Modified: websites/production/commons/content/proper/commons-compress/javadocs/api-release/src-html/org/apache/commons/compress/compressors/lz77support/LZ77Compressor.Block.html ============================================================================== --- websites/production/commons/content/proper/commons-compress/javadocs/api-release/src-html/org/apache/commons/compress/compressors/lz77support/LZ77Compressor.Block.html (original) +++ websites/production/commons/content/proper/commons-compress/javadocs/api-release/src-html/org/apache/commons/compress/compressors/lz77support/LZ77Compressor.Block.html Sat Feb 10 13:44:59 2018 @@ -1,5 +1,5 @@ <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd"> -<html lang="de"> +<html lang="en"> <head> <title>Source code</title> <link rel="stylesheet" type="text/css" href="../../../../../../../stylesheet.css" title="Style"> @@ -92,472 +92,479 @@ <span class="sourceLineNo">084</span> */<a name="line.84"></a> <span class="sourceLineNo">085</span>public class LZ77Compressor {<a name="line.85"></a> <span class="sourceLineNo">086</span><a name="line.86"></a> -<span class="sourceLineNo">087</span> /** Base class representing blocks the compressor may emit. */<a name="line.87"></a> -<span class="sourceLineNo">088</span> public static abstract class Block {<a name="line.88"></a> -<span class="sourceLineNo">089</span> /** Enumeration of the block types the compressor may emit. */<a name="line.89"></a> -<span class="sourceLineNo">090</span> public enum BlockType {<a name="line.90"></a> -<span class="sourceLineNo">091</span> LITERAL, BACK_REFERENCE, EOD<a name="line.91"></a> -<span class="sourceLineNo">092</span> }<a name="line.92"></a> -<span class="sourceLineNo">093</span> public abstract BlockType getType();<a name="line.93"></a> -<span class="sourceLineNo">094</span> }<a name="line.94"></a> -<span class="sourceLineNo">095</span><a name="line.95"></a> -<span class="sourceLineNo">096</span> /**<a name="line.96"></a> -<span class="sourceLineNo">097</span> * Represents a literal block of data.<a name="line.97"></a> -<span class="sourceLineNo">098</span> *<a name="line.98"></a> -<span class="sourceLineNo">099</span> * <p>For performance reasons this encapsulates the real data, not<a name="line.99"></a> -<span class="sourceLineNo">100</span> * a copy of it. Don't modify the data and process it inside of<a name="line.100"></a> -<span class="sourceLineNo">101</span> * {@link Callback#accept} immediately as it will get overwritten<a name="line.101"></a> -<span class="sourceLineNo">102</span> * sooner or later.</p><a name="line.102"></a> -<span class="sourceLineNo">103</span> */<a name="line.103"></a> -<span class="sourceLineNo">104</span> public static final class LiteralBlock extends Block {<a name="line.104"></a> -<span class="sourceLineNo">105</span> private final byte[] data;<a name="line.105"></a> -<span class="sourceLineNo">106</span> private final int offset, length;<a name="line.106"></a> -<span class="sourceLineNo">107</span> public LiteralBlock(byte[] data, int offset, int length) {<a name="line.107"></a> -<span class="sourceLineNo">108</span> this.data = data;<a name="line.108"></a> -<span class="sourceLineNo">109</span> this.offset = offset;<a name="line.109"></a> -<span class="sourceLineNo">110</span> this.length = length;<a name="line.110"></a> -<span class="sourceLineNo">111</span> }<a name="line.111"></a> -<span class="sourceLineNo">112</span> /**<a name="line.112"></a> -<span class="sourceLineNo">113</span> * The literal data.<a name="line.113"></a> -<span class="sourceLineNo">114</span> *<a name="line.114"></a> -<span class="sourceLineNo">115</span> * <p>This returns a life view of the actual data in order to<a name="line.115"></a> -<span class="sourceLineNo">116</span> * avoid copying, modify the array at your own risk.</p><a name="line.116"></a> -<span class="sourceLineNo">117</span> * @return the data<a name="line.117"></a> -<span class="sourceLineNo">118</span> */<a name="line.118"></a> -<span class="sourceLineNo">119</span> public byte[] getData() {<a name="line.119"></a> -<span class="sourceLineNo">120</span> return data;<a name="line.120"></a> -<span class="sourceLineNo">121</span> }<a name="line.121"></a> -<span class="sourceLineNo">122</span> /**<a name="line.122"></a> -<span class="sourceLineNo">123</span> * Offset into data where the literal block starts.<a name="line.123"></a> -<span class="sourceLineNo">124</span> * @return the offset<a name="line.124"></a> +<span class="sourceLineNo">087</span> /**<a name="line.87"></a> +<span class="sourceLineNo">088</span> * Base class representing blocks the compressor may emit.<a name="line.88"></a> +<span class="sourceLineNo">089</span> *<a name="line.89"></a> +<span class="sourceLineNo">090</span> * <p>This class is not supposed to be subclassed by classes<a name="line.90"></a> +<span class="sourceLineNo">091</span> * outside of Commons Compress so it is considered internal and<a name="line.91"></a> +<span class="sourceLineNo">092</span> * changed that would break subclasses may get introduced with<a name="line.92"></a> +<span class="sourceLineNo">093</span> * future releases.</p><a name="line.93"></a> +<span class="sourceLineNo">094</span> */<a name="line.94"></a> +<span class="sourceLineNo">095</span> public static abstract class Block {<a name="line.95"></a> +<span class="sourceLineNo">096</span> /** Enumeration of the block types the compressor may emit. */<a name="line.96"></a> +<span class="sourceLineNo">097</span> public enum BlockType {<a name="line.97"></a> +<span class="sourceLineNo">098</span> LITERAL, BACK_REFERENCE, EOD<a name="line.98"></a> +<span class="sourceLineNo">099</span> }<a name="line.99"></a> +<span class="sourceLineNo">100</span> public abstract BlockType getType();<a name="line.100"></a> +<span class="sourceLineNo">101</span> }<a name="line.101"></a> +<span class="sourceLineNo">102</span><a name="line.102"></a> +<span class="sourceLineNo">103</span> /**<a name="line.103"></a> +<span class="sourceLineNo">104</span> * Represents a literal block of data.<a name="line.104"></a> +<span class="sourceLineNo">105</span> *<a name="line.105"></a> +<span class="sourceLineNo">106</span> * <p>For performance reasons this encapsulates the real data, not<a name="line.106"></a> +<span class="sourceLineNo">107</span> * a copy of it. Don't modify the data and process it inside of<a name="line.107"></a> +<span class="sourceLineNo">108</span> * {@link Callback#accept} immediately as it will get overwritten<a name="line.108"></a> +<span class="sourceLineNo">109</span> * sooner or later.</p><a name="line.109"></a> +<span class="sourceLineNo">110</span> */<a name="line.110"></a> +<span class="sourceLineNo">111</span> public static final class LiteralBlock extends Block {<a name="line.111"></a> +<span class="sourceLineNo">112</span> private final byte[] data;<a name="line.112"></a> +<span class="sourceLineNo">113</span> private final int offset, length;<a name="line.113"></a> +<span class="sourceLineNo">114</span> public LiteralBlock(byte[] data, int offset, int length) {<a name="line.114"></a> +<span class="sourceLineNo">115</span> this.data = data;<a name="line.115"></a> +<span class="sourceLineNo">116</span> this.offset = offset;<a name="line.116"></a> +<span class="sourceLineNo">117</span> this.length = length;<a name="line.117"></a> +<span class="sourceLineNo">118</span> }<a name="line.118"></a> +<span class="sourceLineNo">119</span> /**<a name="line.119"></a> +<span class="sourceLineNo">120</span> * The literal data.<a name="line.120"></a> +<span class="sourceLineNo">121</span> *<a name="line.121"></a> +<span class="sourceLineNo">122</span> * <p>This returns a life view of the actual data in order to<a name="line.122"></a> +<span class="sourceLineNo">123</span> * avoid copying, modify the array at your own risk.</p><a name="line.123"></a> +<span class="sourceLineNo">124</span> * @return the data<a name="line.124"></a> <span class="sourceLineNo">125</span> */<a name="line.125"></a> -<span class="sourceLineNo">126</span> public int getOffset() {<a name="line.126"></a> -<span class="sourceLineNo">127</span> return offset;<a name="line.127"></a> +<span class="sourceLineNo">126</span> public byte[] getData() {<a name="line.126"></a> +<span class="sourceLineNo">127</span> return data;<a name="line.127"></a> <span class="sourceLineNo">128</span> }<a name="line.128"></a> <span class="sourceLineNo">129</span> /**<a name="line.129"></a> -<span class="sourceLineNo">130</span> * Length of literal block.<a name="line.130"></a> -<span class="sourceLineNo">131</span> * @return the length<a name="line.131"></a> +<span class="sourceLineNo">130</span> * Offset into data where the literal block starts.<a name="line.130"></a> +<span class="sourceLineNo">131</span> * @return the offset<a name="line.131"></a> <span class="sourceLineNo">132</span> */<a name="line.132"></a> -<span class="sourceLineNo">133</span> public int getLength() {<a name="line.133"></a> -<span class="sourceLineNo">134</span> return length;<a name="line.134"></a> +<span class="sourceLineNo">133</span> public int getOffset() {<a name="line.133"></a> +<span class="sourceLineNo">134</span> return offset;<a name="line.134"></a> <span class="sourceLineNo">135</span> }<a name="line.135"></a> -<span class="sourceLineNo">136</span> @Override<a name="line.136"></a> -<span class="sourceLineNo">137</span> public BlockType getType() {<a name="line.137"></a> -<span class="sourceLineNo">138</span> return BlockType.LITERAL;<a name="line.138"></a> -<span class="sourceLineNo">139</span> }<a name="line.139"></a> -<span class="sourceLineNo">140</span> @Override<a name="line.140"></a> -<span class="sourceLineNo">141</span> public String toString() {<a name="line.141"></a> -<span class="sourceLineNo">142</span> return "LiteralBlock starting at " + offset + " with length " + length;<a name="line.142"></a> -<span class="sourceLineNo">143</span> }<a name="line.143"></a> -<span class="sourceLineNo">144</span> }<a name="line.144"></a> -<span class="sourceLineNo">145</span><a name="line.145"></a> -<span class="sourceLineNo">146</span> /**<a name="line.146"></a> -<span class="sourceLineNo">147</span> * Represents a back-reference.