writerpoolprocessor.html

来自「网络爬虫开源代码」· HTML 代码 · 共 732 行 · 第 1/5 页

HTML
732
字号
<a name="433" href="#433">433</a>             ((Integer)obj).intValue();<a name="434" href="#434">434</a>     }<a name="435" href="#435">435</a> <a name="436" href="#436">436</a>     <em>/**<em>*</em></em><a name="437" href="#437">437</a> <em>     * @return Returns the poolMaximumWait.</em><a name="438" href="#438">438</a> <em>     */</em><a name="439" href="#439">439</a>     <strong>public</strong> <strong>int</strong> getPoolMaximumWait() {<a name="440" href="#440">440</a>         Object obj = getAttributeUnchecked(ATTR_POOL_MAX_WAIT);<a name="441" href="#441">441</a>         <strong>return</strong> (obj == <strong>null</strong>)? WriterPool.DEFAULT_MAXIMUM_WAIT:<a name="442" href="#442">442</a>             ((Integer)obj).intValue();<a name="443" href="#443">443</a>     }<a name="444" href="#444">444</a> <a name="445" href="#445">445</a>     <strong>public</strong> String getSuffix() {<a name="446" href="#446">446</a>         Object obj = getAttributeUnchecked(ATTR_SUFFIX);<a name="447" href="#447">447</a>         String sfx = (obj == <strong>null</strong>)?<a name="448" href="#448">448</a>             WriterPoolMember.DEFAULT_SUFFIX: (String)obj;<a name="449" href="#449">449</a>         <strong>if</strong> (sfx != <strong>null</strong> &amp;&amp; sfx.trim().<a name="450" href="#450">450</a>                 equals(WriterPoolMember.HOSTNAME_VARIABLE)) {<a name="451" href="#451">451</a>             String str = <span class="string">"localhost.localdomain"</span>;<a name="452" href="#452">452</a>             <strong>try</strong> {<a name="453" href="#453">453</a>                 str = InetAddress.getLocalHost().getHostName();<a name="454" href="#454">454</a>             } <strong>catch</strong> (UnknownHostException ue) {<a name="455" href="#455">455</a>                 logger.severe(<span class="string">"Failed getHostAddress for this host: "</span> + ue);<a name="456" href="#456">456</a>             }<a name="457" href="#457">457</a>             sfx = str;<a name="458" href="#458">458</a>         }<a name="459" href="#459">459</a>         <strong>return</strong> sfx;<a name="460" href="#460">460</a>     }<a name="461" href="#461">461</a>     <a name="462" href="#462">462</a>     <strong>public</strong> <strong>long</strong> getMaxToWrite() {<a name="463" href="#463">463</a>         Object obj = getAttributeUnchecked(ATTR_MAX_BYTES_WRITTEN);<a name="464" href="#464">464</a>         <strong>return</strong> (obj == <strong>null</strong>)? 0: ((Long)obj).longValue();<a name="465" href="#465">465</a>     }<a name="466" href="#466">466</a> <a name="467" href="#467">467</a> 	<strong>public</strong> <strong>void</strong> crawlEnding(String sExitMessage) {<a name="468" href="#468">468</a> 		<strong>this</strong>.pool.close();<a name="469" href="#469">469</a> 	}<a name="470" href="#470">470</a> <a name="471" href="#471">471</a> 	<strong>public</strong> <strong>void</strong> crawlEnded(String sExitMessage) {<a name="472" href="#472">472</a>         <em class="comment">// sExitMessage is unused.</em><a name="473" href="#473">473</a> 	}<a name="474" href="#474">474</a> <a name="475" href="#475">475</a>     <em class="comment">/*<em class="comment"> (non-Javadoc)</em></em><a name="476" href="#476">476</a> <em class="comment">     * @see org.archive.crawler.event.CrawlStatusListener#crawlStarted(java.lang.String)</em><a name="477" href="#477">477</a> <em class="comment">     */</em><a name="478" href="#478">478</a>     <strong>public</strong> <strong>void</strong> crawlStarted(String message) {<a name="479" href="#479">479</a>         <em class="comment">// TODO Auto-generated method stub</em><a name="480" href="#480">480</a>     }<a name="481" href="#481">481</a>     <a name="482" href="#482">482</a>     <strong>protected</strong> String getCheckpointStateFile() {<a name="483" href="#483">483</a>     	<strong>return</strong> <strong>this</strong>.getClass().getName() + <span