abstractfrontier.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="274" href="#274">274</a>                     <span class="string">" assignment queue policy string: "</span> + queueStr);
<a name="275" href="#275">275</a>         }
<a name="276" href="#276">276</a>         t = addElementToDefinition(<strong>new</strong> <a href="../../../../org/archive/crawler/settings/SimpleType.html">SimpleType</a>(ATTR_QUEUE_ASSIGNMENT_POLICY,
<a name="277" href="#277">277</a>                 <span class="string">"Defines how to assign URIs to queues. Can assign by host, "</span> +
<a name="278" href="#278">278</a>                 <span class="string">"by ip, and into one of a fixed set of buckets (1k)."</span>,
<a name="279" href="#279">279</a>                 queues[0], queues));
<a name="280" href="#280">280</a>         t.setExpertSetting(<strong>true</strong>);
<a name="281" href="#281">281</a>         t.setOverrideable(false);
<a name="282" href="#282">282</a> 
<a name="283" href="#283">283</a>         t = addElementToDefinition(<strong>new</strong> <a href="../../../../org/archive/crawler/settings/SimpleType.html">SimpleType</a>(
<a name="284" href="#284">284</a>                 ATTR_FORCE_QUEUE,
<a name="285" href="#285">285</a>                 <span class="string">"The queue name into which to force URIs. Should "</span>
<a name="286" href="#286">286</a>                 + <span class="string">"be left blank at global level.  Specify a "</span>
<a name="287" href="#287">287</a>                 + <span class="string">"per-domain/per-host override to force URIs into "</span>
<a name="288" href="#288">288</a>                 + <span class="string">"a particular named queue, regardless of the assignment "</span>
<a name="289" href="#289">289</a>                 + <span class="string">"policy in effect (domain or ip-based politeness). "</span>
<a name="290" href="#290">290</a>                 + <span class="string">"This could be used on domains known to all be from "</span>
<a name="291" href="#291">291</a>                 + <span class="string">"the same small set of IPs (eg blogspot, dailykos, etc.) "</span>
<a name="292" href="#292">292</a>                 + <span class="string">"to simulate IP-based politeness, or it could be used if "</span>
<a name="293" href="#293">293</a>                 + <span class="string">"you wanted to enforce politeness over a whole domain, even "</span>
<a name="294" href="#294">294</a>                 + <span class="string">"though the subdomains are split across many IPs."</span>,
<a name="295" href="#295">295</a>                 DEFAULT_FORCE_QUEUE));
<a name="296" href="#296">296</a>         t.setOverrideable(<strong>true</strong>);
<a name="297" href="#297">297</a>         t.setExpertSetting(<strong>true</strong>);
<a name="298" href="#298">298</a>         t.addConstraint(<strong>new</strong> <a href="../../../../org/archive/crawler/settings/RegularExpressionConstraint.html">RegularExpressionConstraint</a>(ACCEPTABLE_FORCE_QUEUE,
<a name="299" href="#299">299</a>                 Level.WARNING, <span class="string">"This field must contain only alphanumeric "</span>
<a name="300" href="#300">300</a>                 + <span class="string">"characters plus period, dash, comma, colon, or underscore."</span>));
<a name="301" href="#301">301</a>         t = addElementToDefinition(<strong>new</strong> <a href="../../../../org/archive/crawler/settings/SimpleType.html">SimpleType</a>(
<a name="302" href="#302">302</a>                 ATTR_PAUSE_AT_START,
<a name="303" href="#303">303</a>                 <span class="string">"Whether to pause when the crawl begins, before any URIs "</span> +
<a name="304" href="#304">304</a>                 <span class="string">"are tried. This gives the operator a chance to verify or "</span> +
<a name="305" href="#305">305</a>                 <span class="string">"adjust the crawl before actual work begins. "</span> +
<a name="306" href="#306">306</a>                 <span class="string">"Default is false."</span>, DEFAULT_PAUSE_AT_START));
<a name="307" href="#307">307</a>         t = addElementToDefinition(<strong>new</strong> <a href="../../../../org/archive/crawler/settings/SimpleType.html">SimpleType</a>(
<a name="308" href="#308">308</a>                 ATTR_PAUSE_AT_FINISH,
<a name="309" href="#309">309</a>                 <span class="string">"Whether to pause when the crawl appears finished, rather "</span>
<a name="310" href="#310">310</a>                 + <span class="string">"than immediately end the crawl. This gives the operator an "</span>
<a name="311" href="#311">311</a>                 + <span class="string">"opportunity to view crawl results, and possibly add URIs or "</span>
