bugfix: added content authors to all extractors

build.gradle

@@ -103,6 +103,7 @@ subprojects {
         testImplementation group: 'org.junit.jupiter', name: 'junit-jupiter-engine', version: version_junit
         testImplementation group: 'org.junit.jupiter', name: 'junit-jupiter-params', version: version_junit
         testImplementation group: 'org.junit.platform', name: 'junit-platform-commons', version: version_junit_platform
+        testImplementation(group: "org.jetbrains.kotlinx", "name": "kotlinx-coroutines-test", version: version_kotlinx_coroutines)
     }
 }
 

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

@@ -3,12 +3,17 @@ package org.vitrivr.engine.core.features
 import io.github.oshai.kotlinlogging.KLogger
 import io.github.oshai.kotlinlogging.KotlinLogging
 import kotlinx.coroutines.CoroutineScope
-import kotlinx.coroutines.flow.*
+import kotlinx.coroutines.flow.Flow
+import kotlinx.coroutines.flow.asFlow
+import kotlinx.coroutines.flow.emitAll
+import kotlinx.coroutines.flow.flow
 import org.vitrivr.engine.core.model.content.element.ContentElement
 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.operators.Operator
 import org.vitrivr.engine.core.operators.ingest.Extractor
 
@@ -19,15 +24,27 @@ 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>?, private val bufferSize: Int = 100) : Extractor<C, D> {
-
-    private val logger: KLogger = KotlinLogging.logger {}
+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> {
 
+    companion object {
+        const val BATCH_SIZE_KEY = "batchSize"
+    }
 
     init {
         require(field == null || this.field.analyser == this.analyser) { "Field and analyser do not match! This is a programmer's error!" }
     }
 
+    /** 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:
      *
@@ -101,4 +118,23 @@ abstract class AbstractBatchedExtractor<C : ContentElement<*>, D : Descriptor<*>
      */
     protected abstract fun extract(retrievables: List<Retrievable>): List<List<D>>
 
+    /**
+     * Filters the content of a [Retrievable] based on the [ContentAuthorAttribute] and the [contentSources] parameter.
+     *
+     * @param retrievable [Retrievable] to extract content from.
+     */
+    @Suppress("UNCHECKED_CAST")
+    protected fun filterContent(retrievable: Retrievable): List<C> {
+        val contentIds = this.contentSources?.let {
+            retrievable.filteredAttribute(ContentAuthorAttribute::class.java)?.getContentIds(it)
+        }
+        return retrievable.content.filter { content ->
+            if (this.analyser.contentClasses.none { it.isInstance(content) }) return@filter false
+            if (contentIds == null) {
+                return@filter true
+            } else {
+                return@filter contentIds.contains(content.id)
+            }
+        }.map { it as C }
+    }
 }

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

@@ -10,23 +10,30 @@ 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.operators.Operator
 import org.vitrivr.engine.core.operators.ingest.Extractor
 
 /**
  * An abstract [Extractor] implementation that is suitable for most default [Extractor] implementations.
  *
  * @author Ralph Gasser
- * @version 1.2.0
+ * @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) : Extractor<C, D> {
-
-    protected val logger: KLogger = KotlinLogging.logger {}
+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> {
 
     init {
         require(field == null || this.field.analyser == this.analyser) { "Field and analyser do not match! This is a programmer's error!" }
     }
 
+    /** 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:
      *
@@ -36,7 +43,7 @@ abstract class AbstractExtractor<C : ContentElement<*>, D : Descriptor<*>>(final
      *
      * @return [Flow] of [Retrievable]
      */
-    final override fun toFlow(scope: CoroutineScope): Flow<Retrievable> = this.input.toFlow(scope).onEach { retrievable ->
+    override fun toFlow(scope: CoroutineScope): Flow<Retrievable> = this.input.toFlow(scope).onEach { retrievable ->
         if (this.matches(retrievable)) {
             /* Perform extraction. */
             val descriptors = try {
@@ -74,4 +81,24 @@ abstract class AbstractExtractor<C : ContentElement<*>, D : Descriptor<*>>(final
      * @return List of resulting [Descriptor]s.
      */
     protected abstract fun extract(retrievable: Retrievable): List<D>
+
+    /**
+     * Filters the content of a [Retrievable] based on the [ContentAuthorAttribute] and the [contentSources] parameter.
+     *
+     * @param retrievable [Retrievable] to extract content from.
+     */
+    @Suppress("UNCHECKED_CAST")
+    protected fun filterContent(retrievable: Retrievable): List<C> {
+        val contentIds = this.contentSources?.let {
+            retrievable.filteredAttribute(ContentAuthorAttribute::class.java)?.getContentIds(it)
+        }
+        return retrievable.content.filter { content ->
+            if (this.analyser.contentClasses.none { it.isInstance(content) }) return@filter false
+            if (contentIds == null) {
+                return@filter true
+            } else {
+                return@filter contentIds.contains(content.id)
+            }
+        }.map { it as C }
+    }
 }

