archivereaderfactory.html

来自「网络爬虫开源代码」· HTML 代码 · 共 331 行 · 第 1/2 页

<a name="158" href="#158">158</a>         }
<a name="159" href="#159">159</a>         <em class="comment">// RepositionableInputStream calls mark on each read so can back up at</em>
<a name="160" href="#160">160</a>         <em class="comment">// least the read amount.  Needed for gzip inflater overinflations</em>
<a name="161" href="#161">161</a>         <em class="comment">// reading into the next gzip member.</em>
<a name="162" href="#162">162</a>         <strong>return</strong> <strong>new</strong> <a href="../../../org/archive/io/RepositionableInputStream.html">RepositionableInputStream</a>(is, 16 * 1024);
<a name="163" href="#163">163</a>     }
<a name="164" href="#164">164</a>     
<a name="165" href="#165">165</a>     <strong>protected</strong> <a href="../../../org/archive/io/ArchiveReader.html">ArchiveReader</a> getArchiveReader(<strong>final</strong> String id, 
<a name="166" href="#166">166</a>     		<strong>final</strong> InputStream is, <strong>final</strong> <strong>boolean</strong> atFirstRecord)
<a name="167" href="#167">167</a>     throws IOException {
<a name="168" href="#168">168</a>     	<strong>final</strong> InputStream stream = asRepositionable(is);
<a name="169" href="#169">169</a>         <strong>if</strong> (ARCReaderFactory.isARCSuffix(id)) {
<a name="170" href="#170">170</a>             <strong>return</strong> ARCReaderFactory.get(id, stream, atFirstRecord);
<a name="171" href="#171">171</a>         } <strong>else</strong> <strong>if</strong> (WARCReaderFactory.isWARCSuffix(id)) {
<a name="172" href="#172">172</a>             <strong>return</strong> WARCReaderFactory.get(id, stream, atFirstRecord);
<a name="173" href="#173">173</a>         }
<a name="174" href="#174">174</a>         <strong>throw</strong> <strong>new</strong> IOException(<span class="string">"Unknown extension (Not ARC nor WARC): "</span> + id);
<a name="175" href="#175">175</a>     }
<a name="176" href="#176">176</a>     
<a name="177" href="#177">177</a>     <em>/**<em>*</em></em>
<a name="178" href="#178">178</a> <em>     * Get an Archive Reader aligned at &lt;code>offset&lt;/code>.</em>
<a name="179" href="#179">179</a> <em>     * This version of get will not bring the file local but will try to</em>
<a name="180" href="#180">180</a> <em>     * stream across the net making an HTTP 1.1 Range request on remote</em>
<a name="181" href="#181">181</a> <em>     * http server (RFC1435 Section 14.35).</em>
<a name="182" href="#182">182</a> <em>     * @param u HTTP URL for an Archive file.</em>
<a name="183" href="#183">183</a> <em>     * @param offset Offset into file at which to start fetching.</em>
<a name="184" href="#184">184</a> <em>     * @return An ArchiveReader aligned at offset.</em>
<a name="185" href="#185">185</a> <em>     * @throws IOException</em>
<a name="186" href="#186">186</a> <em>     */</em>
<a name="187" href="#187">187</a>     <strong>public</strong> <strong>static</strong> <a href="../../../org/archive/io/ArchiveReader.html">ArchiveReader</a> get(<strong>final</strong> URL u, <strong>final</strong> <strong>long</strong> offset)
<a name="188" href="#188">188</a>     throws IOException {
<a name="189" href="#189">189</a>     	<strong>return</strong> ArchiveReaderFactory.factory.getArchiveReader(u, offset);
<a name="190" href="#190">190</a>     }
<a name="191" href="#191">191</a>     
<a name="192" href="#192">192</a>     <strong>protected</strong> <a href="../../../org/archive/io/ArchiveReader.html">ArchiveReader</a> getArchiveReader(<strong>final</strong> URL f, <strong>final</strong> <strong>long</strong> offset)
<a name="193" href="#193">193</a>     throws IOException {
<a name="194" href="#194">194</a>         <em class="comment">// Get URL connection.</em>
<a name="195" href="#195">195</a>         URLConnection connection = f.openConnection();
<a name="196" href="#196">196</a>         <strong>if</strong> (!(connection instanceof HttpURLConnection)) {
<a name="197" href="#197">197</a>             <strong>throw</strong> <strong>new</strong> IOException(<span class="string">"This method only handles HTTP connections."</span>);
<a name="198" href="#198">198</a>         }
<a name="199" href="#199">199</a>         addUserAgent((HttpURLConnection)connection);
<a name="200" href="#200">200</a>         <strong>if</strong> (offset != STREAM_ALL) {
<a name="201" href="#201">201</a>         	<em class="comment">// Use a Range request (Assumes HTTP 1.1 on other end). If</em>
<a name="202" href="#202">202</a>         	<em class="comment">// length >= 0, add open-ended range header to the request.  Else,</em>
<a name="203" href="#203">203</a>         	<em class="comment">// because end-byte is inclusive, subtract 1.</em>
<a name="204" href="#204">204</a>         	connection.addRequestProperty(<span class="string">"Range"</span>, <span class="string">"bytes="</span> + offset + <span class="string">"-"</span>);
<a name="205" href="#205">205</a>         }
