bdbfrontier.html

一个开源的网页爬虫一个开源的网页爬虫一个开源的网页爬虫一个开源的网页爬虫一个开源的网页爬虫一个开源的网页爬虫

<a name="228" href="#228">228</a>             <strong>boolean</strong> inCacheOnly) {
<a name="229" href="#229">229</a>         <strong>return</strong> pendingUris.getInitialMarker(regexpr);
<a name="230" href="#230">230</a>     }
<a name="231" href="#231">231</a> 
<a name="232" href="#232">232</a>     <em>/**<em>*</em></em>
<a name="233" href="#233">233</a> <em>     * Return list of urls.</em>
<a name="234" href="#234">234</a> <em>     * @param marker</em>
<a name="235" href="#235">235</a> <em>     * @param numberOfMatches</em>
<a name="236" href="#236">236</a> <em>     * @param verbose </em>
<a name="237" href="#237">237</a> <em>     * @return List of URIs (strings).</em>
<a name="238" href="#238">238</a> <em>     */</em>
<a name="239" href="#239">239</a>     <strong>public</strong> ArrayList getURIsList(<a href="../../../../org/archive/crawler/framework/FrontierMarker.html">FrontierMarker</a> marker, <strong>int</strong> numberOfMatches,
<a name="240" href="#240">240</a>             <strong>final</strong> <strong>boolean</strong> verbose) {
<a name="241" href="#241">241</a>         List curis;
<a name="242" href="#242">242</a>         <strong>try</strong> {
<a name="243" href="#243">243</a>             curis = pendingUris.getFrom(marker, numberOfMatches);
<a name="244" href="#244">244</a>         } <strong>catch</strong> (DatabaseException e) {
<a name="245" href="#245">245</a>             e.printStackTrace();
<a name="246" href="#246">246</a>             <strong>throw</strong> <strong>new</strong> RuntimeException(e);
<a name="247" href="#247">247</a>         }
<a name="248" href="#248">248</a>         ArrayList results = <strong>new</strong> ArrayList(curis.size());
<a name="249" href="#249">249</a>         Iterator iter = curis.iterator();
<a name="250" href="#250">250</a>         <strong>while</strong>(iter.hasNext()) {
<a name="251" href="#251">251</a>             <a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi = (CrawlURI) iter.next();
<a name="252" href="#252">252</a>             results.add(<span class="string">"["</span>+curi.getClassKey()+<span class="string">"] "</span>+curi.singleLineReport());
<a name="253" href="#253">253</a>         }
<a name="254" href="#254">254</a>         <strong>return</strong> results;
<a name="255" href="#255">255</a>     }
<a name="256" href="#256">256</a>     
<a name="257" href="#257">257</a>     <strong>protected</strong> <strong>void</strong> initQueue() throws IOException {
<a name="258" href="#258">258</a>         <strong>try</strong> {
<a name="259" href="#259">259</a>             <strong>this</strong>.pendingUris = createMultipleWorkQueues();
<a name="260" href="#260">260</a>         } <strong>catch</strong>(DatabaseException e) {
<a name="261" href="#261">261</a>             <strong>throw</strong> (IOException)<strong>new</strong> IOException(e.getMessage()).initCause(e);
<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>     <strong>protected</strong> <strong>void</strong> closeQueue() {
<a name="266" href="#266">266</a>         <strong>if</strong> (<strong>this</strong>.pendingUris != <strong>null</strong>) {
<a name="267" href="#267">267</a>             <strong>this</strong>.pendingUris.close();
<a name="268" href="#268">268</a>             <strong>this</strong>.pendingUris = <strong>null</strong>;
<a name="269" href="#269">269</a>         }
<a name="270" href="#270">270</a>     }
<a name="271" href="#271">271</a>         
<a name="272" href="#272">272</a>     <strong>protected</strong> <a href="../../../../org/archive/crawler/frontier/BdbMultipleWorkQueues.html">BdbMultipleWorkQueues</a> getWorkQueues() {
<a name="273" href="#273">273</a>         <strong>return</strong> pendingUris;
<a name="274" href="#274">274</a>     }
<a name="275" href="#275">275</a> 
<a name="276" href="#276">276</a>     <strong>protected</strong> <strong>boolean</strong> workQueueDataOnDisk() {
<a name="277" href="#277">277</a>         <strong>return</strong> <strong>true</strong>;
<a name="278" href="#278">278</a>     }
<a name="279" href="#279">279</a>     
<a name="280" href="#280">280</a>     <strong>public</strong> <strong>void</strong> initialize(<a href="../../../../org/archive/crawler/framework/CrawlController.html">CrawlController</a> c)
<a name="281" href="#281">281</a>     throws FatalConfigurationException, IOException {
<a name="282" href="#282">282</a>         <strong>super</strong>.initialize(c);
<a name="283" href="#283">283</a>         <strong>if</strong> (c.isCheckpointRecover()) {
<a name="284" href="#284">284</a>             <em class="comment">// If a checkpoint recover, copy old values from serialized</em>
<a name="285" href="#285">285</a>             <em class="comment">// instance into this Frontier instance. Do it this way because </em>
<a name="286" href="#286">286</a>             <em class="comment">// though its possible to serialize BdbFrontier, its currently not</em>
<a name="287" href="#287">287</a>             <em class="comment">// possible to set/remove frontier attribute plugging the</em>
<a name="288" href="#288">288</a>             <em class="comment">// deserialized object back into the settings system.</em>
<a name="289" href="#289">289</a>             <em class="comment">// The below copying over is error-prone because its easy</em>
<a name="290" href="#290">290</a>             <em class="comment">// to miss a value.  Perhaps there's a better way?  Introspection?</em>
<a name="291" href="#291">291</a>             <a href="../../../../org/archive/crawler/frontier/BdbFrontier.html">BdbFrontier</a> f = <strong>null</strong>;
<a name="292" href="#292">292</a>             <strong>try</strong> {
<a name="293" href="#293">293</a>                 f = (BdbFrontier)CheckpointUtils.
