* filesrc location=b.png ! pngdec ! videoscale ! imagefreeze ! videoconvert ! video/x-raw,format=RGB,width=100,height=100,framerate=0/1 ! tensor_converter ! mux.sink_1
* filesrc location=b.png ! pngdec ! videoscale ! imagefreeze ! videoconvert ! video/x-raw,format=RGB,width=100,height=100,framerate=0/1 ! tensor_converter ! mux.sink_2
* ]|
+ *
+ * |[
+ * gst-launch -v -m tensormux name=mux ! filesink location=mux.log
+ * multifilesrc location="testsequence_%1d.png" index=0 caps="image/png, framerate=(fraction)30/1" ! pngdec ! tensor_converter ! mux.sink_0
+ * multifilesrc location="testsequence_%1d.png" index=0 caps="image/png, framerate=(fraction)30/1" ! pngdec ! tensor_converter ! mux.sink_1
+ * multifilesrc location="testsequence_%1d.png" index=0 caps="image/png, framerate=(fraction)30/1" ! pngdec ! tensor_converter ! mux.sink_2
+ *
* </refsect2>
*
*/
enum
{
PROP_0,
- PROP_TIMESTAMP_OFFSET,
PROP_SILENT
};
-GMutex buf_mutex;
-GCond buf_cond;
-guint32 buf_count = 0;
-gint num_sink = 0;
-
-#define DEFAULT_TIMESTAMP_OFFSET -1
-
/**
* @brief the capabilities of the inputs and outputs.
* describe the real formats here.
GST_STATIC_CAPS (GST_TENSOR_CAP_DEFAULT)
);
-#define gst_tensor_mux_parent_class parent_class
+static void gst_tensor_mux_finalize (GObject * object);
+static gboolean gst_tensor_mux_handle_src_event (GstPad * pad,
+ GstObject * parent, GstEvent * event);
static GstPad *gst_tensor_mux_request_new_pad (GstElement * element,
GstPadTemplate * templ, const gchar * name, const GstCaps * caps);
-static void gst_tensor_mux_release_pad (GstElement * element, GstPad * pad);
-static GstFlowReturn
-gst_tensor_mux_chain (GstPad * pad, GstObject * parent, GstBuffer * buffer);
-static gboolean
-gst_tensor_mux_setcaps (GstPad * pad, GstTensorMux * tensor_mux,
- GstCaps * caps);
-static gboolean gst_tensor_mux_sink_event (GstPad * pad, GstObject * parent,
- GstEvent * event);
static GstStateChangeReturn gst_tensor_mux_change_state (GstElement * element,
GstStateChange transition);
+static gboolean gst_tensor_mux_sink_event (GstCollectPads * pads,
+ GstCollectData * data, GstEvent * event, GstTensorMux * tensor_mux);
+static GstFlowReturn gst_tensor_mux_collected (GstCollectPads * pads,
+ GstTensorMux * tesnor_mux);
+
static void gst_tensor_mux_set_property (GObject * object, guint prop_id,
const GValue * value, GParamSpec * pspec);
static void gst_tensor_mux_get_property (GObject * object, guint prop_id,
GValue * value, GParamSpec * pspec);
-static void gst_tensor_mux_dispose (GObject * object);
+#define gst_tensor_mux_parent_class parent_class
G_DEFINE_TYPE (GstTensorMux, gst_tensor_mux, GST_TYPE_ELEMENT);
/**
gobject_class = (GObjectClass *) klass;
gstelement_class = (GstElementClass *) klass;
- gst_element_class_add_pad_template (gstelement_class,
- gst_static_pad_template_get (&sink_templ));
- gst_element_class_add_pad_template (gstelement_class,
- gst_static_pad_template_get (&src_templ));
+ parent_class = g_type_class_peek_parent (klass);
+ gobject_class->finalize = gst_tensor_mux_finalize;
gobject_class->get_property = gst_tensor_mux_get_property;
gobject_class->set_property = gst_tensor_mux_set_property;
- gobject_class->dispose = gst_tensor_mux_dispose;
-
- g_object_class_install_property (gobject_class,
- PROP_TIMESTAMP_OFFSET, g_param_spec_int ("timestamp-offset",
- "Timestamp Offset",
- "Offset to add to all outgoing timestamps (-1 = random)", -1,