<a name="206" href="#206">206</a>         
<a name="207" href="#207">207</a>         <strong>return</strong> getArchiveReader(f.toString(), connection.getInputStream(),
<a name="208" href="#208">208</a>             (offset == 0));
<a name="209" href="#209">209</a>     }
<a name="210" href="#210">210</a>     
<a name="211" href="#211">211</a>     <em>/**<em>*</em></em>
<a name="212" href="#212">212</a> <em>     * Get an ARCReader.</em>
<a name="213" href="#213">213</a> <em>     * Pulls the ARC local into whereever the System Property</em>
<a name="214" href="#214">214</a> <em>     * &lt;code>java.io.tmpdir&lt;/code> points. It then hands back an ARCReader that</em>
<a name="215" href="#215">215</a> <em>     * points at this local copy.  A close on this ARCReader instance will</em>
<a name="216" href="#216">216</a> <em>     * remove the local copy.</em>
<a name="217" href="#217">217</a> <em>     * @param u An URL that points at an ARC.</em>
<a name="218" href="#218">218</a> <em>     * @return An ARCReader.</em>
<a name="219" href="#219">219</a> <em>     * @throws IOException </em>
<a name="220" href="#220">220</a> <em>     */</em>
<a name="221" href="#221">221</a>     <strong>public</strong> <strong>static</strong> <a href="../../../org/archive/io/ArchiveReader.html">ArchiveReader</a> get(<strong>final</strong> URL u)
<a name="222" href="#222">222</a>     throws IOException {
<a name="223" href="#223">223</a>     	<strong>return</strong> ArchiveReaderFactory.factory.getArchiveReader(u);
<a name="224" href="#224">224</a>     }
<a name="225" href="#225">225</a>     
<a name="226" href="#226">226</a>     <strong>protected</strong> <a href="../../../org/archive/io/ArchiveReader.html">ArchiveReader</a> getArchiveReader(<strong>final</strong> URL u)
<a name="227" href="#227">227</a>     throws IOException {
<a name="228" href="#228">228</a>         <em class="comment">// If url represents a local file then return file it points to.</em>
<a name="229" href="#229">229</a>         <strong>if</strong> (u.getPath() != <strong>null</strong>) {
<a name="230" href="#230">230</a>             <em class="comment">// TODO: Add scheme check and host check.</em>
<a name="231" href="#231">231</a>             File f = <strong>new</strong> File(u.getPath());
<a name="232" href="#232">232</a>             <strong>if</strong> (f.exists()) {
<a name="233" href="#233">233</a>                 <strong>return</strong> get(f, 0);
<a name="234" href="#234">234</a>             }
<a name="235" href="#235">235</a>         }
<a name="236" href="#236">236</a>        
<a name="237" href="#237">237</a>         String scheme = u.getProtocol();
<a name="238" href="#238">238</a>         <strong>if</strong> (scheme.startsWith(<span class="string">"http"</span>) || scheme.equals(<span class="string">"s3"</span>)) {
<a name="239" href="#239">239</a>             <em class="comment">// Try streaming if http or s3 URLs rather than copying local</em>
<a name="240" href="#240">240</a>         	<em class="comment">// and then reading (Passing an offset will get us an Reader</em>
<a name="241" href="#241">241</a>         	<em class="comment">// that wraps a Stream).</em>
<a name="242" href="#242">242</a>             <strong>return</strong> get(u, STREAM_ALL);
<a name="243" href="#243">243</a>         }
<a name="244" href="#244">244</a>         
<a name="245" href="#245">245</a>         <strong>return</strong> makeARCLocal(u.openConnection());
<a name="246" href="#246">246</a>     }
<a name="247" href="#247">247</a>     
<a name="248" href="#248">248</a>     <strong>protected</strong> <a href="../../../org/archive/io/ArchiveReader.html">ArchiveReader</a> makeARCLocal(<strong>final</strong> URLConnection connection)
<a name="249" href="#249">249</a>     throws IOException {
<a name="250" href="#250">250</a>         File localFile = <strong>null</strong>;
<a name="251" href="#251">251</a>         <strong>if</strong> (connection instanceof HttpURLConnection) {
<a name="252" href="#252">252</a>             <em class="comment">// If http url connection, bring down the resource local.</em>
<a name="253" href="#253">253</a>             String p = connection.getURL().getPath();
<a name="254" href="#254">254</a>             <strong>int</strong> index = p.lastIndexOf('/');
<a name="255" href="#255">255</a>             <strong>if</strong> (index >= 0) {
<a name="256" href="#256">256</a>                 <em class="comment">// Name file for the file we're making local.</em>
<a name="257" href="#257">257</a>                 localFile = <strong>new</strong> File(FileUtils.TMPDIR, p.substring(index + 1));
<a name="258" href="#258">258</a>                 <strong>if</strong> (localFile.exists()) {
<a name="259" href="#259">259</a>                     <em class="comment">// If file of same name already exists in TMPDIR, then</em>
<a name="260" href="#260">260</a>                     <em class="comment">// clean it up (Assuming only reason a file of same name in</em>
