fetchftp.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="155" href="#155">155</a>      + <span class="string">"completed in this number of seconds, give up (and retry later)."</span>;
<a name="156" href="#156">156</a>     
<a name="157" href="#157">157</a>     <em>/**<em>* The default value for the &lt;code>timeout-seconds&lt;/code> attribute. */</em></em>
<a name="158" href="#158">158</a>     <strong>final</strong> <strong>private</strong> <strong>static</strong> <strong>int</strong> DEFAULT_TIMEOUT = 1200;
<a name="159" href="#159">159</a>     
<a name="160" href="#160">160</a> 
<a name="161" href="#161">161</a>     <em>/**<em>*</em></em>
<a name="162" href="#162">162</a> <em>     * Constructs a new &lt;code>FetchFTP&lt;/code>.</em>
<a name="163" href="#163">163</a> <em>     * </em>
<a name="164" href="#164">164</a> <em>     * @param name  the name of this processor</em>
<a name="165" href="#165">165</a> <em>     */</em>
<a name="166" href="#166">166</a>     <strong>public</strong> <a href="../../../../org/archive/crawler/fetcher/FetchFTP.html">FetchFTP</a>(String name) {
<a name="167" href="#167">167</a>         <strong>super</strong>(name, <span class="string">"FTP Fetcher."</span>);
<a name="168" href="#168">168</a>         add(ATTR_USERNAME, DESC_USERNAME, DEFAULT_USERNAME);
<a name="169" href="#169">169</a>         add(ATTR_PASSWORD, DESC_PASSWORD, DEFAULT_PASSWORD);
<a name="170" href="#170">170</a>         add(ATTR_EXTRACT, DESC_EXTRACT, DEFAULT_EXTRACT);
<a name="171" href="#171">171</a>         add(ATTR_EXTRACT_PARENT, DESC_EXTRACT_PARENT, DEFAULT_EXTRACT_PARENT);
<a name="172" href="#172">172</a>         add(ATTR_MAX_LENGTH, DESC_MAX_LENGTH, DEFAULT_MAX_LENGTH);
<a name="173" href="#173">173</a>         add(ATTR_BANDWIDTH, DESC_BANDWIDTH, DEFAULT_BANDWIDTH);
<a name="174" href="#174">174</a>         add(ATTR_TIMEOUT, DESC_TIMEOUT, DEFAULT_TIMEOUT);
<a name="175" href="#175">175</a>     }
<a name="176" href="#176">176</a> 
<a name="177" href="#177">177</a>     
<a name="178" href="#178">178</a>     <em>/**<em>*</em></em>
<a name="179" href="#179">179</a> <em>     * Convenience method for adding an attribute.</em>
<a name="180" href="#180">180</a> <em>     * </em>
<a name="181" href="#181">181</a> <em>     * @param name   The name of the attribute</em>
<a name="182" href="#182">182</a> <em>     * @param desc   The description of the attribute</em>
<a name="183" href="#183">183</a> <em>     * @param def    The default value for the attribute</em>
<a name="184" href="#184">184</a> <em>     */</em>
<a name="185" href="#185">185</a>     <strong>private</strong> <strong>void</strong> add(String name, String desc, Object def) {
<a name="186" href="#186">186</a>         <a href="../../../../org/archive/crawler/settings/SimpleType.html">SimpleType</a> st = <strong>new</strong> <a href="../../../../org/archive/crawler/settings/SimpleType.html">SimpleType</a>(name, desc, def);
<a name="187" href="#187">187</a>         addElementToDefinition(st);
<a name="188" href="#188">188</a>     }
<a name="189" href="#189">189</a>     
<a name="190" href="#190">190</a>     
<a name="191" href="#191">191</a>     <em>/**<em>*</em></em>
<a name="192" href="#192">192</a> <em>     * Convenience method for extracting an attribute.</em>
<a name="193" href="#193">193</a> <em>     * If a value for the specified name cannot be found,</em>
<a name="194" href="#194">194</a> <em>     * a warning is written to the log and the specified</em>
<a name="195" href="#195">195</a> <em>     * default value is returned instead.</em>
<a name="196" href="#196">196</a> <em>     * </em>
<a name="197" href="#197">197</a> <em>     * @param context  The context for the attribute fetch</em>
<a name="198" href="#198">198</a> <em>     * @param name     The name of the attribute to fetch</em>
<a name="199" href="#199">199</a> <em>     * @param def      The value to return if the attribute isn't found</em>
<a name="200" href="#200">200</a> <em>     * @return         The value of that attribute</em>
<a name="201" href="#201">201</a> <em>     */</em>
<a name="202" href="#202">202</a>     <strong>private</strong> Object get(Object context, String name, Object def) {
<a name="203" href="#203">203</a>         <strong>try</strong> {
<a name="204" href="#204">204</a>             <strong>return</strong> getAttribute(context, name);
<a name="205" href="#205">205</a>         } <strong>catch</strong> (AttributeNotFoundException e) {
<a name="206" href="#206">206</a>             logger.warning(<span class="string">"Attribute not found (using default): "</span> + name);
<a name="207" href="#207">207</a>             <strong>return</strong> def;
<a name="208" href="#208">208</a>         }
<a name="209" href="#209">209</a>     }
<a name="210" href="#210">210</a>     
<a name="211" href="#211">211</a> 
