toethread.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="466" href="#466">466</a>         pw.println(getName());
<a name="467" href="#467">467</a> 
<a name="468" href="#468">468</a>         <em class="comment">// Make a local copy of the currentCuri reference in case it gets</em>
<a name="469" href="#469">469</a>         <em class="comment">// nulled while we're using it.  We're doing this because</em>
<a name="470" href="#470">470</a>         <em class="comment">// alternative is synchronizing and we don't want to do this --</em>
<a name="471" href="#471">471</a>         <em class="comment">// it causes hang ups as controller waits on a lock for this thread,</em>
<a name="472" href="#472">472</a>         <em class="comment">// something it gets easily enough on old threading model but something</em>
<a name="473" href="#473">473</a>         <em class="comment">// it can wait interminably for on NPTL threading model.</em>
<a name="474" href="#474">474</a>         <em class="comment">// See [ 994946 ] Pause/Terminate ignored on 2.6 kernel 1.5 JVM.</em>
<a name="475" href="#475">475</a>         <a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> c = currentCuri;
<a name="476" href="#476">476</a>         <strong>if</strong>(c != <strong>null</strong>) {
<a name="477" href="#477">477</a>             pw.print(<span class="string">" "</span>);
<a name="478" href="#478">478</a>             c.singleLineReportTo(pw);
<a name="479" href="#479">479</a>             pw.print(<span class="string">"    "</span>);
<a name="480" href="#480">480</a>             pw.print(c.getFetchAttempts());
<a name="481" href="#481">481</a>             pw.print(<span class="string">" attempts"</span>);
<a name="482" href="#482">482</a>             pw.println();
<a name="483" href="#483">483</a>             pw.print(<span class="string">"    "</span>);
<a name="484" href="#484">484</a>             pw.print(<span class="string">"in processor: "</span>);
<a name="485" href="#485">485</a>             pw.print(currentProcessorName);
<a name="486" href="#486">486</a>         } <strong>else</strong> {
<a name="487" href="#487">487</a>             pw.print(<span class="string">" -no CrawlURI- "</span>);
<a name="488" href="#488">488</a>         }
<a name="489" href="#489">489</a>         pw.println();
<a name="490" href="#490">490</a> 
<a name="491" href="#491">491</a>         <strong>long</strong> now = System.currentTimeMillis();
<a name="492" href="#492">492</a>         <strong>long</strong> time = 0;
<a name="493" href="#493">493</a> 
<a name="494" href="#494">494</a>         pw.print(<span class="string">"    "</span>);
<a name="495" href="#495">495</a>         <strong>if</strong>(lastFinishTime > lastStartTime) {
<a name="496" href="#496">496</a>             <em class="comment">// That means we finished something after we last started something</em>
<a name="497" href="#497">497</a>             <em class="comment">// or in other words we are not working on anything.</em>
<a name="498" href="#498">498</a>             pw.print(<span class="string">"WAITING for "</span>);
<a name="499" href="#499">499</a>             time = now - lastFinishTime;
<a name="500" href="#500">500</a>         } <strong>else</strong> <strong>if</strong>(lastStartTime > 0) {
<a name="501" href="#501">501</a>             <em class="comment">// We are working on something</em>
<a name="502" href="#502">502</a>             pw.print(<span class="string">"ACTIVE for "</span>);
<a name="503" href="#503">503</a>             time = now-lastStartTime;
<a name="504" href="#504">504</a>         }
<a name="505" href="#505">505</a>         pw.print(ArchiveUtils.formatMillisecondsToConventional(time));
<a name="506" href="#506">506</a>         pw.println();
<a name="507" href="#507">507</a> 
<a name="508" href="#508">508</a>         pw.print(<span class="string">"    "</span>);
<a name="509" href="#509">509</a>         pw.print(<span class="string">"step: "</span>);
<a name="510" href="#510">510</a>         pw.print(step);
<a name="511" href="#511">511</a>         pw.print(<span class="string">" for "</span>);
<a name="512" href="#512">512</a>         pw.print(ArchiveUtils.formatMillisecondsToConventional(System.currentTimeMillis()-atStepSince));
<a name="513" href="#513">513</a>         pw.println();
