?? adaptiverevisitfrontier.html
字號:
<a name="397" href="#397">397</a> <em class="comment">// gets immediately redirected to another page. What we're doing</em><a name="398" href="#398">398</a> <em class="comment">// is treating the immediate redirect target as a seed.</em><a name="399" href="#399">399</a> <strong>this</strong>.controller.getScope().addSeed(curi);<a name="400" href="#400">400</a> <em class="comment">// And it needs rapid scheduling.</em><a name="401" href="#401">401</a> curi.setSchedulingDirective(CandidateURI.MEDIUM);<a name="402" href="#402">402</a> }<a name="403" href="#403">403</a> <a name="404" href="#404">404</a> <em class="comment">// Optionally preferencing embeds up to MEDIUM</em><a name="405" href="#405">405</a> <strong>int</strong> prefHops = ((Integer) getUncheckedAttribute(curi,<a name="406" href="#406">406</a> ATTR_PREFERENCE_EMBED_HOPS)).intValue();<a name="407" href="#407">407</a> <strong>boolean</strong> prefEmbed = false;<a name="408" href="#408">408</a> <strong>if</strong> (prefHops > 0) {<a name="409" href="#409">409</a> <strong>int</strong> embedHops = curi.getTransHops();<a name="410" href="#410">410</a> <strong>if</strong> (embedHops > 0 && embedHops <= prefHops<a name="411" href="#411">411</a> && curi.getSchedulingDirective() == CandidateURI.NORMAL) {<a name="412" href="#412">412</a> <em class="comment">// number of embed hops falls within the preferenced range, and</em><a name="413" href="#413">413</a> <em class="comment">// uri is not already MEDIUM -- so promote it</em><a name="414" href="#414">414</a> curi.setSchedulingDirective(CandidateURI.MEDIUM);<a name="415" href="#415">415</a> prefEmbed = <strong>true</strong>;<a name="416" href="#416">416</a> }<a name="417" href="#417">417</a> }<a name="418" href="#418">418</a> <a name="419" href="#419">419</a> <em class="comment">// Finally, allow curi to be fetched right now </em><a name="420" href="#420">420</a> <em class="comment">// (while not overriding overdue items)</em><a name="421" href="#421">421</a> curi.putLong(A_TIME_OF_NEXT_PROCESSING,<a name="422" href="#422">422</a> System.currentTimeMillis());<a name="423" href="#423">423</a> <a name="424" href="#424">424</a> <strong>try</strong> {<a name="425" href="#425">425</a> logger.finest(<span class="string">"scheduling "</span> + curi.toString());<a name="426" href="#426">426</a> <a href="../../../../org/archive/crawler/frontier/AdaptiveRevisitHostQueue.html">AdaptiveRevisitHostQueue</a> hq = getHQ(curi);<a name="427" href="#427">427</a> hq.add(curi,prefEmbed);<a name="428" href="#428">428</a> } <strong>catch</strong> (IOException e) {<a name="429" href="#429">429</a> <em class="comment">// TODO Handle IOExceptions</em><a name="430" href="#430">430</a> e.printStackTrace();<a name="431" href="#431">431</a> }<a name="432" href="#432">432</a> <a name="433" href="#433">433</a> }<a name="434" href="#434">434</a> <a name="435" href="#435">435</a> <em>/**<em>*</em></em><a name="436" href="#436">436</a> <em> * Get the AdaptiveRevisitHostQueue for the given CrawlURI, creating</em><a name="437" href="#437">437</a> <em> * it if necessary. </em><a name="438" href="#438">438</a> <em> * </em><a name="439" href="#439">439</a> <em> * @param curi CrawlURI for which to get a queue</em><a name="440" href="#440">440</a> <em> * @return AdaptiveRevisitHostQueue for given CrawlURI</em><a name="441" href="#441">441</a> <em> * @throws IOException</em><a name="442" href="#442">442</a> <em> */</em><a name="443" href="#443">443</a> <strong>protected</strong> <a href="../../../../org/archive/crawler/frontier/AdaptiveRevisitHostQueue.html">AdaptiveRevisitHostQueue</a> getHQ(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) throws IOException {<a name="444" href="#444">444</a> <a href="../../../../org/archive/crawler/frontier/AdaptiveRevisitHostQueue.html">AdaptiveRevisitHostQueue</a> hq = hostQueues.getHQ(curi.getClassKey());<a name="445" href="#445">445</a> <strong>if</strong>(hq == <strong>null</strong>){<a name="446" href="#446">446</a> <em class="comment">// Need to create it.