Added a bulk data filter, with tests

slicebox · Jun 4, 2017 · 7d797b5 · 7d797b5
1 parent 55b0a43
commit 7d797b5
Show file tree

Hide file tree

Showing 4 changed files with 124 additions and 8 deletions.
diff --git a/src/main/scala/se/nimsa/dcm4che/streams/DicomFlows.scala b/src/main/scala/se/nimsa/dcm4che/streams/DicomFlows.scala
@@ -21,7 +21,7 @@ import java.util.zip.Deflater
 import akka.NotUsed
 import akka.stream.scaladsl.{Flow, Source}
 import akka.util.ByteString
-import org.dcm4che3.data.{StandardElementDictionary, VR}
+import org.dcm4che3.data.{StandardElementDictionary, Tag, VR}
 import org.dcm4che3.io.DicomStreamException
 import se.nimsa.dcm4che.streams.DicomParts._
 
@@ -94,6 +94,14 @@ object DicomFlows {
     */
   def whitelistFilter(tagsWhitelist: Seq[Int]): Flow[DicomPart, DicomPart, NotUsed] = whitelistFilter(tagsWhitelist.contains(_))
 
+  /**
+    * Filter a stream of dicom parts such that attributes with tags in the black list are discarded.
+    *
+    * @param tagsBlacklist list of tags to discard.
+    * @return the associated filter Flow
+    */
+  def blacklistFilter(tagsBlacklist: Seq[Int]): Flow[DicomPart, DicomPart, NotUsed] = blacklistFilter(tagsBlacklist.contains(_))
+
   /**
     * Filter a stream of dicom parts such that all attributes that are group length elements except
     * file meta information group length, will be discarded. Group Length (gggg,0000) Standard Data Elements
@@ -124,7 +132,7 @@ object DicomFlows {
     *
     * @param tagCondition whitelist condition
     * @param keepPreamble true if preamble should be kept, else false
-    * @return Flow  of filtered parts
+    * @return Flow of filtered parts
     */
   def whitelistFilter(tagCondition: (Int) => Boolean, keepPreamble: Boolean = false): Flow[DicomPart, DicomPart, NotUsed] = tagFilter(tagCondition, isWhitelist = true, keepPreamble)
 
@@ -207,9 +215,10 @@ object DicomFlows {
 
   /**
     * Class used to specify modifications to individual attributes of a dataset
-    * @param tag tag number
+    *
+    * @param tag          tag number
     * @param modification a modification function
-    * @param insert if tag is absent in dataset it will be created and inserted when `true`
+    * @param insert       if tag is absent in dataset it will be created and inserted when `true`
     */
   case class TagModification(tag: Int, modification: ByteString => ByteString, insert: Boolean)
 
@@ -274,7 +283,7 @@ object DicomFlows {
                   Nil
                 }
                 .getOrElse(header :: Nil)
-               inserts ::: modify
+              inserts ::: modify
             } else
               header :: Nil
           case chunk: DicomValueChunk if currentModification.isDefined && currentHeader.isDefined =>
@@ -311,7 +320,7 @@ object DicomFlows {
     * attributes. At that stage, in order to maintain valid DICOM information, one can either change the transfer syntax to
     * an appropriate value for non-deflated data, or deflate the data again. This flow helps with the latter.
     *
-    * @return
+    * @return the associated DicomPart Flow
     */
   def deflateDatasetFlow() = Flow[DicomPart]
     .concat(Source.single(DicomEndMarker))
@@ -463,4 +472,51 @@ object DicomFlows {
         }
       }
 
+  /**
+    * Remove attributes from stream that may contain large quantities of data (bulk data)
+    * @return the associated DicomPart Flow
+    */
+  val bulkDataFilter = Flow[DicomPart]
+    .statefulMapConcat {
+
+      def normalizeRepeatingGroup(tag: Int) = {
+        val gg000000 = tag & 0xffe00000
+        if (gg000000 == 0x50000000 || gg000000 == 0x60000000) tag & 0xffe0ffff else tag
+      }
+
+      () =>
+        var sequenceStack = Seq.empty[DicomSequence]
+        var discarding = false
+
+      {
+        case sq: DicomSequence =>
+          sequenceStack = sq +: sequenceStack
+          sq :: Nil
+        case sqd: DicomSequenceDelimitation =>
+          sequenceStack = sequenceStack.drop(1)
+          sqd :: Nil
+        case dh: DicomHeader =>
+          discarding =
+            normalizeRepeatingGroup(dh.tag) match {
+              case Tag.PixelDataProviderURL => true
+              case Tag.AudioSampleData => true
+              case Tag.CurveData => true
+              case Tag.SpectroscopyData => true
+              case Tag.OverlayData => true
+              case Tag.EncapsulatedDocument => true
+              case Tag.FloatPixelData => true
+              case Tag.DoubleFloatPixelData => true
+              case Tag.PixelData => sequenceStack.isEmpty
+              case Tag.WaveformData => sequenceStack.length == 1 && sequenceStack.head.tag == Tag.WaveformSequence
+              case _ => false
+            }
+          if (discarding) Nil else dh :: Nil
+        case dvc: DicomValueChunk =>
+          if (discarding) Nil else dvc :: Nil
+        case p: DicomPart =>
+          discarding = false
+          p :: Nil
+      }
+    }
+
 }
diff --git a/src/main/scala/se/nimsa/dcm4che/streams/DicomParts.scala b/src/main/scala/se/nimsa/dcm4che/streams/DicomParts.scala
@@ -71,7 +71,7 @@ object DicomParts {
   }
 