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

@@ -9,6 +9,7 @@ import org.vitrivr.engine.core.model.content.Content
 import org.vitrivr.engine.core.model.content.element.ImageContent
 import org.vitrivr.engine.core.model.descriptor.vector.FloatVectorDescriptor
 import org.vitrivr.engine.core.model.metamodel.Analyser
+import org.vitrivr.engine.core.model.metamodel.Analyser.Companion.merge
 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
@@ -51,7 +52,7 @@ 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)
+    override fun newExtractor(field: Schema.Field<ImageContent, FloatVectorDescriptor>, input: Operator<Retrievable>, context: IndexContext) = AverageColorExtractor(input, this, field, merge(field, context))
 
     /**
      * Generates and returns a new [AverageColorExtractor] instance for this [AverageColor].
@@ -63,7 +64,7 @@ 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)
+    override fun newExtractor(name: String, input: Operator<Retrievable>, context: IndexContext): Extractor<ImageContent, FloatVectorDescriptor> = AverageColorExtractor(input, this, null, context.local[name] ?: emptyMap())
 
     /**
      * Generates and returns a new [DenseRetriever] instance for this [AverageColor].

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

@@ -8,6 +8,7 @@ import org.vitrivr.engine.core.model.descriptor.Descriptor
 import org.vitrivr.engine.core.model.descriptor.vector.FloatVectorDescriptor
 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.operators.Operator
 import org.vitrivr.engine.core.operators.ingest.Extractor
 import org.vitrivr.engine.core.source.file.FileSource
@@ -20,7 +21,9 @@ import org.vitrivr.engine.core.source.file.FileSource
  * @author Luca Rossetto
  * @version 1.2.0
  */
-class AverageColorExtractor(input: Operator<Retrievable>, analyser: AverageColor, field: Schema.Field<ImageContent, FloatVectorDescriptor>?) : AbstractExtractor<ImageContent, FloatVectorDescriptor>(input, analyser, field) {
+class AverageColorExtractor(input: Operator<Retrievable>, analyser: AverageColor, field: Schema.Field<ImageContent, FloatVectorDescriptor>?, parameters : Map<String, String>) : AbstractExtractor<ImageContent, FloatVectorDescriptor>(input, analyser, field, parameters) {
+
+
     /**
      * Internal method to check, if [Retrievable] matches this [Extractor] and should thus be processed.
      *
@@ -38,7 +41,7 @@ class AverageColorExtractor(input: Operator<Retrievable>, analyser: AverageColor
      * @return List of resulting [Descriptor]s.
      */
     override fun extract(retrievable: Retrievable): List<FloatVectorDescriptor> {
-        val content = retrievable.content.filterIsInstance<ImageContent>()
+        val content = this.filterContent(retrievable)
         return content.map { (this.analyser as AverageColor).analyse(it).copy(retrievableId = retrievable.id, field = this.field) }
     }
 }

vitrivr-engine-core/src/main/kotlin/org/vitrivr/engine/core/features/metadata/source/exif/ExifMetadata.kt

@@ -6,6 +6,7 @@ import org.vitrivr.engine.core.model.content.element.ContentElement
 import org.vitrivr.engine.core.model.descriptor.Attribute
 import org.vitrivr.engine.core.model.descriptor.struct.AnyMapStructDescriptor
 import org.vitrivr.engine.core.model.metamodel.Analyser
+import org.vitrivr.engine.core.model.metamodel.Analyser.Companion.merge
 import org.vitrivr.engine.core.model.metamodel.Schema
 import org.vitrivr.engine.core.model.query.Query
 import org.vitrivr.engine.core.model.query.bool.SimpleBooleanQuery
@@ -39,7 +40,7 @@ class ExifMetadata : Analyser<ContentElement<*>, AnyMapStructDescriptor> {
      *
      * @return A new [Extractor] instance for this [Analyser]
      */
-    override fun newExtractor(name: String, input: Operator<Retrievable>, context: IndexContext) = ExifMetadataExtractor(input, this, null)
+    override fun newExtractor(name: String, input: Operator<Retrievable>, context: IndexContext) = ExifMetadataExtractor(input, this, null, context.local[name] ?: emptyMap())
 
