Fix some chunk types, fix the directory descent, fix the Msg2txt example, and start on fixing core tests

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@897167 13f79535-47bb-0310-9956-ffa450edef68

src/examples/src/org/apache/poi/hsmf/examples/Msg2txt.java

@@ -17,16 +17,14 @@
 
 package org.apache.poi.hsmf.examples;
 
-import java.io.ByteArrayInputStream;
 import java.io.File;
 import java.io.FileOutputStream;
 import java.io.IOException;
 import java.io.OutputStream;
 import java.io.PrintWriter;
-import java.util.Iterator;
-import java.util.Map;
 
 import org.apache.poi.hsmf.MAPIMessage;
+import org.apache.poi.hsmf.datatypes.AttachmentChunks;
 import org.apache.poi.hsmf.exceptions.ChunkNotFoundException;
 
 /**
@@ -35,7 +33,6 @@ import org.apache.poi.hsmf.exceptions.ChunkNotFoundException;
  * attachments.
  * 
  * @author Bruno Girin
- *
  */
 public class Msg2txt {
 	
@@ -105,17 +102,13 @@ public class Msg2txt {
 			} catch (ChunkNotFoundException e) {
 				System.err.println("No message body");
 			}
-			Map attachmentMap = msg.getAttachmentFiles();
+			
-			if(attachmentMap.size() > 0) {
+			AttachmentChunks[] attachments = msg.getAttachmentFiles();
+			if(attachments.length > 0) {
 				File d = new File(attDirName);
 				if(d.mkdir()) {
-					for(
+					for(AttachmentChunks attachment : attachments) {
-							Iterator ii = attachmentMap.entrySet().iterator();
+						processAttachment(attachment, d);
-							ii.hasNext();
-							) {
-						Map.Entry entry = (Map.Entry)ii.next();
-						processAttachment(d, entry.getKey().toString(),
-								(ByteArrayInputStream)entry.getValue());
 					}
 				} else {
 					System.err.println("Can't create directory "+attDirName);
@@ -131,33 +124,26 @@ public class Msg2txt {
 	/**
 	 * Processes a single attachment: reads it from the Outlook MSG file and
 	 * writes it to disk as an individual file.
-	 * 
+	 *
+	 * @param attachment the chunk group describing the attachment
 	 * @param dir the directory in which to write the attachment file
-	 * @param fileName the name of the attachment file
-	 * @param fileIn the input stream that contains the attachment's data
 	 * @throws IOException when any of the file operations fails
 	 */
-	public void processAttachment(File dir, String fileName,
+	public void processAttachment(AttachmentChunks attachment, 
-			ByteArrayInputStream fileIn) throws IOException {
+	      File dir) throws IOException {
+	   String fileName = attachment.attachFileName.toString();
+	   if(attachment.attachLongFileName != null) {
+	      fileName = attachment.attachLongFileName.toString();
+	   }
+	   
 		File f = new File(dir, fileName);
 		OutputStream fileOut = null;
 		try {
 			fileOut = new FileOutputStream(f);
-			byte[] buffer = new byte[2048];
+			fileOut.write(attachment.attachData.getValue());
-			int bNum = fileIn.read(buffer);
-			while(bNum > 0) {
-				fileOut.write(buffer);
-				bNum = fileIn.read(buffer);
-			}
 		} finally {
-			try {
+			if(fileOut != null) {
-				if(fileIn != null) {
+				fileOut.close();
-					fileIn.close();
-				}
-			} finally {
-				if(fileOut != null) {
-					fileOut.close();
-				}
 			}
 		}
 	}

src/scratchpad/src/org/apache/poi/hsmf/datatypes/Chunks.java

@@ -63,7 +63,7 @@ public final class Chunks implements ChunkGroup {
    /** Type of server that the message originated from (SMTP, etc). */
    public StringChunk sentByServerType;
    /** TODO */
-   public StringChunk dateChunk; 
+   public ByteChunk dateChunk; 
    /** TODO */
    public StringChunk emailFromChunk; 
 
