gzippedinputstream.html

来自「网络爬虫开源代码」· HTML 代码 · 共 442 行 · 第 1/3 页

<a name="140" href="#140">140</a>         	is.mark(GzipHeader.MINIMAL_GZIP_HEADER_LENGTH * 4);
<a name="141" href="#141">141</a>         	<strong>return</strong> is;
<a name="142" href="#142">142</a>         }
<a name="143" href="#143">143</a>         <strong>throw</strong> <strong>new</strong> IOException(<span class="string">"Passed stream does not"</span> +
<a name="144" href="#144">144</a>             <span class="string">" implement PositionableStream"</span>);
<a name="145" href="#145">145</a>     }
<a name="146" href="#146">146</a>     
<a name="147" href="#147">147</a>     <em>/**<em>*</em></em>
<a name="148" href="#148">148</a> <em>     * Exhaust current GZIP member content.</em>
<a name="149" href="#149">149</a> <em>     * Call this method when you think you're on the end of the</em>
<a name="150" href="#150">150</a> <em>     * GZIP member.  It will clean out any dross.</em>
<a name="151" href="#151">151</a> <em>     * @param ignore Character to ignore counting characters (Usually</em>
<a name="152" href="#152">152</a> <em>     * trailing new lines).</em>
<a name="153" href="#153">153</a> <em>     * @return Count of characters skipped over.</em>
<a name="154" href="#154">154</a> <em>     * @throws IOException</em>
<a name="155" href="#155">155</a> <em>     */</em>
<a name="156" href="#156">156</a>     <strong>public</strong> <strong>long</strong> gotoEOR(<strong>int</strong> ignore) throws IOException {
<a name="157" href="#157">157</a>         <strong>long</strong> bytesSkipped = 0;
<a name="158" href="#158">158</a>         <strong>if</strong> (<strong>this</strong>.inf.getTotalIn() &lt;= 0) {
<a name="159" href="#159">159</a>             <strong>return</strong> bytesSkipped;
<a name="160" href="#160">160</a>         }
<a name="161" href="#161">161</a>         <strong>if</strong> (!<strong>this</strong>.inf.finished()) {
<a name="162" href="#162">162</a>             <strong>int</strong> read = 0;
<a name="163" href="#163">163</a>             <strong>while</strong> ((read = read()) != -1) {
<a name="164" href="#164">164</a>                 <strong>if</strong> ((byte)read == (byte)ignore) {
<a name="165" href="#165">165</a>                     <strong>continue</strong>;
<a name="166" href="#166">166</a>                 }
<a name="167" href="#167">167</a>                 bytesSkipped = gotoEOR() + 1;
<a name="168" href="#168">168</a>                 <strong>break</strong>;
<a name="169" href="#169">169</a>             }
<a name="170" href="#170">170</a>         }
<a name="171" href="#171">171</a>         <strong>return</strong> bytesSkipped;
<a name="172" href="#172">172</a>     }
<a name="173" href="#173">173</a>     
<a name="174" href="#174">174</a>     <em>/**<em>*</em></em>
<a name="175" href="#175">175</a> <em>     * Exhaust current GZIP member content.</em>
<a name="176" href="#176">176</a> <em>     * Call this method when you think you're on the end of the</em>
<a name="177" href="#177">177</a> <em>     * GZIP member.  It will clean out any dross.</em>
<a name="178" href="#178">178</a> <em>     * @return Count of characters skipped over.</em>
<a name="179" href="#179">179</a> <em>     * @throws IOException</em>
<a name="180" href="#180">180</a> <em>     */</em>
<a name="181" href="#181">181</a>     <strong>public</strong> <strong>long</strong> gotoEOR() throws IOException {
<a name="182" href="#182">182</a>         <strong>long</strong> bytesSkipped = 0;
<a name="183" href="#183">183</a>         <strong>if</strong> (<strong>this</strong>.inf.getTotalIn() &lt;= 0) {
<a name="184" href="#184">184</a>             <strong>return</strong> bytesSkipped;
<a name="185" href="#185">185</a>         }
<a name="186" href="#186">186</a>         <strong>while</strong>(!<strong>this</strong>.inf.finished()) {
<a name="187" href="#187">187</a>             bytesSkipped += skip(Long.MAX_VALUE);
<a name="188" href="#188">188</a>         }
<a name="189" href="#189">189</a>         <strong>return</strong> bytesSkipped;
<a name="190" href="#190">190</a>     }
<a name="191" href="#191">191</a>     
<a name="192" href="#192">192</a>     <em>/**<em>*</em></em>
<a name="193" href="#193">193</a> <em>     * Returns a GZIP Member Iterator.</em>
<a name="194" href="#194">194</a> <em>     * Has limitations. Can only get one Iterator per instance of this class;</em>