<a name="line.147"></a> -<span class="sourceLineNo">148</span> */<a name="line.148"></a> -<span class="sourceLineNo">149</span> public static final class BackReference extends Block {<a name="line.149"></a> -<span class="sourceLineNo">150</span> private final int offset, length;<a name="line.150"></a> -<span class="sourceLineNo">151</span> public BackReference(int offset, int length) {<a name="line.151"></a> -<span class="sourceLineNo">152</span> this.offset = offset;<a name="line.152"></a> -<span class="sourceLineNo">153</span> this.length = length;<a name="line.153"></a> -<span class="sourceLineNo">154</span> }<a name="line.154"></a> -<span class="sourceLineNo">155</span> /**<a name="line.155"></a> -<span class="sourceLineNo">156</span> * Provides the offset of the back-reference.<a name="line.156"></a> -<span class="sourceLineNo">157</span> * @return the offset<a name="line.157"></a> -<span class="sourceLineNo">158</span> */<a name="line.158"></a> -<span class="sourceLineNo">159</span> public int getOffset() {<a name="line.159"></a> -<span class="sourceLineNo">160</span> return offset;<a name="line.160"></a> +<span class="sourceLineNo">136</span> /**<a name="line.136"></a> +<span class="sourceLineNo">137</span> * Length of literal block.<a name="line.137"></a> +<span class="sourceLineNo">138</span> * @return the length<a name="line.138"></a> +<span class="sourceLineNo">139</span> */<a name="line.139"></a> +<span class="sourceLineNo">140</span> public int getLength() {<a name="line.140"></a> +<span class="sourceLineNo">141</span> return length;<a name="line.141"></a> +<span class="sourceLineNo">142</span> }<a name="line.142"></a> +<span class="sourceLineNo">143</span> @Override<a name="line.143"></a> +<span class="sourceLineNo">144</span> public BlockType getType() {<a name="line.144"></a> +<span class="sourceLineNo">145</span> return BlockType.LITERAL;<a name="line.145"></a> +<span class="sourceLineNo">146</span> }<a name="line.146"></a> +<span class="sourceLineNo">147</span> @Override<a name="line.147"></a> +<span class="sourceLineNo">148</span> public String toString() {<a name="line.148"></a> +<span class="sourceLineNo">149</span> return "LiteralBlock starting at " + offset + " with length " + length;<a name="line.149"></a> +<span class="sourceLineNo">150</span> }<a name="line.150"></a> +<span class="sourceLineNo">151</span> }<a name="line.151"></a> +<span class="sourceLineNo">152</span><a name="line.152"></a> +<span class="sourceLineNo">153</span> /**<a name="line.153"></a> +<span class="sourceLineNo">154</span> * Represents a back-reference.<a name="line.154"></a> +<span class="sourceLineNo">155</span> */<a name="line.155"></a> +<span class="sourceLineNo">156</span> public static final class BackReference extends Block {<a name="line.156"></a> +<span class="sourceLineNo">157</span> private final int offset, length;<a name="line.157"></a> +<span class="sourceLineNo">158</span> public BackReference(int offset, int length) {<a name="line.158"></a> +<span class="sourceLineNo">159</span> this.offset = offset;<a name="line.159"></a> +<span class="sourceLineNo">160</span> this.length = length;<a name="line.160"></a> <span class="sourceLineNo">161</span> }<a name="line.161"></a> <span class="sourceLineNo">162</span> /**<a name="line.162"></a> -<span class="sourceLineNo">163</span> * Provides the length of the back-reference.<a name="line.163"></a> -<span class="sourceLineNo">164</span> * @return the length<a name="line.164"></a> +<span class="sourceLineNo">163</span> * Provides the offset of the back-reference.<a name="line.163"></a> +<span class="sourceLineNo">164</span> * @return the offset<a name="line.164"></a> <span class="sourceLineNo">165</span> */<a name="line.165"></a> -<span class="sourceLineNo">166</span> public int getLength() {<a name="line.166"></a> -<span class="sourceLineNo">167</span> return length;<a name="line.167"></a> +<span class="sourceLineNo">166</span> public int getOffset() {<a name="line.166"></a> +<span class="sourceLineNo">167</span> return offset;<a name="line.167"></a> <span class="sourceLineNo">168</span> }<a name="line.168"></a> -<span class="sourceLineNo">169</span> @Override<a name="line.169"></a> -<span class="sourceLineNo">170</span> public BlockType getType() {<a name="line.170"></a> -<span class="sourceLineNo">171</span> return BlockType.BACK_REFERENCE;<a name="line.171"></a> -<span class="sourceLineNo">172</span> }<a name="line.172"></a> -<span class="sourceLineNo">173</span> @Override<a name="line.173"></a> -<span class="sourceLineNo">174</span> public String toString() {<a name="line.174"></a> -<span class="sourceLineNo">175</span> return "BackReference with offset " + offset + " and length " + length;<a name="line.175"></a> -<span class="sourceLineNo">176</span> }<a name="line.176"></a> -<span class="sourceLineNo">177</span> }<a name="line.177"></a> -<span class="sourceLineNo">178</span><a name="line.178"></a> -<span class="sourceLineNo">179</span> /** A simple "we are done" marker. */<a name="line.179"></a> -<span class="sourceLineNo">180</span> public static final class EOD extends Block {<a name="line.180"></a> -<span class="sourceLineNo">181</span> @Override<a name="line.181"></a> -<span class="sourceLineNo">182</span> public BlockType getType() {<a name="line.182"></a> -<span class="sourceLineNo">183</span> return BlockType.EOD;<a name="line.183"></a> -<span class="sourceLineNo">184</span> }<a name="line.184"></a> -<span class="sourceLineNo">185</span> }<a name="line.185"></a> -<span class="sourceLineNo">186</span><a name="line.186"></a> -<span class="sourceLineNo">187</span> private static final Block THE_EOD = new EOD();<a name="line.187"></a> -<span class="sourceLineNo">188</span><a name="line.188"></a> -<span class="sourceLineNo">189</span> /**<a name="line.189"></a> -<span class="sourceLineNo">190</span> * Callback invoked while the compressor processes data.<a name="line.190"></a> -<span class="sourceLineNo">191</span> *<a name="line.191"></a> -<span class="sourceLineNo">192</span> * <p>The callback is invoked on the same thread that receives the<a name="line.192"></a> -<span class="sourceLineNo">193</span> * bytes to compress and may be invoked multiple times during the<a name="line.193"></a> -<span class="sourceLineNo">194</span> * execution of {@link #compress} or {@link #finish}.</p><a name="line.194"></a> -<span class="sourceLineNo">195</span> */<a name="line.195"></a> -<span class="sourceLineNo">196</span> public interface Callback {<a name="line.196"></a> -<span class="sourceLineNo">197</span> /**<a name="line.197"></a> -<span class="sourceLineNo">198</span> * Consumes a block.<a name="line.198"></a> -<span class="sourceLineNo">199</span> * @param b the block to consume<a name="line.199"></a> -<span class="sourceLineNo">200</span> * @throws IOException in case of an error<a name="line.200"></a> -<span class="sourceLineNo">201</span> */<a name="line.201"></a> -<span class="sourceLineNo">202</span> void accept(Block b) throws IOException;<a name="line.202"></a> -<span class="sourceLineNo">203</span> }<a name="line.203"></a> -<span class="sourceLineNo">204</span><a name="line.204"></a> -<span class="sourceLineNo">205</span> static final int NUMBER_OF_BYTES_IN_HASH = 3;<a name="line.205"></a> -<span class="sourceLineNo">206</span> private static final int NO_MATCH = -1;<a name="line.206"></a> -<span class="sourceLineNo">207</span><a name="line.207"></a> -<span class="sourceLineNo">208</span> private final Parameters params;<a name="line.208"></a> -<span class="sourceLineNo">209</span> private final Callback callback;<a name="line.209"></a> -<span class="sourceLineNo">210</span><a name="line.210"></a> -<span class="sourceLineNo">211</span> // the sliding window, twice as big as "windowSize" parameter<a name="line.211"></a> -<span class="sourceLineNo">212</span> private final byte[] window;<a name="line.212"></a> -<span class="sourceLineNo">213</span> // the head of hash-chain - indexed by hash-code, points to the<a name="line.213"></a> -<span class="sourceLineNo">214</span> // location inside of window of the latest sequence of bytes with<a name="line.214"></a> -<span class="sourceLineNo">215</span> // the given hash.<a name="line.215"></a> -<span class="sourceLineNo">216</span> private final int[] head;<a name="line.216"></a> -<span class="sourceLineNo">217</span> // for each window-location points to the latest earlier location<a name="line.217"></a> -<span class="sourceLineNo">218</span> // with the same hash. Only stores values for the latest<a name="line.218"></a> -<span class="sourceLineNo">219</span> // "windowSize" elements, the index is "window location modulo<a name="line.219"></a> -<span class="sourceLineNo">220</span> // windowSize".<a name="line.220"></a> -<span class="sourceLineNo">221</span> private final int[] prev;<a name="line.221"></a> -<span class="sourceLineNo">222</span><a name="line.222"></a> -<span class="sourceLineNo">223</span> // bit mask used when indexing into prev<a name="line.223"></a> -<span class="sourceLineNo">224</span> private final int wMask;<a name="line.224"></a> -<span class="sourceLineNo">225</span><a name="line.225"></a> -<span class="sourceLineNo">226</span> private boolean initialized = false;<a name="line.226"></a> -<span class="sourceLineNo">227</span> // the position inside of window that shall be encoded right now<a name="line.227"></a> -<span class="sourceLineNo">228</span> private int currentPosition;<a name="line.228"></a> -<span class="sourceLineNo">229</span> // the number of bytes available to compress including the one at<a name="line.229"></a> -<span class="sourceLineNo">230</span> // currentPosition<a name="line.230"></a> -<span class="sourceLineNo">231</span> private int lookahead = 0;<a name="line.231"></a> -<span class="sourceLineNo">232</span> // the hash of the three bytes stating at the current position<a name="line.232"></a> -<span class="sourceLineNo">233</span> private int insertHash = 0;<a name="line.233"></a> -<span class="sourceLineNo">234</span> // the position inside of the window where the current literal<a name="line.234"></a> -<span class="sourceLineNo">235</span> // block starts (in case we are inside of a literal block).