warcreaderfactory.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="148" href="#148">148</a>         } <strong>finally</strong> {
<a name="149" href="#149">149</a>             is.close();
<a name="150" href="#150">150</a>         }
<a name="151" href="#151">151</a>         <strong>return</strong> compressed;
<a name="152" href="#152">152</a>     }
<a name="153" href="#153">153</a> 
<a name="154" href="#154">154</a>     <em>/**<em>*</em></em>
<a name="155" href="#155">155</a> <em>     * Uncompressed WARC file reader.</em>
<a name="156" href="#156">156</a> <em>     * @author stack</em>
<a name="157" href="#157">157</a> <em>     */</em>
<a name="158" href="#158">158</a>     <strong>private</strong> <strong>class</strong> UncompressedWARCReader <strong>extends</strong> <a href="../../../../org/archive/io/warc/WARCReader.html">WARCReader</a> {
<a name="159" href="#159">159</a>         <em>/**<em>*</em></em>
<a name="160" href="#160">160</a> <em>         * Constructor.</em>
<a name="161" href="#161">161</a> <em>         * @param f Uncompressed arcfile to read.</em>
<a name="162" href="#162">162</a> <em>         * @throws IOException</em>
<a name="163" href="#163">163</a> <em>         */</em>
<a name="164" href="#164">164</a>         <strong>public</strong> UncompressedWARCReader(<strong>final</strong> File f)
<a name="165" href="#165">165</a>         throws IOException {
<a name="166" href="#166">166</a>             <strong>this</strong>(f, 0);
<a name="167" href="#167">167</a>         }
<a name="168" href="#168">168</a> 
<a name="169" href="#169">169</a>         <em>/**<em>*</em></em>
<a name="170" href="#170">170</a> <em>         * Constructor.</em>
<a name="171" href="#171">171</a> <em>         * </em>
<a name="172" href="#172">172</a> <em>         * @param f Uncompressed file to read.</em>
<a name="173" href="#173">173</a> <em>         * @param offset Offset at which to position Reader.</em>
<a name="174" href="#174">174</a> <em>         * @throws IOException</em>
<a name="175" href="#175">175</a> <em>         */</em>
<a name="176" href="#176">176</a>         <strong>public</strong> UncompressedWARCReader(<strong>final</strong> File f, <strong>final</strong> <strong>long</strong> offset)
<a name="177" href="#177">177</a>         throws IOException {
<a name="178" href="#178">178</a>             <em class="comment">// File has been tested for existence by time it has come to here.</em>
<a name="179" href="#179">179</a>             setIn(getInputStream(f, offset));
<a name="180" href="#180">180</a>             initialize(f.getAbsolutePath());
<a name="181" href="#181">181</a>         }
<a name="182" href="#182">182</a>         
<a name="183" href="#183">183</a>         <em>/**<em>*</em></em>
<a name="184" href="#184">184</a> <em>         * Constructor.</em>
<a name="185" href="#185">185</a> <em>         * </em>
<a name="186" href="#186">186</a> <em>         * @param f Uncompressed file to read.</em>
<a name="187" href="#187">187</a> <em>         * @param is InputStream.</em>
<a name="188" href="#188">188</a> <em>         */</em>
<a name="189" href="#189">189</a>         <strong>public</strong> UncompressedWARCReader(<strong>final</strong> String f, <strong>final</strong> InputStream is) {
<a name="190" href="#190">190</a>             <em class="comment">// Arc file has been tested for existence by time it has come</em>
<a name="191" href="#191">191</a>             <em class="comment">// to here.</em>
<a name="192" href="#192">192</a>             setIn(is);
<a name="193" href="#193">193</a>             initialize(f);
<a name="194" href="#194">194</a>         }
<a name="195" href="#195">195</a>     }
<a name="196" href="#196">196</a>     
<a name="197" href="#197">197</a>     <em>/**<em>*</em></em>
<a name="198" href="#198">198</a> <em>     * Compressed WARC file reader.</em>
<a name="199" href="#199">199</a> <em>     * </em>
<a name="200" href="#200">200</a> <em>     * @author stack</em>
<a name="201" href="#201">201</a> <em>     */</em>
<a name="202" href="#202">202</a>     <strong>private</strong> <strong>class</strong> CompressedWARCReader <strong>extends</strong> <a href="../../../../org/archive/io/warc/WARCReader.html">WARCReader</a> {
