recoveryjournal.html

用JAVA编写的,在做实验的时候留下来的,本来想删的,但是传上来,大家分享吧

<a name="400" href="#400">400</a>                             <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a> caUri = <strong>new</strong> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a>(u, 
<a name="401" href="#401">401</a>                                     pathFromSeed, via, viaContext);
<a name="402" href="#402">402</a>                             frontier.schedule(caUri);
<a name="403" href="#403">403</a>                             
<a name="404" href="#404">404</a>                             queuedDuringRecovery =
<a name="405" href="#405">405</a>                                 frontier.queuedUriCount() - queuedAtStart;
<a name="406" href="#406">406</a>                             <strong>if</strong>(((queuedDuringRecovery + 1) %
<a name="407" href="#407">407</a>                                     ENOUGH_TO_START_CRAWLING) == 0) {
<a name="408" href="#408">408</a>                                 enough.countDown();
<a name="409" href="#409">409</a>                             }
<a name="410" href="#410">410</a>                         } <strong>catch</strong> (URIException e) {
<a name="411" href="#411">411</a>                             e.printStackTrace();
<a name="412" href="#412">412</a>                         }
<a name="413" href="#413">413</a>                     }
<a name="414" href="#414">414</a>                     <strong>if</strong>((qLines%PROGRESS_INTERVAL)==0) {
<a name="415" href="#415">415</a>                         <em class="comment">// every 1 million lines, print progress</em>
<a name="416" href="#416">416</a>                         LOGGER.info(
<a name="417" href="#417">417</a>                                 <span class="string">"through line "</span> 
<a name="418" href="#418">418</a>                                 + qLines + <span class="string">"/"</span> + lines 
<a name="419" href="#419">419</a>                                 + <span class="string">" queued count = "</span> +
<a name="420" href="#420">420</a>                                 frontier.queuedUriCount());
<a name="421" href="#421">421</a>                     }
<a name="422" href="#422">422</a>                 }
<a name="423" href="#423">423</a>             } <strong>catch</strong> (EOFException e) {
<a name="424" href="#424">424</a>                 <em class="comment">// no problem: untidy end of recovery journal</em>
<a name="425" href="#425">425</a>             } <strong>finally</strong> {
<a name="426" href="#426">426</a>             	    is.close(); 
<a name="427" href="#427">427</a>             }
<a name="428" href="#428">428</a>         } <strong>catch</strong> (IOException e) {
<a name="429" href="#429">429</a>             <em class="comment">// TODO Auto-generated catch block</em>
<a name="430" href="#430">430</a>             e.printStackTrace();
<a name="431" href="#431">431</a>         }
<a name="432" href="#432">432</a>         LOGGER.info(<span class="string">"finished recovering frontier from "</span>+source+<span class="string">" "</span>
<a name="433" href="#433">433</a>                 +qLines+<span class="string">" lines processed"</span>);
<a name="434" href="#434">434</a>         enough.countDown();
<a name="435" href="#435">435</a>     }
<a name="436" href="#436">436</a> 
<a name="437" href="#437">437</a>     <em>/**<em>*</em></em>
<a name="438" href="#438">438</a> <em>     * Return an array of the subsequences of the passed-in sequence,</em>
<a name="439" href="#439">439</a> <em>     * split on space runs. </em>
<a name="440" href="#440">440</a> <em>     * </em>
<a name="441" href="#441">441</a> <em>     * @param read</em>
<a name="442" href="#442">442</a> <em>     * @return CharSequence.</em>
<a name="443" href="#443">443</a> <em>     */</em>
<a name="444" href="#444">444</a>     <strong>private</strong> <strong>static</strong> CharSequence[] splitOnSpaceRuns(CharSequence read) {
<a name="445" href="#445">445</a>         <strong>int</strong> lastStart = 0;
<a name="446" href="#446">446</a>         ArrayList segs = <strong>new</strong> ArrayList(5);
<a name="447" href="#447">447</a>         <strong>int</strong> i;
<a name="448" href="#448">448</a>         <strong>for</strong>(i=0;i&lt;read.length();i++) {
<a name="449" href="#449">449</a>             <strong>if</strong> (read.charAt(i)==' ') {
<a name="450" href="#450">450</a>                 segs.add(read.subSequence(lastStart,i));
<a name="451" href="#451">451</a>                 i++;
<a name="452" href="#452">452</a>                 <strong>while</strong>(i &lt; read.length() &amp;&amp; read.charAt(i)==' ') {
<a name="453" href="#453">453</a>                     <em class="comment">// skip any space runs</em>
<a name="454" href="#454">454</a>                     i++;
<a name="455" href="#455">455</a>                 }
<a name="456" href="#456">456</a>                 lastStart = i;
<a name="457" href="#457">457</a>             }
<a name="458" href="#458">458</a>         }
<a name="459" href="#459">459</a>         <strong>if</strong>(lastStart&lt;read.length()) {
<a name="460" href="#460">460</a>             segs.add(read.subSequence(lastStart,i));
<a name="461" href="#461">461</a>         }
<a name="462" href="#462">462</a>         <strong>return</strong> (CharSequence[]) segs.toArray(<strong>new</strong> CharSequence[segs.size()]);        
<a name="463" href="#463">463</a>     }
<a name="464" href="#464">464</a> 
<a name="465" href="#465">465</a>     <em>/**<em>*</em></em>
<a name="466" href="#466">466</a> <em>     * @param source</em>
<a name="467" href="#467">467</a> <em>     * @return Recover log buffered reader.</em>
<a name="468" href="#468">468</a> <em>     * @throws IOException</em>
<a name="469" href="#469">469</a> <em>     */</em>
<a name="470" href="#470">470</a>     <strong>public</strong> <strong>static</strong> BufferedReader getBufferedReader(File source)
<a name="471" href="#471">471</a>     throws IOException {
<a name="472" href="#472">472</a>         <strong>boolean</strong> isGzipped = source.getName().toLowerCase().
