programm now recognize, which type of extractor is usefull

Author: P1tt187

build.sbt

@@ -2,7 +2,7 @@ import com.typesafe.sbt.SbtNativePackager.autoImport._
 
 name := "doc2text"
 
-version := "1.0"
+version := "1.1"
 
 scalaVersion := "2.10.6"
 

src/main/scala/p1tt187/doc2text/Doc2TextApp.scala

@@ -3,24 +3,23 @@ package p1tt187.doc2text
 import java.io.{File, FileInputStream, FileNotFoundException, FileOutputStream}
 import java.util.Formatter
 
-import org.apache.poi.hwpf.extractor.WordExtractor
+import p1tt187.doc2text.extractor.ExtractorChooser
 
 /**
  * @author fabian 
  *         on 18.10.15.
  */
-
-
 object Doc2TextApp extends App {
+
   val results = args.map { arg =>
     try {
-      val wordExtractor = new WordExtractor(new FileInputStream(arg))
+      val wordExtractor = ExtractorChooser(new FileInputStream(arg))
       val tmpFile = File.createTempFile(arg, ".txt")
       val output = new Formatter(new FileOutputStream(tmpFile))
       output.format("%s", wordExtractor.getText)
       output.flush()
       output.close()
-      wordExtractor.close()
+      wordExtractor.close
       tmpFile.getAbsolutePath
     } catch {
       case ex: FileNotFoundException =>

src/main/scala/p1tt187/doc2text/extractor/ExtractorChooser.scala

@@ -0,0 +1,18 @@
+package p1tt187.doc2text.extractor
+
+import java.io.FileInputStream
+
+/**
+  * @author fabian 
+  *         on 07.11.15.
+  */
+object ExtractorChooser {
+  def apply( fileInputStream: FileInputStream ):TExtractor={
+    try {
+      new WExtractor(fileInputStream)
+    } catch {
+      case _: Exception =>
+        new W6Extractor(fileInputStream)
+    }
+  }
+}

src/main/scala/p1tt187/doc2text/extractor/TExtractor.scala

@@ -0,0 +1,13 @@
+package p1tt187.doc2text.extractor
+
+/**
+  * @author fabian 
+  *         on 07.11.15.
+  */
+trait TExtractor {
+
+  def getText:String
+
+  def close:Unit
+
+}

src/main/scala/p1tt187/doc2text/extractor/W6Extractor.scala

@@ -0,0 +1,20 @@
+package p1tt187.doc2text.extractor
+
+import java.io.FileInputStream
+
+import org.apache.poi.hwpf.extractor.Word6Extractor
+
+/**
+  * @author fabian 
+  *         on 07.11.15.
+  */
+class W6Extractor(fileInputStream: FileInputStream) extends TExtractor{
+
+  private val word6Extractor = new Word6Extractor(fileInputStream)
+
+  override def close: Unit = {word6Extractor.close()}
+
+  override def getText: String = {
+    word6Extractor.getText
+  }
+}

src/main/scala/p1tt187/doc2text/extractor/WExtractor.scala

@@ -0,0 +1,22 @@
+package p1tt187.doc2text.extractor
+
+import java.io.FileInputStream
+
+import org.apache.poi.hwpf.extractor.WordExtractor
+
+/**
+  * @author fabian 
+  *         on 07.11.15.
+  */
+class WExtractor(fileInputStream: FileInputStream)  extends TExtractor{
+
+  private val wordExtractor: WordExtractor = new WordExtractor(fileInputStream)
+
+  override def close: Unit = {
+    wordExtractor.close()
+  }
+
+  override def getText: String = {
+    wordExtractor.getText
+  }
+}