feat(api-v1): Add support for PDF files (DSP-1267)

test_data/all_data/permissions-data.ttl

@@ -51,7 +51,18 @@
 
                       knora-admin:forProperty knora-base:hasStillImageFileValue ;
 
-                      knora-base:hasPermissions "M knora-admin:Creator,knora-admin:ProjectMember|V knora-admin:KnownUser|RV knora-admin:UnknownUser"^^xsd:string .
+                      knora-base:hasPermissions "M knora-admin:Creator,knora-admin:ProjectMember|V knora-admin:KnownUser|V knora-admin:UnknownUser"^^xsd:string .
+
+### Default Object Access Permissions on knora-base:hasDocumentFileValue property
+<http://rdfh.ch/permissions/0000/001-d4>
+
+                      rdf:type knora-admin:DefaultObjectAccessPermission ;
+
+                      knora-admin:forProject knora-admin:SystemProject ;
+
+                      knora-admin:forProperty knora-base:hasDocumentFileValue ;
+
+                      knora-base:hasPermissions "M knora-admin:Creator,knora-admin:ProjectMember|V knora-admin:KnownUser|V knora-admin:UnknownUser"^^xsd:string .
 
 
 

webapi/src/main/scala/org/knora/webapi/messages/util/ValueUtilV1.scala

@@ -88,6 +88,8 @@ class ValueUtilV1(private val settings: KnoraSettingsImpl) {
         makeStillImageValue(valueProps, projectShortcode, responderManager, userProfile)
       case OntologyConstants.KnoraBase.TextFileValue =>
         makeTextFileValue(valueProps, projectShortcode, responderManager, userProfile)
+      case OntologyConstants.KnoraBase.DocumentFileValue =>
+        makeDocumentFileValue(valueProps, projectShortcode, responderManager, userProfile)
       case OntologyConstants.KnoraBase.LinkValue => makeLinkValue(valueProps, responderManager, userProfile)
     }
   }
@@ -106,6 +108,16 @@ class ValueUtilV1(private val settings: KnoraSettingsImpl) {
     s"${settings.externalSipiIIIFGetUrl}/${imageFileValueV1.projectShortcode}/${imageFileValueV1.internalFilename}/full/${imageFileValueV1.dimX},${imageFileValueV1.dimY}/0/default.jpg"
   }
 
+  /**
+    * Creates a URL for accessing a document file via Sipi.
+    *
+    * @param documentFileValueV1 the document file value.
+    * @return a Sipi  URL.
+    */
+  def makeSipiDocumentGetUrlFromFilename(documentFileValueV1: DocumentFileValueV1): String = {
+    s"${settings.externalSipiIIIFGetUrl}/${documentFileValueV1.projectShortcode}/${documentFileValueV1.internalFilename}/file"
+  }
+
   /**
     * Creates a URL for accessing a text file via Sipi.
     *
@@ -168,6 +180,16 @@ class ValueUtilV1(private val settings: KnoraSettingsImpl) {
           ny = Some(stillImageFileValueV1.dimY),
           path = makeSipiImageGetUrlFromFilename(stillImageFileValueV1)
         )
+
+      case documentFileValueV1: DocumentFileValueV1 =>
+        LocationV1(
+          format_name = mimeType2V1Format(documentFileValueV1.internalMimeType),
+          origname = documentFileValueV1.originalFilename,
+          nx = documentFileValueV1.dimX,
+          ny = documentFileValueV1.dimY,
+          path = makeSipiDocumentGetUrlFromFilename(documentFileValueV1)
+        )
+
       case textFileValue: TextFileValueV1 =>
         LocationV1(
           format_name = mimeType2V1Format(textFileValue.internalMimeType),
@@ -343,6 +365,8 @@ class ValueUtilV1(private val settings: KnoraSettingsImpl) {
 
       case _: TextFileValueV1 => basicObjectResponse
 
+      case _: DocumentFileValueV1 => basicObjectResponse
+
       case _: HierarchicalListValueV1 => basicObjectResponse
 
       case _: ColorValueV1 => basicObjectResponse
@@ -804,6 +828,31 @@ class ValueUtilV1(private val settings: KnoraSettingsImpl) {
       ))
   }
 