- G_MAXINT, DEFAULT_TIMESTAMP_OFFSET,
- G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
+
g_object_class_install_property (gobject_class, PROP_SILENT,
g_param_spec_boolean ("silent", "Silent", "Produce verbose output ?",
FALSE, G_PARAM_READWRITE));
gstelement_class->request_new_pad =
GST_DEBUG_FUNCPTR (gst_tensor_mux_request_new_pad);
- gstelement_class->release_pad =
- GST_DEBUG_FUNCPTR (gst_tensor_mux_release_pad);
gstelement_class->change_state =
GST_DEBUG_FUNCPTR (gst_tensor_mux_change_state);
+ gst_element_class_add_pad_template (gstelement_class,
+ gst_static_pad_template_get (&sink_templ));
+ gst_element_class_add_pad_template (gstelement_class,
+ gst_static_pad_template_get (&src_templ));
+
gst_element_class_set_details_simple (gstelement_class,
"tensormux",
"Mux multiple tensor stream",
}
/**
- * @brief dispose tensor mux / tensor mux pad object
- * @param object GstTensorMux*
- */
-static void
-gst_tensor_mux_dispose (GObject * object)
-{
- GstTensorMux *tensor_mux = GST_TENSOR_MUX (object);
- GList *item;
-
- g_clear_object (&tensor_mux->last_pad);
-
-restart:
- for (item = GST_ELEMENT_PADS (object); item; item = g_list_next (item)) {
- GstPad *pad = GST_PAD (item->data);
- if (GST_PAD_IS_SINK (pad)) {
- gst_element_release_request_pad (GST_ELEMENT (object), pad);
- goto restart;
- }
- }
-
- G_OBJECT_CLASS (gst_tensor_mux_parent_class)->dispose (object);
-}
-
-
-/**
* @brief initialize the new element
* instantiate pads and add them to element
* set pad calback functions
gst_tensor_mux_init (GstTensorMux * tensor_mux)
{
GstElementClass *klass = GST_ELEMENT_GET_CLASS (tensor_mux);
+
tensor_mux->srcpad =
gst_pad_new_from_template (gst_element_class_get_pad_template (klass,
"src"), "src");
- gst_pad_use_fixed_caps (tensor_mux->srcpad);
+ gst_pad_set_event_function (tensor_mux->srcpad,
+ gst_tensor_mux_handle_src_event);
+
gst_element_add_pad (GST_ELEMENT (tensor_mux), tensor_mux->srcpad);
+ tensor_mux->collect = gst_collect_pads_new ();
+ gst_collect_pads_set_event_function (tensor_mux->collect,
+ (GstCollectPadsEventFunction)
+ GST_DEBUG_FUNCPTR (gst_tensor_mux_sink_event), tensor_mux);
+ gst_collect_pads_set_function (tensor_mux->collect,
+ (GstCollectPadsFunction) GST_DEBUG_FUNCPTR (gst_tensor_mux_collected),
+ tensor_mux);
+
tensor_mux->num_tensors = 0;
- tensor_mux->byte_count = 0;
tensor_mux->silent = FALSE;
tensor_mux->rank = 0;
- tensor_mux->dimensions = g_string_new (NULL);
- tensor_mux->types = g_string_new (NULL);
- tensor_mux->outbuffer = gst_buffer_new ();
- gst_make_tensors (tensor_mux->outbuffer);
- g_cond_init (&buf_cond);
- g_mutex_init (&buf_mutex);
}
/**
- * @brief Setup Sink Pad for tensor mux
- * Initialize pad private data & set functions for pad
- * @param tensor_mux GstTensorMux Pointer
- * @param sinkpad Sink pad
+ * @brief finalize vmethod
*/
static void
-gst_tensor_mux_setup_sinkpad (GstTensorMux * tensor_mux, GstPad * sinkpad)
+gst_tensor_mux_finalize (GObject * object)
{
- GstTensorMuxPadPrivate *padpriv = g_slice_new0 (GstTensorMuxPadPrivate);
- padpriv->done = FALSE;
- gst_pad_set_chain_function (sinkpad,
- GST_DEBUG_FUNCPTR (gst_tensor_mux_chain));
- gst_pad_set_event_function (sinkpad,
- GST_DEBUG_FUNCPTR (gst_tensor_mux_sink_event));
-
- gst_pad_set_element_private (sinkpad, padpriv);
+ GstTensorMux *tensor_mux;
+ tensor_mux = GST_TENSOR_MUX (object);
- gst_pad_set_active (sinkpad, TRUE);