<a name="line.235"></a> -<span class="sourceLineNo">236</span> private int blockStart = 0;<a name="line.236"></a> -<span class="sourceLineNo">237</span> // position of the current match<a name="line.237"></a> -<span class="sourceLineNo">238</span> private int matchStart = NO_MATCH;<a name="line.238"></a> -<span class="sourceLineNo">239</span> // number of missed insertString calls for the up to three last<a name="line.239"></a> -<span class="sourceLineNo">240</span> // bytes of the last match that can only be performed once more<a name="line.240"></a> -<span class="sourceLineNo">241</span> // data has been read<a name="line.241"></a> -<span class="sourceLineNo">242</span> private int missedInserts = 0;<a name="line.242"></a> -<span class="sourceLineNo">243</span><a name="line.243"></a> -<span class="sourceLineNo">244</span> /**<a name="line.244"></a> -<span class="sourceLineNo">245</span> * Initializes a compressor with parameters and a callback.<a name="line.245"></a> -<span class="sourceLineNo">246</span> * @param params the parameters<a name="line.246"></a> -<span class="sourceLineNo">247</span> * @param callback the callback<a name="line.247"></a> -<span class="sourceLineNo">248</span> * @throws NullPointerException if either parameter is <code>null</code><a name="line.248"></a> -<span class="sourceLineNo">249</span> */<a name="line.249"></a> -<span class="sourceLineNo">250</span> public LZ77Compressor(Parameters params, Callback callback) {<a name="line.250"></a> -<span class="sourceLineNo">251</span> if (params == null) {<a name="line.251"></a> -<span class="sourceLineNo">252</span> throw new NullPointerException("params must not be null");<a name="line.252"></a> -<span class="sourceLineNo">253</span> }<a name="line.253"></a> -<span class="sourceLineNo">254</span> if (callback == null) {<a name="line.254"></a> -<span class="sourceLineNo">255</span> throw new NullPointerException("callback must not be null");<a name="line.255"></a> -<span class="sourceLineNo">256</span> }<a name="line.256"></a> -<span class="sourceLineNo">257</span> this.params = params;<a name="line.257"></a> -<span class="sourceLineNo">258</span> this.callback = callback;<a name="line.258"></a> -<span class="sourceLineNo">259</span><a name="line.259"></a> -<span class="sourceLineNo">260</span> final int wSize = params.getWindowSize();<a name="line.260"></a> -<span class="sourceLineNo">261</span> window = new byte[wSize * 2];<a name="line.261"></a> -<span class="sourceLineNo">262</span> wMask = wSize - 1;<a name="line.262"></a> -<span class="sourceLineNo">263</span> head = new int[HASH_SIZE];<a name="line.263"></a> -<span class="sourceLineNo">264</span> Arrays.fill(head, NO_MATCH);<a name="line.264"></a> -<span class="sourceLineNo">265</span> prev = new int[wSize];<a name="line.265"></a> -<span class="sourceLineNo">266</span> }<a name="line.266"></a> -<span class="sourceLineNo">267</span><a name="line.267"></a> -<span class="sourceLineNo">268</span> /**<a name="line.268"></a> -<span class="sourceLineNo">269</span> * Feeds bytes into the compressor which in turn may emit zero or<a name="line.269"></a> -<span class="sourceLineNo">270</span> * more blocks to the callback during the execution of this<a name="line.270"></a> -<span class="sourceLineNo">271</span> * method.<a name="line.271"></a> -<span class="sourceLineNo">272</span> * @param data the data to compress - must not be null<a name="line.272"></a> -<span class="sourceLineNo">273</span> * @throws IOException if the callback throws an exception<a name="line.273"></a> -<span class="sourceLineNo">274</span> */<a name="line.274"></a> -<span class="sourceLineNo">275</span> public void compress(byte[] data) throws IOException {<a name="line.275"></a> -<span class="sourceLineNo">276</span> compress(data, 0, data.length);<a name="line.276"></a> -<span class="sourceLineNo">277</span> }<a name="line.277"></a> -<span class="sourceLineNo">278</span><a name="line.278"></a> -<span class="sourceLineNo">279</span> /**<a name="line.279"></a> -<span class="sourceLineNo">280</span> * Feeds bytes into the compressor which in turn may emit zero or<a name="line.280"></a> -<span class="sourceLineNo">281</span> * more blocks to the callback during the execution of this<a name="line.281"></a> -<span class="sourceLineNo">282</span> * method.<a name="line.282"></a> -<span class="sourceLineNo">283</span> * @param data the data to compress - must not be null<a name="line.283"></a> -<span class="sourceLineNo">284</span> * @param off the start offset of the data<a name="line.284"></a> -<span class="sourceLineNo">285</span> * @param len the number of bytes to compress<a name="line.285"></a> -<span class="sourceLineNo">286</span> * @throws IOException if the callback throws an exception<a name="line.286"></a> -<span class="sourceLineNo">287</span> */<a name="line.287"></a> -<span class="sourceLineNo">288</span> public void compress(byte[] data, int off, int len) throws IOException {<a name="line.288"></a> -<span class="sourceLineNo">289</span> final int wSize = params.getWindowSize();<a name="line.289"></a> -<span class="sourceLineNo">290</span> while (len > wSize) { // chop into windowSize sized chunks<a name="line.290"></a> -<span class="sourceLineNo">291</span> doCompress(data, off, wSize);<a name="line.291"></a> -<span class="sourceLineNo">292</span> off += wSize;<a name="line.292"></a> -<span class="sourceLineNo">293</span> len -= wSize;<a name="line.293"></a> -<span class="sourceLineNo">294</span> }<a name="line.294"></a> -<span class="sourceLineNo">295</span> if (len > 0) {<a name="line.295"></a> -<span class="sourceLineNo">296</span> doCompress(data, off, len);<a name="line.296"></a> -<span class="sourceLineNo">297</span> }<a name="line.297"></a> -<span class="sourceLineNo">298</span> }<a name="line.298"></a> -<span class="sourceLineNo">299</span><a name="line.299"></a> -<span class="sourceLineNo">300</span> /**<a name="line.300"></a> -<span class="sourceLineNo">301</span> * Tells the compressor to process all remaining data and signal<a name="line.301"></a> -<span class="sourceLineNo">302</span> * end of data to the callback.<a name="line.302"></a> -<span class="sourceLineNo">303</span> *<a name="line.303"></a> -<span class="sourceLineNo">304</span> * <p>The compressor will in turn emit at least one block ({@link<a name="line.304"></a> -<span class="sourceLineNo">305</span> * EOD}) but potentially multiple blocks to the callback during<a name="line.305"></a> -<span class="sourceLineNo">306</span> * the execution of this method.</p><a name="line.306"></a> -<span class="sourceLineNo">307</span> * @throws IOException if the callback throws an exception<a name="line.307"></a> -<span class="sourceLineNo">308</span> */<a name="line.308"></a> -<span class="sourceLineNo">309</span> public void finish() throws IOException {<a name="line.309"></a> -<span class="sourceLineNo">310</span> if (blockStart != currentPosition || lookahead > 0) {<a name="line.310"></a> -<span class="sourceLineNo">311</span> currentPosition += lookahead;<a name="line.311"></a> -<span class="sourceLineNo">312</span> flushLiteralBlock();<a name="line.312"></a> -<span class="sourceLineNo">313</span> }<a name="line.313"></a> -<span class="sourceLineNo">314</span> callback.accept(THE_EOD);<a name="line.314"></a> -<span class="sourceLineNo">315</span> }<a name="line.315"></a> -<span class="sourceLineNo">316</span><a name="line.316"></a> -<span class="sourceLineNo">317</span> /**<a name="line.317"></a> -<span class="sourceLineNo">318</span> * Adds some initial data to fill the window with.<a name="line.318"></a> -<span class="sourceLineNo">319</span> *<a name="line.319"></a> -<span class="sourceLineNo">320</span> * <p>This is used if the stream has been cut into blocks and<a name="line.320"></a> -<span class="sourceLineNo">321</span> * back-references of one block may refer to data of the previous<a name="line.321"></a> -<span class="sourceLineNo">322</span> * block(s). One such example is the LZ4 frame format using block<a name="line.322"></a> -<span class="sourceLineNo">323</span> * dependency.</p><a name="line.323"></a> -<span class="sourceLineNo">324</span> *<a name="line.324"></a> -<span class="sourceLineNo">325</span> * @param data the data to fill the window with.<a name="line.325"></a> -<span class="sourceLineNo">326</span> * @throws IllegalStateException if the compressor has already started to accept data<a name="line.326"></a> -<span class="sourceLineNo">327</span> */<a name="line.327"></a> -<span class="sourceLineNo">328</span> public void prefill(byte[] data) {<a name="line.328"></a> -<span class="sourceLineNo">329</span> if (currentPosition != 0 || lookahead != 0) {<a name="line.329"></a> -<span class="sourceLineNo">330</span> throw new IllegalStateException("the compressor has already started to accept data, can't prefill anymore");<a name="line.330"></a> -<span class="sourceLineNo">331</span> }<a name="line.331"></a> -<span class="sourceLineNo">332</span><a name="line.332"></a> -<span class="sourceLineNo">333</span> // don't need more than windowSize for back-references<a name="line.333"></a> -<span class="sourceLineNo">334</span> final int len = Math.min(params.getWindowSize(), data.length);<a name="line.334"></a> -<span class="sourceLineNo">335</span> System.arraycopy(data, data.length - len, window, 0, len);<a name="line.335"></a> -<span class="sourceLineNo">336</span><a name="line.336"></a> -<span class="sourceLineNo">337</span> if (len >= NUMBER_OF_BYTES_IN_HASH) {<a name="line.337"></a> -<span class="sourceLineNo">338</span> initialize();<a name="line.338"></a> -<span class="sourceLineNo">339</span> final int stop = len - NUMBER_OF_BYTES_IN_HASH + 1;<a name="line.339"></a> -<span class="sourceLineNo">340</span> for (int i = 0; i < stop; i++) {<a name="line.340"></a> -<span class="sourceLineNo">341</span> insertString(i);<a name="line.341"></a> -<span class="sourceLineNo">342</span> }<a name="line.342"></a> -<span class="sourceLineNo">343</span> missedInserts = NUMBER_OF_BYTES_IN_HASH - 1;<a name="line.343"></a> -<span class="sourceLineNo">344</span> } else { // not enough data to hash anything<a name="line.344"></a> -<span class="sourceLineNo">345</span> missedInserts = len;<a name="line.345"></a> -<span class="sourceLineNo">346</span> }<a name="line.346"></a> -<span class="sourceLineNo">347</span> blockStart = currentPosition = len;<a name="line.347"></a> -<span class="sourceLineNo">348</span> }<a name="line.348"></a> -<span class="sourceLineNo">349</span><a name="line.349"></a> -<span class="sourceLineNo">350</span> // we use a 15 bit hashcode as calculated in updateHash<a name="line.350"></a> -<span class="sourceLineNo">351</span> private static final int HASH_SIZE = 1 << 15;<a name="line.351"></a> -<span class="sourceLineNo">352</span> private static final int HASH_MASK = HASH_SIZE - 1;<a name="line.352"></a> -<span class="sourceLineNo">353</span> private static final int H_SHIFT = 5;<a name="line.353"></a> -<span class="sourceLineNo">354</span><a name="line.354"></a> -<span class="sourceLineNo">355</span> /**<a name="line.355"></a> -<span class="sourceLineNo">356</span> * Assumes we are calculating the hash for three consecutive bytes<a name="line.356"></a> -<span class="sourceLineNo">357</span> * as a rolling hash, i.e. for bytes ABCD if H is the hash of ABC<a name="line.357"></a> -<span class="sourceLineNo">358</span> * the new hash for BCD is nextHash(H, D).