<a name="312" href="#312">312</a>                 + <span class="string">"adjust settings, while the crawl state is still available. "</span>
<a name="313" href="#313">313</a>                 + <span class="string">"Default is false."</span>, DEFAULT_PAUSE_AT_FINISH));
<a name="314" href="#314">314</a>         t.setOverrideable(false);
<a name="315" href="#315">315</a>         
<a name="316" href="#316">316</a>         t = addElementToDefinition(<strong>new</strong> <a href="../../../../org/archive/crawler/settings/SimpleType.html">SimpleType</a>(
<a name="317" href="#317">317</a>                 ATTR_SOURCE_TAG_SEEDS,
<a name="318" href="#318">318</a>                 <span class="string">"Whether to tag seeds with their own URI as a heritable "</span> +
<a name="319" href="#319">319</a>                 <span class="string">"'source' String, which will be carried-forward to all URIs "</span> +
<a name="320" href="#320">320</a>                 <span class="string">"discovered on paths originating from that seed. When "</span> +
<a name="321" href="#321">321</a>                 <span class="string">"present, such source tags appear in the second-to-last "</span> +
<a name="322" href="#322">322</a>                 <span class="string">"crawl.log field."</span>, DEFAULT_SOURCE_TAG_SEEDS));
<a name="323" href="#323">323</a>         t.setOverrideable(false);
<a name="324" href="#324">324</a>         
<a name="325" href="#325">325</a>         t = addElementToDefinition(<strong>new</strong> <a href="../../../../org/archive/crawler/settings/SimpleType.html">SimpleType</a>(ATTR_RECOVERY_ENABLED,
<a name="326" href="#326">326</a>                 <span class="string">"Set to false to disable recovery log writing.  Do this if "</span> +
<a name="327" href="#327">327</a>                 <span class="string">"you you are using the checkpoint feature for recovering "</span> +
<a name="328" href="#328">328</a>                 <span class="string">"crashed crawls."</span>, DEFAULT_ATTR_RECOVERY_ENABLED));
<a name="329" href="#329">329</a>         t.setExpertSetting(<strong>true</strong>);
<a name="330" href="#330">330</a>         <em class="comment">// No sense in it being overrideable.</em>
<a name="331" href="#331">331</a>         t.setOverrideable(false);
<a name="332" href="#332">332</a>     }
<a name="333" href="#333">333</a> 
<a name="334" href="#334">334</a>     <strong>public</strong> <strong>void</strong> start() {
<a name="335" href="#335">335</a>         <strong>if</strong> (((Boolean)getUncheckedAttribute(<strong>null</strong>, ATTR_PAUSE_AT_START))
<a name="336" href="#336">336</a>                 .booleanValue()) {
<a name="337" href="#337">337</a>             <em class="comment">// trigger crawl-wide pause</em>
<a name="338" href="#338">338</a>             controller.requestCrawlPause();
<a name="339" href="#339">339</a>         } <strong>else</strong> {
<a name="340" href="#340">340</a>             <em class="comment">// simply begin</em>
<a name="341" href="#341">341</a>             unpause(); 
<a name="342" href="#342">342</a>         }
<a name="343" href="#343">343</a>     }
<a name="344" href="#344">344</a>     
<a name="345" href="#345">345</a>     <strong>synchronized</strong> <strong>public</strong> <strong>void</strong> pause() {
<a name="346" href="#346">346</a>         shouldPause = <strong>true</strong>;
<a name="347" href="#347">347</a>     }
<a name="348" href="#348">348</a> 
<a name="349" href="#349">349</a>     <strong>synchronized</strong> <strong>public</strong> <strong>void</strong> unpause() {
<a name="350" href="#350">350</a>         shouldPause = false;
<a name="351" href="#351">351</a>         notifyAll();
<a name="352" href="#352">352</a>     }
<a name="353" href="#353">353</a> 
<a name="354" href="#354">354</a>     <strong>public</strong> <strong>void</strong> initialize(<a href="../../../../org/archive/crawler/framework/CrawlController.html">CrawlController</a> c)
<a name="355" href="#355">355</a>             throws FatalConfigurationException, IOException {
<a name="356" href="#356">356</a>         c.addCrawlStatusListener(<strong>this</strong>);
<a name="357" href="#357">357</a>         File logsDisk = <strong>null</strong>;
<a name="358" href="#358">358</a>         <strong>try</strong> {
<a name="359" href="#359">359</a>             logsDisk = c.getSettingsDir(CrawlOrder.ATTR_LOGS_PATH);
<a name="360" href="#360">360</a>         } <strong>catch</strong> (AttributeNotFoundException e) {
<a name="361" href="#361">361</a>             logger.log(Level.SEVERE, <span class="string">"Failed to get logs directory"</span>, e);