<a name="212" href="#212">212</a>     <em>/**<em>*</em></em>
<a name="213" href="#213">213</a> <em>     * Processes the given URI.  If the given URI is not an FTP URI, then</em>
<a name="214" href="#214">214</a> <em>     * this method does nothing.  Otherwise an attempt is made to connect</em>
<a name="215" href="#215">215</a> <em>     * to the FTP server.</em>
<a name="216" href="#216">216</a> <em>     * </em>
<a name="217" href="#217">217</a> <em>     * &lt;p>If the connection is successful, an attempt will be made to CD to </em>
<a name="218" href="#218">218</a> <em>     * the path specified in the URI.  If the remote CD command succeeds, </em>
<a name="219" href="#219">219</a> <em>     * then it is assumed that the URI represents a directory.  If the</em>
<a name="220" href="#220">220</a> <em>     * CD command fails, then it is assumed that the URI represents</em>
<a name="221" href="#221">221</a> <em>     * a file.</em>
<a name="222" href="#222">222</a> <em>     * </em>
<a name="223" href="#223">223</a> <em>     * &lt;p>For directories, the directory listing will be fetched using</em>
<a name="224" href="#224">224</a> <em>     * the FTP LIST command, and saved to the HttpRecorder.  If the</em>
<a name="225" href="#225">225</a> <em>     * &lt;code>extract.from.dirs&lt;/code> attribute is set to true, then</em>
<a name="226" href="#226">226</a> <em>     * the files in the fetched list will be added to the curi as</em>
<a name="227" href="#227">227</a> <em>     * extracted FTP links.  (It was easier to do that here, rather</em>
<a name="228" href="#228">228</a> <em>     * than writing a separate FTPExtractor.)</em>
<a name="229" href="#229">229</a> <em>     * </em>
<a name="230" href="#230">230</a> <em>     * &lt;p>For files, the file will be fetched using the FTP RETR</em>
<a name="231" href="#231">231</a> <em>     * command, and saved to the HttpRecorder.</em>
<a name="232" href="#232">232</a> <em>     * </em>
<a name="233" href="#233">233</a> <em>     * &lt;p>All file transfers (including directory listings) occur using</em>
<a name="234" href="#234">234</a> <em>     * Binary mode transfer.  Also, the local passive transfer mode</em>
<a name="235" href="#235">235</a> <em>     * is always used, to play well with firewalls.</em>
<a name="236" href="#236">236</a> <em>     * </em>
<a name="237" href="#237">237</a> <em>     * @param curi  the curi to process</em>
<a name="238" href="#238">238</a> <em>     * @throws InterruptedException  if the thread is interrupted during</em>
<a name="239" href="#239">239</a> <em>     *   processing</em>
<a name="240" href="#240">240</a> <em>     */</em>
<a name="241" href="#241">241</a>     <strong>public</strong> <strong>void</strong> innerProcess(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) throws InterruptedException {
<a name="242" href="#242">242</a>         <strong>if</strong> (!curi.getUURI().getScheme().equals(<span class="string">"ftp"</span>)) {
<a name="243" href="#243">243</a>             <strong>return</strong>;
<a name="244" href="#244">244</a>         }
<a name="245" href="#245">245</a>         
<a name="246" href="#246">246</a>         curi.putLong(A_FETCH_BEGAN_TIME, System.currentTimeMillis());
<a name="247" href="#247">247</a>         <a href="../../../../org/archive/util/HttpRecorder.html">HttpRecorder</a> recorder = HttpRecorder.getHttpRecorder();
<a name="248" href="#248">248</a>         <a href="../../../../org/archive/net/ClientFTP.html">ClientFTP</a> client = <strong>new</strong> <a href="../../../../org/archive/net/ClientFTP.html">ClientFTP</a>();
<a name="249" href="#249">249</a>         
<a name="250" href="#250">250</a>         <strong>try</strong> {
<a name="251" href="#251">251</a>             fetch(curi, client, recorder);
<a name="252" href="#252">252</a>         } <strong>catch</strong> (FTPException e) {
<a name="253" href="#253">253</a>             logger.log(Level.SEVERE, <span class="string">"FTP server reported problem."</span>, e);
<a name="254" href="#254">254</a>             curi.setFetchStatus(e.getReplyCode());
<a name="255" href="#255">255</a>         } <strong>catch</strong> (IOException e) {
<a name="256" href="#256">256</a>             logger.log(Level.SEVERE, <span class="string">"IO Error during FTP fetch."</span>, e);
<a name="257" href="#257">257</a>             curi.setFetchStatus(FetchStatusCodes.S_CONNECT_LOST);
<a name="258" href="#258">258</a>         } <strong>finally</strong> {
<a name="259" href="#259">259</a>             disconnect(client);
<a name="260" href="#260">260</a>             curi.setContentSize(recorder.getRecordedInput().getSize());
<a name="261" href="#261">261</a>             curi.putLong(A_FETCH_COMPLETED_TIME, System.currentTimeMillis());
<a name="262" href="#262">262</a>         }