<a name="195" href="#195">195</a> <em>     * you must get new instance if you want to get Iterator again.</em>
<a name="196" href="#196">196</a> <em>     * @return Iterator over GZIP Members.</em>
<a name="197" href="#197">197</a> <em>     */</em>
<a name="198" href="#198">198</a>     <strong>public</strong> Iterator iterator() {
<a name="199" href="#199">199</a>         <strong>final</strong> Logger logger = Logger.getLogger(<strong>this</strong>.getClass().getName());
<a name="200" href="#200">200</a>         
<a name="201" href="#201">201</a>         <strong>try</strong> {
<a name="202" href="#202">202</a>             <em class="comment">// We know its a RepositionableStream else we'd have failed</em>
<a name="203" href="#203">203</a>         	<em class="comment">// construction.  On iterator construction, set file back to</em>
<a name="204" href="#204">204</a>         	<em class="comment">// initial position so we're ready to read GZIP Members</em>
<a name="205" href="#205">205</a>         	<em class="comment">// (May not always work dependent on how the</em>
<a name="206" href="#206">206</a>         	<em class="comment">// RepositionableStream was implemented).</em>
<a name="207" href="#207">207</a>             ((RepositionableStream)<strong>this</strong>.in).position(<strong>this</strong>.initialOffset);
<a name="208" href="#208">208</a>         } <strong>catch</strong> (IOException e) {
<a name="209" href="#209">209</a>             <strong>throw</strong> <strong>new</strong> RuntimeException(e);
<a name="210" href="#210">210</a>         }
<a name="211" href="#211">211</a>         <strong>return</strong> <strong>new</strong> Iterator() {
<a name="212" href="#212">212</a>             <strong>private</strong> <a href="../../../org/archive/io/GzippedInputStream.html">GzippedInputStream</a> compressedStream =
<a name="213" href="#213">213</a>                 GzippedInputStream.<strong>this</strong>;
<a name="214" href="#214">214</a>             
<a name="215" href="#215">215</a>             <strong>public</strong> <strong>boolean</strong> hasNext() {
<a name="216" href="#216">216</a>                 <strong>try</strong> {
<a name="217" href="#217">217</a>                     gotoEOR();
<a name="218" href="#218">218</a>                 } <strong>catch</strong> (IOException e) {
<a name="219" href="#219">219</a>                     <strong>if</strong> ((e instanceof ZipException) ||
<a name="220" href="#220">220</a>                         (e.getMessage() != <strong>null</strong> &amp;&amp;
<a name="221" href="#221">221</a>                          e.getMessage().startsWith(<span class="string">"Corrupt GZIP trailer"</span>))) {
<a name="222" href="#222">222</a>                         <em class="comment">// Try skipping end of bad record; try moving to next.</em>
<a name="223" href="#223">223</a>                         logger.info(<span class="string">"Skipping exception "</span> + e.getMessage());
<a name="224" href="#224">224</a>                     } <strong>else</strong> {
<a name="225" href="#225">225</a>                         <strong>throw</strong> <strong>new</strong> RuntimeException(e);
<a name="226" href="#226">226</a>                     }
<a name="227" href="#227">227</a>                 }
<a name="228" href="#228">228</a>                 <strong>return</strong> moveToNextGzipMember();
<a name="229" href="#229">229</a>             }
<a name="230" href="#230">230</a>             
<a name="231" href="#231">231</a>             <em>/**<em>*</em></em>
<a name="232" href="#232">232</a> <em>             * @return An InputStream onto a GZIP Member.</em>
<a name="233" href="#233">233</a> <em>             */</em>
<a name="234" href="#234">234</a>             <strong>public</strong> Object next() {
<a name="235" href="#235">235</a>                 <strong>try</strong> {
<a name="236" href="#236">236</a>                     gzipMemberSeek();
<a name="237" href="#237">237</a>                 } <strong>catch</strong> (IOException e) {
<a name="238" href="#238">238</a>                     <strong>throw</strong> <strong>new</strong> RuntimeException(<span class="string">"Failed move to EOR or "</span> +
<a name="239" href="#239">239</a>                         <span class="string">"failed header read: "</span> + e.getMessage());
<a name="240" href="#240">240</a>                 }
