Stub out the HDGF compression routine, and tests for it. Document the first slab of the compressed data in the test, so it's easy to see if we're generating it properly. (Have yet to implement the compression support yet though)

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@584534 13f79535-47bb-0310-9956-ffa450edef68

src/scratchpad/src/org/apache/poi/hdgf/HDGFLZW.java

@@ -40,7 +40,7 @@ public class HDGFLZW {
  *  the wrapping.
  * This is a convenience method
  */
-public byte fromInt(int b) {
+public static byte fromInt(int b) {
 	if(b < 128) return (byte)b;
 	return (byte)(b - 256);
 }
@@ -49,11 +49,21 @@ public byte fromInt(int b) {
  *  and 255 (i.e. handle the unwrapping).
  * This is a convenience method
  */
-public int fromByte(byte b) {
+public static int fromByte(byte b) {
 	if(b >= 0) return (int)b;
 	return (int)(b + 256);
 }
 
+/**
+ * Compress the given input stream, returning the array of bytes
+ *  of the compressed input
+ */
+public byte[] compress(InputStream src) throws IOException {
+	ByteArrayOutputStream res = new ByteArrayOutputStream();
+	compress(src,res);
+    return res.toByteArray();
+}
+
 /**
  * Decompresses the given input stream, returning the array of bytes
  *  of the decompressed input.
@@ -158,4 +168,88 @@ public void decode(InputStream src, OutputStream res) throws IOException {
     }
 }
 
+/**
+ * Performs the Visio compatible streaming LZW compression.
+ * Works by:
+ * 1) ???
+ * 2) ???
+ * TODO - Finish
+ */
+public void compress(InputStream src, OutputStream res) throws IOException {
+	// We use 12 bit codes:
+	// * 0-255 are real bytes
+	// * 256-4095 are the substring codes
+	// Java handily initialises our buffer / dictionary
+	//  to all zeros
+	byte[] dict = new byte[4096];
+	// The next block of data to be written out, minus
+	//  its mask byte
+	byte[] buffer = new byte[16];
+	// And how long it is
+	// (Un-compressed codes are 1 byte each, compressed codes
+	//   are two)
+	int bufferLen = 0;
+	
+	// How far through the input and output streams we are
+	int posInp = 0;
+	int posOut = 0;
+	
+	// What the next mask byte to output will be
+	int nextMask = 0;
+	// And how many bits we've already set
+	int maskBitsSet = 0;
+	
+	// This is a byte as looked up in the dictionary
+	// It needs to be signed, as it'll get passed on to
+	//  the output stream
+	byte dataB;
+	// This is an unsigned byte read from the stream
+	// It needs to be unsigned, so that bit stuff works
+	int dataI;
+	
+	// Have we hit the end of the file yet?
+	boolean going = true;
+	
+	while( going ) {
+		dataI = src.read();
+		posInp++;
+		if(dataI == -1) { going = false; }
+		
+		// Decide if we're going to output uncompressed or compressed
+		//  for this byte
+		// (It takes 2 bytes to hold a compressed code, so it's only
+		//  worth doing for 3+ byte long sequences)
+		// TODO
+		
+		boolean compressThis = true;
+		if(compressThis) {
+			// Set the mask bit for us 
+			nextMask += (1<<maskBitsSet);
+			
+			// And add us to the buffer + dictionary
+			buffer[bufferLen] = fromInt(dataI);
+			bufferLen++;
+			dict[(posOut&4095)] = fromInt(dataI);
+			posOut++;
+		} else {
+			// ????
+		}
+		// Increment the mask bit count, we've done another code
+		maskBitsSet++;
+		
+		// If we've just done the 8th bit, or reached the end
+		//  of the stream, output our mask and data
+		if(maskBitsSet == 8 || !going) {
+			// Output
+			res.write(new byte[] { fromInt(nextMask) } );
+			res.write(buffer, 0, bufferLen);
+			
+			// Reset things
+			nextMask = 0;
+			maskBitsSet = 0;
+			bufferLen = 0;
+		}
+	}
+}
+
 }

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

@@ -22,10 +22,34 @@ import junit.framework.TestCase;
 