+  /**
+    * Converts a [[ValueProps]] into a [[DocumentFileValueV1]].
+    *
+    * @param valueProps a [[ValueProps]] representing the SPARQL query results to be converted.
+    * @return a [[DocumentFileValueV1]].
+    */
+  private def makeDocumentFileValue(
+      valueProps: ValueProps,
+      projectShortcode: String,
+      responderManager: ActorRef,
+      userProfile: UserADM)(implicit timeout: Timeout, executionContext: ExecutionContext): Future[ApiValueV1] = {
+    val predicates = valueProps.literalData
+
+    Future(
+      DocumentFileValueV1(
+        internalMimeType = predicates(OntologyConstants.KnoraBase.InternalMimeType).literals.head,
+        internalFilename = predicates(OntologyConstants.KnoraBase.InternalFilename).literals.head,
+        originalFilename = predicates.get(OntologyConstants.KnoraBase.OriginalFilename).map(_.literals.head),
+        projectShortcode = projectShortcode,
+        pageCount = predicates(OntologyConstants.KnoraBase.PageCount).literals.head.toInt,
+        dimX = predicates.get(OntologyConstants.KnoraBase.DimX).flatMap(_.literals.headOption.map(_.toInt)),
+        dimY = predicates.get(OntologyConstants.KnoraBase.DimY).flatMap(_.literals.headOption.map(_.toInt))
+      ))
+  }
+
   /**
     * Converts a [[ValueProps]] into a [[LinkValueV1]].
     *

webapi/src/main/scala/org/knora/webapi/messages/v1/responder/valuemessages/ValueMessagesV1.scala

@@ -35,12 +35,7 @@ import org.knora.webapi.messages.v1.responder.resourcemessages.LocationV1
 import org.knora.webapi.messages.v1.responder.{KnoraRequestV1, KnoraResponseV1}
 import org.knora.webapi.messages.v2.responder.UpdateResultInProject
 import org.knora.webapi.messages.v2.responder.standoffmessages._
-import org.knora.webapi.messages.v2.responder.valuemessages.{
-  FileValueContentV2,
-  FileValueV2,
-  StillImageFileValueContentV2,
-  TextFileValueContentV2
-}
+import org.knora.webapi.messages.v2.responder.valuemessages._
 import org.knora.webapi.messages.{OntologyConstants, StringFormatter}
 import spray.json._
 
@@ -1574,6 +1569,75 @@ case class StillImageFileValueV1(internalMimeType: String,
   }
 }
 
+/**
+  * A representation of a document in a binary format.
+  *
+  * @param internalMimeType the MIME-type of the internal representation.
+  * @param internalFilename the internal filename of the object.
+  * @param originalFilename the original filename of the object at the time of the import.
+  * @param pageCount        the number of pages in the document.
+  * @param dimX             the X dimension of the object.
+  * @param dimY             the Y dimension of the object.
+  */
+case class DocumentFileValueV1(internalMimeType: String,
+                               internalFilename: String,
+                               originalFilename: Option[String] = None,
+                               originalMimeType: Option[String] = None,
+                               projectShortcode: String,
+                               pageCount: Int,
+                               dimX: Option[Int],
+                               dimY: Option[Int])
+    extends FileValueV1 {
+  def valueTypeIri: IRI = OntologyConstants.KnoraBase.DocumentFileValue
+
+  def toJsValue: JsValue = ApiValueV1JsonProtocol.documentFileValueV1Format.write(this)
+
+  override def toString: String = internalFilename
+
+  /**
+    * Checks if a new document file value would duplicate an existing document file value.
+    *
+    * @param other another [[ValueV1]].
+    * @return `true` if `other` is a duplicate of `this`.
+    */
+  override def isDuplicateOfOtherValue(other: ApiValueV1): Boolean = {
+    other match {
+      case documentFileValueV1: DocumentFileValueV1 => documentFileValueV1 == this
+      case otherValue =>
+        throw InconsistentRepositoryDataException(s"Cannot compare a $valueTypeIri to a ${otherValue.valueTypeIri}")
+    }
+  }
+
+  /**
+    * Checks if a new version of a document file value would be redundant given the current version of the value.
+    *
+    * @param currentVersion the current version of the value.
+    * @return `true` if this [[UpdateValueV1]] is redundant given `currentVersion`.
+    */
+  override def isRedundant(currentVersion: ApiValueV1): Boolean = {
+    currentVersion match {
+      case documentFileValueV1: DocumentFileValueV1 => documentFileValueV1 == this
+      case other =>
+        throw InconsistentRepositoryDataException(s"Cannot compare a $valueTypeIri to a ${other.valueTypeIri}")
+    }
+  }
+
+  override def toFileValueContentV2: FileValueContentV2 = {
+    DocumentFileValueContentV2(
+      ontologySchema = InternalSchema,
+      fileValue = FileValueV2(
+        internalFilename = internalFilename,
+        internalMimeType = internalMimeType,
+        originalFilename = originalFilename,
+        originalMimeType = Some(internalMimeType)
+      ),
+      pageCount = pageCount,
+      dimX = dimX,
+      dimY = dimY
+    )
+  }
+}
+
 case class MovingImageFileValueV1(internalMimeType: String,
                                   internalFilename: String,
                                   originalFilename: Option[String],
@@ -1756,6 +1820,7 @@ object ApiValueV1JsonProtocol extends SprayJsonSupport with DefaultJsonProtocol
   implicit val valueGetResponseV1Format: RootJsonFormat[ValueGetResponseV1] = jsonFormat7(ValueGetResponseV1)
   implicit val dateValueV1Format: JsonFormat[DateValueV1] = jsonFormat5(DateValueV1)
   implicit val stillImageFileValueV1Format: JsonFormat[StillImageFileValueV1] = jsonFormat7(StillImageFileValueV1)
+  implicit val documentFileValueV1Format: JsonFormat[DocumentFileValueV1] = jsonFormat8(DocumentFileValueV1)
   implicit val textFileValueV1Format: JsonFormat[TextFileValueV1] = jsonFormat5(TextFileValueV1)
   implicit val movingImageFileValueV1Format: JsonFormat[MovingImageFileValueV1] = jsonFormat5(MovingImageFileValueV1)
   implicit val valueVersionV1Format: JsonFormat[ValueVersionV1] = jsonFormat3(ValueVersionV1)

