dynamic aggregation of descriptors

vitrivr-engine-core/src/main/kotlin/org/vitrivr/engine/core/features/AbstractBatchedExtractor.kt

@@ -12,8 +12,8 @@ import org.vitrivr.engine.core.model.descriptor.Descriptor
 import org.vitrivr.engine.core.model.metamodel.Analyser
 import org.vitrivr.engine.core.model.metamodel.Schema
 import org.vitrivr.engine.core.model.retrievable.Retrievable
-import org.vitrivr.engine.core.model.retrievable.attributes.CONTENT_AUTHORS_KEY
 import org.vitrivr.engine.core.model.retrievable.attributes.ContentAuthorAttribute
+import org.vitrivr.engine.core.model.retrievable.attributes.DescriptorAuthorAttribute
 import org.vitrivr.engine.core.operators.Operator
 import org.vitrivr.engine.core.operators.ingest.Extractor
 
@@ -24,7 +24,37 @@ import org.vitrivr.engine.core.operators.ingest.Extractor
  * @author Ralph Gasser
  * @version 1.0.0
  */
-abstract class AbstractBatchedExtractor<C : ContentElement<*>, D : Descriptor<*>>(final override val input: Operator<Retrievable>, final override val analyser: Analyser<C, D>, final override val field: Schema.Field<C, D>?, protected val parameters: Map<String, String>) : Extractor<C, D> {
+abstract class AbstractBatchedExtractor<C : ContentElement<*>, D : Descriptor<*>>
+    private constructor(
+        final override val input: Operator<Retrievable>,
+        final override val analyser: Analyser<C, D>,
+        final override val field: Schema.Field<C, D>? = null,
+        protected val contentSources : Set<String>? = null,
+        final override val name: String,
+        private val bufferSize: Int
+) :
+    Extractor<C, D> {
+
+    constructor(input: Operator<Retrievable>, analyser: Analyser<C, D>, contentSources : Set<String>?, field: Schema.Field<C, D>, bufferSize: Int = 100) : this(
+        input,
+        analyser,
+        field,
+        contentSources,
+        field.fieldName,
+        bufferSize
+    )
+
+    constructor(input: Operator<Retrievable>, analyser: Analyser<C, D>, contentSources : Set<String>?, name: String, bufferSize: Int = 100) : this(
+        input,
+        analyser,
+        null,
+        contentSources,
+        name,
+        bufferSize
+    )
+
+
+
 
     companion object {
         const val BATCH_SIZE_KEY = "batchSize"
@@ -37,13 +67,6 @@ abstract class AbstractBatchedExtractor<C : ContentElement<*>, D : Descriptor<*>
     /** The [KLogger] instance used by this [AbstractExtractor]. */
     protected val logger: KLogger = KotlinLogging.logger {}
 
-    /** The names of the content source to consider during processing. */
-    protected val contentSources : Set<String>?
-        get() = this.parameters[CONTENT_AUTHORS_KEY]?.split(",")?.toSet()
-
-    /** The buffer- and batch size. */
-    private val bufferSize : Int
-        get() = this.parameters[BATCH_SIZE_KEY]?.toIntOrNull() ?: 1
 
     /**
      * A default [Extractor] implementation for batched extraction. It executes the following steps:
@@ -71,9 +94,12 @@ abstract class AbstractBatchedExtractor<C : ContentElement<*>, D : Descriptor<*>
                         logger.debug { "Batch size reached for field ${field?.fieldName}, extracting descriptors" }
                         val descriptors = extract(batch)
                         batch.forEachIndexed { i, r ->
+                            val sourceAttribute = DescriptorAuthorAttribute()
                             descriptors[i].forEach { d ->
                                 r.addDescriptor(d)
+                                sourceAttribute.add(d, name)
                             }
+                            r.addAttribute(sourceAttribute)
                         }
                         emitAll(batch.asFlow())
                         batch.clear()

vitrivr-engine-core/src/main/kotlin/org/vitrivr/engine/core/features/AbstractExtractor.kt

@@ -10,8 +10,8 @@ import org.vitrivr.engine.core.model.descriptor.Descriptor
 import org.vitrivr.engine.core.model.metamodel.Analyser
 import org.vitrivr.engine.core.model.metamodel.Schema
 import org.vitrivr.engine.core.model.retrievable.Retrievable
-import org.vitrivr.engine.core.model.retrievable.attributes.CONTENT_AUTHORS_KEY
 import org.vitrivr.engine.core.model.retrievable.attributes.ContentAuthorAttribute
+import org.vitrivr.engine.core.model.retrievable.attributes.DescriptorAuthorAttribute
 import org.vitrivr.engine.core.operators.Operator
 import org.vitrivr.engine.core.operators.ingest.Extractor
 
@@ -21,7 +21,29 @@ import org.vitrivr.engine.core.operators.ingest.Extractor
  * @author Ralph Gasser
  * @version 1.3.0
  */
-abstract class AbstractExtractor<C : ContentElement<*>, D : Descriptor<*>>(final override val input: Operator<Retrievable>, final override val analyser: Analyser<C, D>, final override val field: Schema.Field<C, D>? = null, protected val parameters: Map<String, String>) : Extractor<C, D> {
+abstract class AbstractExtractor<C : ContentElement<*>, D : Descriptor<*>> private constructor(
+    final override val input: Operator<Retrievable>,
+    final override val analyser: Analyser<C, D>,
+    final override val field: Schema.Field<C, D>? = null,
+    protected val contentSources : Set<String>? = null,
+    final override val name: String
+) : Extractor<C, D> {
+
+    constructor(input: Operator<Retrievable>, analyser: Analyser<C, D>, contentSources : Set<String>?, field: Schema.Field<C, D>) : this(
+        input,
+        analyser,
+        field,
+        contentSources,
+        field.fieldName
+    )
+
+    constructor(input: Operator<Retrievable>, analyser: Analyser<C, D>, contentSources : Set<String>?, name: String) : this(
+        input,
+        analyser,
+        null,
+        contentSources,
+        name
+    )
 
     init {
         require(field == null || this.field.analyser == this.analyser) { "Field and analyser do not match! This is a programmer's error!" }
@@ -30,10 +52,6 @@ abstract class AbstractExtractor<C : ContentElement<*>, D : Descriptor<*>>(final
     /** The [KLogger] instance used by this [AbstractExtractor]. */
     protected val logger: KLogger = KotlinLogging.logger {}
 