- gst_element_add_pad (GST_ELEMENT (tensor_mux), sinkpad);
+ if (tensor_mux->collect)
+ gst_object_unref (tensor_mux->collect);
+ G_OBJECT_CLASS (parent_class)->finalize (object);
}
-
/**
* @brief making new request pad (gst element vmethod)
*/
{
GstPad *newpad;
GstTensorMux *tensor_mux;
+ gchar *name;
g_return_val_if_fail (templ != NULL, NULL);
g_return_val_if_fail (GST_IS_TENSOR_MUX (element), NULL);
tensor_mux = GST_TENSOR_MUX (element);
- if (templ->direction != GST_PAD_SINK) {
- GST_WARNING_OBJECT (tensor_mux, "request pad that is not a SINK pad");
- return NULL;
- }
-
- newpad = gst_pad_new_from_template (templ, req_name);
+ name = g_strdup_printf ("sink_%u", tensor_mux->num_tensors);
+ newpad = gst_pad_new_from_template (templ, name);
+ g_free (name);
if (newpad) {
- GST_OBJECT_LOCK (tensor_mux);
+ GstTensorMuxPadData *tensormuxpad;
+ tensormuxpad =
+ (GstTensorMuxPadData *) gst_collect_pads_add_pad (tensor_mux->collect,
+ newpad, sizeof (GstTensorMuxPadData), NULL, TRUE);
+ tensormuxpad->pad = newpad;
+ gst_pad_set_element_private (newpad, tensormuxpad);
tensor_mux->num_tensors++;
- GST_OBJECT_UNLOCK (tensor_mux);
- gst_tensor_mux_setup_sinkpad (tensor_mux, newpad);
+ gst_element_add_pad (element, newpad);
} else {
GST_WARNING_OBJECT (tensor_mux, "failed to create request pad");
}
}
/**
- * @brief release request pad (gst element vmethod)
+ * @brief src event vmethod
*/
-static void
-gst_tensor_mux_release_pad (GstElement * element, GstPad * pad)
+static gboolean
+gst_tensor_mux_handle_src_event (GstPad * pad, GstObject * parent,
+ GstEvent * event)
{
- GstTensorMuxPadPrivate *padpriv;
- GstTensorMux *tensor_mux = GST_TENSOR_MUX_CAST (element);
-
- GST_DEBUG_OBJECT (tensor_mux, "releaseing pad %s:%s",
- GST_DEBUG_PAD_NAME (pad));
-
- gst_pad_set_active (pad, FALSE);
-
- GST_OBJECT_LOCK (element);
- padpriv = gst_pad_get_element_private (pad);
- gst_pad_set_element_private (pad, NULL);
- GST_OBJECT_UNLOCK (element);
- gst_element_remove_pad (GST_ELEMENT_CAST (tensor_mux), pad);
- if (padpriv) {
- g_slice_free (GstTensorMuxPadPrivate, padpriv);
+ GstEventType type;
+ type = event ? GST_EVENT_TYPE (event) : GST_EVENT_UNKNOWN;
+ switch (type) {
+ case GST_EVENT_SEEK:
+ return FALSE;
+ default:
+ break;
}
+
+ return gst_pad_event_default (pad, parent, event);
}
/**
- * @brief resend events for sticky events
+ * @brief sink event vmethod
*/
static gboolean
-resend_events (GstPad * pad, GstEvent ** event, gpointer user_data)
+gst_tensor_mux_sink_event (GstCollectPads * pads, GstCollectData * data,
+ GstEvent * event, GstTensorMux * tensor_mux)
{
- GstTensorMux *tensor_mux = user_data;
-
- if (GST_EVENT_TYPE (*event) == GST_EVENT_CAPS) {
- GstCaps *caps;
-
- gst_event_parse_caps (*event, &caps);
- gst_tensor_mux_setcaps (pad, tensor_mux, caps);
- } else if (GST_EVENT_TYPE (*event) == GST_EVENT_SEGMENT) {
- GstSegment new_segment;
- gst_segment_init (&new_segment, GST_FORMAT_TIME);
- gst_pad_push_event (tensor_mux->srcpad,
- gst_event_new_segment (&new_segment));
- } else {
- gst_pad_push_event (tensor_mux->srcpad, gst_event_ref (*event));
+ gboolean ret;
+ switch (GST_EVENT_TYPE (event)) {
+ case GST_EVENT_FLUSH_STOP:
+ {
+ tensor_mux->need_segment = TRUE;
+ break;
+ }
+ default:
+ break;
}
- return TRUE;
+ ret = gst_collect_pads_event_default (pads, data, event, FALSE);
+ return ret;
}
/**
- * @brief push tensor in tensors
- * @param pad Sink pad
- * @param tensor_mux Tensor Mux
