replaycharsequencefactory.html

用JAVA编写的,在做实验的时候留下来的,本来想删的,但是传上来,大家分享吧

<a name="140" href="#140">140</a> <em>     *</em>
<a name="141" href="#141">141</a> <em>     * @param encoding Encoding to use reading the passed prefix buffer and</em>
<a name="142" href="#142">142</a> <em>     * backing file.  For now, should be java canonical name for the encoding.</em>
<a name="143" href="#143">143</a> <em>     * (If null is passed, we will default to ByteReplayCharSequence).</em>
<a name="144" href="#144">144</a> <em>     *</em>
<a name="145" href="#145">145</a> <em>     * @return True if multibyte encoding.</em>
<a name="146" href="#146">146</a> <em>     */</em>
<a name="147" href="#147">147</a>     <strong>private</strong> <strong>boolean</strong> isMultibyteEncoding(String encoding) {
<a name="148" href="#148">148</a>         <strong>boolean</strong> isMultibyte = false;
<a name="149" href="#149">149</a>         <strong>final</strong> Charset cs;
<a name="150" href="#150">150</a>         <strong>try</strong> {
<a name="151" href="#151">151</a>             <strong>if</strong> (encoding != <strong>null</strong> &amp;&amp; encoding.length() > 0) {
<a name="152" href="#152">152</a>                 cs = Charset.forName(encoding);
<a name="153" href="#153">153</a>                 <strong>if</strong>(cs.canEncode()) {
<a name="154" href="#154">154</a>                     isMultibyte = cs.newEncoder().maxBytesPerChar() > 1;
<a name="155" href="#155">155</a>                 } <strong>else</strong> {
<a name="156" href="#156">156</a>                     isMultibyte = false;
<a name="157" href="#157">157</a>                     logger.info(<span class="string">"Encoding not fully supported: "</span> + encoding
<a name="158" href="#158">158</a>                             + <span class="string">".  Defaulting to single byte."</span>);
<a name="159" href="#159">159</a>                 }
<a name="160" href="#160">160</a>             }
<a name="161" href="#161">161</a>         } <strong>catch</strong> (IllegalCharsetNameException e) {
<a name="162" href="#162">162</a>             <em class="comment">// Unsupported encoding.  Default to singlebyte.</em>
<a name="163" href="#163">163</a>             isMultibyte = false;
<a name="164" href="#164">164</a> 
<a name="165" href="#165">165</a>             logger.info(<span class="string">"Illegal encoding name: "</span> + encoding
<a name="166" href="#166">166</a>                 + <span class="string">".  Defaulting to single byte."</span>);
<a name="167" href="#167">167</a>         } <strong>catch</strong> (UnsupportedCharsetException e) {
<a name="168" href="#168">168</a>             <em class="comment">// Unsupported encoding.  Default to singlebyte.</em>
<a name="169" href="#169">169</a>             isMultibyte = false;
<a name="170" href="#170">170</a> 
<a name="171" href="#171">171</a>             logger.info(<span class="string">"Unsupported encoding "</span> + encoding
<a name="172" href="#172">172</a>                 + <span class="string">".  Defaulting to single byte."</span>);
<a name="173" href="#173">173</a>         }
<a name="174" href="#174">174</a>         logger.fine(<span class="string">"Encoding "</span> + encoding + <span class="string">" is multibyte: "</span>
<a name="175" href="#175">175</a>             + ((isMultibyte) ? Boolean.TRUE : Boolean.FALSE));
<a name="176" href="#176">176</a>         
<a name="177" href="#177">177</a>         <strong>return</strong> isMultibyte;
<a name="178" href="#178">178</a>     }
<a name="179" href="#179">179</a> 
<a name="180" href="#180">180</a>     <em>/**<em>*</em></em>
<a name="181" href="#181">181</a> <em>     * Test passed arguments.</em>
<a name="182" href="#182">182</a> <em>     *</em>
<a name="183" href="#183">183</a> <em>     * @param buffer In-memory buffer of recordings prefix.  We read from here</em>
<a name="184" href="#184">184</a> <em>     * first and will only go to the backing file if &lt;code>size&lt;/code> requested</em>
<a name="185" href="#185">185</a> <em>     * is greater than &lt;code>buffer.length&lt;/code>.</em>
<a name="186" href="#186">186</a> <em>     * @param size Total size of stream to replay in bytes.  Used to find EOS.</em>
<a name="187" href="#187">187</a> <em>     * This is total length of content including HTTP headers if present.</em>
<a name="188" href="#188">188</a> <em>     * @param responseBodyStart Where the response body starts in bytes. Used to</em>
