📄 crawljob.html
字号:
<a name="400" href="#400">400</a> <span class="string">" is corrupt (length is zero)"</span>);<a name="401" href="#401">401</a> }<a name="402" href="#402">402</a> <a name="403" href="#403">403</a> <em class="comment">// Open file. Read data and set up class variables accordingly...</em><a name="404" href="#404">404</a> BufferedReader jobReader =<a name="405" href="#405">405</a> <strong>new</strong> BufferedReader(<strong>new</strong> FileReader(jobFile), 4096);<a name="406" href="#406">406</a> <em class="comment">// UID</em><a name="407" href="#407">407</a> <strong>this</strong>.UID = jobReader.readLine();<a name="408" href="#408">408</a> <em class="comment">// name</em><a name="409" href="#409">409</a> <strong>this</strong>.name = jobReader.readLine();<a name="410" href="#410">410</a> <em class="comment">// status</em><a name="411" href="#411">411</a> <strong>this</strong>.status = jobReader.readLine();<a name="412" href="#412">412</a> <strong>if</strong>(status.equals(STATUS_ABORTED)==false<a name="413" href="#413">413</a> && status.equals(STATUS_CREATED)==false<a name="414" href="#414">414</a> && status.equals(STATUS_DELETED)==false<a name="415" href="#415">415</a> && status.equals(STATUS_FINISHED)==false<a name="416" href="#416">416</a> && status.equals(STATUS_FINISHED_ABNORMAL)==false<a name="417" href="#417">417</a> && status.equals(STATUS_FINISHED_DATA_LIMIT)==false<a name="418" href="#418">418</a> && status.equals(STATUS_FINISHED_DOCUMENT_LIMIT)==false<a name="419" href="#419">419</a> && status.equals(STATUS_FINISHED_TIME_LIMIT)==false<a name="420" href="#420">420</a> && status.equals(STATUS_MISCONFIGURED)==false<a name="421" href="#421">421</a> && status.equals(STATUS_PAUSED)==false<a name="422" href="#422">422</a> && status.equals(STATUS_CHECKPOINTING)==false<a name="423" href="#423">423</a> && status.equals(STATUS_PENDING)==false<a name="424" href="#424">424</a> && status.equals(STATUS_RUNNING)==false<a name="425" href="#425">425</a> && status.equals(STATUS_WAITING_FOR_PAUSE)==false<a name="426" href="#426">426</a> && status.equals(STATUS_PREPARING)==false){<a name="427" href="#427">427</a> <em class="comment">// status is invalid. Must be one of the above</em><a name="428" href="#428">428</a> <strong>throw</strong> <strong>new</strong> <a href="../../../../org/archive/crawler/admin/InvalidJobFileException.html">InvalidJobFileException</a>(<span class="string">"Status (line 3) in job file "</span> +<a name="429" href="#429">429</a> <span class="string">"is not valid: '"</span> + status + <span class="string">"'"</span>);<a name="430" href="#430">430</a> }<a name="431" href="#431">431</a> <em class="comment">// isReadOnly</em><a name="432" href="#432">432</a> String tmp = jobReader.readLine();<a name="433" href="#433">433</a> <strong>if</strong>(tmp.equals(<span class="string">"true"</span>)){<a name="434" href="#434">434</a> isReadOnly = <strong>true</strong>;<a name="435" href="#435">435</a> } <strong>else</strong> <strong>if</strong>(tmp.equals(<span class="string">"false"</span>)){<a name="436" href="#436">436</a> isReadOnly = false;<a name="437" href="#437">437</a> } <strong>else</strong> {<a name="438" href="#438">438</a> <strong>throw</strong> <strong>new</strong> <a href="../../../../org/archive/crawler/admin/InvalidJobFileException.html">InvalidJobFileException</a>(<span class="string">"isReadOnly (line 4) in job"</span> +<a name="439" href="#439">439</a> <span class="string">" file '"</span> + jobFile.getAbsolutePath() + <span class="string">"' is not "</span> +<a name="440" href="#440">440</a> <span class="string">"valid: '"</span> + tmp + <span class="string">"'"</span>);<a name="441" href="#441">441</a> }<a name="442" href="#442">442</a> <em class="comment">// isRunning</em><a name="443" href="#443">443</a> tmp = jobReader.readLine();<a name="444" href="#444">444</a> <strong>if</strong>(tmp.equals(<span class="string">"true"</span>)){<a