- * @param buffer input buffer
+ * @brief Compare dts & pts time and find earliest
+ * @param tensor_mux tensor muxer
+ * @param old previous mux pad data
+ * @param new current mux pad data
+ * @return if > 0, new is earlier than old
*/
-static gboolean
-gst_push_tensor (GstPad * pad, GstTensorMux * tensor_mux, GstBuffer * buffer,
- gint nth)
+static gint
+gst_tensor_mux_compare_pads (GstTensorMux * tensor_mux,
+ GstTensorMuxPadData * old, GstTensorMuxPadData * new)
{
- tensor_dim dim;
- GstMemory *mem;
- gboolean ret = FALSE;
- tensor_type tensor_type;
- GstTensor_Filter_CheckStatus status;
-
- GstCaps *caps = gst_pad_get_current_caps (pad);
+ guint64 oldtime, newtime;
+ if (old == NULL || old->buffer == NULL)
+ return 1;
+ if (new == NULL || new->buffer == NULL)
+ return -1;
+ if (GST_CLOCK_TIME_IS_VALID (old->dts_timestamp) &&
+ GST_CLOCK_TIME_IS_VALID (new->dts_timestamp)) {
+ oldtime = old->dts_timestamp;
+ newtime = new->dts_timestamp;
+ } else {
+ oldtime = old->pts_timestamp;
+ newtime = new->pts_timestamp;
+ }
- status = get_tensor_from_padcap (caps, dim, &tensor_type, NULL, NULL);
- g_assert ((status & _TFC_ALL) == _TFC_ALL);
+ if (oldtime == GST_CLOCK_TIME_NONE)
+ return -1;
+ if (newtime == GST_CLOCK_TIME_NONE)
+ return 1;
- mem = gst_buffer_get_memory (buffer, 0);
- gst_memory_ref (mem);
- if (gst_append_tensor (tensor_mux->outbuffer, mem, dim, tensor_type, nth))
- ret = TRUE;
+ if (newtime < oldtime)
+ return 1;
+ else if (newtime > oldtime)
+ return -1;
- return ret;
+ return 0;
}
/**
- * @brief chain function (gst element vmethod)
+ * @brief Looping to generete outbut buffer for srcpad
+ * @param tensor_mux tensor muxer
+ * @param tensors_buf output buffer for srcpad
+ * @param dimensions collected dimensions as string
+ * @param types collected types as string
+ * @param pts_time earliest pts time (present timestamp)
+ * @param dts_time earliest dts time (decoding timestamp)
+ * @return isEOS boolean EOS ( End of Stream )
*/
-static GstFlowReturn
-gst_tensor_mux_chain (GstPad * pad, GstObject * parent, GstBuffer * buffer)
+gboolean
+gst_tensor_mux_collect_buffer (GstTensorMux * tensor_mux,
+ GstBuffer * tensors_buf, GString * dimensions, GString * types,
+ GstClockTime * pts_time, GstClockTime * dts_time)
{
- GstTensorMux *tensor_mux;
- GstFlowReturn ret;
- GstTensorMuxPadPrivate *padpriv;
- gboolean changed = FALSE;
- tensor_mux = GST_TENSOR_MUX (parent);
-
- GST_DEBUG_OBJECT (pad, "recevied %" GST_PTR_FORMAT, buffer);
-
- if (gst_pad_check_reconfigure (tensor_mux->srcpad)) {
- GstCaps *current_caps = gst_pad_get_current_caps (pad);
- if (!gst_tensor_mux_setcaps (pad, tensor_mux, current_caps)) {
- gst_pad_mark_reconfigure (tensor_mux->srcpad);
- if (GST_PAD_IS_FLUSHING (tensor_mux->srcpad))
- ret = GST_FLOW_FLUSHING;
- else
- ret = GST_FLOW_NOT_NEGOTIATED;
- goto out;
- }
- gst_caps_unref (current_caps);
- }
- GST_OBJECT_LOCK (tensor_mux);
- padpriv = gst_pad_get_element_private (pad);
+ GSList *walk = NULL;
+ GstTensorMuxPadData *bestpad = NULL;
- if (!padpriv) {
- GST_OBJECT_UNLOCK (tensor_mux);
- gst_buffer_unref (buffer);
- return GST_FLOW_NOT_LINKED;
- }
-
- buffer = gst_buffer_make_writable (buffer);
-
- if (pad != tensor_mux->last_pad) {
- changed = TRUE;
- g_clear_object (&tensor_mux->last_pad);
- tensor_mux->last_pad = g_object_ref (pad);
-
- if (!gst_push_tensor (pad, tensor_mux, buffer, padpriv->nth))
- GST_ERROR_OBJECT (tensor_mux, "Cannot append GstMemory\n");