<a name="line.358"></a> -<span class="sourceLineNo">359</span> *<a name="line.359"></a> -<span class="sourceLineNo">360</span> * <p>The hash is shifted by five bits on each update so all<a name="line.360"></a> -<span class="sourceLineNo">361</span> * effects of A have been swapped after the third update.</p><a name="line.361"></a> -<span class="sourceLineNo">362</span> */<a name="line.362"></a> -<span class="sourceLineNo">363</span> private int nextHash(int oldHash, byte nextByte) {<a name="line.363"></a> -<span class="sourceLineNo">364</span> final int nextVal = nextByte & 0xFF;<a name="line.364"></a> -<span class="sourceLineNo">365</span> return ((oldHash << H_SHIFT) ^ nextVal) & HASH_MASK;<a name="line.365"></a> -<span class="sourceLineNo">366</span> }<a name="line.366"></a> -<span class="sourceLineNo">367</span><a name="line.367"></a> -<span class="sourceLineNo">368</span> // performs the actual algorithm with the pre-condition len <= windowSize<a name="line.368"></a> -<span class="sourceLineNo">369</span> private void doCompress(byte[] data, int off, int len) throws IOException {<a name="line.369"></a> -<span class="sourceLineNo">370</span> int spaceLeft = window.length - currentPosition - lookahead;<a name="line.370"></a> -<span class="sourceLineNo">371</span> if (len > spaceLeft) {<a name="line.371"></a> -<span class="sourceLineNo">372</span> slide();<a name="line.372"></a> -<span class="sourceLineNo">373</span> }<a name="line.373"></a> -<span class="sourceLineNo">374</span> System.arraycopy(data, off, window, currentPosition + lookahead, len);<a name="line.374"></a> -<span class="sourceLineNo">375</span> lookahead += len;<a name="line.375"></a> -<span class="sourceLineNo">376</span> if (!initialized && lookahead >= params.getMinBackReferenceLength()) {<a name="line.376"></a> -<span class="sourceLineNo">377</span> initialize();<a name="line.377"></a> -<span class="sourceLineNo">378</span> }<a name="line.378"></a> -<span class="sourceLineNo">379</span> if (initialized) {<a name="line.379"></a> -<span class="sourceLineNo">380</span> compress();<a name="line.380"></a> -<span class="sourceLineNo">381</span> }<a name="line.381"></a> -<span class="sourceLineNo">382</span> }<a name="line.382"></a> -<span class="sourceLineNo">383</span><a name="line.383"></a> -<span class="sourceLineNo">384</span> private void slide() throws IOException {<a name="line.384"></a> -<span class="sourceLineNo">385</span> final int wSize = params.getWindowSize();<a name="line.385"></a> -<span class="sourceLineNo">386</span> if (blockStart != currentPosition && blockStart < wSize) {<a name="line.386"></a> -<span class="sourceLineNo">387</span> flushLiteralBlock();<a name="line.387"></a> -<span class="sourceLineNo">388</span> blockStart = currentPosition;<a name="line.388"></a> -<span class="sourceLineNo">389</span> }<a name="line.389"></a> -<span class="sourceLineNo">390</span> System.arraycopy(window, wSize, window, 0, wSize);<a name="line.390"></a> -<span class="sourceLineNo">391</span> currentPosition -= wSize;<a name="line.391"></a> -<span class="sourceLineNo">392</span> matchStart -= wSize;<a name="line.392"></a> -<span class="sourceLineNo">393</span> blockStart -= wSize;<a name="line.393"></a> -<span class="sourceLineNo">394</span> for (int i = 0; i < HASH_SIZE; i++) {<a name="line.394"></a> -<span class="sourceLineNo">395</span> int h = head[i];<a name="line.395"></a> -<span class="sourceLineNo">396</span> head[i] = h >= wSize ? h - wSize : NO_MATCH;<a name="line.396"></a> -<span class="sourceLineNo">397</span> }<a name="line.397"></a> -<span class="sourceLineNo">398</span> for (int i = 0; i < wSize; i++) {<a name="line.398"></a> -<span class="sourceLineNo">399</span> int p = prev[i];<a name="line.399"></a> -<span class="sourceLineNo">400</span> prev[i] = p >= wSize ? p - wSize : NO_MATCH;<a name="line.400"></a> -<span class="sourceLineNo">401</span> }<a name="line.401"></a> -<span class="sourceLineNo">402</span> }<a name="line.402"></a> -<span class="sourceLineNo">403</span><a name="line.403"></a> -<span class="sourceLineNo">404</span> private void initialize() {<a name="line.404"></a> -<span class="sourceLineNo">405</span> for (int i = 0; i < NUMBER_OF_BYTES_IN_HASH - 1; i++) {<a name="line.405"></a> -<span class="sourceLineNo">406</span> insertHash = nextHash(insertHash, window[i]);<a name="line.406"></a> -<span class="sourceLineNo">407</span> }<a name="line.407"></a> -<span class="sourceLineNo">408</span> initialized = true;<a name="line.408"></a> +<span class="sourceLineNo">169</span> /**<a name="line.169"></a> +<span class="sourceLineNo">170</span> * Provides the length of the back-reference.<a name="line.170"></a> +<span class="sourceLineNo">171</span> * @return the length<a name="line.171"></a> +<span class="sourceLineNo">172</span> */<a name="line.172"></a> +<span class="sourceLineNo">173</span> public int getLength() {<a name="line.173"></a> +<span class="sourceLineNo">174</span> return length;<a name="line.174"></a> +<span class="sourceLineNo">175</span> }<a name="line.175"></a> +<span class="sourceLineNo">176</span> @Override<a name="line.176"></a> +<span class="sourceLineNo">177</span> public BlockType getType() {<a name="line.177"></a> +<span class="sourceLineNo">178</span> return BlockType.BACK_REFERENCE;<a name="line.178"></a> +<span class="sourceLineNo">179</span> }<a name="line.179"></a> +<span class="sourceLineNo">180</span> @Override<a name="line.180"></a> +<span class="sourceLineNo">181</span> public String toString() {<a name="line.181"></a> +<span class="sourceLineNo">182</span> return "BackReference with offset " + offset + " and length " + length;<a name="line.182"></a> +<span class="sourceLineNo">183</span> }<a name="line.183"></a> +<span class="sourceLineNo">184</span> }<a name="line.184"></a> +<span class="sourceLineNo">185</span><a name="line.185"></a> +<span class="sourceLineNo">186</span> /** A simple "we are done" marker. */<a name="line.186"></a> +<span class="sourceLineNo">187</span> public static final class EOD extends Block {<a name="line.187"></a> +<span class="sourceLineNo">188</span> @Override<a name="line.188"></a> +<span class="sourceLineNo">189</span> public BlockType getType() {<a name="line.189"></a> +<span class="sourceLineNo">190</span> return BlockType.EOD;<a name="line.190"></a> +<span class="sourceLineNo">191</span> }<a name="line.191"></a> +<span class="sourceLineNo">192</span> }<a name="line.192"></a> +<span class="sourceLineNo">193</span><a name="line.193"></a> +<span class="sourceLineNo">194</span> private static final Block THE_EOD = new EOD();<a name="line.194"></a> +<span class="sourceLineNo">195</span><a name="line.195"></a> +<span class="sourceLineNo">196</span> /**<a name="line.196"></a> +<span class="sourceLineNo">197</span> * Callback invoked while the compressor processes data.<a name="line.197"></a> +<span class="sourceLineNo">198</span> *<a name="line.198"></a> +<span class="sourceLineNo">199</span> * <p>The callback is invoked on the same thread that receives the<a name="line.199"></a> +<span class="sourceLineNo">200</span> * bytes to compress and may be invoked multiple times during the<a name="line.200"></a> +<span class="sourceLineNo">201</span> * execution of {@link #compress} or {@link #finish}.</p><a name="line.201"></a> +<span class="sourceLineNo">202</span> */<a name="line.202"></a> +<span class="sourceLineNo">203</span> public interface Callback {<a name="line.203"></a> +<span class="sourceLineNo">204</span> /**<a name="line.204"></a> +<span class="sourceLineNo">205</span> * Consumes a block.<a name="line.205"></a> +<span class="sourceLineNo">206</span> * @param b the block to consume<a name="line.206"></a> +<span class="sourceLineNo">207</span> * @throws IOException in case of an error<a name="line.207"></a> +<span class="sourceLineNo">208</span> */<a name="line.208"></a> +<span class="sourceLineNo">209</span> void accept(Block b) throws IOException;<a name="line.209"></a> +<span class="sourceLineNo">210</span> }<a name="line.210"></a> +<span class="sourceLineNo">211</span><a name="line.211"></a> +<span class="sourceLineNo">212</span> static final int NUMBER_OF_BYTES_IN_HASH = 3;<a name="line.212"></a> +<span class="sourceLineNo">213</span> private static final int NO_MATCH = -1;<a name="line.213"></a> +<span class="sourceLineNo">214</span><a name="line.214"></a> +<span class="sourceLineNo">215</span> private final Parameters params;<a name="line.215"></a> +<span class="sourceLineNo">216</span> private final Callback callback;<a name="line.216"></a> +<span class="sourceLineNo">217</span><a name="line.217"></a> +<span class="sourceLineNo">218</span> // the sliding window, twice as big as "windowSize" parameter<a name="line.218"></a> +<span class="sourceLineNo">219</span> private final byte[] window;<a name="line.219"></a> +<span class="sourceLineNo">220</span> // the head of hash-chain - indexed by hash-code, points to the<a name="line.220"></a> +<span class="sourceLineNo">221</span> // location inside of window of the latest sequence of bytes with<a name="line.221"></a> +<span class="sourceLineNo">222</span> // the given hash.<a name="line.222"></a> +<span class="sourceLineNo">223</span> private final int[] head;<a name="line.223"></a> +<span class="sourceLineNo">224</span> // for each window-location points to the latest earlier location<a name="line.224"></a> +<span class="sourceLineNo">225</span> // with the same hash. Only stores values for the latest<a name="line.225"></a> +<span class="sourceLineNo">226</span> // "windowSize" elements, the index is "window location modulo<a name="line.226"></a> +<span class="sourceLineNo">227</span> // windowSize".