<a name="362" href="#362">362</a>         }
<a name="363" href="#363">363</a>         <strong>if</strong> (logsDisk != <strong>null</strong>) {
<a name="364" href="#364">364</a>             String logsPath = logsDisk.getAbsolutePath() + File.separatorChar;
<a name="365" href="#365">365</a>             <strong>if</strong> (((Boolean)getUncheckedAttribute(<strong>null</strong>, ATTR_RECOVERY_ENABLED))
<a name="366" href="#366">366</a>                     .booleanValue()) {
<a name="367" href="#367">367</a>                 <strong>this</strong>.recover = <strong>new</strong> <a href="../../../../org/archive/crawler/frontier/RecoveryJournal.html">RecoveryJournal</a>(logsPath,
<a name="368" href="#368">368</a>                     FrontierJournal.LOGNAME_RECOVER);
<a name="369" href="#369">369</a>             }
<a name="370" href="#370">370</a>         }
<a name="371" href="#371">371</a>         <strong>try</strong> {
<a name="372" href="#372">372</a>             <strong>final</strong> Class qapClass = Class.forName((String)getUncheckedAttribute(
<a name="373" href="#373">373</a>                     <strong>null</strong>, ATTR_QUEUE_ASSIGNMENT_POLICY));
<a name="374" href="#374">374</a> 
<a name="375" href="#375">375</a>             queueAssignmentPolicy =
<a name="376" href="#376">376</a>                 (<a href="../../../../org/archive/crawler/frontier/QueueAssignmentPolicy.html">QueueAssignmentPolicy</a>)qapClass.newInstance();
<a name="377" href="#377">377</a>         } <strong>catch</strong> (Exception e) {
<a name="378" href="#378">378</a>             logger.log(Level.SEVERE, <span class="string">"Bad queue assignment policy class"</span>, e);
<a name="379" href="#379">379</a>             <strong>throw</strong> <strong>new</strong> <a href="../../../../org/archive/crawler/framework/exceptions/FatalConfigurationException.html">FatalConfigurationException</a>(e.getMessage());
<a name="380" href="#380">380</a>         }
<a name="381" href="#381">381</a>     }
<a name="382" href="#382">382</a> 
<a name="383" href="#383">383</a>     <strong>synchronized</strong> <strong>public</strong> <strong>void</strong> terminate() {
<a name="384" href="#384">384</a>         shouldTerminate = <strong>true</strong>;
<a name="385" href="#385">385</a>         <strong>if</strong> (<strong>this</strong>.recover != <strong>null</strong>) {
<a name="386" href="#386">386</a>             <strong>this</strong>.recover.close();
<a name="387" href="#387">387</a>             <strong>this</strong>.recover = <strong>null</strong>;
<a name="388" href="#388">388</a>         }
<a name="389" href="#389">389</a>         unpause();
<a name="390" href="#390">390</a>     }
<a name="391" href="#391">391</a> 
<a name="392" href="#392">392</a>     <strong>protected</strong> <strong>void</strong> doJournalFinishedSuccess(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> c) {
<a name="393" href="#393">393</a>         <strong>if</strong> (<strong>this</strong>.recover != <strong>null</strong>) {
<a name="394" href="#394">394</a>             <strong>this</strong>.recover.finishedSuccess(c);
<a name="395" href="#395">395</a>         }
<a name="396" href="#396">396</a>     }
<a name="397" href="#397">397</a> 
<a name="398" href="#398">398</a>     <strong>protected</strong> <strong>void</strong> doJournalAdded(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> c) {
<a name="399" href="#399">399</a>         <strong>if</strong> (<strong>this</strong>.recover != <strong>null</strong>) {
<a name="400" href="#400">400</a>             <strong>this</strong>.recover.added(c);
<a name="401" href="#401">401</a>         }
<a name="402" href="#402">402</a>     }
<a name="403" href="#403">403</a> 
<a name="404" href="#404">404</a>     <strong>protected</strong> <strong>void</strong> doJournalRescheduled(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> c) {
<a name="405" href="#405">405</a>         <strong>if</strong> (<strong>this</strong>.recover != <strong>null</strong>) {
<a name="406" href="#406">406</a>             <strong>this</strong>.recover.rescheduled(c);
<a name="407" href="#407">407</a>         }
<a name="408" href="#408">408</a>     }
<a name="409" href="#409">409</a> 
<a name="410" href="#410">410</a>     <strong>protected</strong> <strong>void</strong> doJournalFinishedFailure(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> c) {
<a name="411" href="#411">411</a>         <strong>if</strong> (<strong>this</strong>.recover != <strong>null</strong>) {
<a name="412" href="#412">412</a>             <strong>this</strong>.recover.finishedFailure(c);
<a name="413" href="#413">413</a>         }
<a name="414" href="#414">414</a>     }