@@ -86,7 +86,7 @@ public final class Chunks implements ChunkGroup {
          subjectChunk = (StringChunk)chunk;
          break;
       case DATE:
-         dateChunk = (StringChunk)chunk;
+         dateChunk = (ByteChunk)chunk;
          break;
       case CONVERSATION_TOPIC:
          conversationTopic = (StringChunk)chunk;

src/scratchpad/src/org/apache/poi/hsmf/datatypes/RecipientChunks.java

@@ -32,7 +32,7 @@ public final class RecipientChunks implements ChunkGroup {
    public static final int RECIPIENT_EMAIL  = 0x39FE;
    
    /** TODO */
-   public StringChunk recipientSearchChunk; 
+   public ByteChunk recipientSearchChunk; 
    /** TODO */
    public StringChunk recipientEmailChunk;
    
@@ -53,7 +53,7 @@ public final class RecipientChunks implements ChunkGroup {
    public void record(Chunk chunk) {
       switch(chunk.getChunkId()) {
       case RECIPIENT_SEARCH:
-         recipientSearchChunk = (StringChunk)chunk;
+         recipientSearchChunk = (ByteChunk)chunk;
          break;
       case RECIPIENT_EMAIL:
          recipientEmailChunk = (StringChunk)chunk;

src/scratchpad/src/org/apache/poi/hsmf/datatypes/StringChunk.java

@@ -55,7 +55,7 @@ public class StringChunk extends Chunk {
 	   switch(type) {
 	   case Types.ASCII_STRING:
          try {
-            tmpValue = new String(data, "UTF-16LE");
+            tmpValue = new String(data, "CP1252");
          } catch (UnsupportedEncodingException e) {
             throw new RuntimeException("Core encoding not found, JVM broken?", e);
          }
@@ -77,7 +77,7 @@ public class StringChunk extends Chunk {
       switch(type) {
       case Types.ASCII_STRING:
          try {
-            data = value.getBytes("UTF-16LE");
+            data = value.getBytes("CP1252");
          } catch (UnsupportedEncodingException e) {
             throw new RuntimeException("Core encoding not found, JVM broken?", e);
          }

src/scratchpad/src/org/apache/poi/hsmf/parsers/POIFSChunkParser.java

@@ -55,7 +55,7 @@ public final class POIFSChunkParser {
       //  there doesn't seem to be any use of that in Outlook
       for(Entry entry : node) {
          if(entry instanceof DirectoryNode) {
-            DirectoryNode dir = (DirectoryNode)node;
+            DirectoryNode dir = (DirectoryNode)entry;
             ChunkGroup group = null;
             
             // Do we know what to do with it?
@@ -66,7 +66,7 @@ public final class POIFSChunkParser {
                group = new NameIdChunks();
             }
             if(dir.getName().startsWith(RecipientChunks.PREFIX)) {
-               group = new NameIdChunks();
+               group = new RecipientChunks();
             }
             
             if(group != null) {

src/scratchpad/testcases/org/apache/poi/hsmf/parsers/TestPOIFSChunkParser.java

@@ -18,8 +18,16 @@
 package org.apache.poi.hsmf.parsers;
 
 import java.io.FileInputStream;
+import java.io.FileNotFoundException;
 import java.io.IOException;
 
+import org.apache.poi.hsmf.MAPIMessage;
+import org.apache.poi.hsmf.datatypes.AttachmentChunks;
+import org.apache.poi.hsmf.datatypes.ChunkGroup;
+import org.apache.poi.hsmf.datatypes.Chunks;
+import org.apache.poi.hsmf.datatypes.NameIdChunks;
+import org.apache.poi.hsmf.datatypes.RecipientChunks;
+import org.apache.poi.hsmf.exceptions.ChunkNotFoundException;
 import org.apache.poi.poifs.filesystem.POIFSFileSystem;
 import org.apache.poi.POIDataSamples;
 