-
- g_mutex_lock (&buf_mutex);
- buf_count++;
- g_cond_signal (&buf_cond);
- g_mutex_unlock (&buf_mutex);
- }
-
- if (GST_BUFFER_DURATION_IS_VALID (buffer) && GST_BUFFER_PTS_IS_VALID (buffer))
- tensor_mux->last_stop =
- GST_BUFFER_PTS (buffer) + GST_BUFFER_DURATION (buffer);
- else
- tensor_mux->last_stop = GST_CLOCK_TIME_NONE;
-
- gst_buffer_unref (buffer);
- GST_OBJECT_UNLOCK (tensor_mux);
+ tensor_dim dim;
+ tensor_type type;
+ GstMemory *mem;
+ GstTensor_Filter_CheckStatus status;
+ gboolean isEOS = FALSE;
+ gint i;
+
+ gint old_numerator = 2147483647, old_denominator = 2147483647;
+ gint new_numerator, new_denominator;
+ gint counting = 0;
+ tensor_mux->rank = NNS_TENSOR_RANK_LIMIT;
+
+ walk = tensor_mux->collect->data;
+
+ while (walk) {
+ GstCollectData *data = (GstCollectData *) walk->data;
+ GstTensorMuxPadData *pad = (GstTensorMuxPadData *) data;
+
+ GstCaps *caps = gst_pad_get_current_caps (pad->pad);
+ status =
+ get_tensor_from_padcap (caps, dim, &type, &new_numerator,
+ &new_denominator);
+ g_assert ((status & _TFC_ALL) == _TFC_ALL);
+
+ if (new_denominator < old_denominator)
+ old_denominator = new_denominator;
+ if (new_numerator < old_numerator)
+ old_numerator = new_numerator;
+
+ if (dimensions->len != 0)
+ dimensions = g_string_append (dimensions, ",");
+ for (i = 0; i < NNS_TENSOR_RANK_LIMIT; i++) {
+ dimensions = g_string_append (dimensions, g_strdup_printf ("%d", dim[i]));
+ if (i < NNS_TENSOR_RANK_LIMIT - 1)
+ dimensions = g_string_append (dimensions, ":");
+ }
+ if (types->len != 0)
+ types = g_string_append (types, ",");
+ types = g_string_append (types, tensor_element_typename[type]);
+ gst_caps_unref (caps);
+ walk = g_slist_next (walk);
+ GstBuffer *buf = NULL;
+ buf = gst_collect_pads_pop (tensor_mux->collect, data);
+ if (buf && GST_BUFFER_PTS_IS_VALID (buf)) {
+ pad->pts_timestamp =
+ gst_segment_to_running_time (&data->segment, GST_FORMAT_TIME,
+ GST_BUFFER_PTS (buf));
+ } else {
+ pad->pts_timestamp = GST_CLOCK_TIME_NONE;
+ }
+ if (buf && GST_BUFFER_DTS_IS_VALID (buf)) {
+ pad->dts_timestamp =
+ gst_segment_to_running_time (&data->segment, GST_FORMAT_TIME,
+ GST_BUFFER_DTS (buf));
+ } else {
+ pad->dts_timestamp = GST_CLOCK_TIME_NONE;
+ }
- g_mutex_lock (&buf_mutex);
- while (buf_count != tensor_mux->num_tensors) {
- g_cond_wait (&buf_cond, &buf_mutex);
- g_cond_signal (&buf_cond);
+ pad->buffer = buf;
+ if (GST_IS_BUFFER (buf)) {
+ gst_buffer_unref (pad->buffer);
+ mem = gst_buffer_get_memory (buf, 0);
+ gst_memory_ref (mem);
+ if (!gst_append_tensor (tensors_buf, mem, dim, type, counting))
+ return FALSE;
+ if (pad->buffer != NULL)
+ if (gst_tensor_mux_compare_pads (tensor_mux, bestpad, pad) > 0) {
+ bestpad = pad;
+ *pts_time = bestpad->pts_timestamp;
+ *dts_time = bestpad->dts_timestamp;
+ }
+ } else {
+ isEOS = TRUE;
+ }
+ counting++;
}
- g_mutex_unlock (&buf_mutex);
- if (changed)
- gst_pad_sticky_events_foreach (pad, resend_events, tensor_mux);
+ tensor_mux->framerate_denominator = old_denominator;
+ tensor_mux->framerate_numerator = old_numerator;
- if (pad == tensor_mux->last_pad) {
- gst_buffer_ref (tensor_mux->outbuffer);
- ret = gst_pad_push (tensor_mux->srcpad, tensor_mux->outbuffer);
- } else {
- ret = GST_FLOW_OK;
- }
-
-out:
- return ret;
+ return isEOS;
}
-
/**
- * @brief set caps (gst element vmethod)
+ * @brief Gst Collect Pads Function which is called once collect pads done.