<a name="203" href="#203">203</a>         <em>/**<em>*</em></em>
<a name="204" href="#204">204</a> <em>         * Constructor.</em>
<a name="205" href="#205">205</a> <em>         * </em>
<a name="206" href="#206">206</a> <em>         * @param f Compressed file to read.</em>
<a name="207" href="#207">207</a> <em>         * @throws IOException</em>
<a name="208" href="#208">208</a> <em>         */</em>
<a name="209" href="#209">209</a>         <strong>public</strong> CompressedWARCReader(<strong>final</strong> File f) throws IOException {
<a name="210" href="#210">210</a>             <strong>this</strong>(f, 0);
<a name="211" href="#211">211</a>         }
<a name="212" href="#212">212</a> 
<a name="213" href="#213">213</a>         <em>/**<em>*</em></em>
<a name="214" href="#214">214</a> <em>         * Constructor.</em>
<a name="215" href="#215">215</a> <em>         * </em>
<a name="216" href="#216">216</a> <em>         * @param f Compressed arcfile to read.</em>
<a name="217" href="#217">217</a> <em>         * @param offset Position at where to start reading file.</em>
<a name="218" href="#218">218</a> <em>         * @throws IOException</em>
<a name="219" href="#219">219</a> <em>         */</em>
<a name="220" href="#220">220</a>         <strong>public</strong> CompressedWARCReader(<strong>final</strong> File f, <strong>final</strong> <strong>long</strong> offset)
<a name="221" href="#221">221</a>                 throws IOException {
<a name="222" href="#222">222</a>             <em class="comment">// File has been tested for existence by time it has come to here.</em>
<a name="223" href="#223">223</a>             setIn(<strong>new</strong> <a href="../../../../org/archive/io/GzippedInputStream.html">GzippedInputStream</a>(getInputStream(f, offset)));
<a name="224" href="#224">224</a>             setCompressed((offset == 0));
<a name="225" href="#225">225</a>             initialize(f.getAbsolutePath());
<a name="226" href="#226">226</a>         }
<a name="227" href="#227">227</a>         
<a name="228" href="#228">228</a>         <em>/**<em>*</em></em>
<a name="229" href="#229">229</a> <em>         * Constructor.</em>
<a name="230" href="#230">230</a> <em>         * </em>
<a name="231" href="#231">231</a> <em>         * @param f Compressed arcfile.</em>
<a name="232" href="#232">232</a> <em>         * @param is InputStream to use.</em>
<a name="233" href="#233">233</a> <em>         * @param atFirstRecord</em>
<a name="234" href="#234">234</a> <em>         * @throws IOException</em>
<a name="235" href="#235">235</a> <em>         */</em>
<a name="236" href="#236">236</a>         <strong>public</strong> CompressedWARCReader(<strong>final</strong> String f, <strong>final</strong> InputStream is,
<a name="237" href="#237">237</a>             <strong>final</strong> <strong>boolean</strong> atFirstRecord)
<a name="238" href="#238">238</a>         throws IOException {
<a name="239" href="#239">239</a>             <em class="comment">// Arc file has been tested for existence by time it has come</em>
<a name="240" href="#240">240</a>             <em class="comment">// to here.</em>
<a name="241" href="#241">241</a>             setIn(<strong>new</strong> <a href="../../../../org/archive/io/GzippedInputStream.html">GzippedInputStream</a>(is));
<a name="242" href="#242">242</a>             setCompressed(<strong>true</strong>);
<a name="243" href="#243">243</a>             initialize(f);
<a name="244" href="#244">244</a>             <em class="comment">// TODO: Ignore atFirstRecord. Probably doesn't apply in WARC world.</em>
<a name="245" href="#245">245</a>         }
<a name="246" href="#246">246</a>         
<a name="247" href="#247">247</a>         <em>/**<em>*</em></em>
<a name="248" href="#248">248</a> <em>         * Get record at passed &lt;code>offset&lt;/code>.</em>
<a name="249" href="#249">249</a> <em>         * </em>
<a name="250" href="#250">250</a> <em>         * @param offset Byte index into file at which a record starts.</em>
<a name="251" href="#251">251</a> <em>         * @return A WARCRecord reference.</em>
<a name="252" href="#252">252</a> <em>         * @throws IOException</em>
<a name="253" href="#253">253</a> <em>         */</em>