name="445" href="#445">445</a> <strong>this</strong>.isRunning = <strong>true</strong>;<a name="446" href="#446">446</a> } <strong>else</strong> <strong>if</strong>(tmp.equals(<span class="string">"false"</span>)){<a name="447" href="#447">447</a> <strong>this</strong>.isRunning = false;<a name="448" href="#448">448</a> } <strong>else</strong> {<a name="449" href="#449">449</a> <strong>throw</strong> <strong>new</strong> <a href="../../../../org/archive/crawler/admin/InvalidJobFileException.html">InvalidJobFileException</a>(<span class="string">"isRunning (line 5) in job "</span> +<a name="450" href="#450">450</a> <span class="string">"file '"</span> + jobFile.getAbsolutePath() + <span class="string">"' is not valid: "</span> +<a name="451" href="#451">451</a> <span class="string">"'"</span> + tmp + <span class="string">"'"</span>);<a name="452" href="#452">452</a> }<a name="453" href="#453">453</a> <em class="comment">// priority</em><a name="454" href="#454">454</a> tmp = jobReader.readLine();<a name="455" href="#455">455</a> <strong>try</strong>{<a name="456" href="#456">456</a> <strong>this</strong>.priority = Integer.parseInt(tmp);<a name="457" href="#457">457</a> } <strong>catch</strong>(NumberFormatException e){<a name="458" href="#458">458</a> <strong>throw</strong> <strong>new</strong> <a href="../../../../org/archive/crawler/admin/InvalidJobFileException.html">InvalidJobFileException</a>(<span class="string">"priority (line 5) in job "</span> +<a name="459" href="#459">459</a> <span class="string">"file '"</span> + jobFile.getAbsolutePath() + <span class="string">"' is not valid: "</span> +<a name="460" href="#460">460</a> <span class="string">"'"</span> + tmp + <span class="string">"'"</span>);<a name="461" href="#461">461</a> }<a name="462" href="#462">462</a> <em class="comment">// numberOfJournalEntries</em><a name="463" href="#463">463</a> tmp = jobReader.readLine();<a name="464" href="#464">464</a> <strong>try</strong>{<a name="465" href="#465">465</a> <strong>this</strong>.numberOfJournalEntries = Integer.parseInt(tmp);<a name="466" href="#466">466</a> } <strong>catch</strong>(NumberFormatException e){<a name="467" href="#467">467</a> <strong>throw</strong> <strong>new</strong> <a href="../../../../org/archive/crawler/admin/InvalidJobFileException.html">InvalidJobFileException</a>(<span class="string">"numberOfJournalEntries "</span> +<a name="468" href="#468">468</a> <span class="string">"(line 5) in job file '"</span> + jobFile.getAbsolutePath() +<a name="469" href="#469">469</a> <span class="string">"' is not valid: "</span> + <span class="string">"'"</span> + tmp + <span class="string">"'"</span>);<a name="470" href="#470">470</a> }<a name="471" href="#471">471</a> <em class="comment">// settingsHandler</em><a name="472" href="#472">472</a> tmp = jobReader.readLine();<a name="473" href="#473">473</a> <strong>try</strong> {<a name="474" href="#474">474</a> File f = <strong>new</strong> File(tmp);<a name="475" href="#475">475</a> <strong>this</strong>.settingsHandler = <strong>new</strong> <a href="../../../../org/archive/crawler/settings/XMLSettingsHandler.html">XMLSettingsHandler</a>((f.isAbsolute())?<a name="476" href="#476">476</a> f: <strong>new</strong> File(jobDir, f.getName()));<a name="477" href="#477">477</a> <strong>if</strong>(<strong>this</strong>.errorHandler != <strong>null</strong>){<a name="478" href="#478">478</a> <strong>this</strong>.settingsHandler.registerValueErrorHandler(errorHandler);<a name="479" href="#479">479</a> }<a name="480" href="#480">480</a> <strong>this</strong>.settingsHandler.initialize();<a name="481" href="#481">481</a> } <strong>catch</strong> (InvalidAttributeValueException e1) {<a name="482" href="#482">482</a> <strong>throw</strong> <strong>new</strong> <a href="../../../../org/archive/crawler/admin/InvalidJobFileException.html">InvalidJobFileException</a>(<span class="string">"Problem reading from settings "</span> +<a name="483" href="#483">483</a> <span class="string">"file ("</span> + tmp + <span class="string">") specified in job file '"</span> +<a name="484" href="#484">484</a> jobFile.getAbsolutePath() + <span class="string">"'\n"</span> + e1.getMessage());<a name="485" href="#485">485</a> }<a name="486" href="#486">486</a> <em class="comment">// Statistics tracker.