+ * @param pads GstCollectPads
+ * @param tensor_mux Muxer
+ * @return GstFlowReturn
*/
-static gboolean
-gst_tensor_mux_setcaps (GstPad * pad, GstTensorMux * tensor_mux, GstCaps * caps)
+static GstFlowReturn
+gst_tensor_mux_collected (GstCollectPads * pads, GstTensorMux * tensor_mux)
{
- GstStructure *structure;
- gboolean ret = FALSE;
- GstTensorMuxPadPrivate *padpriv;
- GstCaps *peercaps, *src_caps;
- gint dim;
- const gchar *t;
-
- padpriv = gst_pad_get_element_private (pad);
- if (padpriv->done)
- return TRUE;
- if (!gst_caps_is_fixed (caps))
- return FALSE;
-
- peercaps = gst_pad_peer_query_caps (tensor_mux->srcpad, NULL);
- if (peercaps) {
- GstCaps *tcaps, *othercaps;
- tcaps = gst_pad_get_pad_template_caps (pad);
- othercaps =
- gst_caps_intersect_full (peercaps, tcaps, GST_CAPS_INTERSECT_FIRST);
-
- if (gst_caps_get_size (othercaps) > 0) {
- structure = gst_caps_get_structure (othercaps, 0);
- GST_OBJECT_LOCK (tensor_mux);
- GST_OBJECT_UNLOCK (tensor_mux);
- }
-
- gst_caps_unref (othercaps);
- gst_caps_unref (peercaps);
- gst_caps_unref (tcaps);
- }
-
-
- structure = gst_caps_get_structure (caps, 0);
- if (!structure)
- return FALSE;
-
- GST_OBJECT_LOCK (tensor_mux);
- gst_structure_get_int (structure, "rank", &tensor_mux->rank);
- gst_structure_get_int (structure, "dim1", &dim);
- if (tensor_mux->dimensions->len != 0)
- tensor_mux->dimensions = g_string_append (tensor_mux->dimensions, ",");
- tensor_mux->dimensions =
- g_string_append (tensor_mux->dimensions, g_strdup_printf ("%d", dim));
- tensor_mux->dimensions = g_string_append (tensor_mux->dimensions, ":");
- gst_structure_get_int (structure, "dim2", &dim);
- tensor_mux->dimensions =
- g_string_append (tensor_mux->dimensions, g_strdup_printf ("%d", dim));
- tensor_mux->dimensions = g_string_append (tensor_mux->dimensions, ":");
- gst_structure_get_int (structure, "dim3", &dim);
- tensor_mux->dimensions =
- g_string_append (tensor_mux->dimensions, g_strdup_printf ("%d", dim));
- tensor_mux->dimensions = g_string_append (tensor_mux->dimensions, ":");
- gst_structure_get_int (structure, "dim4", &dim);
- tensor_mux->dimensions =
- g_string_append (tensor_mux->dimensions, g_strdup_printf ("%d", dim));
-
- if (tensor_mux->types->len != 0)
- tensor_mux->types = g_string_append (tensor_mux->types, ",");
- t = gst_structure_get_string (structure, "type");
- tensor_mux->types = g_string_append (tensor_mux->types, t);
-
-
- gst_structure_get_fraction (structure, "framerate",
- &tensor_mux->framerate_numerator, &tensor_mux->framerate_denominator);
-
- padpriv = gst_pad_get_element_private (pad);
- if (padpriv
- && gst_structure_get_uint (structure, "timestamps-offset",
- &padpriv->timestamp_offset)) {
- padpriv->have_timestamp_offset = TRUE;
- }
-
- padpriv->done = TRUE;
- padpriv->nth = num_sink;
- num_sink++;
- src_caps = gst_caps_new_simple ("other/tensors",
- "rank", G_TYPE_INT, tensor_mux->rank,
- "num_tensors", G_TYPE_INT, tensor_mux->num_tensors,