<a name="254" href="#254">254</a>         <strong>public</strong> <a href="../../../../org/archive/io/warc/WARCRecord.html">WARCRecord</a> get(<strong>long</strong> offset) throws IOException {
<a name="255" href="#255">255</a>             cleanupCurrentRecord();
<a name="256" href="#256">256</a>             ((<a href="../../../../org/archive/io/GzippedInputStream.html">GzippedInputStream</a>)getIn()).gzipMemberSeek(offset);
<a name="257" href="#257">257</a>             <strong>return</strong> (WARCRecord) createArchiveRecord(getIn(), offset);
<a name="258" href="#258">258</a>         }
<a name="259" href="#259">259</a>         
<a name="260" href="#260">260</a>         <strong>public</strong> Iterator&lt;ArchiveRecord> iterator() {
<a name="261" href="#261">261</a>             <em>/**<em>*</em></em>
<a name="262" href="#262">262</a> <em>             * Override ArchiveRecordIterator so can base returned iterator on</em>
<a name="263" href="#263">263</a> <em>             * GzippedInputStream iterator.</em>
<a name="264" href="#264">264</a> <em>             */</em>
<a name="265" href="#265">265</a>             <strong>return</strong> <strong>new</strong> ArchiveRecordIterator() {
<a name="266" href="#266">266</a>                 <strong>private</strong> <a href="../../../../org/archive/io/GzippedInputStream.html">GzippedInputStream</a> gis =
<a name="267" href="#267">267</a>                     (<a href="../../../../org/archive/io/GzippedInputStream.html">GzippedInputStream</a>)getInputStream();
<a name="268" href="#268">268</a> 
<a name="269" href="#269">269</a>                 <strong>private</strong> Iterator gzipIterator = <strong>this</strong>.gis.iterator();
<a name="270" href="#270">270</a> 
<a name="271" href="#271">271</a>                 <strong>protected</strong> <strong>boolean</strong> innerHasNext() {
<a name="272" href="#272">272</a>                     <strong>return</strong> <strong>this</strong>.gzipIterator.hasNext();
<a name="273" href="#273">273</a>                 }
<a name="274" href="#274">274</a> 
<a name="275" href="#275">275</a>                 <strong>protected</strong> <a href="../../../../org/archive/io/ArchiveRecord.html">ArchiveRecord</a> innerNext() throws IOException {
<a name="276" href="#276">276</a>                     <em class="comment">// Get the positoin before gzipIterator.next moves</em>
<a name="277" href="#277">277</a>                     <em class="comment">// it on past the gzip header.</em>
<a name="278" href="#278">278</a>                     <strong>long</strong> p = <strong>this</strong>.gis.position();
<a name="279" href="#279">279</a>                     InputStream is = (InputStream) <strong>this</strong>.gzipIterator.next();
<a name="280" href="#280">280</a>                     <strong>return</strong> createArchiveRecord(is, p);
<a name="281" href="#281">281</a>                 }
<a name="282" href="#282">282</a>             };
<a name="283" href="#283">283</a>         }
<a name="284" href="#284">284</a>         
<a name="285" href="#285">285</a>         <strong>protected</strong> <strong>void</strong> gotoEOR(<a href="../../../../org/archive/io/ArchiveRecord.html">ArchiveRecord</a> rec) throws IOException {
<a name="286" href="#286">286</a>         	<em class="comment">// TODO</em>
<a name="287" href="#287">287</a>         }
<a name="288" href="#288">288</a>     }
<a name="289" href="#289">289</a>     
<a name="290" href="#290">290</a>     <strong>public</strong> <strong>static</strong> <strong>boolean</strong> isWARCSuffix(<strong>final</strong> String f) {
<a name="291" href="#291">291</a>     	<strong>return</strong> (f == <strong>null</strong>)?
<a name="292" href="#292">292</a>     		false:
<a name="293" href="#293">293</a>     		(f.toLowerCase().endsWith(DOT_COMPRESSED_WARC_FILE_EXTENSION))?
<a name="294" href="#294">294</a>     		    <strong>true</strong>:
<a name="295" href="#295">295</a>     			(f.toLowerCase().endsWith(DOT_WARC_FILE_EXTENSION))?
<a name="296" href="#296">296</a>     			<strong>true</strong>: false;
<a name="297" href="#297">297</a>     }
<a name="298" href="#298">298</a> }
</pre>
<hr/><div id="footer">This page was automatically generated by <a href="http://maven.apache.org/">Maven</a></div></body>
</html>