webapi/src/main/scala/org/knora/webapi/responders/v1/ResourcesResponderV1.scala

@@ -2764,8 +2764,8 @@ class ResourcesResponderV1(responderData: ResponderData) extends Responder(respo
             valueUtilV1.fileValueV12LocationV1(fullSizeImageFileValueToPreview(fullSizeImageFileValue))
         }
 
-        // Convert the full-resolution file values into LocationV1 objects as required by Knora API v1.
-        locations: Seq[LocationV1] = preview.toVector ++ fullSizeImageFileValues.flatMap { fileValueV1 =>
+        // Convert the file values into LocationV1 objects as required by Knora API v1.
+        locations: Seq[LocationV1] = preview.toVector ++ fileValues.flatMap { fileValueV1 =>
           createMultipleImageResolutions(fileValueV1).map(oneResolution =>
             valueUtilV1.fileValueV12LocationV1(oneResolution))
         }

webapi/src/main/scala/org/knora/webapi/routing/RouteUtilV1.scala

@@ -33,7 +33,12 @@ import org.knora.webapi.messages.admin.responder.usersmessages.UserADM
 import org.knora.webapi.messages.store.sipimessages.GetFileMetadataResponse
 import org.knora.webapi.messages.util.standoff.StandoffTagUtilV2
 import org.knora.webapi.messages.util.standoff.StandoffTagUtilV2.TextWithStandoffTagsV2
-import org.knora.webapi.messages.v1.responder.valuemessages.{FileValueV1, StillImageFileValueV1, TextFileValueV1}
+import org.knora.webapi.messages.v1.responder.valuemessages.{
+  DocumentFileValueV1,
+  FileValueV1,
+  StillImageFileValueV1,
+  TextFileValueV1
+}
 import org.knora.webapi.messages.v1.responder.{KnoraRequestV1, KnoraResponseV1}
 import org.knora.webapi.messages.v2.responder.standoffmessages.{GetMappingRequestV2, GetMappingResponseV2}
 import org.knora.webapi.settings.KnoraSettingsImpl
@@ -257,6 +262,13 @@ object RouteUtilV1 {
     "text/plain"
   )
 
+  /**
+    * MIME types used in Sipi to store document files.
+    */
+  private val documentMimeTypes: Set[String] = Set(
+    "application/pdf"
+  )
+
   /**
     * Converts file metadata from Sipi into a [[FileValueV1]].
     *
@@ -287,6 +299,18 @@ object RouteUtilV1 {
         originalMimeType = fileMetadataResponse.originalMimeType,
         projectShortcode = projectShortcode
       )
+    } else if (documentMimeTypes.contains(fileMetadataResponse.internalMimeType)) {
+      DocumentFileValueV1(
+        internalFilename = filename,
+        internalMimeType = fileMetadataResponse.internalMimeType,
+        originalFilename = fileMetadataResponse.originalFilename,
+        originalMimeType = fileMetadataResponse.originalMimeType,
+        projectShortcode = projectShortcode,
+        pageCount = fileMetadataResponse.pageCount.getOrElse(
+          throw SipiException(s"Sipi did not return the page count of the document")),
+        dimX = fileMetadataResponse.width,
+        dimY = fileMetadataResponse.height
+      )
     } else {
       throw BadRequestException(s"MIME type ${fileMetadataResponse.internalMimeType} not supported in Knora API v1")
     }

webapi/src/main/scala/org/knora/webapi/store/iiif/SipiConnector.scala

@@ -69,6 +69,12 @@ class SipiConnector extends Actor with ActorLogging {
 
   private val httpClient: CloseableHttpClient = HttpClients.custom.setDefaultRequestConfig(sipiRequestConfig).build
 