-    /** The names of the content source to consider during processing. */
-    protected val contentSources : Set<String>?
-        get() = this.parameters[CONTENT_AUTHORS_KEY]?.split(",")?.toSet()
-
     /**
      * A default [Extractor] implementation. It executes the following steps:
      *
@@ -43,24 +61,31 @@ abstract class AbstractExtractor<C : ContentElement<*>, D : Descriptor<*>>(final
      *
      * @return [Flow] of [Retrievable]
      */
-    override fun toFlow(scope: CoroutineScope): Flow<Retrievable> = this.input.toFlow(scope).onEach { retrievable ->
-        if (this.matches(retrievable)) {
-            /* Perform extraction. */
-            val descriptors = try {
-                logger.debug{"Extraction on field ${field?.fieldName} for retrievable: $retrievable" }
-                extract(retrievable)
-            } catch (e: Throwable) {
-                logger.error(e) { "Error during extraction of $retrievable" }
-                emptyList()
-            }
+    override fun toFlow(scope: CoroutineScope): Flow<Retrievable> =
+        this.input.toFlow(scope).onEach { retrievable ->
+            if (this.matches(retrievable)) {
+                /* Perform extraction. */
+                val descriptors = try {
+                    logger.debug{"Extraction on field ${field?.fieldName} for retrievable: $retrievable" }
+                    extract(retrievable)
+                } catch (e: Throwable) {
+                    logger.error(e) { "Error during extraction of $retrievable" }
+                    emptyList()
+                }
+
+                if (descriptors.isNotEmpty()) {
+                    /* Append descriptor. */
+                    logger.trace { "Extracted descriptors for retrievable ($retrievable): $descriptors" }
+                    val authorAttribute = DescriptorAuthorAttribute()
+                    for (d in descriptors) {
+                        retrievable.addDescriptor(d)
+                        authorAttribute.add(d, this.name)
+                    }
+                    retrievable.addAttribute(authorAttribute)
+                }
 
-            /* Append descriptor. */
-            logger.trace { "Extracted descriptors for retrievable ($retrievable): $descriptors" }
-            for (d in descriptors) {
-                retrievable.addDescriptor(d)
             }
         }
-    }
 
     /**
      * Internal method to check, if [Retrievable] matches this [Extractor] and should thus be processed.

vitrivr-engine-core/src/main/kotlin/org/vitrivr/engine/core/features/averagecolor/AverageColor.kt

@@ -14,6 +14,7 @@ import org.vitrivr.engine.core.model.metamodel.Schema
 import org.vitrivr.engine.core.model.query.Query
 import org.vitrivr.engine.core.model.query.proximity.ProximityQuery
 import org.vitrivr.engine.core.model.retrievable.Retrievable
+import org.vitrivr.engine.core.model.retrievable.attributes.CONTENT_AUTHORS_KEY
 import org.vitrivr.engine.core.model.types.Value
 import org.vitrivr.engine.core.operators.Operator
 import org.vitrivr.engine.core.operators.ingest.Extractor
@@ -52,7 +53,7 @@ class AverageColor : Analyser<ImageContent, FloatVectorDescriptor> {
      * @return A new [Extractor] instance for this [Analyser]
      * @throws [UnsupportedOperationException], if this [Analyser] does not support the creation of an [Extractor] instance.
      */
-    override fun newExtractor(field: Schema.Field<ImageContent, FloatVectorDescriptor>, input: Operator<Retrievable>, context: IndexContext) = AverageColorExtractor(input, this, field, merge(field, context))
+    override fun newExtractor(field: Schema.Field<ImageContent, FloatVectorDescriptor>, input: Operator<Retrievable>, context: IndexContext) = AverageColorExtractor(input, this, context[field.fieldName, CONTENT_AUTHORS_KEY]?.split(",")?.toSet(), field)
 
     /**
      * Generates and returns a new [AverageColorExtractor] instance for this [AverageColor].
@@ -64,7 +65,7 @@ class AverageColor : Analyser<ImageContent, FloatVectorDescriptor> {
      * @return A new [Extractor] instance for this [Analyser]
      * @throws [UnsupportedOperationException], if this [Analyser] does not support the creation of an [Extractor] instance.
      */
-    override fun newExtractor(name: String, input: Operator<Retrievable>, context: IndexContext): Extractor<ImageContent, FloatVectorDescriptor> = AverageColorExtractor(input, this, null, context.local[name] ?: emptyMap())
+    override fun newExtractor(name: String, input: Operator<Retrievable>, context: IndexContext): Extractor<ImageContent, FloatVectorDescriptor> = AverageColorExtractor(input, this, context[name, CONTENT_AUTHORS_KEY]?.split(",")?.toSet(), name)
 
     /**
      * Generates and returns a new [DenseRetriever] instance for this [AverageColor].