Two more differences between the LZW in HDGF and HMEF:

 * Little Endian vs Big Endian storage of the code position
 * Initial dictionary position is the end of pre-fill, if there is one, rather than always being position 0


git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@1078300 13f79535-47bb-0310-9956-ffa450edef68

3 files changed, 23 insertions, 12 deletions

diff --git a/src/scratchpad/src/org/apache/poi/hdgf/HDGFLZW.java b/src/scratchpad/src/org/apache/poi/hdgf/HDGFLZW.java
index f122c40f17..e6d4aa2e65 100644
--- a/src/scratchpad/src/org/apache/poi/hdgf/HDGFLZW.java
+++ b/src/scratchpad/src/org/apache/poi/hdgf/HDGFLZW.java
@@ -37,8 +37,10 @@ import org.apache.poi.util.LZWDecompresser;
  */
 public class HDGFLZW extends LZWDecompresser {
    public HDGFLZW() {
-      // We're the wrong way round!
-      super(false, 3);
+      // Out flag is the wrong way round!
+      // Length wise, we're 3 longer than we say, so the max len is 19
+      // Endian wise, we're little endian, so 0x1234 is pos 0x312
+      super(false, 3, false);
    }
 
    /**
@@ -63,12 +65,13 @@ public class HDGFLZW extends LZWDecompresser {
       }
       return pntr;
    }
-
+   
    /**
     * We want an empty dictionary, so do nothing
     */
    @Override
-   protected void populateDictionary(byte[] dict) {
+   protected int populateDictionary(byte[] dict) {
+      return 0;
    }
 
    /**
diff --git a/src/scratchpad/src/org/apache/poi/hmef/CompressedRTF.java b/src/scratchpad/src/org/apache/poi/hmef/CompressedRTF.java
index 81218bc9ba..70c3e59295 100644
--- a/src/scratchpad/src/org/apache/poi/hmef/CompressedRTF.java
+++ b/src/scratchpad/src/org/apache/poi/hmef/CompressedRTF.java
@@ -54,7 +54,10 @@ public final class CompressedRTF extends LZWDecompresser {
       "{\\colortbl\\red0\\green0\\blue0\n\r\\par \\pard\\plain\\f0\\fs20\\b\\i\\u\\tab\\tx";
    
    public CompressedRTF() {
-      super(true, 2);
+      // Out flag has the normal meaning
+      // Length wise, we're 2 longer than we say, so the max len is 18
+      // Endian wise, we're big endian, so 0x1234 is pos 0x123
+      super(true, 2, true);
    }
 
    public void decompress(InputStream src, OutputStream res) throws IOException {
@@ -80,17 +83,24 @@ public final class CompressedRTF extends LZWDecompresser {
       super.decompress(src, res);
    }
 
+   /**
+    * We use regular dictionary offsets, so no
+    *  need to change anything
+    */
    @Override
    protected int adjustDictionaryOffset(int offset) {
-      // TODO Do we need to change anything?
-      return 0;
+      return offset;
    }
 
    @Override
-   protected void populateDictionary(byte[] dict) {
+   protected int populateDictionary(byte[] dict) {
       try {
+         // Copy in the RTF constants 
          byte[] preload = LZW_RTF_PRELOAD.getBytes("US-ASCII");
          System.arraycopy(preload, 0, dict, 0, preload.length);
+         
+         // Start adding new codes after the constants
+         return preload.length;
       } catch(UnsupportedEncodingException e) {
          throw new RuntimeException("Your JVM is broken as it doesn't support US ASCII");
       }
diff --git a/src/scratchpad/testcases/org/apache/poi/hmef/TestCompressedRTF.java b/src/scratchpad/testcases/org/apache/poi/hmef/TestCompressedRTF.java
index ad8f6692b1..36991c43cf 100644
--- a/src/scratchpad/testcases/org/apache/poi/hmef/TestCompressedRTF.java
+++ b/src/scratchpad/testcases/org/apache/poi/hmef/TestCompressedRTF.java
@@ -93,7 +93,7 @@ public final class TestCompressedRTF extends TestCase {
      * Check that we can decode the first 8 codes
      * (1 flag byte + 8 codes)  
      */
-    public void DISABLEDtestFirstBlock() throws Exception {
+    public void testFirstBlock() throws Exception {
        HMEFMessage msg = new HMEFMessage(
              _samples.openResourceAsStream("quick-winmail.dat")
        );
@@ -112,7 +112,6 @@ public final class TestCompressedRTF extends TestCase {
        String decompStr = new String(decomp, "ASCII");
        
        // Test
-System.err.println(decompStr);       
        assertEquals(block1.length(), decomp.length);
        assertEquals(block1, decompStr);
     }
@@ -121,7 +120,7 @@ System.err.println(decompStr);
      * Check that we can decode the first 16 codes
      * (flag + 8 codes, flag + 8 codes)  
      */
-    public void DISABLEDtestFirstTwoBlocks() throws Exception {
+    public void testFirstTwoBlocks() throws Exception {
        HMEFMessage msg = new HMEFMessage(
              _samples.openResourceAsStream("quick-winmail.dat")
        );
@@ -140,7 +139,6 @@ System.err.println(decompStr);
        String decompStr = new String(decomp, "ASCII");
        
        // Test
-System.err.println(decompStr);       
        assertEquals(block2.length(), decomp.length);
        assertEquals(block2, decompStr);
     }