+  // Sipi's /knora.json route only returns the correct original filename for images.
+  private val internalMimeTypesForWhichSipiReturnsTheCorrectOriginalFilename: Set[String] = Set(
+    "image/jpx",
+    "image/jp2"
+  )
+
   override def receive: Receive = {
     case getFileMetadataRequest: GetFileMetadataRequest =>
       try2Message(sender(), getFileMetadata(getFileMetadataRequest), log)
@@ -128,7 +134,10 @@ class SipiConnector extends Actor with ActorLogging {
       sipiResponse: SipiKnoraJsonResponse = sipiResponseStr.parseJson.convertTo[SipiKnoraJsonResponse]
     } yield
       GetFileMetadataResponse(
-        originalFilename = sipiResponse.originalFilename,
+        originalFilename =
+          if (internalMimeTypesForWhichSipiReturnsTheCorrectOriginalFilename.contains(sipiResponse.internalMimeType))
+            sipiResponse.originalFilename
+          else None,
         originalMimeType = sipiResponse.originalMimeType,
         internalMimeType = sipiResponse.internalMimeType,
         width = sipiResponse.width,

webapi/src/main/twirl/org/knora/webapi/messages/twirl/queries/sparql/v1/addValueVersion.scala.txt

@@ -265,6 +265,44 @@ DELETE {
                 }
             }
 
+            case documentFileValue: DocumentFileValueV1 => {
+                ?newValue knora-base:internalFilename """@documentFileValue.internalFilename""" .
+                ?newValue knora-base:internalMimeType """@documentFileValue.internalMimeType""" .
+                ?newValue knora-base:pageCount @documentFileValue.pageCount .
+
+                @documentFileValue.dimX match {
+                    case Some(definedDimX) => {
+                        ?newValue knora-base:dimX @definedDimX .
+                    }
+
+                    case None => {}
+                }
+
+                @documentFileValue.dimY match {
+                    case Some(definedDimY) => {
+                        ?newValue knora-base:dimY @definedDimY .
+                    }
+
+                    case None => {}
+                }
+
+                @documentFileValue.originalFilename match {
+                    case Some(definedOriginalFilename) => {
+                        ?newValue knora-base:originalFilename """@definedOriginalFilename""" .
+                    }
+
+                    case None => {}
+                }
+
+                @documentFileValue.originalMimeType match {
+                    case Some(definedOriginalMimeType) => {
+                        ?newValue knora-base:originalMimeType """@definedOriginalMimeType""" .
+                    }
+
+                    case None => {}
+                }
+            }
+
             case listValue: HierarchicalListValueV1 => {
 
                 ?newValue knora-base:valueHasListNode <@listValue.hierarchicalListIri> .

webapi/src/main/twirl/org/knora/webapi/messages/twirl/queries/sparql/v1/generateInsertStatementsForCreateValue.scala.txt

@@ -221,6 +221,44 @@
                 }
             }
 
+            case documentFileValue: DocumentFileValueV1 => {
+                <@newValueIri> knora-base:internalFilename """@documentFileValue.internalFilename""" ;
+                               knora-base:internalMimeType """@documentFileValue.internalMimeType""" ;
+                               knora-base:pageCount @documentFileValue.pageCount .
+
+                @documentFileValue.dimX match {
+                    case Some(definedDimX) => {
+                        <@newValueIri> knora-base:dimX @definedDimX .
+                    }
+
+                    case None => {}
+                }
+
+                @documentFileValue.dimY match {
+                    case Some(definedDimY) => {
+                        <@newValueIri> knora-base:dimY @definedDimY .
+                    }
+
+                    case None => {}
+                }
+
+                @documentFileValue.originalFilename match {
+                    case Some(definedOriginalFilename) => {
+                        <@newValueIri> knora-base:originalFilename """@definedOriginalFilename""" .
+                    }
+
+                    case None => {}
+                }
+
+                @documentFileValue.originalMimeType match {
+                    case Some(definedOriginalMimeType) => {
+                        <@newValueIri> knora-base:originalMimeType """@definedOriginalMimeType""" .
+                    }
+
+                    case None => {}
+                }
+            }
+
             case listValue: HierarchicalListValueV1 => {
 
                 <@newValueIri> knora-base:valueHasListNode <@listValue.hierarchicalListIri> .