<a name="line.227"></a> +<span class="sourceLineNo">228</span> private final int[] prev;<a name="line.228"></a> +<span class="sourceLineNo">229</span><a name="line.229"></a> +<span class="sourceLineNo">230</span> // bit mask used when indexing into prev<a name="line.230"></a> +<span class="sourceLineNo">231</span> private final int wMask;<a name="line.231"></a> +<span class="sourceLineNo">232</span><a name="line.232"></a> +<span class="sourceLineNo">233</span> private boolean initialized = false;<a name="line.233"></a> +<span class="sourceLineNo">234</span> // the position inside of window that shall be encoded right now<a name="line.234"></a> +<span class="sourceLineNo">235</span> private int currentPosition;<a name="line.235"></a> +<span class="sourceLineNo">236</span> // the number of bytes available to compress including the one at<a name="line.236"></a> +<span class="sourceLineNo">237</span> // currentPosition<a name="line.237"></a> +<span class="sourceLineNo">238</span> private int lookahead = 0;<a name="line.238"></a> +<span class="sourceLineNo">239</span> // the hash of the three bytes stating at the current position<a name="line.239"></a> +<span class="sourceLineNo">240</span> private int insertHash = 0;<a name="line.240"></a> +<span class="sourceLineNo">241</span> // the position inside of the window where the current literal<a name="line.241"></a> +<span class="sourceLineNo">242</span> // block starts (in case we are inside of a literal block).<a name="line.242"></a> +<span class="sourceLineNo">243</span> private int blockStart = 0;<a name="line.243"></a> +<span class="sourceLineNo">244</span> // position of the current match<a name="line.244"></a> +<span class="sourceLineNo">245</span> private int matchStart = NO_MATCH;<a name="line.245"></a> +<span class="sourceLineNo">246</span> // number of missed insertString calls for the up to three last<a name="line.246"></a> +<span class="sourceLineNo">247</span> // bytes of the last match that can only be performed once more<a name="line.247"></a> +<span class="sourceLineNo">248</span> // data has been read<a name="line.248"></a> +<span class="sourceLineNo">249</span> private int missedInserts = 0;<a name="line.249"></a> +<span class="sourceLineNo">250</span><a name="line.250"></a> +<span class="sourceLineNo">251</span> /**<a name="line.251"></a> +<span class="sourceLineNo">252</span> * Initializes a compressor with parameters and a callback.<a name="line.252"></a> +<span class="sourceLineNo">253</span> * @param params the parameters<a name="line.253"></a> +<span class="sourceLineNo">254</span> * @param callback the callback<a name="line.254"></a> +<span class="sourceLineNo">255</span> * @throws NullPointerException if either parameter is <code>null</code><a name="line.255"></a> +<span class="sourceLineNo">256</span> */<a name="line.256"></a> +<span class="sourceLineNo">257</span> public LZ77Compressor(Parameters params, Callback callback) {<a name="line.257"></a> +<span class="sourceLineNo">258</span> if (params == null) {<a name="line.258"></a> +<span class="sourceLineNo">259</span> throw new NullPointerException("params must not be null");<a name="line.259"></a> +<span class="sourceLineNo">260</span> }<a name="line.260"></a> +<span class="sourceLineNo">261</span> if (callback == null) {<a name="line.261"></a> +<span class="sourceLineNo">262</span> throw new NullPointerException("callback must not be null");<a name="line.262"></a> +<span class="sourceLineNo">263</span> }<a name="line.263"></a> +<span class="sourceLineNo">264</span> this.params = params;<a name="line.264"></a> +<span class="sourceLineNo">265</span> this.callback = callback;<a name="line.265"></a> +<span class="sourceLineNo">266</span><a name="line.266"></a> +<span class="sourceLineNo">267</span> final int wSize = params.getWindowSize();<a name="line.267"></a> +<span class="sourceLineNo">268</span> window = new byte[wSize * 2];<a name="line.268"></a> +<span class="sourceLineNo">269</span> wMask = wSize - 1;<a name="line.269"></a> +<span class="sourceLineNo">270</span> head = new int[HASH_SIZE];<a name="line.270"></a> +<span class="sourceLineNo">271</span> Arrays.fill(head, NO_MATCH);<a name="line.271"></a> +<span class="sourceLineNo">272</span> prev = new int[wSize];<a name="line.272"></a> +<span class="sourceLineNo">273</span> }<a name="line.273"></a> +<span class="sourceLineNo">274</span><a name="line.274"></a> +<span class="sourceLineNo">275</span> /**<a name="line.275"></a> +<span class="sourceLineNo">276</span> * Feeds bytes into the compressor which in turn may emit zero or<a name="line.276"></a> +<span class="sourceLineNo">277</span> * more blocks to the callback during the execution of this<a name="line.277"></a> +<span class="sourceLineNo">278</span> * method.<a name="line.278"></a> +<span class="sourceLineNo">279</span> * @param data the data to compress - must not be null<a name="line.279"></a> +<span class="sourceLineNo">280</span> * @throws IOException if the callback throws an exception<a name="line.280"></a> +<span class="sourceLineNo">281</span> */<a name="line.281"></a> +<span class="sourceLineNo">282</span> public void compress(byte[] data) throws IOException {<a name="line.282"></a> +<span class="sourceLineNo">283</span> compress(data, 0, data.length);<a name="line.283"></a> +<span class="sourceLineNo">284</span> }<a name="line.284"></a> +<span class="sourceLineNo">285</span><a name="line.285"></a> +<span class="sourceLineNo">286</span> /**<a name="line.286"></a> +<span class="sourceLineNo">287</span> * Feeds bytes into the compressor which in turn may emit zero or<a name="line.287"></a> +<span class="sourceLineNo">288</span> * more blocks to the callback during the execution of this<a name="line.288"></a> +<span class="sourceLineNo">289</span> * method.<a name="line.289"></a> +<span class="sourceLineNo">290</span> * @param data the data to compress - must not be null<a name="line.290"></a> +<span class="sourceLineNo">291</span> * @param off the start offset of the data<a name="line.291"></a> +<span class="sourceLineNo">292</span> * @param len the number of bytes to compress<a name="line.292"></a> +<span class="sourceLineNo">293</span> * @throws IOException if the callback throws an exception<a name="line.293"></a> +<span class="sourceLineNo">294</span> */<a name="line.294"></a> +<span class="sourceLineNo">295</span> public void compress(byte[] data, int off, int len) throws IOException {<a name="line.295"></a> +<span class="sourceLineNo">296</span> final int wSize = params.getWindowSize();<a name="line.296"></a> +<span class="sourceLineNo">297</span> while (len > wSize) { // chop into windowSize sized chunks<a name="line.297"></a> +<span class="sourceLineNo">298</span> doCompress(data, off, wSize);<a name="line.298"></a> +<span class="sourceLineNo">299</span> off += wSize;<a name="line.299"></a> +<span class="sourceLineNo">300</span> len -= wSize;<a name="line.300"></a> +<span class="sourceLineNo">301</span> }<a name="line.301"></a> +<span class="sourceLineNo">302</span> if (len > 0) {<a name="line.302"></a> +<span class="sourceLineNo">303</span> doCompress(data, off, len);<a name="line.303"></a> +<span class="sourceLineNo">304</span> }<a name="line.304"></a> +<span class="sourceLineNo">305</span> }<a name="line.305"></a> +<span class="sourceLineNo">306</span><a name="line.306"></a> +<span class="sourceLineNo">307</span> /**<a name="line.307"></a> +<span class="sourceLineNo">308</span> * Tells the compressor to process all remaining data and signal<a name="line.308"></a> +<span class="sourceLineNo">309</span> * end of data to the callback.<a name="line.309"></a> +<span class="sourceLineNo">310</span> *<a name="line.310"></a> +<span class="sourceLineNo">311</span> * <p>The compressor will in turn emit at least one block ({@link<a name="line.311"></a> +<span class="sourceLineNo">312</span> * EOD}) but potentially multiple blocks to the callback during<a name="line.312"></a> +<span class="sourceLineNo">313</span> * the execution of this method.</p><a name="line.313"></a> +<span class="sourceLineNo">314</span> * @throws IOException if the callback throws an exception<a name="line.314"></a> +<span class="sourceLineNo">315</span> */<a name="line.315"></a> +<span class="sourceLineNo">316</span> public void finish() throws IOException {<a name="line.316"></a> +<span class="sourceLineNo">317</span> if (blockStart != currentPosition || lookahead > 0) {<a name="line.317"></a> +<span class="sourceLineNo">318</span> currentPosition += lookahead;<a name="line.318"></a> +<span class="sourceLineNo">319</span> flushLiteralBlock();<a name="line.319"></a> +<span class="sourceLineNo">320</span> }<a name="line.320"></a> +<span class="sourceLineNo">321</span> callback.accept(THE_EOD);<a name="line.321"></a> +<span class="sourceLineNo">322</span> }<a name="line.322"></a> +<span class="sourceLineNo">323</span><a name="line.323"></a> +<span class="sourceLineNo">324</span> /**<a name="line.324"></a> +<span class="sourceLineNo">325</span> * Adds some initial data to fill the window with.<a name="line.325"></a> +<span class="sourceLineNo">326</span> *<a name="line.326"></a> +<span class="sourceLineNo">327</span> * <p>This is used if the stream has been cut into blocks and<a name="line.327"></a> +<span class="sourceLineNo">328</span> * back-references of one block may refer to data of the previous<a name="line.328"></a> +<span class="sourceLineNo">329</span> * block(s). One such example is the LZ4 frame format using block<a name="line.329"></a> +<span class="sourceLineNo">330</span> * dependency.</p><a name="line.330"></a> +<span class="sourceLineNo">331</span> *<a name="line.331"></a> +<span class="sourceLineNo">332</span> * @param data the data to fill the window with.