_boolean (&amp;value, TRUE);
  g_value_array_append (array, &amp;value);

  /* add FALSE */
  g_value_set_boolean (&amp;value, FALSE);
  g_value_array_append (array, &amp;value);

  g_value_unset (&amp;value);

  return array;
}

static GValueArray *
gst_my_filter_probe_get_values (GstPropertyProbe *probe,
				guint             prop_id,
				const GParamSpec *pspec)
{
  GstMyFilter *filter = GST_MY_FILTER (probe);
  GValueArray *array = NULL;

  switch (prop_id) {
    case ARG_SILENT:
      array = gst_my_filter_get_silent_values (filter);
      break;
    default:
      G_OBJECT_WARN_INVALID_PROPERTY_ID (probe, prop_id, pspec);
      break;
  }

  return array;
}

static void
gst_my_filter_probe_interface_init (GstPropertyProbeInterface *iface)
{
  iface->get_properties = gst_my_filter_probe_get_properties;
  iface->needs_probe    = gst_my_filter_probe_needs_probe;
  iface->probe_property = gst_my_filter_probe_probe_property;
  iface->get_values     = gst_my_filter_probe_get_values;
}
    </programlisting>
    <para>
      You don't need to support any functions for getting or setting values.
      All that is handled via the standard <classname>GObject</classname>
      <function>_set_property ()</function> and <function>_get_property ()</function>
      functions.
    </para>
  </sect1>

  <sect1 id="section-iface-profile" xreflabel="Profile Interface">
    <title>Profile Interface</title>
    <para>
      WRITEME
    </para>
  </sect1>
                                                                                
  <sect1 id="section-iface-xoverlay" xreflabel="X Overlay Interface">
    <title>X Overlay Interface</title>
    <para>
      An X Overlay is basically a video output in a XFree86 drawable. Elements
      implementing this interface will draw video in a X11 window. Through this
      interface, applications will be proposed 2 different modes to work with
      a plugin implemeting it. The first mode is a passive mode where the plugin
      owns, creates and destroys the X11 window. The second mode is an active
      mode where the application handles the X11 window creation and then tell
      the plugin where it should output video. Let's get a bit deeper in those
      modes...
    </para>
    <para>
      A plugin drawing video output in a X11 window will need to have that
      window at one stage or another. Passive mode simply means that no window
      has been given to the plugin before that stage, so the plugin created the
      window by itself. In that case the plugin is responsible of destroying
      that window when it's not needed anymore and it has to tell the 
      applications that a window has been created so that the application can
      use it. This is done using the <classname>have_xwindow_id</classname>
      signal that can be emitted from the plugin with the
      <function>gst_x_overlay_got_xwindow_id</function> method.
    </para>
    <para>
      As you probably guessed already active mode just means sending a X11
      window to the plugin so that video output goes there. This is done using
      the <function>gst_x_overlay_set_xwindow_id</function> method.
    </para>
    <para>
      It is possible to switch from one mode to another at any moment, so the
      plugin implementing this interface has to handle all cases. There are only
      2 methods that plugins writers have to implement and they most probably
      look like that :
    </para>
    <programlisting><![CDATA[
static void
gst_my_filter_set_xwindow_id (GstXOverlay *overlay, XID xwindow_id)
{
  GstMyFilter *my_filter = GST_MY_FILTER (overlay);

  if (my_filter->window)
    gst_my_filter_destroy_window (my_filter->window);
    
  my_filter->window = xwindow_id;
}

static void
gst_my_filter_get_desired_size (GstXOverlay *overlay,
                                guint *width, guint *height)
{
  GstMyFilter *my_filter = GST_MY_FILTER (overlay);

  *width = my_filter->width;
  *height = my_filter->height;
}

static void
gst_my_filter_xoverlay_init (GstXOverlayClass *iface)
{
  iface->set_xwindow_id = gst_my_filter_set_xwindow_id;
  iface->get_desired_size = gst_my_filter_get_desired_size;
}
    ]]></programlisting>
    <para>
      You will also need to use the interface methods to fire signals when 
      needed such as in the pad link function where you will know the video
      geometry and maybe create the window.
    </para>
    <programlisting><![CDATA[
static MyFilterWindow *
gst_my_filter_window_create (GstMyFilter *my_filter, gint width, gint height)
{
  MyFilterWindow *window = g_new (MyFilterWindow, 1);
  ...
  gst_x_overlay_got_xwindow_id (GST_X_OVERLAY (my_filter), window->win);
}

static GstPadLinkReturn
gst_my_filter_sink_link (GstPad *pad, const GstCaps *caps)
{
  GstMyFilter *my_filter = GST_MY_FILTER (overlay);
  gint width, height;
  gboolean ret;
  ...
  ret = gst_structure_get_int (structure, "width", &width);
  ret &= gst_structure_get_int (structure, "height", &height);
  if (!ret) return GST_PAD_LINK_REFUSED;
  
  if (!my_filter->window)
    my_filter->window = gst_my_filter_create_window (my_filter, width, height);

  gst_x_overlay_got_desired_size (GST_X_OVERLAY (my_filter),
                                  width, height);
  ...
}
    ]]></programlisting>
  </sect1>

  <sect1 id="section-iface-navigation" xreflabel="Navigation Interface">
    <title>Navigation Interface</title>
    <para>
      WRITEME
    </para>
  </sect1>
</chapter>