Bugzilla 53205 - Fix some parsing errors and encoding issues in HDGF

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@1365638 13f79535-47bb-0310-9956-ffa450edef68

src/documentation/content/xdocs/status.xml

@@ -34,6 +34,7 @@
 
     <changes>
         <release version="3.9-beta1" date="2012-??-??">
+          <action dev="poi-developers" type="fix">53205 - Fix some parsing errors and encoding issues in HDGF </action>
           <action dev="poi-developers" type="add">53204 - Improved performanceof PageSettingsBlock in HSSF </action>
           <action dev="poi-developers" type="add">53500 - Getter for repeating rows and columns</action>
           <action dev="poi-developers" type="fix">53369 - Fixed tests failing on JDK 1.7</action>

src/scratchpad/src/org/apache/poi/hdgf/chunks/Chunk.java

@@ -161,6 +161,7 @@ public final class Chunk {
 				continue;
 			}
 
+			try {
 				// Process
 				switch(type) {
 				// Types 0->7 = a flat at bit 0->7
@@ -199,8 +200,8 @@ public final class Chunk {
 						endsAt = contents.length;
 					}
 
-				int strLen = (endsAt-startsAt) / 2;
-				command.value = StringUtil.getFromUnicodeLE(contents, startsAt, strLen);
+					int strLen = endsAt - startsAt;
+					command.value = new String(contents, startsAt, strLen, header.getChunkCharset().name());
 					break;
 				case 25:
 					command.value = Short.valueOf(
@@ -226,6 +227,11 @@ public final class Chunk {
 					logger.log(POILogger.INFO,
 							"Command of type " + type + " not processed!");
 				}
+			}
+			catch (Exception e) {
+				logger.log(POILogger.ERROR, "Unexpected error processing command, ignoring and continuing. Command: " +
+						command, e);
+			}
 
 			// Add to the array
 			commands.add(command);

src/scratchpad/src/org/apache/poi/hdgf/chunks/ChunkHeader.java

@@ -19,6 +19,8 @@ package org.apache.poi.hdgf.chunks;
 
 import org.apache.poi.util.LittleEndian;
 
+import java.nio.charset.Charset;
+
 /**
  * A chunk header
  */
@@ -80,6 +82,7 @@ public abstract class ChunkHeader {
 	public abstract int getSizeInBytes();
 	public abstract boolean hasTrailer();
 	public abstract boolean hasSeparator();
+	public abstract Charset getChunkCharset();
 
 	/**
 	 * Returns the ID/IX of the chunk

src/scratchpad/src/org/apache/poi/hdgf/chunks/ChunkHeaderV11.java

@@ -17,6 +17,8 @@
 
 package org.apache.poi.hdgf.chunks;
 
+import java.nio.charset.Charset;
+
 /**
  * A chunk header from v11+
  */
@@ -42,4 +44,9 @@ public final class ChunkHeaderV11 extends ChunkHeaderV6 {
 
 		return false;
 	}
+
+	@Override
+	public Charset getChunkCharset() {
+		return Charset.forName("UTF-16LE");
+	}
 }

src/scratchpad/src/org/apache/poi/hdgf/chunks/ChunkHeaderV4V5.java

@@ -17,6 +17,8 @@
 
 package org.apache.poi.hdgf.chunks;
 
+import java.nio.charset.Charset;
+
 /**
  * A chunk header from v4 or v5
  */
@@ -54,4 +56,9 @@ public final class ChunkHeaderV4V5 extends ChunkHeader {
 		// V4 and V5 never has separators
 		return false;
 	}
+
+	@Override
+	public Charset getChunkCharset() {
+		return Charset.forName("ASCII");
+	}
 }

src/scratchpad/src/org/apache/poi/hdgf/chunks/ChunkHeaderV6.java

@@ -17,6 +17,8 @@
 
 package org.apache.poi.hdgf.chunks;
 
+import java.nio.charset.Charset;
+
 /**
  * A chunk header from v6
  */
@@ -59,4 +61,9 @@ public class ChunkHeaderV6 extends ChunkHeader {
 		// V6 never has separators
 		return false;
 	}
+
+	@Override
+	public Charset getChunkCharset() {
+		return Charset.forName("ASCII");
+	}
 }

src/scratchpad/src/org/apache/poi/hdgf/streams/ChunkStream.java

@@ -52,6 +52,7 @@ public final class ChunkStream extends Stream {
 
 		int pos = 0;
 		byte[] contents = getStore().getContents();
+		try {
 			while(pos < contents.length) {
 				// Ensure we have enough data to create a chunk from
 				int headerSize = ChunkHeader.getHeaderSize(chunkFactory.getVersion());
@@ -65,6 +66,11 @@ public final class ChunkStream extends Stream {
 					pos = contents.length;
 				}
 			}
+		}
+		catch (Exception e)
+		{
+			System.err.println("Failed to create chunk at " + pos + ", ignoring rest of data." + e);
+		}
 
 		chunks = chunksA.toArray(new Chunk[chunksA.size()]);
 	}

src/scratchpad/testcases/org/apache/poi/hdgf/TestHDGFCore.java

@@ -17,6 +17,7 @@
 
 package org.apache.poi.hdgf;
 
+import org.apache.poi.hdgf.extractor.VisioTextExtractor;
 import org.apache.poi.hdgf.streams.PointerContainingStream;
 import org.apache.poi.hdgf.streams.TrailerStream;
 import org.apache.poi.poifs.filesystem.POIFSFileSystem;
@@ -88,4 +89,28 @@ public final class TestHDGFCore extends TestCase {
       HDGFDiagram hdgf = new HDGFDiagram(fs);
       assertNotNull(hdgf);
 	}
+
+    public void testV6NonUtf16LE() throws Exception {
+   		fs = new POIFSFileSystem(_dgTests.openResourceAsStream("v6-non-utf16le.vsd"));
+
+   		HDGFDiagram hdgf = new HDGFDiagram(fs);
+   		assertNotNull(hdgf);
+
+        VisioTextExtractor textExtractor = new VisioTextExtractor(hdgf);
+        String text = textExtractor.getText().replace("\u0000", "").trim();
+
+        assertEquals("Table\n\n\nPropertySheet\n\n\n\nPropertySheetField", text);
+   	}
+
+    public void testUtf16LE() throws Exception {
+   		fs = new POIFSFileSystem(_dgTests.openResourceAsStream("Test_Visio-Some_Random_Text.vsd"));
+
+   		HDGFDiagram hdgf = new HDGFDiagram(fs);
+   		assertNotNull(hdgf);
+
+        VisioTextExtractor textExtractor = new VisioTextExtractor(hdgf);
+        String text = textExtractor.getText().trim();
+
+        assertEquals("text\nView\nTest View\nI am a test view\nSome random text, on a page", text);
+   	}
 }