Improve HSMF MAPIMessage access to the HTML and RTF versions of the message body (where available)

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@1087782 13f79535-47bb-0310-9956-ffa450edef68

src/documentation/content/xdocs/status.xml

@@ -34,6 +34,7 @@
 
     <changes>
         <release version="3.8-beta3" date="2011-??-??">
+           <action dev="poi-developers" type="add">Improve HSMF MAPIMessage access to the HTML and RTF versions of the message body (where available)</action>
            <action dev="poi-developers" type="add">Add new method to HSMF of MAPIMessage.has7BitEncodingStrings() to make it easier to decide when encoding guessing is needed</action>
            <action dev="poi-developers" type="fix">OutlookTextExtractor now requests 7 bit encoding guessing</action>
            <action dev="poi-developers" type="add">Improve HSMF encoding guessing for 7 bit fields in MAPIMessage</action>

src/scratchpad/src/org/apache/poi/hsmf/MAPIMessage.java

@@ -29,11 +29,14 @@ import java.util.regex.Matcher;
 import java.util.regex.Pattern;
 
 import org.apache.poi.POIDocument;
+import org.apache.poi.hmef.attribute.MAPIRtfAttribute;
 import org.apache.poi.hsmf.datatypes.AttachmentChunks;
 import org.apache.poi.hsmf.datatypes.AttachmentChunks.AttachmentChunksSorter;
+import org.apache.poi.hsmf.datatypes.ByteChunk;
 import org.apache.poi.hsmf.datatypes.Chunk;
 import org.apache.poi.hsmf.datatypes.ChunkGroup;
 import org.apache.poi.hsmf.datatypes.Chunks;
+import org.apache.poi.hsmf.datatypes.MAPIProperty;
 import org.apache.poi.hsmf.datatypes.NameIdChunks;
 import org.apache.poi.hsmf.datatypes.RecipientChunks;
 import org.apache.poi.hsmf.datatypes.Types;
@@ -184,7 +187,36 @@ public class MAPIMessage extends POIDocument {
     * @throws ChunkNotFoundException
     */
    public String getHmtlBody() throws ChunkNotFoundException {
-      return getStringFromChunk(mainChunks.htmlBodyChunk);
+      if(mainChunks.htmlBodyChunkBinary != null) {
+         return mainChunks.htmlBodyChunkBinary.getAs7bitString();
+      }
+      return getStringFromChunk(mainChunks.htmlBodyChunkString);
+   }
+
+   /**
+    * Gets the RTF Rich Message body of this Outlook Message, if this email
+    *  contains a RTF (rich) version.
+    * @return The string representation of the 'RTF' version of the body, if available.
+    * @throws ChunkNotFoundException
+    */
+   public String getRtfBody() throws ChunkNotFoundException {
+      ByteChunk chunk = mainChunks.rtfBodyChunk;
+      if(chunk == null) {
+         if(returnNullOnMissingChunk) {
+            return null;
+         } else {
+            throw new ChunkNotFoundException();
+         }
+      }
+      
+      try {
+         MAPIRtfAttribute rtf = new MAPIRtfAttribute(
+               MAPIProperty.RTF_COMPRESSED, Types.BINARY, chunk.getValue()
+         );
+         return rtf.getDataString();
+      } catch(IOException e) {
+         throw new RuntimeException("Shouldn't happen", e);
+      }
    }
 
    /**

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

@@ -38,7 +38,10 @@ public final class Chunks implements ChunkGroup {
    /** BODY Chunk, for plain/text messages */
    public StringChunk textBodyChunk;
    /** BODY Html Chunk, for html messages */
-   public StringChunk htmlBodyChunk;
+   public StringChunk htmlBodyChunkString;
+   public ByteChunk htmlBodyChunkBinary;
+   /** BODY Rtf Chunk, for Rtf (Rich) messages */
+   public ByteChunk rtfBodyChunk;
    /** Subject link chunk, in plain/text */
    public StringChunk subjectChunk;
    /** Value that is in the TO field (not actually the addresses as they are stored in recip directory nodes */
@@ -119,9 +122,16 @@ public final class Chunks implements ChunkGroup {
       else if(chunk.getChunkId() == MAPIProperty.BODY.id) {
          textBodyChunk = (StringChunk)chunk;
       }
-      else if(chunk.getChunkId() == MAPIProperty.BODY_HTML.id && 
+      else if(chunk.getChunkId() == MAPIProperty.BODY_HTML.id) {
-              chunk instanceof StringChunk) {
+         if(chunk instanceof StringChunk) {
-         htmlBodyChunk = (StringChunk)chunk;
+            htmlBodyChunkString = (StringChunk)chunk;
+         }
+         if(chunk instanceof ByteChunk) {
+            htmlBodyChunkBinary = (ByteChunk)chunk;
+         }
+      }
+      else if(chunk.getChunkId() == MAPIProperty.RTF_COMPRESSED.id) {
+         rtfBodyChunk = (ByteChunk)chunk;
       }
       
       // And add to the main list

src/scratchpad/testcases/org/apache/poi/hsmf/TestOutlook30FileRead.java

@@ -120,7 +120,6 @@ private MAPIMessage mapiMessage;
 		TestCase.assertEquals("IN-SPIRE servers going down for a bit, back up around 8am", obtained);
 	}
 
-
 	/**
 	 * Check if we can read the subject line of the blank message, we expect ""
 	 *
@@ -131,6 +130,16 @@ private MAPIMessage mapiMessage;
 		TestCase.assertEquals("IPM.Note", obtained);
 	}
    
+   /**
+    * Ensure we can get the HTML and RTF versions
+    */
+   public void testReadBodyContents() throws Exception {
+      String html = mapiMessage.getHmtlBody();
+      String rtf = mapiMessage.getRtfBody();
+      assertNotNull(html);
+      assertNotNull(rtf);
       
-
+      assertTrue("Wrong text:\n" + html, html.startsWith("<!DOCTYPE"));
+      assertTrue("Wrong text:\n" + rtf,  rtf.startsWith("{\\rtf1"));
+   }
 }