     /**
      * Generates and returns a new [ExifMetadataExtractor] instance for this [ExifMetadata].
@@ -50,7 +51,7 @@ class ExifMetadata : Analyser<ContentElement<*>, AnyMapStructDescriptor> {
      *
      * @return A new [Extractor] instance for this [Analyser]
      */
-    override fun newExtractor(field: Schema.Field<ContentElement<*>, AnyMapStructDescriptor>, input: Operator<Retrievable>, context: IndexContext) = ExifMetadataExtractor(input, this, field)
+    override fun newExtractor(field: Schema.Field<ContentElement<*>, AnyMapStructDescriptor>, input: Operator<Retrievable>, context: IndexContext) = ExifMetadataExtractor(input, this, field, merge(field, context))
 
     /**
      *

vitrivr-engine-core/src/main/kotlin/org/vitrivr/engine/core/features/metadata/source/exif/ExifMetadataExtractor.kt

@@ -87,7 +87,7 @@ private fun JsonObject.convertType(type: Type): Value<*>? {
     }
 }
 
-class ExifMetadataExtractor(input: Operator<Retrievable>, analyser: ExifMetadata, field: Schema.Field<ContentElement<*>, AnyMapStructDescriptor>?) : AbstractExtractor<ContentElement<*>, AnyMapStructDescriptor>(input, analyser, field) {
+class ExifMetadataExtractor(input: Operator<Retrievable>, analyser: ExifMetadata, field: Schema.Field<ContentElement<*>, AnyMapStructDescriptor>?, parameters: Map<String,String>) : AbstractExtractor<ContentElement<*>, AnyMapStructDescriptor>(input, analyser, field, parameters) {
 
 
     override fun matches(retrievable: Retrievable): Boolean =

vitrivr-engine-core/src/main/kotlin/org/vitrivr/engine/core/features/metadata/source/file/FileSourceMetadata.kt

@@ -5,6 +5,7 @@ import org.vitrivr.engine.core.context.QueryContext
 import org.vitrivr.engine.core.model.content.element.ContentElement
 import org.vitrivr.engine.core.model.descriptor.struct.metadata.source.FileSourceMetadataDescriptor
 import org.vitrivr.engine.core.model.metamodel.Analyser
+import org.vitrivr.engine.core.model.metamodel.Analyser.Companion.merge
 import org.vitrivr.engine.core.model.metamodel.Schema
 import org.vitrivr.engine.core.model.query.Query
 import org.vitrivr.engine.core.model.query.bool.SimpleBooleanQuery
@@ -39,7 +40,7 @@ class FileSourceMetadata : Analyser<ContentElement<*>, FileSourceMetadataDescrip
      *
      * @return [FileSourceMetadataExtractor]
      */
-    override fun newExtractor(field: Schema.Field<ContentElement<*>, FileSourceMetadataDescriptor>, input: Operator<Retrievable>, context: IndexContext) = FileSourceMetadataExtractor(input, this, field)
+    override fun newExtractor(field: Schema.Field<ContentElement<*>, FileSourceMetadataDescriptor>, input: Operator<Retrievable>, context: IndexContext) = FileSourceMetadataExtractor(input, this, field, merge(field, context))
 
     /**
      * Generates and returns a new [FileSourceMetadataExtractor] for the provided [Schema.Field].
@@ -50,7 +51,7 @@ class FileSourceMetadata : Analyser<ContentElement<*>, FileSourceMetadataDescrip
      *
      * @return [FileSourceMetadataExtractor]
      */
-    override fun newExtractor(name: String, input: Operator<Retrievable>, context: IndexContext) = FileSourceMetadataExtractor(input, this, null)
+    override fun newExtractor(name: String, input: Operator<Retrievable>, context: IndexContext) = FileSourceMetadataExtractor(input, this, null, context.local[name] ?: emptyMap())
 
     /**
      * Generates and returns a new [FileSourceMetadataRetriever] for the provided [Schema.Field].