<a name="294" href="#294">294</a>                     readObjectFromFile(<strong>this</strong>.getClass(),
<a name="295" href="#295">295</a>                         <strong>this</strong>.controller.getCheckpointRecover().getDirectory());
<a name="296" href="#296">296</a>             } <strong>catch</strong> (FileNotFoundException e) {
<a name="297" href="#297">297</a>                 <strong>throw</strong> <strong>new</strong> <a href="../../../../org/archive/crawler/framework/exceptions/FatalConfigurationException.html">FatalConfigurationException</a>(<span class="string">"Failed checkpoint "</span> +
<a name="298" href="#298">298</a>                     <span class="string">"recover: "</span> + e.getMessage());
<a name="299" href="#299">299</a>             } <strong>catch</strong> (IOException e) {
<a name="300" href="#300">300</a>                 <strong>throw</strong> <strong>new</strong> <a href="../../../../org/archive/crawler/framework/exceptions/FatalConfigurationException.html">FatalConfigurationException</a>(<span class="string">"Failed checkpoint "</span> +
<a name="301" href="#301">301</a>                     <span class="string">"recover: "</span> + e.getMessage());
<a name="302" href="#302">302</a>             } <strong>catch</strong> (ClassNotFoundException e) {
<a name="303" href="#303">303</a>                 <strong>throw</strong> <strong>new</strong> <a href="../../../../org/archive/crawler/framework/exceptions/FatalConfigurationException.html">FatalConfigurationException</a>(<span class="string">"Failed checkpoint "</span> +
<a name="304" href="#304">304</a>                     <span class="string">"recover: "</span> + e.getMessage());
<a name="305" href="#305">305</a>             }
<a name="306" href="#306">306</a> 
<a name="307" href="#307">307</a>             <strong>this</strong>.nextOrdinal = f.nextOrdinal;
<a name="308" href="#308">308</a>             <strong>this</strong>.totalProcessedBytes = f.totalProcessedBytes;
<a name="309" href="#309">309</a>             <strong>this</strong>.disregardedUriCount = f.disregardedUriCount;
<a name="310" href="#310">310</a>             <strong>this</strong>.failedFetchCount = f.failedFetchCount;
<a name="311" href="#311">311</a>             <strong>this</strong>.processedBytesAfterLastEmittedURI =
<a name="312" href="#312">312</a>                 f.processedBytesAfterLastEmittedURI;
<a name="313" href="#313">313</a>             <strong>this</strong>.queuedUriCount = f.queuedUriCount;
<a name="314" href="#314">314</a>             <strong>this</strong>.succeededFetchCount = f.succeededFetchCount;
<a name="315" href="#315">315</a>             <strong>this</strong>.lastMaxBandwidthKB = f.lastMaxBandwidthKB;
<a name="316" href="#316">316</a>             <strong>this</strong>.readyClassQueues = f.readyClassQueues;
<a name="317" href="#317">317</a>             <strong>this</strong>.inactiveQueues = f.inactiveQueues;
<a name="318" href="#318">318</a>             <strong>this</strong>.retiredQueues = f.retiredQueues;
<a name="319" href="#319">319</a>             <strong>this</strong>.snoozedClassQueues = f.snoozedClassQueues;
<a name="320" href="#320">320</a>             <strong>this</strong>.inProcessQueues = f.inProcessQueues;
<a name="321" href="#321">321</a>             wakeQueues();
<a name="322" href="#322">322</a>         }
<a name="323" href="#323">323</a>     }
<a name="324" href="#324">324</a> 
<a name="325" href="#325">325</a>     <strong>public</strong> <strong>void</strong> crawlCheckpoint(File checkpointDir) throws Exception {
<a name="326" href="#326">326</a>         <strong>super</strong>.crawlCheckpoint(checkpointDir);
<a name="327" href="#327">327</a>         logger.fine(<span class="string">"Started serializing already seen as part "</span>
<a name="328" href="#328">328</a>             + <span class="string">"of checkpoint. Can take some time."</span>);
<a name="329" href="#329">329</a>         <em class="comment">// An explicit sync on the any deferred write dbs is needed to make the</em>
<a name="330" href="#330">330</a>         <em class="comment">// db recoverable. Sync'ing the environment doesn't work.</em>
<a name="331" href="#331">331</a>         <strong>if</strong> (<strong>this</strong>.pendingUris != <strong>null</strong>) {
<a name="332" href="#332">332</a>         	<strong>this</strong>.pendingUris.sync();
<a name="333" href="#333">333</a>         }
<a name="334" href="#334">334</a>         <a href="../../../../org/archive/crawler/util/CheckpointUtils.html">CheckpointUtils</a> .writeObjectToFile(<strong>this</strong>.alreadyIncluded, checkpointDir);
<a name="335" href="#335">335</a>         logger.fine(<span class="string">"Finished serializing already seen as part "</span>
<a name="336" href="#336">336</a>             + <span class="string">"of checkpoint."</span>);
<a name="337" href="#337">337</a>         <em class="comment">// Serialize ourselves.</em>
<a name="338" href="#338">338</a>         CheckpointUtils.writeObjectToFile(<strong>this</strong>, checkpointDir);
<a name="339" href="#339">339</a>     }
<a name="340" href="#340">340</a> }
</pre>
<hr/><div id="footer">This page was automatically generated by <a href="http://maven.apache.org/">Maven</a></div></body>
</html>