<a name="189" href="#189">189</a> <em>     * skip over the HTTP headers if present.</em>
<a name="190" href="#190">190</a> <em>     *</em>
<a name="191" href="#191">191</a> <em>     * @throws IllegalArgumentException Thrown if passed an illegal argument.</em>
<a name="192" href="#192">192</a> <em>     */</em>
<a name="193" href="#193">193</a>     <strong>protected</strong> <strong>void</strong> checkParameters(byte[] buffer, <strong>long</strong> size,
<a name="194" href="#194">194</a>             <strong>long</strong> responseBodyStart)
<a name="195" href="#195">195</a>         throws IllegalArgumentException {
<a name="196" href="#196">196</a> 
<a name="197" href="#197">197</a>         <strong>if</strong> (responseBodyStart > size) {
<a name="198" href="#198">198</a>             <strong>throw</strong> <strong>new</strong> IllegalArgumentException(<span class="string">"Illegal response body offset"</span> +
<a name="199" href="#199">199</a>                 <span class="string">" of "</span> + responseBodyStart + <span class="string">" whereas size is only "</span> + size);
<a name="200" href="#200">200</a>         }
<a name="201" href="#201">201</a> 
<a name="202" href="#202">202</a>         <strong>if</strong> (responseBodyStart > Integer.MAX_VALUE) {
<a name="203" href="#203">203</a>             <em class="comment">// A value of this size will mess up math below.</em>
<a name="204" href="#204">204</a>             <strong>throw</strong> <strong>new</strong> IllegalArgumentException(<span class="string">"Response body start "</span> +
<a name="205" href="#205">205</a>                 <span class="string">" of "</span> + responseBodyStart + <span class="string">" > Integer.MAX_VALUE."</span>);
<a name="206" href="#206">206</a>         }
<a name="207" href="#207">207</a> 
<a name="208" href="#208">208</a>         <strong>if</strong> (responseBodyStart > buffer.length) {
<a name="209" href="#209">209</a>             logger.log(Level.WARNING,
<a name="210" href="#210">210</a>                 <span class="string">"Unexpected response body offset "</span> + responseBodyStart + <span class="string">",\n"</span> +
<a name="211" href="#211">211</a>                 <span class="string">"beyond the first buffer of length "</span>+buffer.length+<span class="string">".\n"</span> +
<a name="212" href="#212">212</a>                 <span class="string">"Thread: "</span>+ Thread.currentThread().getName() + <span class="string">"\n"</span>);
<a name="213" href="#213">213</a>         }
<a name="214" href="#214">214</a> 
<a name="215" href="#215">215</a>         <strong>if</strong> ((size - responseBodyStart) > Integer.MAX_VALUE) {
<a name="216" href="#216">216</a>             <strong>throw</strong> <strong>new</strong> IllegalArgumentException(<span class="string">"Length is bigger than we  can"</span> +
<a name="217" href="#217">217</a>                <span class="string">" handle: "</span> + (size - responseBodyStart));
<a name="218" href="#218">218</a>         }
<a name="219" href="#219">219</a>     }
<a name="220" href="#220">220</a> 
<a name="221" href="#221">221</a>     <em>/**<em>*</em></em>
<a name="222" href="#222">222</a> <em>     * Provides a (Replay)CharSequence view on recorded stream bytes (a prefix</em>
<a name="223" href="#223">223</a> <em>     * buffer and overflow backing file).</em>
<a name="224" href="#224">224</a> <em>     *</em>
<a name="225" href="#225">225</a> <em>     * Treats the byte stream as 8-bit.</em>
<a name="226" href="#226">226</a> <em>     *</em>
<a name="227" href="#227">227</a> <em>     * &lt;p>Uses a wraparound rolling buffer of the last windowSize bytes read</em>
<a name="228" href="#228">228</a> <em>     * from disk in memory; as long as the 'random access' of a CharSequence</em>
<a name="229" href="#229">229</a> <em>     * user stays within this window, access should remain fairly efficient.</em>
<a name="230" href="#230">230</a> <em>     * (So design any regexps pointed at these CharSequences to work within</em>
<a name="231" href="#231">231</a> <em>     * that range!)</em>
<a name="232" href="#232">232</a> <em>     *</em>
<a name="233" href="#233">233</a> <em>     * &lt;p>When rereading of a location is necessary, the whole window is</em>
<a name="234" href="#234">234</a> <em>     * recentered around the location requested. (TODO: More research</em>
<a name="235" href="#235">235</a> <em>     * into whether this is the best strategy.)</em>