</em><a name="447" href="#447">447</a> <strong>int</strong> valence = DEFAULT_HOST_VALENCE.<strong>int</strong>Value();<a name="448" href="#448">448</a> <strong>try</strong> {<a name="449" href="#449">449</a> valence = ((Integer)getAttribute(curi,ATTR_HOST_VALENCE)).intValue();<a name="450" href="#450">450</a> } <strong>catch</strong> (AttributeNotFoundException e2) {<a name="451" href="#451">451</a> logger.severe(<span class="string">"Unable to load valence."</span>);<a name="452" href="#452">452</a> }<a name="453" href="#453">453</a> hq = hostQueues.createHQ(curi.getClassKey(),valence);<a name="454" href="#454">454</a> }<a name="455" href="#455">455</a> <strong>return</strong> hq;<a name="456" href="#456">456</a> }<a name="457" href="#457">457</a> <a name="458" href="#458">458</a> <strong>protected</strong> <strong>void</strong> batchSchedule(<a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a> caUri) {<a name="459" href="#459">459</a> threadWaiting.getQueue().enqueue(caUri);<a name="460" href="#460">460</a> }<a name="461" href="#461">461</a> <a name="462" href="#462">462</a> <strong>protected</strong> <strong>void</strong> batchFlush() {<a name="463" href="#463">463</a> innerBatchFlush();<a name="464" href="#464">464</a> }<a name="465" href="#465">465</a> <a name="466" href="#466">466</a> <strong>private</strong> <strong>void</strong> innerBatchFlush() {<a name="467" href="#467">467</a> <a href="../../../../org/archive/queue/Queue.html">Queue</a> q = threadWaiting.getQueue();<a name="468" href="#468">468</a> <strong>while</strong>(!q.isEmpty()) {<a name="469" href="#469">469</a> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a> caUri = (CandidateURI)q.dequeue();<a name="470" href="#470">470</a> <strong>if</strong>(alreadyIncluded != <strong>null</strong>){<a name="471" href="#471">471</a> String cannon = canonicalize(caUri);<a name="472" href="#472">472</a> System.out.println(<span class="string">"Cannon of "</span> + caUri + <span class="string">" is "</span> + cannon);<a name="473" href="#473">473</a> <strong>if</strong> (caUri.forceFetch()) {<a name="474" href="#474">474</a> alreadyIncluded.addForce(cannon, caUri);<a name="475" href="#475">475</a> } <strong>else</strong> {<a name="476" href="#476">476</a> alreadyIncluded.add(cannon, caUri);<a name="477" href="#477">477</a> }<a name="478" href="#478">478</a> } <strong>else</strong> {<a name="479" href="#479">479</a> innerSchedule(caUri);<a name="480" href="#480">480</a> }<a name="481" href="#481">481</a> }<a name="482" href="#482">482</a> }<a name="483" href="#483">483</a> <a name="484" href="#484">484</a> <em>/**<em>*</em></em><a name="485" href="#485">485</a> <em> * @param curi</em><a name="486" href="#486">486</a> <em> * @return the CrawlServer to be associated with this CrawlURI</em><a name="487" href="#487">487</a> <em> */</em><a name="488" href="#488">488</a> <strong>protected</strong> <a href="../../../../org/archive/crawler/datamodel/CrawlServer.html">CrawlServer</a> getServer(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) {<a name="489" href="#489">489</a> <strong>return</strong> <strong>this</strong>.controller.getServerCache().getServerFor(curi);<a