- "types", G_TYPE_STRING, tensor_mux->types->str,
- "framerate", GST_TYPE_FRACTION, tensor_mux->framerate_numerator,
- tensor_mux->framerate_denominator, "dimensions", G_TYPE_STRING,
- tensor_mux->dimensions->str, NULL);
-
- GST_OBJECT_UNLOCK (tensor_mux);
-
- if (tensor_mux->send_stream_start) {
+ GstFlowReturn ret = GST_FLOW_OK;
+ GstBuffer *tensors_buf;
+ GString *dimensions = g_string_new (NULL);
+ GString *types = g_string_new (NULL);
+ GstClockTime pts_time, dts_time;
+ GstClockTime time = 0;
+ gboolean isEOS = FALSE;
+ GST_DEBUG_OBJECT (tensor_mux, " all pads are collected ");
+ if (tensor_mux->need_stream_start) {
gchar s_id[32];
-
- g_snprintf (s_id, sizeof (s_id), "interleave-%08x", g_random_int ());
+ g_snprintf (s_id, sizeof (s_id), " tensormux - %08x ", g_random_int ());
gst_pad_push_event (tensor_mux->srcpad, gst_event_new_stream_start (s_id));
-
- tensor_mux->send_stream_start = FALSE;
+ tensor_mux->need_stream_start = FALSE;
}
- GST_DEBUG_OBJECT (tensor_mux,
- "setting caps %" GST_PTR_FORMAT " on src pad..", src_caps);
-
- ret = gst_pad_set_caps (tensor_mux->srcpad, src_caps);
- gst_caps_unref (src_caps);
-
- return ret;
-}
+ tensors_buf = gst_buffer_new ();
+ gst_make_tensors (tensors_buf);
+ isEOS =
+ gst_tensor_mux_collect_buffer (tensor_mux, tensors_buf, dimensions,
+ types, &pts_time, &dts_time);
+ if (isEOS) {
+ if (tensors_buf)
+ gst_buffer_unref (tensors_buf);
+ gst_pad_push_event (tensor_mux->srcpad, gst_event_new_eos ());
+ ret = GST_FLOW_EOS;
+ goto beach;
+ }
-/**
- * @brief event function (gst element vmethod)
- */
-static gboolean
-gst_tensor_mux_sink_event (GstPad * pad, GstObject * parent, GstEvent * event)
-{
- GstTensorMux *mux = GST_TENSOR_MUX (parent);
- gboolean is_pad;
- gboolean ret = TRUE;
+ if (!tensor_mux->negotiated) {
+ GstCaps *newcaps;
+ newcaps =
+ gst_caps_new_simple ("other/tensors", "rank", G_TYPE_INT,
+ tensor_mux->rank, "num_tensors", G_TYPE_INT,
+ tensor_mux->num_tensors, "types", G_TYPE_STRING,
+ types->str, "framerate", GST_TYPE_FRACTION,
+ tensor_mux->framerate_numerator, tensor_mux->framerate_denominator,
+ "dimensions", G_TYPE_STRING, dimensions->str, NULL);
+ if (!gst_pad_set_caps (tensor_mux->srcpad, newcaps)) {
+ gst_caps_unref (newcaps);
+ goto nego_error;
+ }
- GST_OBJECT_LOCK (mux);
- is_pad = (pad == mux->last_pad);
- GST_OBJECT_UNLOCK (mux);
+ gst_caps_unref (newcaps);
+ tensor_mux->negotiated = TRUE;
+ }
- switch (GST_EVENT_TYPE (event)) {
- case GST_EVENT_CAPS:
- {
- GstCaps *caps;
- gst_event_parse_caps (event, &caps);
- GST_LOG_OBJECT (pad, "Received caps-event with caps: %"
- GST_PTR_FORMAT, caps);
- ret = gst_tensor_mux_setcaps (pad, mux, caps);
- gst_event_unref (event);
- return ret;
- }
- break;
- case GST_EVENT_FLUSH_STOP:
- {
- GST_OBJECT_LOCK (mux);
- mux->last_stop = GST_CLOCK_TIME_NONE;
- GST_OBJECT_UNLOCK (mux);
- break;
- }
- case GST_EVENT_SEGMENT:
- {
- GstTensorMuxPadPrivate *padpriv;
- GST_OBJECT_LOCK (mux);
- padpriv = gst_pad_get_element_private (pad);
-