<a name="514" href="#514">514</a> 
<a name="515" href="#515">515</a>         StackTraceElement[] ste = <strong>this</strong>.getStackTrace();
<a name="516" href="#516">516</a>         <strong>for</strong>(<strong>int</strong> i=0;i&lt;ste.length;i++) {
<a name="517" href="#517">517</a>             pw.print(<span class="string">"    "</span>);
<a name="518" href="#518">518</a>             pw.print(ste[i].toString());
<a name="519" href="#519">519</a>             pw.println();
<a name="520" href="#520">520</a>         }
<a name="521" href="#521">521</a>         pw.print(<span class="string">"]"</span>);
<a name="522" href="#522">522</a>         pw.println();
<a name="523" href="#523">523</a>         
<a name="524" href="#524">524</a>         pw.flush();
<a name="525" href="#525">525</a>     }
<a name="526" href="#526">526</a> 
<a name="527" href="#527">527</a>     <em>/**<em>*</em></em>
<a name="528" href="#528">528</a> <em>     * @param w PrintWriter to write to.</em>
<a name="529" href="#529">529</a> <em>     */</em>
<a name="530" href="#530">530</a>     <strong>public</strong> <strong>void</strong> singleLineReportTo(PrintWriter w)
<a name="531" href="#531">531</a>     {
<a name="532" href="#532">532</a>         w.print(<span class="string">"#"</span>);
<a name="533" href="#533">533</a>         w.print(<strong>this</strong>.serialNumber);
<a name="534" href="#534">534</a> 
<a name="535" href="#535">535</a>         <em class="comment">// Make a local copy of the currentCuri reference in case it gets</em>
<a name="536" href="#536">536</a>         <em class="comment">// nulled while we're using it.  We're doing this because</em>
<a name="537" href="#537">537</a>         <em class="comment">// alternative is synchronizing and we don't want to do this --</em>
<a name="538" href="#538">538</a>         <em class="comment">// it causes hang ups as controller waits on a lock for this thread,</em>
<a name="539" href="#539">539</a>         <em class="comment">// something it gets easily enough on old threading model but something</em>
<a name="540" href="#540">540</a>         <em class="comment">// it can wait interminably for on NPTL threading model.</em>
<a name="541" href="#541">541</a>         <em class="comment">// See [ 994946 ] Pause/Terminate ignored on 2.6 kernel 1.5 JVM.</em>
<a name="542" href="#542">542</a>         <a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> c = currentCuri;
<a name="543" href="#543">543</a>         <strong>if</strong>(c != <strong>null</strong>) {
<a name="544" href="#544">544</a>             w.print(<span class="string">" "</span>);
<a name="545" href="#545">545</a>             w.print(currentProcessorName);
<a name="546" href="#546">546</a>             w.print(<span class="string">" "</span>);
<a name="547" href="#547">547</a>             w.print(c.toString());
<a name="548" href="#548">548</a>             w.print(<span class="string">" ("</span>);
<a name="549" href="#549">549</a>             w.print(c.getFetchAttempts());
<a name="550" href="#550">550</a>             w.print(<span class="string">") "</span>);
<a name="551" href="#551">551</a>         } <strong>else</strong> {
<a name="552" href="#552">552</a>             w.print(<span class="string">" [no CrawlURI] "</span>);
<a name="553" href="#553">553</a>         }
<a name="554" href="#554">554</a>         
<a name="555" href="#555">555</a>         <strong>long</strong> now = System.currentTimeMillis();
<a name="556" href="#556">556</a>         <strong>long</strong> time = 0;
<a name="557" href="#557">557</a> 
<a name="558" href="#558">558</a>         <strong>if</strong>(lastFinishTime > lastStartTime) {
<a name="559" href="#559">559</a>             <em class="comment">// That means we finished something after we last started something</em>
<a name="560" href="#560">560</a>             <em class="comment">// or in other words we are not working on anything.</em>
<a name="561" href="#561">561</a>             w.print(<span class="string">"WAITING for "</span>);
<a name="562" href="#562">562</a>             time = now - lastFinishTime;
<a name="563" href="#563">563</a>         } <strong>else</strong> <strong>if</strong>(lastStartTime > 0) {
<a name="564" href="#564">564</a>             <em class="comment">// We are working on something</em>