name="490" href="#490">490</a> }<a name="491" href="#491">491</a> <a name="492" href="#492">492</a> <em class="comment">/*<em class="comment"> (non-Javadoc)</em></em><a name="493" href="#493">493</a> <em class="comment"> * @see org.archive.crawler.framework.Frontier#next()</em><a name="494" href="#494">494</a> <em class="comment"> */</em><a name="495" href="#495">495</a> <strong>public</strong> <strong>synchronized</strong> <a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> next() <a name="496" href="#496">496</a> throws InterruptedException, <a href="../../../../org/archive/crawler/framework/exceptions/EndedException.html">EndedException</a> {<a name="497" href="#497">497</a> controller.checkFinish();<a name="498" href="#498">498</a> <a name="499" href="#499">499</a> <strong>while</strong>(shouldPause){<a name="500" href="#500">500</a> controller.toePaused();<a name="501" href="#501">501</a> wait();<a name="502" href="#502">502</a> }<a name="503" href="#503">503</a> <a name="504" href="#504">504</a> <strong>if</strong>(shouldTerminate){<a name="505" href="#505">505</a> <strong>throw</strong> <strong>new</strong> <a href="../../../../org/archive/crawler/framework/exceptions/EndedException.html">EndedException</a>(<span class="string">"terminated"</span>);<a name="506" href="#506">506</a> }<a name="507" href="#507">507</a> <a name="508" href="#508">508</a> <a href="../../../../org/archive/crawler/frontier/AdaptiveRevisitHostQueue.html">AdaptiveRevisitHostQueue</a> hq = hostQueues.getTopHQ();<a name="509" href="#509">509</a> <a name="510" href="#510">510</a> <strong>while</strong>(hq.getState() != AdaptiveRevisitHostQueue.HQSTATE_READY){<a name="511" href="#511">511</a> <em class="comment">// Ok, so we don't have a ready queue, wait until the top one</em><a name="512" href="#512">512</a> <em class="comment">// will become available.</em><a name="513" href="#513">513</a> <strong>long</strong> waitTime = hq.getNextReadyTime() - System.currentTimeMillis();<a name="514" href="#514">514</a> <strong>if</strong>(waitTime > 0){<a name="515" href="#515">515</a> wait(waitTime);<a name="516" href="#516">516</a> }<a name="517" href="#517">517</a> <em class="comment">// The top HQ may have changed, so get it again</em><a name="518" href="#518">518</a> hq = hostQueues.getTopHQ(); <a name="519" href="#519">519</a> } <a name="520" href="#520">520</a> <a name="521" href="#521">521</a> <strong>if</strong>(shouldTerminate){<a name="522" href="#522">522</a> <em class="comment">// May have been terminated while thread was waiting for IO</em><a name="523" href="#523">523</a> <strong>throw</strong> <strong>new</strong> <a href="../../../../org/archive/crawler/framework/exceptions/EndedException.html">EndedException</a>(<span class="string">"terminated"</span>);<a name="524" href="#524">524</a> }<a name="525" href="#525">525</a> <a name="526" href="#526">526</a> <strong>try</strong> {<a name="527" href="#527">527</a> <a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi = hq.next();<a name="528" href="#528">528</a> <em class="comment">// Populate CURI with 'transient' variables such as server.</em><a name="529" href="#529">529</a> logger.fine(<span class="string">"Issuing "</span> + curi.toString());<a name="530" href="#530">530</a> <strong>long</strong> temp = curi.getLong(A_TIME_OF_NEXT_PROCESSING);<a name="531" href="#531">531</a> <strong>long</strong> currT = System.currentTimeMillis();
?? 快捷鍵說明
復制代碼
Ctrl + C
搜索代碼
Ctrl + F
全屏模式
F11
切換主題
Ctrl + Shift + D
顯示快捷鍵
?
增大字號
Ctrl + =
減小字號
Ctrl + -