<a name="263" href="#263">263</a>     }
<a name="264" href="#264">264</a> 
<a name="265" href="#265">265</a> 
<a name="266" href="#266">266</a>     <em>/**<em>*</em></em>
<a name="267" href="#267">267</a> <em>     * Fetches a document from an FTP server.</em>
<a name="268" href="#268">268</a> <em>     * </em>
<a name="269" href="#269">269</a> <em>     * @param curi      the URI of the document to fetch</em>
<a name="270" href="#270">270</a> <em>     * @param client    the FTPClient to use for the fetch</em>
<a name="271" href="#271">271</a> <em>     * @param recorder  the recorder to preserve the document in</em>
<a name="272" href="#272">272</a> <em>     * @throws IOException  if a network or protocol error occurs</em>
<a name="273" href="#273">273</a> <em>     * @throws InterruptedException  if the thread is interrupted</em>
<a name="274" href="#274">274</a> <em>     */</em>
<a name="275" href="#275">275</a>     <strong>private</strong> <strong>void</strong> fetch(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi, <a href="../../../../org/archive/net/ClientFTP.html">ClientFTP</a> client, <a href="../../../../org/archive/util/HttpRecorder.html">HttpRecorder</a> recorder) 
<a name="276" href="#276">276</a>     throws IOException, InterruptedException {
<a name="277" href="#277">277</a>         <em class="comment">// Connect to the FTP server.</em>
<a name="278" href="#278">278</a>         <a href="../../../../org/archive/net/UURI.html">UURI</a> uuri = curi.getUURI();
<a name="279" href="#279">279</a>         <strong>int</strong> port = uuri.getPort();
<a name="280" href="#280">280</a>         <strong>if</strong> (port == -1) {
<a name="281" href="#281">281</a>             port = 21;
<a name="282" href="#282">282</a>         }
<a name="283" href="#283">283</a>         client.connectStrict(uuri.getHost(), port);
<a name="284" href="#284">284</a>         
<a name="285" href="#285">285</a>         <em class="comment">// Authenticate.</em>
<a name="286" href="#286">286</a>         String[] auth = getAuth(curi);
<a name="287" href="#287">287</a>         client.loginStrict(auth[0], auth[1]);
<a name="288" href="#288">288</a>         
<a name="289" href="#289">289</a>         <em class="comment">// The given resource may or may not be a directory.</em>
<a name="290" href="#290">290</a>         <em class="comment">// To figure out which is which, execute a CD command to</em>
<a name="291" href="#291">291</a>         <em class="comment">// the UURI's path.  If CD works, it's a directory.</em>
<a name="292" href="#292">292</a>         <strong>boolean</strong> dir = client.changeWorkingDirectory(uuri.getPath());
<a name="293" href="#293">293</a>         <strong>if</strong> (dir) {
<a name="294" href="#294">294</a>             curi.setContentType(<span class="string">"text/plain"</span>);
<a name="295" href="#295">295</a>         }
<a name="296" href="#296">296</a>         
<a name="297" href="#297">297</a>         <em class="comment">// TODO: A future version of this class could use the system string to</em>
<a name="298" href="#298">298</a>         <em class="comment">// set up custom directory parsing if the FTP server doesn't support </em>
<a name="299" href="#299">299</a>         <em class="comment">// the nlist command.</em>
<a name="300" href="#300">300</a>         <strong>if</strong> (logger.isLoggable(Level.FINE)) {
<a name="301" href="#301">301</a>             String system = client.getSystemName();
<a name="302" href="#302">302</a>             logger.fine(system);
<a name="303" href="#303">303</a>         }
<a name="304" href="#304">304</a>         
<a name="305" href="#305">305</a>         <em class="comment">// Get a data socket.  This will either be the result of a NLIST</em>
<a name="306" href="#306">306</a>         <em class="comment">// command for a directory, or a RETR command for a file.</em>
<a name="307" href="#307">307</a>         <strong>int</strong> command = dir ? FTPCommand.NLST : FTPCommand.RETR;
<a name="308" href="#308">308</a>         String path = dir ? <span class="string">"."</span> : uuri.getPath();
<a name="309" href="#309">309</a>         client.enterLocalPassiveMode();
<a name="310" href="#310">310</a>         client.setBinary();
<a name="311" href="#311">311</a>         Socket socket = client.openDataConnection(command, path);
<a name="312" href="#312">312</a>         curi.setFetchStatus(client.getReplyCode());
<a name="313" href="#313">313</a> 
<a name="314" href="#314">314</a>         <em class="comment">// Save the streams in the CURI, where downstream processors</em>
<a name="315" href="#315">315</a>         <em class="comment">// expect to find them.</em>
<a name="316" href="#316">316</a>         <strong>try</strong> {
<a name="317" href="#317">317</a>             saveToRecorder(curi, socket, recorder);