GZIP格式解析和Deflate静态Huffman解压缩

GZIP是封装了Deflate压缩的格式文件，Deflate使用了无压缩、Huffman+LZ77进行压缩，Huffman包括静态Huffman和动态Huffman。

Java实现了GZIP格式解析，静态Huffman解压缩，CRC32校验

gzip文件格式解析代码：

	    BinaryInputStream bis = new BinaryInputStream(bytes); // 二进制字节流读取类
	
		Header header = new Header();  // 读取GZIP header
		header.ID1 = bis.ReadUInt8();  // GZIP ID1
		header.ID2 = bis.ReadUInt8();  // GzIP ID2
		header.CM = bis.ReadUInt8();  
		header.Flag = bis.ReadUInt8();
		header.Time = bis.ReadUnix32TimeStamp(); 
		header.Xfl = bis.ReadUInt8();
		header.OS = bis.ReadUInt8();
		
		gzip.header = header;
		
		Console.printlnf("ID=%xb%xb", header.ID1, header.ID2); // 打印
		Console.printlnf("CM=%xb", header.CM);
		Console.printlnf("Flag=%xb", header.Flag);
		Console.printlnf("MTime=%S", header.Time);
		Console.printlnf("XFL=%xb", header.Xfl);
		Console.printlnf("OS=%xb", header.OS);
		
		if ( (header.Flag & 0b00001000) != 0) {  // 如果标志位是文件名
			String filename = bis.ReadCString();
			Console.printlnf("filename=%S", filename);
		}
		else // 其他待实现
			throw new java.lang.IllegalArgumentException(header.Flag +"");
		
		byte[] result ; // 解压后的字节
		int data_len ;  // 解压前数据长度
		int crc;        // 程序计算解压后的CRC32（见上篇文章）
		
		if (header.CM == 8) { // deflate 
			data_len = bytes.length - bis.GetPosition() - 8;			
			byte[] data = bis.ReadBytes(data_len);			
			result = Deflate.uncompress(data);  // 解压缩，返回解压后字节流
			
			crc = CRC.CRC32(result); // 计算循环冗余码
		}
		else // 待实现
			throw new java.lang.IllegalArgumentException(header.CM +""); 
		
		gzip.crc = bis.ReadUInt32(); // gzip文件自身存储的crc32值
		gzip.isize = bis.ReadInt32(); 
		
		System.out.println("gzip crc="+ Long.toHexString(gzip.crc) +",calc-crc=" + Integer.toHexString( crc) );

解析结果如下：

显示了GZIP标志、压缩方法、压缩时间、原始文件名、操作系统类型、CRC校验值

GZIP中存储的CRC32值==解压后计算的CRC32值。(0xa93145a2)

Deflate -静态Huffman解压缩：

	// 读取扩展Code
	private static int ReadExtCode(BitsInputStream bis, int len) {
		bis.setOrder(BitOrder.LeftIsHigh);
		return bis.ReadBits(len);
	}
	
	// 读取距离
	private static int ReadDistance(BitsInputStream bis) {
		bis.setOrder(BitOrder.RightIsHigh);
		return bis.ReadBits(5);
	}
	
	// 读取Code
	private static int ReadCode(BitsInputStream bis) {
		bis.setOrder(BitOrder.RightIsHigh);
		int code = bis.ReadBits(7);
		Integer value = FixHuffmanTable_7.get(code);
		
		if (value == null) {
			int ext = bis.ReadBit();
			code = (code << 1 | ext);
			
			value = FixHuffmanTable_8.get(code);
			
			if (value == null) {
				ext = bis.ReadBit();
				code = (code << 1 | ext);
				value = FixHuffmanTable_9.get(code);
				
				if (value == -1)
					throw new java.lang.IllegalArgumentException(code + "");
			}
		}

		return value;
	}
	
	// Deflate解压缩
	public static byte[] uncompress(final byte[] _input) throws IOException {
		IntArrayBuffer baos = new IntArrayBuffer();
		
		// 位流读取类
		BitsInputStream bis = new BitsInputStream(_input);
		
		
		int bFinal = bis.ReadBits(1); // 读取Deflate头，0 – 还有后续子块；1 – 该子块是最后一块。
		int bType = bis.ReadBits(2);  // 读取Deflate头，00 – 不压缩；01 – 静态Huffman编码压缩；10 – 动态Huffman编码压缩；11 – 保留
		
		while (true) {
			if (bType == 0) { // 无压缩
				int len = bis.ReadBits(16);
				int nlen = bis.ReadBits(16);
				
				assert len + nlen == 65535;
				throw new java.lang.IllegalArgumentException(bType + "");
			}
			else if (bType == 1) { // fixed Huffman
				while (true) {
					int value = ReadCode(bis); // 读取Huffman code
					
					if (value >= 0 && value <= 255) // literal
						baos.Write(value);
					else if (value == 256) // 结束标志
						break ;
					else if (value >= 257 && value <= 285) {
						int length = LengthExtraCodeLengthsTable.get(value);
						int bits = LengthExtraCodeBitsTable.get(value); 
						
						if (bits != 0) {
							int ext =  ReadExtCode(bis, bits); // 读取长度
							length = length + ext; 
						}
						
						value = ReadDistance(bis); // 读取距离
						int distance = DistanceExtraCodeLengthsTable.get(value);
						bits = DistanceExtraCodeBitsTable.get(value);
						
						if (bits != 0) {
							int ext =ReadExtCode(bis , bits); // 读取距离
							distance = distance + ext; 
						}
						
						// LZ77滑动窗口计算获取量
						int[] arr = baos.GetInts();
						int d = arr.length - distance;
						if (d < 0) {
							d = 0;
							length = length + distance - arr.length;
						}
						
						// 读取滑动窗口，写入到结果
						for (int i=0; i<length; i++) {	
							int m = arr[ d + i];
							
							baos.Write(m);
							arr = baos.GetInts();
						}
						
					}
					else 
						throw new java.lang.IllegalArgumentException(value + "");

				}
			}
			else if (bType == 2) { // dynamic huffman 动态Huffman待实现
				throw new java.lang.IllegalArgumentException(bType + "");
			}
			else 
				throw new java.lang.IllegalArgumentException(bType + "");
			
			if (bFinal == 1) // 最后一个块
				break ;
		}
		
		int[] b = baos.GetInts();
		byte[] result = new byte[b.length];
		for (int i=0; i<b.length; i++) 
			result[i] = (byte) b[i];
		
		return result;

测试结果如下：

将字符串"abcabcabcabcabcabcabcabcabcabcabcabc1111"写入到文件

运行gzip程序压缩（系统程序）

使用Java程序对gzip文件解析并显示解压后的内容。 （gzip使用了静态huffman）。