 public class TestHDGFLZW extends TestCase {
 	public static final byte[] testTrailerComp = new byte[] {
-		123, -60, 2, -21, -16, 1, 0, 0, -72, -13, -16, 78, -32, -5, 1, 
+		123,      // *mask bit*
-		0, 3, -21, -16, 10, 5, 4, -21, -16, 21, 9, -21, -16, 103, -21, 
+		-60, 2, 
-		-16, 34, -36, -1, 52, 15, 70, 15, 120, 88, 15, -7, -2, -28, -9, 
+		-21, -16, // 3 @ 4093
-		-123, 21, 0, 44, -122, 1, -4, 104, 15, -24, -13, 40, -98, 32, 
+		1, 0, 0, -72, 
+		-13, -16, // 3 @ 5
+		78,       // *mask bit*
+		-32, -5,  // 14 @ 4082 
+		1, 0, 3, 
+		-21, -16, // 3 @ 4093
+		10, 5,    // 8 @ 28
+		4, 
+		-21, -16, // 3 @ 4093
+		21,       // *mask bit* 
+		9, 
+		-21, -16, // 3 @ 4093 
+		103, -21, -16, 34, 
+		-36, -1,  // 18 @ 4078
+		52, 15,   // 18 @ 70 
+		70, 15,   // 18 @ 88 
+		120,      // *mask bit*
+		88, 15,   // 18 @ 106 
+		-7, -2,   // 17 @ 11 
+		-28, -9,  // 10 @ 4086
+		-123, 21, 0, 44, 
+		-122, 1,  // 4 @ 152 
+		-4,       // *mask bit* 
+		104, 15,  // 18 @ 122 
+		-24, -13, 40, -98, 32, 
 		78, 102, -67, -1, -2, -30, 64, 40, -67, -113, -73, 116, -98, 
 		-85, 2, 66, 123, 9, 109, -85, 2, -89, 14, -56, -69, -83, -79, 
 		-34, -3, 120, 110, 75, -9, -10, 20, -6, -25, -12, 22, -21, -16, 
@@ -81,6 +105,33 @@ public class TestHDGFLZW extends TestCase {
 		0, 0, 42, 1, 0, 0, 84, 0, 0, 0, 0, 0
 	};
 	
+	public void testFromToInt() throws Exception {
+		byte b255 = -1;
+		assertEquals(255, HDGFLZW.fromByte(b255));
+		assertEquals(-1, HDGFLZW.fromInt( HDGFLZW.fromByte(b255) ));
+		assertEquals(-1, HDGFLZW.fromInt( 255 ));
+		
+		byte b11 = 11;
+		assertEquals(11, HDGFLZW.fromByte(b11));
+		assertEquals(11, HDGFLZW.fromInt( HDGFLZW.fromByte(b11) ));
+		assertEquals(11, HDGFLZW.fromInt( 11 ));
+		
+		byte b0 = 0;
+		assertEquals(0, HDGFLZW.fromByte(b0));
+		assertEquals(0, HDGFLZW.fromInt( HDGFLZW.fromByte(b0) ));
+		assertEquals(0, HDGFLZW.fromInt( 0 ));
+		
+		byte b127 = 127;
+		assertEquals(127, HDGFLZW.fromByte(b127));
+		assertEquals(127, HDGFLZW.fromInt( HDGFLZW.fromByte(b127) ));
+		assertEquals(127, HDGFLZW.fromInt( 127 ));
+		
+		byte b128 = -128;
+		assertEquals(128, HDGFLZW.fromByte(b128));
+		assertEquals(-128, HDGFLZW.fromInt( HDGFLZW.fromByte(b128) ));
+		assertEquals(-128, HDGFLZW.fromInt( 128 ));
+	}
+	
 	public void testCounts() throws Exception {
 		assertEquals(339, testTrailerComp.length);
 		assertEquals(632, testTrailerDecomp.length);
@@ -92,10 +143,44 @@ public class TestHDGFLZW extends TestCase {
 		// Check it's of the right size
 		assertEquals(632, dec.length);
 
-		// Now check it matches
+/*		
+		// Encode it again using our engine
+		byte[] comp = lzw.compress(new ByteArrayInputStream(testTrailerDecomp));
+		
+		// Check it's of the right size
+		assertEquals(339, comp.length);
+*/
+	}
+	
+	public void testDecompress() throws Exception {
+		assertEquals(339, testTrailerComp.length);
+		assertEquals(632, testTrailerDecomp.length);
+		
+		// Decode it using our engine
+		HDGFLZW lzw = new HDGFLZW();
+		byte[] dec = lzw.decode(new ByteArrayInputStream(testTrailerComp));
+		
+		// Now check it's the right data
+		assertEquals(632, dec.length);
 		for(int i=0; i<dec.length; i++) {
 			if(dec[i] != testTrailerDecomp[i]) 
 				System.err.println(i + "\t" + dec[i] + "\t" + testTrailerDecomp[i]);
 		}
 	}
+	
+	public void DISABLEDtestCompress() throws Exception {
+		assertEquals(339, testTrailerComp.length);
+		assertEquals(632, testTrailerDecomp.length);
+		
+		// Compress it using our engine
+		HDGFLZW lzw = new HDGFLZW();
+		byte[] comp = lzw.compress(new ByteArrayInputStream(testTrailerDecomp));
+		
+		// Now check it's the right data
+		assertEquals(339, comp.length);
+		for(int i=0; i<comp.length; i++) {
+			if(comp[i] != testTrailerComp[i]) 
+				System.err.println(i + "\t" + comp[i] + "\t" + testTrailerComp[i]);
+		}
+	}
 }