<a name="261" href="#261">261</a>                     <em class="comment">// TMPDIR is because we failed a previous download).</em>
<a name="262" href="#262">262</a>                     localFile.delete();
<a name="263" href="#263">263</a>                 }
<a name="264" href="#264">264</a>             } <strong>else</strong> {
<a name="265" href="#265">265</a>                 localFile = File.createTempFile(ArchiveReader.<strong>class</strong>.getName(),
<a name="266" href="#266">266</a>                     <span class="string">".tmp"</span>, FileUtils.TMPDIR);
<a name="267" href="#267">267</a>             }
<a name="268" href="#268">268</a>             addUserAgent((HttpURLConnection)connection);
<a name="269" href="#269">269</a>             connection.connect();
<a name="270" href="#270">270</a>             <strong>try</strong> {
<a name="271" href="#271">271</a>                 IoUtils.readFullyToFile(connection.getInputStream(), localFile,
<a name="272" href="#272">272</a>                     <strong>new</strong> byte[16 * 1024]);
<a name="273" href="#273">273</a>             } <strong>catch</strong> (IOException ioe) {
<a name="274" href="#274">274</a>                 localFile.delete();
<a name="275" href="#275">275</a>                 <strong>throw</strong> ioe;
<a name="276" href="#276">276</a>             }
<a name="277" href="#277">277</a>         } <strong>else</strong> <strong>if</strong> (connection instanceof RsyncURLConnection) {
<a name="278" href="#278">278</a>             <em class="comment">// Then, connect and this will create a local file.</em>
<a name="279" href="#279">279</a>             <em class="comment">// See implementation of the rsync handler.</em>
<a name="280" href="#280">280</a>             connection.connect();
<a name="281" href="#281">281</a>             localFile = ((RsyncURLConnection)connection).getFile();
<a name="282" href="#282">282</a>         } <strong>else</strong> <strong>if</strong> (connection instanceof Md5URLConnection) {
<a name="283" href="#283">283</a>             <em class="comment">// Then, connect and this will create a local file.</em>
<a name="284" href="#284">284</a>             <em class="comment">// See implementation of the md5 handler.</em>
<a name="285" href="#285">285</a>             connection.connect();
<a name="286" href="#286">286</a>             localFile = ((Md5URLConnection)connection).getFile();
<a name="287" href="#287">287</a>         } <strong>else</strong> {
<a name="288" href="#288">288</a>             <strong>throw</strong> <strong>new</strong> UnsupportedOperationException(<span class="string">"No support for "</span> +
<a name="289" href="#289">289</a>                 connection);
<a name="290" href="#290">290</a>         }
<a name="291" href="#291">291</a>         
<a name="292" href="#292">292</a>         <a href="../../../org/archive/io/ArchiveReader.html">ArchiveReader</a> reader = <strong>null</strong>;
<a name="293" href="#293">293</a>         <strong>try</strong> {
<a name="294" href="#294">294</a>             reader = get(localFile, 0);
<a name="295" href="#295">295</a>         } <strong>catch</strong> (IOException e) {
<a name="296" href="#296">296</a>             localFile.delete();
<a name="297" href="#297">297</a>             <strong>throw</strong> e;
<a name="298" href="#298">298</a>         }
<a name="299" href="#299">299</a>         
<a name="300" href="#300">300</a>         <em class="comment">// Return a delegate that does cleanup of downloaded file on close.</em>
<a name="301" href="#301">301</a>         <strong>return</strong> reader.getDeleteFileOnCloseReader(localFile);
<a name="302" href="#302">302</a>     }
<a name="303" href="#303">303</a>     
<a name="304" href="#304">304</a>     <strong>protected</strong> <strong>void</strong> addUserAgent(<strong>final</strong> HttpURLConnection connection) {
<a name="305" href="#305">305</a>         connection.addRequestProperty(<span class="string">"User-Agent"</span>, <strong>this</strong>.getClass().getName());
<a name="306" href="#306">306</a>     }
<a name="307" href="#307">307</a>     
<a name="308" href="#308">308</a>     <em>/**<em>*</em></em>
<a name="309" href="#309">309</a> <em>     * @param f File to test.</em>
<a name="310" href="#310">310</a> <em>     * @return True if &lt;code>f&lt;/code> is compressed.</em>
<a name="311" href="#311">311</a> <em>     * @throws IOException</em>
<a name="312" href="#312">312</a> <em>     */</em>
<a name="313" href="#313">313</a>     <strong>protected</strong> <strong>boolean</strong> isCompressed(<strong>final</strong> File f) throws IOException {
<a name="314" href="#314">314</a>         <strong>return</strong> f.getName().toLowerCase().
<a name="315" href="#315">315</a>         	endsWith(DOT_COMPRESSED_FILE_EXTENSION);
<a name="316" href="#316">316</a>     }
<a name="317" href="#317">317</a> }
</pre>
<hr/><div id="footer">This page was automatically generated by <a href="http://maven.apache.org/">Maven</a></div></body>
</html>