<a name="241" href="#241">241</a>                 <strong>return</strong> <strong>this</strong>.compressedStream;
<a name="242" href="#242">242</a>             }
<a name="243" href="#243">243</a>             
<a name="244" href="#244">244</a>             <strong>public</strong> <strong>void</strong> remove() {
<a name="245" href="#245">245</a>                 <strong>throw</strong> <strong>new</strong> UnsupportedOperationException();
<a name="246" href="#246">246</a>             }
<a name="247" href="#247">247</a>         };   
<a name="248" href="#248">248</a>     }
<a name="249" href="#249">249</a>     
<a name="250" href="#250">250</a>     <em>/**<em>*</em></em>
<a name="251" href="#251">251</a> <em>     * @return True if we found another record in the stream.</em>
<a name="252" href="#252">252</a> <em>     */</em>
<a name="253" href="#253">253</a>     <strong>protected</strong> <strong>boolean</strong> moveToNextGzipMember() {
<a name="254" href="#254">254</a>         <strong>boolean</strong> result = false;
<a name="255" href="#255">255</a>         <em class="comment">// Move to the next gzip member, if there is one, positioning</em>
<a name="256" href="#256">256</a>         <em class="comment">// ourselves by backing up the stream so we reread any inflater</em>
<a name="257" href="#257">257</a>         <em class="comment">// remaining bytes. Then add 8 bytes to get us past the GZIP</em>
<a name="258" href="#258">258</a>         <em class="comment">// CRC trailer block that ends all gzip members.</em>
<a name="259" href="#259">259</a>         <strong>try</strong> {
<a name="260" href="#260">260</a>             RepositionableStream ps = (RepositionableStream)getInputStream();
<a name="261" href="#261">261</a>             <em class="comment">// 8 is sizeof gzip CRC block thats on tail of gzipped</em>
<a name="262" href="#262">262</a>             <em class="comment">// record. If remaining is &lt; 8 then experience indicates</em>
<a name="263" href="#263">263</a>             <em class="comment">// we're seeking past the gzip header -- don't backup the</em>
<a name="264" href="#264">264</a>             <em class="comment">// stream.</em>
<a name="265" href="#265">265</a>             <strong>if</strong> (getInflater().getRemaining() > GZIP_TRAILER_LENGTH) {
<a name="266" href="#266">266</a>                 ps.position(position() - getInflater().getRemaining() +
<a name="267" href="#267">267</a>                     GZIP_TRAILER_LENGTH);
<a name="268" href="#268">268</a>             }
<a name="269" href="#269">269</a>             <strong>for</strong> (<strong>int</strong> read = -1, headerRead = 0; <strong>true</strong>; headerRead = 0) {
<a name="270" href="#270">270</a>                 <em class="comment">// Give a hint to underlying stream that we're going to want to</em>
<a name="271" href="#271">271</a>                 <em class="comment">// do some backing up.</em>
<a name="272" href="#272">272</a>                 getInputStream().mark(3);
<a name="273" href="#273">273</a>                 <strong>if</strong> ((read = getInputStream().read()) == -1) {
<a name="274" href="#274">274</a>                     <strong>break</strong>;
<a name="275" href="#275">275</a>                 }
<a name="276" href="#276">276</a>                 <strong>if</strong>(compareBytes(read, GZIPInputStream.GZIP_MAGIC)) {
<a name="277" href="#277">277</a>                     headerRead++;
<a name="278" href="#278">278</a>                     <strong>if</strong> ((read = getInputStream().read()) == -1) {
<a name="279" href="#279">279</a>                     	<strong>break</strong>;
<a name="280" href="#280">280</a>                     }
<a name="281" href="#281">281</a>                     <strong>if</strong>(compareBytes(read, GZIPInputStream.GZIP_MAGIC >> 8)) {
<a name="282" href="#282">282</a>                         headerRead++;
<a name="283" href="#283">283</a>                         <strong>if</strong> ((read = getInputStream().read()) == -1) {
<a name="284" href="#284">284</a>                         	<strong>break</strong>;
<a name="285" href="#285">285</a>                         }
<a name="286" href="#286">286</a>                         <strong>if</strong> (compareBytes(read, Deflater.DEFLATED)) {
<a name="287" href="#287">287</a>                             headerRead++;