<a name="line.332"></a> +<span class="sourceLineNo">333</span> * @throws IllegalStateException if the compressor has already started to accept data<a name="line.333"></a> +<span class="sourceLineNo">334</span> */<a name="line.334"></a> +<span class="sourceLineNo">335</span> public void prefill(byte[] data) {<a name="line.335"></a> +<span class="sourceLineNo">336</span> if (currentPosition != 0 || lookahead != 0) {<a name="line.336"></a> +<span class="sourceLineNo">337</span> throw new IllegalStateException("the compressor has already started to accept data, can't prefill anymore");<a name="line.337"></a> +<span class="sourceLineNo">338</span> }<a name="line.338"></a> +<span class="sourceLineNo">339</span><a name="line.339"></a> +<span class="sourceLineNo">340</span> // don't need more than windowSize for back-references<a name="line.340"></a> +<span class="sourceLineNo">341</span> final int len = Math.min(params.getWindowSize(), data.length);<a name="line.341"></a> +<span class="sourceLineNo">342</span> System.arraycopy(data, data.length - len, window, 0, len);<a name="line.342"></a> +<span class="sourceLineNo">343</span><a name="line.343"></a> +<span class="sourceLineNo">344</span> if (len >= NUMBER_OF_BYTES_IN_HASH) {<a name="line.344"></a> +<span class="sourceLineNo">345</span> initialize();<a name="line.345"></a> +<span class="sourceLineNo">346</span> final int stop = len - NUMBER_OF_BYTES_IN_HASH + 1;<a name="line.346"></a> +<span class="sourceLineNo">347</span> for (int i = 0; i < stop; i++) {<a name="line.347"></a> +<span class="sourceLineNo">348</span> insertString(i);<a name="line.348"></a> +<span class="sourceLineNo">349</span> }<a name="line.349"></a> +<span class="sourceLineNo">350</span> missedInserts = NUMBER_OF_BYTES_IN_HASH - 1;<a name="line.350"></a> +<span class="sourceLineNo">351</span> } else { // not enough data to hash anything<a name="line.351"></a> +<span class="sourceLineNo">352</span> missedInserts = len;<a name="line.352"></a> +<span class="sourceLineNo">353</span> }<a name="line.353"></a> +<span class="sourceLineNo">354</span> blockStart = currentPosition = len;<a name="line.354"></a> +<span class="sourceLineNo">355</span> }<a name="line.355"></a> +<span class="sourceLineNo">356</span><a name="line.356"></a> +<span class="sourceLineNo">357</span> // we use a 15 bit hashcode as calculated in updateHash<a name="line.357"></a> +<span class="sourceLineNo">358</span> private static final int HASH_SIZE = 1 << 15;<a name="line.358"></a> +<span class="sourceLineNo">359</span> private static final int HASH_MASK = HASH_SIZE - 1;<a name="line.359"></a> +<span class="sourceLineNo">360</span> private static final int H_SHIFT = 5;<a name="line.360"></a> +<span class="sourceLineNo">361</span><a name="line.361"></a> +<span class="sourceLineNo">362</span> /**<a name="line.362"></a> +<span class="sourceLineNo">363</span> * Assumes we are calculating the hash for three consecutive bytes<a name="line.363"></a> +<span class="sourceLineNo">364</span> * as a rolling hash, i.e. for bytes ABCD if H is the hash of ABC<a name="line.364"></a> +<span class="sourceLineNo">365</span> * the new hash for BCD is nextHash(H, D).<a name="line.365"></a> +<span class="sourceLineNo">366</span> *<a name="line.366"></a> +<span class="sourceLineNo">367</span> * <p>The hash is shifted by five bits on each update so all<a name="line.367"></a> +<span class="sourceLineNo">368</span> * effects of A have been swapped after the third update.</p><a name="line.368"></a> +<span class="sourceLineNo">369</span> */<a name="line.369"></a> +<span class="sourceLineNo">370</span> private int nextHash(int oldHash, byte nextByte) {<a name="line.370"></a> +<span class="sourceLineNo">371</span> final int nextVal = nextByte & 0xFF;<a name="line.371"></a> +<span class="sourceLineNo">372</span> return ((oldHash << H_SHIFT) ^ nextVal) & HASH_MASK;<a name="line.372"></a> +<span class="sourceLineNo">373</span> }<a name="line.373"></a> +<span class="sourceLineNo">374</span><a name="line.374"></a> +<span class="sourceLineNo">375</span> // performs the actual algorithm with the pre-condition len <= windowSize<a name="line.375"></a> +<span class="sourceLineNo">376</span> private void doCompress(byte[] data, int off, int len) throws IOException {<a name="line.376"></a> +<span class="sourceLineNo">377</span> int spaceLeft = window.length - currentPosition - lookahead;<a name="line.377"></a> +<span class="sourceLineNo">378</span> if (len > spaceLeft) {<a name="line.378"></a> +<span class="sourceLineNo">379</span> slide();<a name="line.379"></a> +<span class="sourceLineNo">380</span> }<a name="line.380"></a> +<span class="sourceLineNo">381</span> System.arraycopy(data, off, window, currentPosition + lookahead, len);<a name="line.381"></a> +<span class="sourceLineNo">382</span> lookahead += len;<a name="line.382"></a> +<span class="sourceLineNo">383</span> if (!initialized && lookahead >= params.getMinBackReferenceLength()) {<a name="line.383"></a> +<span class="sourceLineNo">384</span> initialize();<a name="line.384"></a> +<span class="sourceLineNo">385</span> }<a name="line.385"></a> +<span class="sourceLineNo">386</span> if (initialized) {<a name="line.386"></a> +<span class="sourceLineNo">387</span> compress();<a name="line.387"></a> +<span class="sourceLineNo">388</span> }<a name="line.388"></a> +<span class="sourceLineNo">389</span> }<a name="line.389"></a> +<span class="sourceLineNo">390</span><a name="line.390"></a> +<span class="sourceLineNo">391</span> private void slide() throws IOException {<a name="line.391"></a> +<span class="sourceLineNo">392</span> final int wSize = params.getWindowSize();<a name="line.392"></a> +<span class="sourceLineNo">393</span> if (blockStart != currentPosition && blockStart < wSize) {<a name="line.393"></a> +<span class="sourceLineNo">394</span> flushLiteralBlock();<a name="line.394"></a> +<span class="sourceLineNo">395</span> blockStart = currentPosition;<a name="line.395"></a> +<span class="sourceLineNo">396</span> }<a name="line.396"></a> +<span class="sourceLineNo">397</span> System.arraycopy(window, wSize, window, 0, wSize);<a name="line.397"></a> +<span class="sourceLineNo">398</span> currentPosition -= wSize;<a name="line.398"></a> +<span class="sourceLineNo">399</span> matchStart -= wSize;<a name="line.399"></a> +<span class="sourceLineNo">400</span> blockStart -= wSize;<a name="line.400"></a> +<span class="sourceLineNo">401</span> for (int i = 0; i < HASH_SIZE; i++) {<a name="line.401"></a> +<span class="sourceLineNo">402</span> int h = head[i];<a name="line.402"></a> +<span class="sourceLineNo">403</span> head[i] = h >= wSize ? h - wSize : NO_MATCH;<a name="line.403"></a> +<span class="sourceLineNo">404</span> }<a name="line.404"></a> +<span class="sourceLineNo">405</span> for (int i = 0; i < wSize; i++) {<a name="line.405"></a> +<span class="sourceLineNo">406</span> int p = prev[i];<a name="line.406"></a> +<span class="sourceLineNo">407</span> prev[i] = p >= wSize ? p - wSize : NO_MATCH;<a name="line.407"></a> +<span class="sourceLineNo">408</span> }<a name="line.408"></a> <span class="sourceLineNo">409</span> }<a name="line.409"></a> <span class="sourceLineNo">410</span><a name="line.410"></a> -<span class="sourceLineNo">411</span> private void compress() throws IOException {<a name="line.411"></a> -<span class="sourceLineNo">412</span> final int minMatch = params.getMinBackReferenceLength();<a name="line.412"></a> -<span class="sourceLineNo">413</span> final boolean lazy = params.getLazyMatching();<a name="line.413"></a> -<span class="sourceLineNo">414</span> final int lazyThreshold = params.getLazyMatchingThreshold();<a name="line.414"></a> -<span class="sourceLineNo">415</span><a name="line.415"></a> -<span class="sourceLineNo">416</span> while (lookahead >= minMatch) {<a name="line.416"></a> -<span class="sourceLineNo">417</span> catchUpMissedInserts();<a name="line.417"></a> -<span class="sourceLineNo">418</span> int matchLength = 0;<a name="line.418"></a> -<span class="sourceLineNo">419</span> int hashHead = insertString(currentPosition);<a name="line.419"></a> -<span class="sourceLineNo">420</span> if (hashHead != NO_MATCH && hashHead - currentPosition <= params.getMaxOffset()) {<a name="line.420"></a> -<span class="sourceLineNo">421</span> // sets matchStart as a side effect<a name="line.421"></a> -<span class="sourceLineNo">422</span> matchLength = longestMatch(hashHead);<a name="line.422"></a> -<span class="sourceLineNo">423</span><a name="line.423"></a> -<span class="sourceLineNo">424</span> if (lazy && matchLength <= lazyThreshold && lookahead > minMatch) {<a name="line.424"></a> -<span class="sourceLineNo">425</span> // try to find a longer match using the next position<a name="line.425"></a> -<span class="sourceLineNo">426</span> matchLength = longestMatchForNextPosition(matchLength);<a name="line.426"></a> -<span class="sourceLineNo">427</span> }<a name="line.427"></a> -<span class="sourceLineNo">428</span> }<a name="line.428"></a> -<span class="sourceLineNo">429</span> if (matchLength >= minMatch) {<a name="line.429"></a> -<span class="sourceLineNo">430</span> if (blockStart != currentPosition) {<a name="line.430"></a> -<span class="sourceLineNo">431</span> // emit preceeding literal block<a name="line.431"></a> -<span class="sourceLineNo">432</span> flushLiteralBlock();<a name="line.432"></a> -<span class="sourceLineNo">433</span> blockStart = NO_MATCH;<a name="line.433"></a> +<span class="sourceLineNo">411</span> private void initialize() {<a name="line.411"></a> +<span class="sourceLineNo">412</span> for (int i = 0; i < NUMBER_OF_BYTES_IN_HASH - 1; i++) {<a name="line.412"></a> +<span class="sourceLineNo">413</span> insertHash = nextHash(insertHash, window[i]);<a name="line.413"></a> +<span class="sourceLineNo">414</span> }<a name="line.414"></a> +<span class="sourceLineNo">415</span> initialized = true;<a name="line.415"></a> +<span class="sourceLineNo">416</span> }<a name="line.416"></a> +<span class="sourceLineNo">417</span><a name="line.417"></a> +<span class="sourceLineNo">418</span> private void compress() throws IOException {<a name="line.418"></a> +<span class="sourceLineNo">419</span> final int minMatch = params.getMinBackReferenceLength();<a name="line.419"></a> +<span class="sourceLineNo">420</span> final boolean lazy = params.getLazyMatching();<a name="line.420"></a> +<span class="sourceLineNo">421</span> final int lazyThreshold = params.getLazyMatchingThreshold();<a name="line.421"></a> +<span class="sourceLineNo">422</span><a name="line.422"></a> +<span class="sourceLineNo">423</span> while (lookahead >= minMatch) {<a name="line.423"></a> +<span class="sourceLineNo">424</span> catchUpMissedInserts();<a name="line.424"></a> +<span class="sourceLineNo">425</span> int matchLength = 0;<a name="line.425"></a> +<span class="sourceLineNo">426</span> int hashHead = insertString(currentPosition);<a name="line.426"></a> +<span class="sourceLineNo">427</span> if (hashHead != NO_MATCH && hashHead - currentPosition <= params.getMaxOffset()) {<a name="line.427"></a> +<span class="sourceLineNo">428</span> // sets matchStart as a side effect<a name="line.428"></a> +<span class="sourceLineNo">429</span> matchLength = longestMatch(hashHead);<a