- if (padpriv) {
- gst_event_copy_segment (event, &padpriv->segment);
- }
- GST_OBJECT_UNLOCK (mux);
-
- if (is_pad) {
- GstSegment new_segment;
- gst_segment_init (&new_segment, GST_FORMAT_TIME);
- gst_event_unref (event);
- event = gst_event_new_segment (&new_segment);
- }
- break;
+ if (tensor_mux->need_segment) {
+ GstSegment segment;
+ if (dts_time != GST_CLOCK_TIME_NONE) {
+ time = dts_time;
+ } else if (pts_time != GST_CLOCK_TIME_NONE) {
+ time = pts_time;
+ } else {
+ time = 0;
}
- default:
- break;
+
+ gst_segment_init (&segment, GST_FORMAT_TIME);
+ segment.start = time;
+ gst_pad_push_event (tensor_mux->srcpad, gst_event_new_segment (&segment));
+ tensor_mux->need_segment = FALSE;
}
- if (is_pad) {
- return gst_pad_push_event (mux->srcpad, event);
- } else {
- gst_event_unref (event);
- return ret;
+ ret = gst_pad_push (tensor_mux->srcpad, tensors_buf);
+ if (ret != GST_FLOW_OK)
+ goto beach;
+beach:
+ return ret;
+nego_error:
+ {
+ GST_WARNING_OBJECT (tensor_mux, "failed to set caps");
+ GST_ELEMENT_ERROR (tensor_mux, CORE, NEGOTIATION, (NULL), (NULL));
+ return GST_FLOW_NOT_NEGOTIATED;
}
}
static void
gst_tensor_mux_ready_to_paused (GstTensorMux * tensor_mux)
{
- GST_OBJECT_LOCK (tensor_mux);
- g_clear_object (&tensor_mux->last_pad);
- tensor_mux->send_stream_start = TRUE;
- GST_OBJECT_UNLOCK (tensor_mux);
+ tensor_mux->need_stream_start = TRUE;
+ tensor_mux->need_segment = TRUE;
+ tensor_mux->negotiated = FALSE;
+ gst_collect_pads_start (tensor_mux->collect);
}
/**
{
GstTensorMux *tensor_mux;
GstStateChangeReturn ret;
-
tensor_mux = GST_TENSOR_MUX (element);
-
switch (transition) {
case GST_STATE_CHANGE_READY_TO_PAUSED:
gst_tensor_mux_ready_to_paused (tensor_mux);
break;
+ case GST_STATE_CHANGE_PAUSED_TO_READY:
+ gst_collect_pads_stop (tensor_mux->collect);
+ break;
default:
break;
}
- ret = GST_ELEMENT_CLASS (gst_tensor_mux_parent_class)->change_state (element,
- transition);
-
+ ret = GST_ELEMENT_CLASS (parent_class)->change_state (element, transition);
+ if (ret == GST_STATE_CHANGE_FAILURE)
+ return ret;
switch (transition) {
case GST_STATE_CHANGE_PAUSED_TO_READY:
- g_clear_object (&tensor_mux->last_pad);
break;
default:
break;
const GValue * value, GParamSpec * pspec)
{
GstTensorMux *filter = GST_TENSOR_MUX (object);
-
switch (prop_id) {
case PROP_SILENT:
filter->silent = g_value_get_boolean (value);
GValue * value, GParamSpec * pspec)
{
GstTensorMux *filter = GST_TENSOR_MUX (object);
-
switch (prop_id) {
case PROP_SILENT:
g_value_set_boolean (value, filter->silent);
* initialize the plug-in itself
* register the element factories and other features
*/
-static gboolean
+gboolean
gst_tensor_mux_plugin_init (GstPlugin * tensormux)
{
/** debug category for fltering log messages
*/
GST_DEBUG_CATEGORY_INIT (gst_tensor_mux_debug, "tensormux", 0,
"Tensor Muxer");
-
return gst_element_register (tensormux, "tensormux",
GST_RANK_NONE, GST_TYPE_TENSOR_MUX);
}
projects / platform / upstream / nnstreamer.git / commitdiff