Add new method to HSMF of MAPIMessage.has7BitEncodingStrings() to make it easier to decide when encoding guessing is needed

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@1087746 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">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>
            <action dev="poi-developers" type="add">Allow HSMF access to the HTML body contents in MAPIMessage</action>

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

@@ -36,6 +36,7 @@ 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.datatypes.Types;
 import org.apache.poi.hsmf.datatypes.RecipientChunks.RecipientChunksSorter;
 import org.apache.poi.hsmf.datatypes.StringChunk;
 import org.apache.poi.hsmf.exceptions.ChunkNotFoundException;
@@ -396,6 +397,37 @@ public class MAPIMessage extends POIDocument {
       }
    }
    
+   /**
+    * Does this file contain any strings that
+    *  are stored as 7 bit rather than unicode?
+    */
+   public boolean has7BitEncodingStrings() {
+      for(Chunk c : mainChunks.getAll()) {
+         if(c instanceof StringChunk) {
+            if( ((StringChunk)c).getType() == Types.ASCII_STRING ) {
+               return true;
+            }
+         }
+      }
+      for(Chunk c : nameIdChunks.getAll()) {
+         if(c instanceof StringChunk) {
+            if( ((StringChunk)c).getType() == Types.ASCII_STRING ) {
+               return true;
+            }
+         }
+      }
+      for(RecipientChunks rc : recipientChunks) {
+         for(Chunk c : rc.getAll()) {
+            if(c instanceof StringChunk) {
+               if( ((StringChunk)c).getType() == Types.ASCII_STRING ) {
+                  return true;
+               }
+            }
+         }
+      }
+      return false;
+   }
+   
    /**
     * Returns all the headers, one entry per line
     */

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

@@ -34,6 +34,7 @@ public final class TestBasics extends TestCase {
    private MAPIMessage outlook30;
    private MAPIMessage attachments;
    private MAPIMessage noRecipientAddress;
+   private MAPIMessage unicode;
    private MAPIMessage cyrillic;
    private MAPIMessage chinese;
 
@@ -48,6 +49,7 @@ public final class TestBasics extends TestCase {
       outlook30  = new MAPIMessage(samples.openResourceAsStream("outlook_30_msg.msg"));
       attachments = new MAPIMessage(samples.openResourceAsStream("attachment_test_msg.msg"));
       noRecipientAddress = new MAPIMessage(samples.openResourceAsStream("no_recipient_address.msg"));
+      unicode = new MAPIMessage(samples.openResourceAsStream("example_received_unicode.msg"));
       cyrillic = new MAPIMessage(samples.openResourceAsStream("cyrillic_message.msg"));
       chinese = new MAPIMessage(samples.openResourceAsStream("chinese-traditional.msg"));
 	}
@@ -182,6 +184,16 @@ public final class TestBasics extends TestCase {
       noRecipientAddress.setReturnNullOnMissingChunk(false);
 	}
 	
+	/**
+	 * Test the 7 bit detection
+	 */
+	public void test7BitDetection() throws Exception {
+	   assertEquals(false, unicode.has7BitEncodingStrings());
+      assertEquals(true, simple.has7BitEncodingStrings());
+      assertEquals(true, chinese.has7BitEncodingStrings());
+      assertEquals(true, cyrillic.has7BitEncodingStrings());
+	}
+	
    /**
     * We default to CP1252, but can sometimes do better
     *  if needed.