name="line.429"></a> +<span class="sourceLineNo">430</span><a name="line.430"></a> +<span class="sourceLineNo">431</span> if (lazy && matchLength <= lazyThreshold && lookahead > minMatch) {<a name="line.431"></a> +<span class="sourceLineNo">432</span> // try to find a longer match using the next position<a name="line.432"></a> +<span class="sourceLineNo">433</span> matchLength = longestMatchForNextPosition(matchLength);<a name="line.433"></a> <span class="sourceLineNo">434</span> }<a name="line.434"></a> -<span class="sourceLineNo">435</span> flushBackReference(matchLength);<a name="line.435"></a> -<span class="sourceLineNo">436</span> insertStringsInMatch(matchLength);<a name="line.436"></a> -<span class="sourceLineNo">437</span> lookahead -= matchLength;<a name="line.437"></a> -<span class="sourceLineNo">438</span> currentPosition += matchLength;<a name="line.438"></a> -<span class="sourceLineNo">439</span> blockStart = currentPosition;<a name="line.439"></a> -<span class="sourceLineNo">440</span> } else {<a name="line.440"></a> -<span class="sourceLineNo">441</span> // no match, append to current or start a new literal<a name="line.441"></a> -<span class="sourceLineNo">442</span> lookahead--;<a name="line.442"></a> -<span class="sourceLineNo">443</span> currentPosition++;<a name="line.443"></a> -<span class="sourceLineNo">444</span> if (currentPosition - blockStart >= params.getMaxLiteralLength()) {<a name="line.444"></a> -<span class="sourceLineNo">445</span> flushLiteralBlock();<a name="line.445"></a> -<span class="sourceLineNo">446</span> blockStart = currentPosition;<a name="line.446"></a> -<span class="sourceLineNo">447</span> }<a name="line.447"></a> -<span class="sourceLineNo">448</span> }<a name="line.448"></a> -<span class="sourceLineNo">449</span> }<a name="line.449"></a> -<span class="sourceLineNo">450</span> }<a name="line.450"></a> -<span class="sourceLineNo">451</span><a name="line.451"></a> -<span class="sourceLineNo">452</span> /**<a name="line.452"></a> -<span class="sourceLineNo">453</span> * Inserts the current three byte sequence into the dictionary and<a name="line.453"></a> -<span class="sourceLineNo">454</span> * returns the previous head of the hash-chain.<a name="line.454"></a> -<span class="sourceLineNo">455</span> *<a name="line.455"></a> -<span class="sourceLineNo">456</span> * <p>Updates <code>insertHash</code> and <code>prev</code> as a<a name="line.456"></a> -<span class="sourceLineNo">457</span> * side effect.</p><a name="line.457"></a> -<span class="sourceLineNo">458</span> */<a name="line.458"></a> -<span class="sourceLineNo">459</span> private int insertString(int pos) {<a name="line.459"></a> -<span class="sourceLineNo">460</span> insertHash = nextHash(insertHash, window[pos - 1 + NUMBER_OF_BYTES_IN_HASH]);<a name="line.460"></a> -<span class="sourceLineNo">461</span> int hashHead = head[insertHash];<a name="line.461"></a> -<span class="sourceLineNo">462</span> prev[pos & wMask] = hashHead;<a name="line.462"></a> -<span class="sourceLineNo">463</span> head[insertHash] = pos;<a name="line.463"></a> -<span class="sourceLineNo">464</span> return hashHead;<a name="line.464"></a> -<span class="sourceLineNo">465</span> }<a name="line.465"></a> -<span class="sourceLineNo">466</span><a name="line.466"></a> -<span class="sourceLineNo">467</span> private int longestMatchForNextPosition(final int prevMatchLength) {<a name="line.467"></a> -<span class="sourceLineNo">468</span> // save a bunch of values to restore them if the next match isn't better than the current one<a name="line.468"></a> -<span class="sourceLineNo">469</span> final int prevMatchStart = matchStart;<a name="line.469"></a> -<span class="sourceLineNo">470</span> final int prevInsertHash = insertHash;<a name="line.470"></a> -<span class="sourceLineNo">471</span><a name="line.471"></a> -<span class="sourceLineNo">472</span> lookahead--;<a name="line.472"></a> -<span class="sourceLineNo">473</span> currentPosition++;<a name="line.473"></a> -<span class="sourceLineNo">474</span> int hashHead = insertString(currentPosition);<a name="line.474"></a> -<span class="sourceLineNo">475</span> final int prevHashHead = prev[currentPosition & wMask];<a name="line.475"></a> -<span class="sourceLineNo">476</span> int matchLength = longestMatch(hashHead);<a name="line.476"></a> -<span class="sourceLineNo">477</span><a name="line.477"></a> -<span class="sourceLineNo">478</span> if (matchLength <= prevMatchLength) {<a name="line.478"></a> -<span class="sourceLineNo">479</span> // use the first match, as the next one isn't any better<a name="line.479"></a> -<span class="sourceLineNo">480</span> matchLength = prevMatchLength;<a name="line.480"></a> -<span class="sourceLineNo">481</span> matchStart = prevMatchStart;<a name="line.481"></a> -<span class="sourceLineNo">482</span><a name="line.482"></a> -<span class="sourceLineNo">483</span> // restore modified values<a name="line.483"></a> -<span class="sourceLineNo">484</span> head[insertHash] = prevHashHead;<a name="line.484"></a> -<span class="sourceLineNo">485</span> insertHash = prevInsertHash;<a name="line.485"></a> -<span class="sourceLineNo">486</span> currentPosition--;<a name="line.486"></a> -<span class="sourceLineNo">487</span> lookahead++;<a name="line.487"></a> -<span class="sourceLineNo">488</span> }<a name="line.488"></a> -<span class="sourceLineNo">489</span> return matchLength;<a name="line.489"></a> -<span class="sourceLineNo">490</span> }<a name="line.490"></a> -<span class="sourceLineNo">491</span><a name="line.491"></a> -<span class="sourceLineNo">492</span> private void insertStringsInMatch(int matchLength) {<a name="line.492"></a> -<span class="sourceLineNo">493</span> // inserts strings contained in current match<a name="line.493"></a> -<span class="sourceLineNo">494</span> // insertString inserts the byte 2 bytes after position, which may not yet be available -> missedInserts<a name="line.494"></a> -<span class="sourceLineNo">495</span> final int stop = Math.min(matchLength - 1, lookahead - NUMBER_OF_BYTES_IN_HASH);<a name="line.495"></a> -<span class="sourceLineNo">496</span> // currentPosition has been inserted already<a name="line.496"></a> -<span class="sourceLineNo">497</span> for (int i = 1; i <= stop; i++) {<a name="line.497"></a> -<span class="sourceLineNo">498</span> insertString(currentPosition + i);<a name="line.498"></a> -<span class="sourceLineNo">499</span> }<a name="line.499"></a> -<span class="sourceLineNo">500</span> missedInserts = matchLength - stop - 1;<a name="line.500"></a> -<span class="sourceLineNo">501</span> }<a name="line.501"></a> -<span class="sourceLineNo">502</span><a name="line.502"></a> -<span class="sourceLineNo">503</span> private void catchUpMissedInserts() {<a name="line.503"></a> -<span class="sourceLineNo">504</span> while (missedInserts > 0) {<a name="line.504"></a> -<span class="sourceLineNo">505</span> insertString(currentPosition - missedInserts--);<a name="line.505"></a> +<span class="sourceLineNo">435</span> }<a name="line.435"></a> +<span class="sourceLineNo">436</span> if (matchLength >= minMatch) {<a name="line.436"></a> +<span class="sourceLineNo">437</span> if (blockStart != currentPosition) {<a name="line.437"></a> +<span class="sourceLineNo">438</span> // emit preceeding literal block<a name="line.438"></a> +<span class="sourceLineNo">439</span> flushLiteralBlock();<a name="line.439"></a> +<span class="sourceLineNo">440</span> blockStart = NO_MATCH;<a name="line.440"></a> +<span class="sourceLineNo">441</span> }<a name="line.441"></a> +<span class="sourceLineNo">442</span> flushBackReference(matchLength);<a name="line.442"></a> +<span class="sourceLineNo">443</span> insertStringsInMatch(matchLength);<a name="line.443"></a> +<span class="sourceLineNo">444</span> lookahead -= matchLength;<a name="line.444"></a> +<span class="sourceLineNo">445</span> currentPosition += matchLength;<a name="line.445"></a> +<span class="sourceLineNo">446</span> blockStart = currentPosition;<a name="line.446"></a> +<span class="sourceLineNo">447</span> } else {<a name="line.447"></a> +<span class="sourceLineNo">448</span> // no match, append to current or start a new literal<a name="line.448"></a> +<span class="sourceLineNo">449</span> lookahead--;<a name="line.449"></a> +<span class="sourceLineNo">450</span> currentPosition++;<a name="line.450"></a> +<span class="sourceLineNo">451</span> if (currentPosition - blockStart >= params.getMaxLiteralLength()) {<a name="line.451"></a> +<span class="sourceLineNo">452</span> flushLiteralBlock();<a name="line.452"></a> +<span class="sourceLineNo">453</span> blockStart = currentPosition;<a name="line.453"></a> +<span class="sourceLineNo">454</span> }<a name="line.454"></a> +<span class="sourceLineNo">455</span> }<a name="line.455"></a> +<span class="sourceLineNo">456</span> }<a name="line.456"></a> +<span class="sourceLineNo">457</span> }<a name="line.457"></a> +<span class="sourceLineNo">458</span><a name="line.458"></a> +<span class="sourceLineNo">459</span> /**<a name="line.459"></a> +<span class="sourceLineNo">460</span> * Inserts the current three byte sequence into the dictionary and<a name="line.460"></a> +<span class="sourceLineNo">461</span> * returns the previous head of the hash-chain.<a name="line.461"></a> +<span class="sourceLineNo">462</span> *<a name="line.462"></a> +<span class="sourceLineNo">463</span> * <p>Updates <code>insertHash</code> and <code>prev</code> as a<a name="line.463"></a> +<span class="sourceLineNo">464</span> * side effect.