<a name="565" href="#565">565</a>             w.print(<span class="string">"ACTIVE for "</span>);
<a name="566" href="#566">566</a>             time = now-lastStartTime;
<a name="567" href="#567">567</a>         }
<a name="568" href="#568">568</a>         w.print(ArchiveUtils.formatMillisecondsToConventional(time));
<a name="569" href="#569">569</a>         w.print(<span class="string">" at "</span>);
<a name="570" href="#570">570</a>         w.print(step);
<a name="571" href="#571">571</a>         w.print(<span class="string">" for "</span>);
<a name="572" href="#572">572</a>         w.print(ArchiveUtils.formatMillisecondsToConventional(now-atStepSince));
<a name="573" href="#573">573</a>         w.print(<span class="string">"\n"</span>);
<a name="574" href="#574">574</a>         w.flush();
<a name="575" href="#575">575</a>     }
<a name="576" href="#576">576</a> 
<a name="577" href="#577">577</a>     <em class="comment">/*<em class="comment"> (non-Javadoc)</em></em>
<a name="578" href="#578">578</a> <em class="comment">     * @see org.archive.util.Reporter#singleLineLegend()</em>
<a name="579" href="#579">579</a> <em class="comment">     */</em>
<a name="580" href="#580">580</a>     <strong>public</strong> String singleLineLegend() {
<a name="581" href="#581">581</a>         <strong>return</strong> <span class="string">"#serialNumber processorName currentUri (fetchAttempts) threadState threadStep"</span>;
<a name="582" href="#582">582</a>     }
<a name="583" href="#583">583</a>     
<a name="584" href="#584">584</a>     <em class="comment">/*<em class="comment"> (non-Javadoc)</em></em>
<a name="585" href="#585">585</a> <em class="comment">     * @see org.archive.util.Reporter#getReports()</em>
<a name="586" href="#586">586</a> <em class="comment">     */</em>
<a name="587" href="#587">587</a>     <strong>public</strong> String[] getReports() {
<a name="588" href="#588">588</a>         <em class="comment">// for now none but the default</em>
<a name="589" href="#589">589</a>         <strong>return</strong> <strong>new</strong> String[] {};
<a name="590" href="#590">590</a>     }
<a name="591" href="#591">591</a> 
<a name="592" href="#592">592</a>     <strong>public</strong> <strong>void</strong> reportTo(PrintWriter writer) {
<a name="593" href="#593">593</a>         reportTo(<strong>null</strong>, writer);
<a name="594" href="#594">594</a>     }
<a name="595" href="#595">595</a> 
<a name="596" href="#596">596</a>     <em class="comment">/*<em class="comment"> (non-Javadoc)</em></em>
<a name="597" href="#597">597</a> <em class="comment">     * @see org.archive.util.Reporter#singleLineReport()</em>
<a name="598" href="#598">598</a> <em class="comment">     */</em>
<a name="599" href="#599">599</a>     <strong>public</strong> String singleLineReport() {
<a name="600" href="#600">600</a>         <strong>return</strong> ArchiveUtils.singleLineReport(<strong>this</strong>);
<a name="601" href="#601">601</a>     }
<a name="602" href="#602">602</a> 
<a name="603" href="#603">603</a>     <strong>public</strong> <strong>void</strong> progressStatisticsLine(PrintWriter writer) {
<a name="604" href="#604">604</a>         writer.print(getController().getStatistics()
<a name="605" href="#605">605</a>             .getProgressStatisticsLine());
<a name="606" href="#606">606</a>         writer.print(<span class="string">"\n"</span>);
<a name="607" href="#607">607</a>     }
<a name="608" href="#608">608</a> 
<a name="609" href="#609">609</a>     <strong>public</strong> <strong>void</strong> progressStatisticsLegend(PrintWriter writer) {
<a name="610" href="#610">610</a>         writer.print(getController().getStatistics()
<a name="611" href="#611">611</a>             .progressStatisticsLegend());
<a name="612" href="#612">612</a>         writer.print(<span class="string">"\n"</span>);
<a name="613" href="#613">613</a>     }
<a name="614" href="#614">614</a>     
<a name="615" href="#615">615</a>     <strong>public</strong> String getCurrentProcessorName() {
<a name="616" href="#616">616</a>         <strong>return</strong> currentProcessorName;
<a name="617" href="#617">617</a>     }
<a name="618" href="#618">618</a> }
</pre>
<hr/><div id="footer">This page was automatically generated by <a href="http://maven.apache.org/">Maven</a></div></body>
</html>