   case class DicomValueChunk(bigEndian: Boolean, bytes: ByteString, last: Boolean) extends DicomPart {
-    override def toString = s"DicomValueChunk ${if (last) "(last) " else ""}length = ${bytes.length} ${if (bigEndian) "(big endian) " else ""}ASCII = '${new String(bytes.toArray, "US-ASCII")}' $bytes"
+    override def toString = s"DicomValueChunk ${if (last) "(last) " else ""}length = ${bytes.length} ${if (bigEndian) "(big endian) " else ""}ASCII = '${bytes.utf8String}' $bytes"
   }
 
   case class DicomDeflatedChunk(bigEndian: Boolean, bytes: ByteString) extends DicomPart

diff --git a/src/test/scala/se/nimsa/dcm4che/streams/DicomData.scala b/src/test/scala/se/nimsa/dcm4che/streams/DicomData.scala
@@ -64,6 +64,7 @@ object DicomData {
   val pixeDataFragments = ByteString(224, 127, 16, 0, 79, 87, 0, 0, 255, 255, 255, 255) // VR = OW, length = -1
 
   val seqStart = ByteString(0x08, 0x00, 0x15, 0x92, 'S', 'Q', 0, 0, -1, -1, -1, -1)
+  val waveformSeqStart = ByteString(0x00, 0x54, 0x00, 0x01, 'S', 'Q', 0, 0, -1, -1, -1, -1)
 
   // file meta with wrong transfer syntax:
   // implicit little endian (not conforming to standard)
@@ -73,6 +74,7 @@ object DicomData {
   val tsuidExplicitLEImplicitLE = ByteString(2, 0, 16, 0, 20, 0, 0, 0, '1', '.', '2', '.', '8', '4', '0', '.', '1', '0', '0', '0', '8', '.', '1', '.', '2', '.', '1', 0)
 
   def pixelData(length: Int) = ByteString(0xe0, 0x7f, 0x10, 0x00, 0x4f, 0x42, 0, 0) ++ DicomParsing.intToBytes(length, bigEndian = false) ++ ByteString(new Array[Byte](length))
+  def waveformData(length: Int) = ByteString(0x00, 0x54, 0x10, 0x10, 0x4f, 0x42, 0, 0) ++ DicomParsing.intToBytes(length, bigEndian = false) ++ ByteString(new Array[Byte](length))
 
 
   implicit class DicomPartProbe(probe: TestSubscriber.Probe[DicomPart]) {

diff --git a/src/test/scala/se/nimsa/dcm4che/streams/DicomFlowsTest.scala b/src/test/scala/se/nimsa/dcm4che/streams/DicomFlowsTest.scala
@@ -209,7 +209,7 @@ class DicomFlowsTest extends TestKit(ActorSystem("DicomAttributesSinkSpec")) wit
     val file = new File(getClass.getResource("CT0055.dcm").toURI)
     val source = FileIO.fromPath(file.toPath)
       .via(new DicomPartFlow())
-      .via(blacklistFilter(DicomParsing.isPrivateAttribute))
+      .via(blacklistFilter(DicomParsing.isPrivateAttribute _))
 
     source.runWith(TestSink.probe[DicomPart])
       .expectPreamble()
@@ -478,5 +478,63 @@ class DicomFlowsTest extends TestKit(ActorSystem("DicomAttributesSinkSpec")) wit
       .expectDicomComplete()
   }
 
+  "The buld data filter flow" should "remove pixel data" in {
+    val bytes = preamble ++ fmiGroupLength(tsuidExplicitLE) ++ tsuidExplicitLE ++ patientNameJohnDoe ++ pixelData(1000)
+
+    val source = Source.single(bytes)
+      .via(DicomPartFlow.partFlow)
+      .via(bulkDataFilter)
+
+    source.runWith(TestSink.probe[DicomPart])
+      .expectPreamble()
+      .expectHeader(Tag.FileMetaInformationGroupLength)
+      .expectValueChunk()
+      .expectHeader(Tag.TransferSyntaxUID)
+      .expectValueChunk()
+      .expectHeader(Tag.PatientName)
+      .expectValueChunk()
+      .expectDicomComplete()
+  }
+
+  it should "not remove pixel data in sequences" in {
+    val bytes = seqStart ++ itemNoLength ++ patientNameJohnDoe ++ pixelData(100) ++ itemEnd ++ seqEnd
+
+    val source = Source.single(bytes)
+      .via(DicomPartFlow.partFlow)
+      .via(bulkDataFilter)
+
+    source.runWith(TestSink.probe[DicomPart])
+      .expectSequence(Tag.DerivationCodeSequence)
+      .expectItem()
+      .expectHeader(Tag.PatientName)
+      .expectValueChunk()
+      .expectHeader(Tag.PixelData)
+      .expectValueChunk()
+      .expectItemDelimitation()
+      .expectSequenceDelimitation()
+      .expectDicomComplete()
+  }
+
+  it should "only remove waveform data when inside waveform sequence" in {
+    val bytes = waveformSeqStart ++ itemNoLength ++ patientNameJohnDoe ++ waveformData(100) ++ itemEnd ++ seqEnd ++ patientNameJohnDoe ++ waveformData(100)
+
+    val source = Source.single(bytes)
+      .via(DicomPartFlow.partFlow)
+      .via(bulkDataFilter)
+
+    source.runWith(TestSink.probe[DicomPart])
+      .expectSequence(Tag.WaveformSequence)
+      .expectItem()
+      .expectHeader(Tag.PatientName)
+      .expectValueChunk()
+      .expectItemDelimitation()
+      .expectSequenceDelimitation()
+      .expectHeader(Tag.PatientName)
+      .expectValueChunk()
+      .expectHeader(Tag.WaveformData)
+      .expectValueChunk()
+      .expectDicomComplete()
+  }
+
 }