class="string">".state"</span>;<a name="484" href="#484">484</a>     }<a name="485" href="#485">485</a>     <a name="486" href="#486">486</a>     <strong>public</strong> <strong>void</strong> crawlCheckpoint(File checkpointDir) throws IOException {<a name="487" href="#487">487</a>         <strong>int</strong> serial = getSerialNo().get();<a name="488" href="#488">488</a>         <strong>if</strong> (<strong>this</strong>.pool.getNumActive() > 0) {<a name="489" href="#489">489</a>             <em class="comment">// If we have open active Archive files, up the serial number</em><a name="490" href="#490">490</a>             <em class="comment">// so after checkpoint, we start at one past current number and</em><a name="491" href="#491">491</a>             <em class="comment">// so the number we serialize, is one past current serialNo.</em><a name="492" href="#492">492</a>             <em class="comment">// All this serial number manipulation should be fine in here since</em><a name="493" href="#493">493</a>             <em class="comment">// we're paused checkpointing (Revisit if this assumption changes).</em><a name="494" href="#494">494</a>             serial = getSerialNo().incrementAndGet();<a name="495" href="#495">495</a>         }<a name="496" href="#496">496</a>         saveCheckpointSerialNumber(checkpointDir, serial);<a name="497" href="#497">497</a>         <em class="comment">// Close all ARCs on checkpoint.</em><a name="498" href="#498">498</a>         <strong>try</strong> {<a name="499" href="#499">499</a>             <strong>this</strong>.pool.close();<a name="500" href="#500">500</a>         } <strong>finally</strong> {<a name="501" href="#501">501</a>             <em class="comment">// Reopen on checkpoint.</em><a name="502" href="#502">502</a>             setupPool(<strong>new</strong> AtomicInteger(serial));<a name="503" href="#503">503</a>         }<a name="504" href="#504">504</a>     }<a name="505" href="#505">505</a>     <a name="506" href="#506">506</a> 	<strong>public</strong> <strong>void</strong> crawlPausing(String statusMessage) {<a name="507" href="#507">507</a>         <em class="comment">// sExitMessage is unused.</em><a name="508" href="#508">508</a> 	}<a name="509" href="#509">509</a> <a name="510" href="#510">510</a> 	<strong>public</strong> <strong>void</strong> crawlPaused(String statusMessage) {<a name="511" href="#511">511</a>         <em class="comment">// sExitMessage is unused.</em><a name="512" href="#512">512</a> 	}<a name="513" href="#513">513</a> <a name="514" href="#514">514</a> 	<strong>public</strong> <strong>void</strong> crawlResuming(String statusMessage) {<a name="515" href="#515">515</a>         <em class="comment">// sExitMessage is unused.</em><a name="516" href="#516">516</a> 	}<a name="517" href="#517">517</a> 	<a name="518" href="#518">518</a>     <strong>private</strong> <strong>void</strong> readObject(ObjectInputStream stream)<a name="519" href="#519">519</a>     throws IOException, ClassNotFoundException {<a name="520" href="#520">520</a>         stream.defaultReadObject();<a name="521" href="#521">521</a>         <a href="../../../../org/archive/io/ObjectPlusFilesInputStream.html">ObjectPlusFilesInputStream</a> coistream =<a name="522" href="#522">522</a>             (<a href="../../../../org/archive/io/ObjectPlusFilesInputStream.html">ObjectPlusFilesInputStream</a>)stream;<a name="523" href="#523">523</a>         coistream.registerFinishTask( <strong>new</strong> Runnable() {<a name="524" href="#524">524</a>             <strong>public</strong> <strong>void</strong> run() {<a name="525" href="#525">525</a>             	setupPool(<strong>new</strong> AtomicInteger());<a name="526" href="#526">526</a>             }<a name="527" href="#527">527</a>         });<a name="528" href="#528">528</a>     }<a name="529" href="#529">529</a> <a name="530" href="#530">530</a> 	<strong>protected</strong> <a href="../../../../org/archive/io/WriterPool.html">WriterPool</a> getPool() {<a