@@ -44,13 +52,83 @@ public final class TestPOIFSChunkParser extends TestCase {
 	         new FileInputStream(samples.getFile("attachment_test_msg.msg"))
 	   );
       POIFSFileSystem without = new POIFSFileSystem(
-            new FileInputStream(samples.getFile("simple_test_msg.msg"))
+            new FileInputStream(samples.getFile("quick.msg"))
       );
+      AttachmentChunks attachment;
       
-      // Check details on the one with
+      
-	      
+      // Check raw details on the one with
-	   // One with, from the top
+      with.getRoot().getEntry("__attach_version1.0_#00000000");
+      with.getRoot().getEntry("__attach_version1.0_#00000001");
+      POIFSChunkParser.parse(with.getRoot());
+      
+      ChunkGroup[] groups = POIFSChunkParser.parse(with.getRoot());
+      assertEquals(5, groups.length);
+      assertTrue(groups[0] instanceof Chunks);
+      assertTrue(groups[1] instanceof RecipientChunks);
+      assertTrue(groups[2] instanceof AttachmentChunks);
+      assertTrue(groups[3] instanceof AttachmentChunks);
+      assertTrue(groups[4] instanceof NameIdChunks);
+      
+      attachment = (AttachmentChunks)groups[2];
+      assertEquals("TEST-U~1.DOC", attachment.attachFileName.toString());
+      assertEquals("test-unicode.doc", attachment.attachLongFileName.toString());
+      assertEquals(24064, attachment.attachData.getValue().length);
+      
+      attachment = (AttachmentChunks)groups[3];
+      assertEquals("pj1.txt", attachment.attachFileName.toString());
+      assertEquals("pj1.txt", attachment.attachLongFileName.toString());
+      assertEquals(89, attachment.attachData.getValue().length);
 	   
+      
+      // Check raw details on one without
+      try {
+         without.getRoot().getEntry("__attach_version1.0_#00000000");
+         fail();
+      } catch(FileNotFoundException e) {}
+      try {
+         without.getRoot().getEntry("__attach_version1.0_#00000001");
+         fail();
+      } catch(FileNotFoundException e) {}
+      
+      
+	   // One with, from the top
+      MAPIMessage msgWith = new MAPIMessage(with);
+      assertEquals(2, msgWith.getAttachmentFiles().length);
+
+      attachment = msgWith.getAttachmentFiles()[0];
+      assertEquals("TEST-U~1.DOC", attachment.attachFileName.toString());
+      assertEquals("test-unicode.doc", attachment.attachLongFileName.toString());
+      assertEquals(24064, attachment.attachData.getValue().length);
+      
+      attachment = msgWith.getAttachmentFiles()[1];
+      assertEquals("pj1.txt", attachment.attachFileName.toString());
+      assertEquals("pj1.txt", attachment.attachLongFileName.toString());
+      assertEquals(89, attachment.attachData.getValue().length);
+      
+      // Plus check core details are there
+      try {
+         assertEquals("'nicolas1.23456@free.fr'", msgWith.getDisplayTo());
+         assertEquals("Nicolas1 23456", msgWith.getDisplayFrom());
+         assertEquals("test pi\u00e8ce jointe 1", msgWith.getSubject());
+      } catch(ChunkNotFoundException e) {
+         fail();
+      }
+	   
+      
 	   // One without, from the top
+      MAPIMessage msgWithout = new MAPIMessage(without);
+      
+      // No attachments
+      assertEquals(0, msgWithout.getAttachmentFiles().length);
+      
+      // But has core details
+      try {
+         assertEquals("Kevin Roast", msgWithout.getDisplayTo());
+         assertEquals("Kevin Roast", msgWithout.getDisplayFrom());
+         assertEquals("Test the content transformer", msgWithout.getSubject());
+      } catch(ChunkNotFoundException e) {
+         fail();
+      }
 	}
 }