<a name="236" href="#236">236</a> <em>     *</em>
<a name="237" href="#237">237</a> <em>     * &lt;p>An implementation of a ReplayCharSequence done with ByteBuffers -- one</em>
<a name="238" href="#238">238</a> <em>     * to wrap the passed prefix buffer and the second, a memory-mapped</em>
<a name="239" href="#239">239</a> <em>     * ByteBuffer view into the backing file -- was consistently slower: ~10%.</em>
<a name="240" href="#240">240</a> <em>     * My tests did the following. Made a buffer filled w/ regular content.</em>
<a name="241" href="#241">241</a> <em>     * This buffer was used as the prefix buffer.  The buffer content was</em>
<a name="242" href="#242">242</a> <em>     * written MULTIPLER times to a backing file.  I then did accesses w/ the</em>
<a name="243" href="#243">243</a> <em>     * following pattern: Skip forward 32 bytes, then back 16 bytes, and then</em>
<a name="244" href="#244">244</a> <em>     * read forward from byte 16-32.  Repeat.  Though I varied the size of the</em>
<a name="245" href="#245">245</a> <em>     * buffer to the size of the backing file,from 3-10, the difference of 10%</em>
<a name="246" href="#246">246</a> <em>     * or so seemed to persist.  Same if I tried to favor get() over get(index).</em>
<a name="247" href="#247">247</a> <em>     * I used a profiler, JMP, to study times taken (St.Ack did above comment).</em>
<a name="248" href="#248">248</a> <em>     *</em>
<a name="249" href="#249">249</a> <em>     * &lt;p>TODO determine in memory mapped files is better way to do this;</em>
<a name="250" href="#250">250</a> <em>     * probably not -- they don't offer the level of control over</em>
<a name="251" href="#251">251</a> <em>     * total memory used that this approach does.</em>
<a name="252" href="#252">252</a> <em>     *</em>
<a name="253" href="#253">253</a> <em>     * @author Gordon Mohr</em>
<a name="254" href="#254">254</a> <em>     * @version $Revision: 1.39 $, $Date: 2006/06/01 05:58:37 $</em>
<a name="255" href="#255">255</a> <em>     */</em>
<a name="256" href="#256">256</a>     <strong>private</strong> <strong>class</strong> ByteReplayCharSequence implements <a href="../../../org/archive/io/ReplayCharSequence.html">ReplayCharSequence</a> {
<a name="257" href="#257">257</a> 
<a name="258" href="#258">258</a>         <em>/**<em>*</em></em>
<a name="259" href="#259">259</a> <em>         * Buffer that holds the first bit of content.</em>
<a name="260" href="#260">260</a> <em>         *</em>
<a name="261" href="#261">261</a> <em>         * Once this is exhausted we go to the backing file.</em>
<a name="262" href="#262">262</a> <em>         */</em>
<a name="263" href="#263">263</a>         <strong>private</strong> byte[] prefixBuffer;
<a name="264" href="#264">264</a> 
<a name="265" href="#265">265</a>         <em>/**<em>*</em></em>
<a name="266" href="#266">266</a> <em>         * Total length of character stream to replay minus the HTTP headers</em>
<a name="267" href="#267">267</a> <em>         * if present.</em>
<a name="268" href="#268">268</a> <em>         *</em>
<a name="269" href="#269">269</a> <em>         * Used to find EOS.</em>
<a name="270" href="#270">270</a> <em>         */</em>
<a name="271" href="#271">271</a>         <strong>protected</strong> <strong>int</strong> length;
<a name="272" href="#272">272</a> 
<a name="273" href="#273">273</a>         <em>/**<em>*</em></em>
<a name="274" href="#274">274</a> <em>         * Absolute length of the stream.</em>
<a name="275" href="#275">275</a> <em>         *</em>
<a name="276" href="#276">276</a> <em>         * Includes HTTP headers.  Needed doing calc. in the below figuring</em>
<a name="277" href="#277">277</a> <em>         * how much to load into buffer.</em>
<a name="278" href="#278">278</a> <em>         */</em>
<a name="279" href="#279">279</a>         <strong>private</strong> <strong>int</strong> absoluteLength = -1;
<a name="280" href="#280">280</a> 
<a name="281" href="#281">281</a>         <em>/**<em>*</em></em>
<a name="282" href="#282">282</a> <em>         * Buffer window on to backing file.</em>
<a name="283" href="#283">283</a> <em>         */</em>
<a name="284" href="#284">284</a>         <strong>private</strong> byte[] wraparoundBuffer;
<a name="285" href="#285">285</a> 
<a name="286" href="#286">286</a>         <em>/**<em>*</em></em>
<a name="287" href="#287">287</a> <em>         * Absolute index into underlying bytestream where wrap starts.</em>