vitrivr-engine-core/src/main/kotlin/org/vitrivr/engine/core/features/metadata/source/file/FileSourceMetadataExtractor.kt

@@ -22,8 +22,8 @@ import kotlin.io.path.absolutePathString
  * @author Ralph Gasser
  * @version 1.2.0
  */
-class FileSourceMetadataExtractor(input: Operator<Retrievable>, analyser: FileSourceMetadata, field: Schema.Field<ContentElement<*>, FileSourceMetadataDescriptor>?) :
-    AbstractExtractor<ContentElement<*>, FileSourceMetadataDescriptor>(input, analyser, field) {
+class FileSourceMetadataExtractor(input: Operator<Retrievable>, analyser: FileSourceMetadata, field: Schema.Field<ContentElement<*>, FileSourceMetadataDescriptor>?, parameters: Map<String,String>) :
+    AbstractExtractor<ContentElement<*>, FileSourceMetadataDescriptor>(input, analyser, field, parameters) {
     /**
      * 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/metadata/source/video/VideoSourceMetadata.kt

@@ -9,6 +9,7 @@ import org.vitrivr.engine.core.model.content.element.ContentElement
 import org.vitrivr.engine.core.model.descriptor.struct.metadata.source.FileSourceMetadataDescriptor
 import org.vitrivr.engine.core.model.descriptor.struct.metadata.source.VideoSourceMetadataDescriptor
 import org.vitrivr.engine.core.model.metamodel.Analyser
+import org.vitrivr.engine.core.model.metamodel.Analyser.Companion.merge
 import org.vitrivr.engine.core.model.metamodel.Schema
 import org.vitrivr.engine.core.model.query.Query
 import org.vitrivr.engine.core.model.query.bool.BooleanQuery
@@ -43,7 +44,7 @@ class VideoSourceMetadata : Analyser<ContentElement<*>, VideoSourceMetadataDescr
      *
      * @return [FileSourceMetadataExtractor]
      */
-    override fun newExtractor(field: Schema.Field<ContentElement<*>, VideoSourceMetadataDescriptor>, input: Operator<Retrievable>, context: IndexContext) = VideoSourceMetadataExtractor(input, this, field)
+    override fun newExtractor(field: Schema.Field<ContentElement<*>, VideoSourceMetadataDescriptor>, input: Operator<Retrievable>, context: IndexContext) = VideoSourceMetadataExtractor(input, this, field, merge(field, context))
 
     /**
      * Generates and returns a new [FileSourceMetadataExtractor] for the provided [Schema.Field].
@@ -54,7 +55,7 @@ class VideoSourceMetadata : Analyser<ContentElement<*>, VideoSourceMetadataDescr
      *
      * @return [FileSourceMetadataExtractor]
      */
-    override fun newExtractor(name: String, input: Operator<Retrievable>, context: IndexContext): Extractor<ContentElement<*>, VideoSourceMetadataDescriptor> = VideoSourceMetadataExtractor(input, this, null)
+    override fun newExtractor(name: String, input: Operator<Retrievable>, context: IndexContext): Extractor<ContentElement<*>, VideoSourceMetadataDescriptor> = VideoSourceMetadataExtractor(input, this, null, context.local[name] ?: emptyMap())
 
     /**
      * Generates and returns a new [VideoSourceMetadataRetriever] for the provided [Schema.Field].

vitrivr-engine-core/src/main/kotlin/org/vitrivr/engine/core/features/metadata/source/video/VideoSourceMetadataExtractor.kt

@@ -21,8 +21,8 @@ import java.util.*
  * @author Ralph Gasser
  * @version 1.1.0
  */
-class VideoSourceMetadataExtractor(input: Operator<Retrievable>, analyser: VideoSourceMetadata, field: Schema.Field<ContentElement<*>, VideoSourceMetadataDescriptor>?) :
-    AbstractExtractor<ContentElement<*>, VideoSourceMetadataDescriptor>(input, analyser, field) {
+class VideoSourceMetadataExtractor(input: Operator<Retrievable>, analyser: VideoSourceMetadata, field: Schema.Field<ContentElement<*>, VideoSourceMetadataDescriptor>?, parameters: Map<String,String>) :
+    AbstractExtractor<ContentElement<*>, VideoSourceMetadataDescriptor>(input, analyser, field, parameters) {
     /**
      * Internal method to check, if [Retrievable] matches this [Extractor] and should thus be processed.
      *