name="531" href="#531">531</a> 		<strong>return</strong> pool;<a name="532" href="#532">532</a> 	}<a name="533" href="#533">533</a> <a name="534" href="#534">534</a> 	<strong>protected</strong> <strong>void</strong> setPool(<a href="../../../../org/archive/io/WriterPool.html">WriterPool</a> pool) {<a name="535" href="#535">535</a> 		<strong>this</strong>.pool = pool;<a name="536" href="#536">536</a> 	}<a name="537" href="#537">537</a> <a name="538" href="#538">538</a> 	<strong>protected</strong> <strong>long</strong> getTotalBytesWritten() {<a name="539" href="#539">539</a> 		<strong>return</strong> totalBytesWritten;<a name="540" href="#540">540</a> 	}<a name="541" href="#541">541</a> <a name="542" href="#542">542</a> 	<strong>protected</strong> <strong>void</strong> setTotalBytesWritten(<strong>long</strong> totalBytesWritten) {<a name="543" href="#543">543</a>         <strong>this</strong>.totalBytesWritten = totalBytesWritten;<a name="544" href="#544">544</a>     }<a name="545" href="#545">545</a> 	<a name="546" href="#546">546</a>     <em>/**<em>*</em></em><a name="547" href="#547">547</a> <em>     * Called out of {@link #initialTasks()} when recovering a checkpoint.</em><a name="548" href="#548">548</a> <em>     * Restore state.</em><a name="549" href="#549">549</a> <em>     */</em><a name="550" href="#550">550</a>     <strong>protected</strong> <strong>void</strong> checkpointRecover() {<a name="551" href="#551">551</a>         <strong>int</strong> serialNo = loadCheckpo<strong>int</strong>SerialNumber();<a name="552" href="#552">552</a>         <strong>if</strong> (serialNo != -1) {<a name="553" href="#553">553</a>             getSerialNo().set(serialNo);<a name="554" href="#554">554</a>         }<a name="555" href="#555">555</a>     }<a name="556" href="#556">556</a> <a name="557" href="#557">557</a>     <em>/**<em>*</em></em><a name="558" href="#558">558</a> <em>     * @return Serial number from checkpoint state file or if unreadable, -1</em><a name="559" href="#559">559</a> <em>     * (Client should check for -1).</em><a name="560" href="#560">560</a> <em>     */</em><a name="561" href="#561">561</a>     <strong>protected</strong> <strong>int</strong> loadCheckpo<strong>int</strong>SerialNumber() {<a name="562" href="#562">562</a>         <strong>int</strong> result = -1;<a name="563" href="#563">563</a>         <a name="564" href="#564">564</a>         <em class="comment">// If in recover mode, read in the Writer serial number saved</em><a name="565" href="#565">565</a>         <em class="comment">// off when we checkpointed.</em><a name="566" href="#566">566</a>         File stateFile = <strong>new</strong> File(getSettingsHandler().getOrder()<a name="567" href="#567">567</a>                 .getController().getCheckpointRecover().getDirectory(),<a name="568" href="#568">568</a>                 getCheckpointStateFile());<a name="569" href="#569">569</a>         <strong>if</strong> (!stateFile.exists()) {<a name="570" href="#570">570</a>             logger.info(stateFile.getAbsolutePath()<a name="571" href="#571">571</a>                     + <span class="string">" doesn't exist so cannot restore Writer serial number."</span>);<a name="572" href="#572">572</a>         } <strong>else</strong> {<a name="573" href="#573">573</a>             DataInputStream dis = <strong>null</strong>;<a name="574" href="#574">574</a>             <strong>try</strong> {<a name="575" href="#575">575</a>                 dis = <strong>new</strong> DataInputStream(<strong>new</strong> FileInputStream(stateFile));<a name="576" href="#576">576</a>                 result = dis.readShort();<a name="577" href="#577">577</a>             } <strong>catch</strong> (FileNotFoundException e) {<a name="578" href="#578">578</a>                 e.printStackTrace();<a name="579" href="#579">579</a>             } <strong>catch</strong> (IOException e) {

⌨️ 快捷键说明

复制代码Ctrl + C
搜索代码Ctrl + F
全屏模式F11
增大字号Ctrl + =
减小字号Ctrl + -
显示快捷键?