</em><a name="487" href="#487">487</a> jobReader.readLine();<a name="488" href="#488">488</a> <em class="comment">// errorMessage</em><a name="489" href="#489">489</a> <em class="comment">// TODO: Multilines</em><a name="490" href="#490">490</a> tmp = jobReader.readLine();<a name="491" href="#491">491</a> errorMessage = <span class="string">""</span>;<a name="492" href="#492">492</a> <strong>while</strong>(tmp!=<strong>null</strong>){<a name="493" href="#493">493</a> errorMessage+=tmp+'\n';<a name="494" href="#494">494</a> tmp = jobReader.readLine();<a name="495" href="#495">495</a> }<a name="496" href="#496">496</a> <strong>if</strong>(errorMessage.length()==0){<a name="497" href="#497">497</a> <em class="comment">// Empty error message should be null</em><a name="498" href="#498">498</a> errorMessage = <strong>null</strong>;<a name="499" href="#499">499</a> }<a name="500" href="#500">500</a> <em class="comment">// TODO: Load stattrack if needed.</em><a name="501" href="#501">501</a> <a name="502" href="#502">502</a> <em class="comment">// TODO: This should be inside a finally block.</em><a name="503" href="#503">503</a> jobReader.close();<a name="504" href="#504">504</a> }<a name="505" href="#505">505</a> <a name="506" href="#506">506</a> <em>/**<em>*</em></em><a name="507" href="#507">507</a> <em> * Cause the job to be written to persistent storage.</em><a name="508" href="#508">508</a> <em> * This will also save the statistics tracker if it is not null and the</em><a name="509" href="#509">509</a> <em> * job status is finished (regardless of how it's finished)</em><a name="510" href="#510">510</a> <em> */</em><a name="511" href="#511">511</a> <strong>private</strong> <strong>void</strong> writeJobFile() {<a name="512" href="#512">512</a> <strong>if</strong> (isProfile) {<a name="513" href="#513">513</a> <strong>return</strong>;<a name="514" href="#514">514</a> }<a name="515" href="#515">515</a> <a name="516" href="#516">516</a> <strong>final</strong> String jobDirAbsolute = jobDir.getAbsolutePath();<a name="517" href="#517">517</a> <strong>if</strong> (!jobDir.exists() || !jobDir.canWrite()) {<a name="518" href="#518">518</a> logger.warning(<span class="string">"Can't update status on "</span> +<a name="519" href="#519">519</a> jobDirAbsolute + <span class="string">" because file does not"</span> +<a name="520" href="#520">520</a> <span class="string">" exist (or is unwriteable)"</span>);<a name="521" href="#521">521</a> <strong>return</strong>;<a name="522" href="#522">522</a> }<a name="523" href="#523">523</a> File f = <strong>new</strong> File(jobDirAbsolute, <span class="string">"state.job"</span>);<a name="524" href="#524">524</a> <a name="525" href="#525">525</a> String settingsFile = getSettingsDirectory();<a name="526" href="#526">526</a> <em class="comment">// Make settingsFile's path relative if order.xml is somewhere in the</em><a name="527" href="#527">527</a> <em class="comment">// job's directory tree</em><a name="528" href="#528">528</a> <strong>if</strong>(settingsFile.startsWith(jobDirAbsolute.concat(File.separator))) {<a name="529" href="#529">529</a> settingsFile = settingsFile.substring(jobDirAbsolute.length()+1);<a name="530" href="#530">530</a> }<a name="531" href="#531">531</a> <strong>try</strong> {<a name="532" href="#532">532</a> FileWriter jobWriter = <strong>new</strong> FileWriter(f, false);<a name="533" href="#533">533</a> <strong>try</strong> {<a name="534" href="#534">534</a> jobWriter.write(UID + <span class="string">"\n"</span>);<a name="535" href="#535">535</a> jobWriter.write(name + <span class="string">"\n"</span>);
⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -