Integrated Pipeline: refactored TreeParser implementations, moved common logic to trait

integrated-pipeline/src/main/scala/eu/nomad_lab/integrated_pipeline/ParsingTaskGenerator.scala

@@ -2,6 +2,7 @@ package eu.nomad_lab.integrated_pipeline
 
 import java.io.InputStream
 import java.nio.file.Path
+import java.util.NoSuchElementException
 
 import eu.nomad_lab.integrated_pipeline.messages.{ FileParsingTask, FileTreeScanTask }
 import eu.nomad_lab.parsers.{ CandidateParser, ParserCollection }
@@ -12,6 +13,33 @@ import eu.nomad_lab.parsers.{ CandidateParser, ParserCollection }
  */
 trait ParsingTaskGenerator extends Iterator[FileParsingTask] {
 
+  protected[this] def findNextParsingCandidate(): Option[FileParsingTask]
+
+  private var nextRequest: Option[FileParsingTask] = None
+  private var searchStarted: Boolean = false
+
+  def hasNext: Boolean = {
+    if (!searchStarted) {
+      nextRequest = findNextParsingCandidate()
+      searchStarted = true
+    }
+    nextRequest.isDefined
+  }
+
+  def next(): FileParsingTask = {
+    if (!searchStarted) {
+      nextRequest = findNextParsingCandidate()
+      searchStarted = true
+    }
+    val toReturn = nextRequest
+    nextRequest = findNextParsingCandidate()
+    if (toReturn.isDefined) {
+      toReturn.get
+    } else {
+      throw new NoSuchElementException("No more elements in this iterator")
+    }
+  }
+
   def scanInputStream(parsers: ParserCollection, input: InputStream,
     internalFilePath: String): Seq[CandidateParser] = {
     val buf = Array.fill[Byte](8 * 1024)(0)

integrated-pipeline/src/main/scala/eu/nomad_lab/integrated_pipeline/io_integrations/DirectoryTreeParsingTaskGenerator.scala

@@ -18,21 +18,8 @@ class DirectoryTreeParsingTaskGenerator(request: FileTreeScanTask, parserCollect
   require(request.treeType == TreeType.Directory, "file tree to process must be a directory")
   private val basePath = request.treeBasePath
   private val fileIterator = Files.walk(basePath).iterator().asScala.filter(Files.isRegularFile(_))
-  private var nextRequest: Option[FileParsingTask] = findNextParsingCandidate()
 
-  override def hasNext = nextRequest.isDefined
-
-  override def next(): FileParsingTask = {
-    val toReturn = nextRequest
-    nextRequest = findNextParsingCandidate()
-    if (toReturn.isDefined) {
-      toReturn.get
-    } else {
-      throw new NoSuchElementException("No more elements in this iterator")
-    }
-  }
-
-  @tailrec private def findNextParsingCandidate(): Option[FileParsingTask] = {
+  @tailrec final protected[this] override def findNextParsingCandidate(): Option[FileParsingTask] = {
     if (fileIterator.hasNext) {
       val file = fileIterator.next()
       val internalFilePath = basePath.relativize(file)

integrated-pipeline/src/main/scala/eu/nomad_lab/integrated_pipeline/io_integrations/ZipTreeParsingTaskGenerator.scala

@@ -18,25 +18,12 @@ class ZipTreeParsingTaskGenerator(request: FileTreeScanTask, parserCollection: P
 
   private val zipFile = new ZipFile(request.treeBasePath.toFile)
   private val zipEntries: java.util.Enumeration[ZipArchiveEntry] = zipFile.getEntries
-  private var nextRequest: Option[FileParsingTask] = findNextParsingCandidate()
-
-  override def hasNext: Boolean = nextRequest.isDefined
 
   override def finalize(): Unit = {
     zipFile.close()
   }
 
-  override def next(): FileParsingTask = {
-    val toReturn = nextRequest
-    nextRequest = findNextParsingCandidate()
-    if (toReturn.isDefined) {
-      toReturn.get
-    } else {
-      throw new NoSuchElementException("No more elements in this iterator")
-    }
-  }
-
-  @tailrec private def findNextParsingCandidate(): Option[FileParsingTask] = {
+  @tailrec protected[this] final override def findNextParsingCandidate(): Option[FileParsingTask] = {
     if (zipEntries.hasMoreElements) {
       val zipEntry: ZipArchiveEntry = zipEntries.nextElement()
       val internalFilePath = Paths.get(zipEntry.getName)