<a name="473" href="#473">473</a>             endsWith(GZIP_SUFFIX);
<a name="474" href="#474">474</a>         FileInputStream fis = <strong>new</strong> FileInputStream(source);
<a name="475" href="#475">475</a>         <strong>return</strong> <strong>new</strong> BufferedReader(isGzipped?
<a name="476" href="#476">476</a>             <strong>new</strong> InputStreamReader(<strong>new</strong> GZIPInputStream(fis)):
<a name="477" href="#477">477</a>             <strong>new</strong> InputStreamReader(fis));   
<a name="478" href="#478">478</a>     }
<a name="479" href="#479">479</a> 
<a name="480" href="#480">480</a>     <em>/**<em>*</em></em>
<a name="481" href="#481">481</a> <em>     * Get a BufferedInputStream on the recovery file given.</em>
<a name="482" href="#482">482</a> <em>     *</em>
<a name="483" href="#483">483</a> <em>     * @param source file to open</em>
<a name="484" href="#484">484</a> <em>     * @return Recover log buffered input stream.</em>
<a name="485" href="#485">485</a> <em>     * @throws IOException</em>
<a name="486" href="#486">486</a> <em>     */</em>
<a name="487" href="#487">487</a>     <strong>public</strong> <strong>static</strong> BufferedInputStream getBufferedInput(File source)
<a name="488" href="#488">488</a>     throws IOException {
<a name="489" href="#489">489</a>         <strong>boolean</strong> isGzipped = source.getName().toLowerCase().
<a name="490" href="#490">490</a>             endsWith(GZIP_SUFFIX);
<a name="491" href="#491">491</a>         FileInputStream fis = <strong>new</strong> FileInputStream(source);
<a name="492" href="#492">492</a>         <strong>return</strong> isGzipped ? <strong>new</strong> BufferedInputStream(<strong>new</strong> GZIPInputStream(fis))
<a name="493" href="#493">493</a>                 : <strong>new</strong> BufferedInputStream(fis);
<a name="494" href="#494">494</a>     }
<a name="495" href="#495">495</a>     
<a name="496" href="#496">496</a>     <em>/**<em>*</em></em>
<a name="497" href="#497">497</a> <em>     * Flush and close the underlying IO objects.</em>
<a name="498" href="#498">498</a> <em>     */</em>
<a name="499" href="#499">499</a>     <strong>public</strong> <strong>void</strong> close() {
<a name="500" href="#500">500</a>         <strong>if</strong> (<strong>this</strong>.out == <strong>null</strong>) {
<a name="501" href="#501">501</a>             <strong>return</strong>;
<a name="502" href="#502">502</a>         }
<a name="503" href="#503">503</a>         <strong>try</strong> {
<a name="504" href="#504">504</a>             <strong>this</strong>.out.flush();
<a name="505" href="#505">505</a>             <strong>this</strong>.out.close();
<a name="506" href="#506">506</a>             <strong>this</strong>.out = <strong>null</strong>;
<a name="507" href="#507">507</a>         } <strong>catch</strong> (IOException e) {
<a name="508" href="#508">508</a>             e.printStackTrace();
<a name="509" href="#509">509</a>         }
<a name="510" href="#510">510</a>     }
<a name="511" href="#511">511</a> 
<a name="512" href="#512">512</a>     <strong>public</strong> <strong>void</strong> seriousError(String err) {
<a name="513" href="#513">513</a>         writeLine(<span class="string">"\n"</span>+LOG_ERROR+ArchiveUtils.getLog14Date()+<span class="string">" "</span>+err);
<a name="514" href="#514">514</a>     }
<a name="515" href="#515">515</a> 
<a name="516" href="#516">516</a>     <strong>public</strong> <strong>synchronized</strong> <strong>void</strong> checkpoint(<strong>final</strong> File checkpointDir)
<a name="517" href="#517">517</a>     throws IOException {
<a name="518" href="#518">518</a>         <strong>if</strong> (<strong>this</strong>.out == <strong>null</strong> || !<strong>this</strong>.gzipFile.exists()) {
<a name="519" href="#519">519</a>             <strong>return</strong>;
<a name="520" href="#520">520</a>         }
<a name="521" href="#521">521</a>         close();
<a name="522" href="#522">522</a>         <em class="comment">// Rename gzipFile with the checkpoint name as suffix.</em>
<a name="523" href="#523">523</a>         <strong>this</strong>.gzipFile.renameTo(<strong>new</strong> File(<strong>this</strong>.gzipFile.getParentFile(),
<a name="524" href="#524">524</a>                 <strong>this</strong>.gzipFile.getName() + <span class="string">"."</span> + checkpointDir.getName()));
<a name="525" href="#525">525</a>         <em class="comment">// Open new gzip file.</em>
<a name="526" href="#526">526</a>         <strong>this</strong>.out = initialize(<strong>this</strong>.gzipFile);
<a name="527" href="#527">527</a>     }
<a name="528" href="#528">528</a> }
</pre>
<hr/><div id="footer">This page was automatically generated by <a href="http://maven.apache.org/">Maven</a></div></body>
</html>