</p><a name="line.464"></a> +<span class="sourceLineNo">465</span> */<a name="line.465"></a> +<span class="sourceLineNo">466</span> private int insertString(int pos) {<a name="line.466"></a> +<span class="sourceLineNo">467</span> insertHash = nextHash(insertHash, window[pos - 1 + NUMBER_OF_BYTES_IN_HASH]);<a name="line.467"></a> +<span class="sourceLineNo">468</span> int hashHead = head[insertHash];<a name="line.468"></a> +<span class="sourceLineNo">469</span> prev[pos & wMask] = hashHead;<a name="line.469"></a> +<span class="sourceLineNo">470</span> head[insertHash] = pos;<a name="line.470"></a> +<span class="sourceLineNo">471</span> return hashHead;<a name="line.471"></a> +<span class="sourceLineNo">472</span> }<a name="line.472"></a> +<span class="sourceLineNo">473</span><a name="line.473"></a> +<span class="sourceLineNo">474</span> private int longestMatchForNextPosition(final int prevMatchLength) {<a name="line.474"></a> +<span class="sourceLineNo">475</span> // save a bunch of values to restore them if the next match isn't better than the current one<a name="line.475"></a> +<span class="sourceLineNo">476</span> final int prevMatchStart = matchStart;<a name="line.476"></a> +<span class="sourceLineNo">477</span> final int prevInsertHash = insertHash;<a name="line.477"></a> +<span class="sourceLineNo">478</span><a name="line.478"></a> +<span class="sourceLineNo">479</span> lookahead--;<a name="line.479"></a> +<span class="sourceLineNo">480</span> currentPosition++;<a name="line.480"></a> +<span class="sourceLineNo">481</span> int hashHead = insertString(currentPosition);<a name="line.481"></a> +<span class="sourceLineNo">482</span> final int prevHashHead = prev[currentPosition & wMask];<a name="line.482"></a> +<span class="sourceLineNo">483</span> int matchLength = longestMatch(hashHead);<a name="line.483"></a> +<span class="sourceLineNo">484</span><a name="line.484"></a> +<span class="sourceLineNo">485</span> if (matchLength <= prevMatchLength) {<a name="line.485"></a> +<span class="sourceLineNo">486</span> // use the first match, as the next one isn't any better<a name="line.486"></a> +<span class="sourceLineNo">487</span> matchLength = prevMatchLength;<a name="line.487"></a> +<span class="sourceLineNo">488</span> matchStart = prevMatchStart;<a name="line.488"></a> +<span class="sourceLineNo">489</span><a name="line.489"></a> +<span class="sourceLineNo">490</span> // restore modified values<a name="line.490"></a> +<span class="sourceLineNo">491</span> head[insertHash] = prevHashHead;<a name="line.491"></a> +<span class="sourceLineNo">492</span> insertHash = prevInsertHash;<a name="line.492"></a> +<span class="sourceLineNo">493</span> currentPosition--;<a name="line.493"></a> +<span class="sourceLineNo">494</span> lookahead++;<a name="line.494"></a> +<span class="sourceLineNo">495</span> }<a name="line.495"></a> +<span class="sourceLineNo">496</span> return matchLength;<a name="line.496"></a> +<span class="sourceLineNo">497</span> }<a name="line.497"></a> +<span class="sourceLineNo">498</span><a name="line.498"></a> +<span class="sourceLineNo">499</span> private void insertStringsInMatch(int matchLength) {<a name="line.499"></a> +<span class="sourceLineNo">500</span> // inserts strings contained in current match<a name="line.500"></a> +<span class="sourceLineNo">501</span> // insertString inserts the byte 2 bytes after position, which may not yet be available -> missedInserts<a name="line.501"></a> +<span class="sourceLineNo">502</span> final int stop = Math.min(matchLength - 1, lookahead - NUMBER_OF_BYTES_IN_HASH);<a name="line.502"></a> +<span class="sourceLineNo">503</span> // currentPosition has been inserted already<a name="line.503"></a> +<span class="sourceLineNo">504</span> for (int i = 1; i <= stop; i++) {<a name="line.504"></a> +<span class="sourceLineNo">505</span> insertString(currentPosition + i);<a name="line.505"></a> <span class="sourceLineNo">506</span> }<a name="line.506"></a> -<span class="sourceLineNo">507</span> }<a name="line.507"></a> -<span class="sourceLineNo">508</span><a name="line.508"></a> -<span class="sourceLineNo">509</span> private void flushBackReference(int matchLength) throws IOException {<a name="line.509"></a> -<span class="sourceLineNo">510</span> callback.accept(new BackReference(currentPosition - matchStart, matchLength));<a name="line.510"></a> -<span class="sourceLineNo">511</span> }<a name="line.511"></a> -<span class="sourceLineNo">512</span><a name="line.512"></a> -<span class="sourceLineNo">513</span> private void flushLiteralBlock() throws IOException {<a name="line.513"></a> -<span class="sourceLineNo">514</span> callback.accept(new LiteralBlock(window, blockStart, currentPosition - blockStart));<a name="line.514"></a> -<span class="sourceLineNo">515</span> }<a name="line.515"></a> -<span class="sourceLineNo">516</span><a name="line.516"></a> -<span class="sourceLineNo">517</span> /**<a name="line.517"></a> -<span class="sourceLineNo">518</span> * Searches the hash chain for real matches and returns the length<a name="line.518"></a> -<span class="sourceLineNo">519</span> * of the longest match (0 if none were found) that isn't too far<a name="line.519"></a> -<span class="sourceLineNo">520</span> * away (WRT maxOffset).<a name="line.520"></a> -<span class="sourceLineNo">521</span> *<a name="line.521"></a> -<span class="sourceLineNo">522</span> * <p>Sets matchStart to the index of the start position of the<a name="line.522"></a> -<span class="sourceLineNo">523</span> * longest match as a side effect.</p><a name="line.523"></a> -<span class="sourceLineNo">524</span> */<a name="line.524"></a> -<span class="sourceLineNo">525</span> private int longestMatch(int matchHead) {<a name="line.525"></a> -<span class="sourceLineNo">526</span> final int minLength = params.getMinBackReferenceLength();<a name="line.526"></a> -<span class="sourceLineNo">527</span> int longestMatchLength = minLength - 1;<a name="line.527"></a> -<span class="sourceLineNo">528</span> final int maxPossibleLength = Math.min(params.getMaxBackReferenceLength(), lookahead);<a name="line.528"></a> -<span class="sourceLineNo">529</span> final int minIndex = Math.max(0, currentPosition - params.getMaxOffset());<a name="line.529"></a> -<span class="sourceLineNo">530</span> final int niceBackReferenceLength = Math.min(maxPossibleLength, params.getNiceBackReferenceLength());<a name="line.530"></a> -<span class="sourceLineNo">531</span> final int maxCandidates = params.getMaxCandidates();<a name="line.531"></a> -<span class="sourceLineNo">532</span> for (int candidates = 0; candidates < maxCandidates && matchHead >= minIndex; candidates++) {<a name="line.532"></a> -<span class="sourceLineNo">533</span> int currentLength = 0;<a name="line.533"></a> -<span class="sourceLineNo">534</span> for (int i = 0; i < maxPossibleLength; i++) {<a name="line.534"></a> -<span class="sourceLineNo">535</span> if (window[matchHead + i] != window[currentPosition + i]) {<a name="line.535"></a> -<span class="sourceLineNo">536</span> break;<a name="line.536"></a> -<span class="sourceLineNo">537</span> }<a name="line.537"></a> -<span class="sourceLineNo">538</span> currentLength++;<a name="line.538"></a> -<span class="sourceLineNo">539</span> }<a name="line.539"></a> -<span class="sourceLineNo">540</span> if (currentLength > longestMatchLength) {<a name="line.540"></a> -<span class="sourceLineNo">541</span> longestMatchLength = currentLength;<a name="line.541"></a> -<span class="sourceLineNo">542</span> matchStart = matchHead;<a name="line.542"></a> -<span class="sourceLineNo">543</span> if (currentLength >= niceBackReferenceLength) {<a name="line.543"></a> -<span class="sourceLineNo">544</span> // no need to search any further<a name="line.544"></a> -<span class="sourceLineNo">545</span> break;<a name="line.545"></a> -<span class="sourceLineNo">546</span> }<a name="line.546"></a> -<span class="sourceLineNo">547</span> }<a name="line.547"></a> -<span class="sourceLineNo">548</span> matchHead = prev[matchHead & wMask];<a name="line.548"></a> -<span class="sourceLineNo">549</span> }<a name="line.549"></a> -<span class="sourceLineNo">550</span> return longestMatchLength; // < minLength if no matches have been found, will be ignored in compress()<a name="line.550"></a> -<span class="sourceLineNo">551</span> }<a name="line.551"></a> -<span class="sourceLineNo">552</span>}<a name="line.552"></a> +<span class="sourceLineNo">507</span> missedInserts = matchLength - stop - 1;<a name="line.507"></a> +<span class="sourceLineNo">508</span> }<a name="line.508"></a> +<span class="sourceLineNo">509</span><a name="line.509"></a> +<span class="sourceLineNo">510</span> private void catchUpMissedInserts() {<a name="line.510"></a> +<span class="sourceLineNo">511</span> while (missedInserts > 0) {<a name="line.511"></a> +<span class="sourceLineNo">512</span> insertString(currentPosition - missedInserts--);<a name="line.512"></a> +<span class="sourceLineNo">513</span> }<a name="line.513"></a> +<span class="sourceLineNo">514</span> }<a name="line.514"></a> +<span class="sourceLineNo">515</span><a name="line.515"></a> +<span class="sourceLineNo">516</span> private void flushBackReference(int matchLength) throws IOException {<a name="line.516"></a> +<span class="sourceLineNo">517</span> callback.accept(new BackReference(currentPosition - matchStart, matchLength));<a name="line.517"></a> +<span class="sourceLineNo">518</span> }<a name="line.518"></a> +<span class="sourceLineNo">519</span><a name="line.519"></a> +<span class="sourceLineNo">520</span> private void flushLiteralBlock() throws IOException {<a name="line.520"></a> +<span class="sourceLineNo">521</span> callback.accept(new LiteralBlock(window, blockStart, currentPosition - blockStart));<a name="line.521"></a> +<span class="sourceLineNo">522</span> }<a name="line.522"></a> +<span class="sourceLineNo">523</span><a name="line.523"></a> +<span class="sourceLineNo">524</span> /**<a name="line.524"></a> +<span class="sourceLineNo">525</span> * Searches the hash chain for real matches and returns the length<a name="line.525"></a> +<span class="sourceLineNo">526</span> * of the longest match (0 if none were found) that isn't too far<a name="line.526"></a> +<span class="sourceLineNo">527</span> * away (WRT maxOffset).<a name="line.527"></a> +<span class="sourceLineNo">528</span> *<a name="line.528"></a> +<span class="sourceLineNo">529</span> * <p>Sets matchStart to the index of the start position of the<a name="line.529"></a> +<span class="sourceLineNo">530</span> * longest match as a side effect.</p><a name="line.530"></a> +<span class="sourceLineNo">531</span> */<a name="line.531"></a> +<span class="sourceLineNo">532</span> private int longestMatch(int matchHead) {<a name="line.532"></a> +<span class="sourceLineNo">533</span> final int minLength = params.getMinBackReferenceLength();<a name="line.533"></a> +<span class="sourceLineNo">534</span> int longestMatchLength = minLength - 1;<a name="line.534"></a> +<span class="sourceLineNo">535</span> final int maxPossibleLength = Math.min(params.getMaxBackReferenceLength(), lookahead);<a name="line.535"></a> +<span class="sourceLineNo">536</span> final int minIndex = Math.max(0, currentPosition - params.getMaxOffset());<a name="line.536"></a>
[... 27 lines stripped ...]