crawljob.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="536" href="#536">536</a>                 jobWriter.write(status + <span class="string">"\n"</span>);
<a name="537" href="#537">537</a>                 jobWriter.write(isReadOnly + <span class="string">"\n"</span>);
<a name="538" href="#538">538</a>                 jobWriter.write(isRunning + <span class="string">"\n"</span>);
<a name="539" href="#539">539</a>                 jobWriter.write(priority + <span class="string">"\n"</span>);
<a name="540" href="#540">540</a>                 jobWriter.write(numberOfJournalEntries + <span class="string">"\n"</span>);
<a name="541" href="#541">541</a>                 jobWriter.write(settingsFile + <span class="string">"\n"</span>);
<a name="542" href="#542">542</a>                 jobWriter.write(statisticsFileSave + <span class="string">"\n"</span>);<em class="comment">// TODO: Is this</em>
<a name="543" href="#543">543</a>                                                             <em class="comment">// right?</em>
<a name="544" href="#544">544</a>                 <em class="comment">// Can be multiple lines so we keep it last</em>
<a name="545" href="#545">545</a>                 <strong>if</strong> (errorMessage != <strong>null</strong>) {
<a name="546" href="#546">546</a>                     jobWriter.write(errorMessage + <span class="string">"\n"</span>);
<a name="547" href="#547">547</a>                 }
<a name="548" href="#548">548</a>             } <strong>finally</strong> {
<a name="549" href="#549">549</a>                 <strong>if</strong> (jobWriter != <strong>null</strong>) {
<a name="550" href="#550">550</a>                     jobWriter.close();
<a name="551" href="#551">551</a>                 }
<a name="552" href="#552">552</a>             }
<a name="553" href="#553">553</a>         } <strong>catch</strong> (IOException e) {
<a name="554" href="#554">554</a>             logger.log(Level.WARNING, <span class="string">"An IOException occured saving job "</span> +
<a name="555" href="#555">555</a>                     name + <span class="string">" ("</span> + UID + <span class="string">")"</span>, e);
<a name="556" href="#556">556</a>         }
<a name="557" href="#557">557</a>     }
<a name="558" href="#558">558</a>   
<a name="559" href="#559">559</a>     <em>/**<em>*</em></em>
<a name="560" href="#560">560</a> <em>     * Returns this jobs unique ID (UID) that was issued by the</em>
<a name="561" href="#561">561</a> <em>     * CrawlJobHandler() when this job was first created.</em>
<a name="562" href="#562">562</a> <em>     * </em>
<a name="563" href="#563">563</a> <em>     * @return Job This jobs UID.</em>
<a name="564" href="#564">564</a> <em>     * @see CrawlJobHandler#getNextJobUID()</em>
<a name="565" href="#565">565</a> <em>     */</em>
<a name="566" href="#566">566</a>     <strong>public</strong> String getUID(){
<a name="567" href="#567">567</a>         <strong>return</strong> UID;
<a name="568" href="#568">568</a>     }
<a name="569" href="#569">569</a> 
<a name="570" href="#570">570</a>     <em>/**<em>*</em></em>
<a name="571" href="#571">571</a> <em>     * Returns this job's 'name'. The name comes from the settings for this job,</em>
<a name="572" href="#572">572</a> <em>     * need not be unique and may change. For a unique identifier use</em>
<a name="573" href="#573">573</a> <em>     * {@link #getUID() getUID()}.</em>
<a name="574" href="#574">574</a> <em>     * &lt;p></em>
<a name="575" href="#575">575</a> <em>     * The name corrisponds to the value of the 'name' tag in the 'meta' section</em>
<a name="576" href="#576">576</a> <em>     * of the settings file.</em>
<a name="577" href="#577">577</a> <em>     *</em>
<a name="578" href="#578">578</a> <em>     * @return This job's 'name'</em>
<a name="579" href="#579">579</a> <em>     */</em>
<a name="580" href="#580">580</a>     <strong>public</strong> String getJobName(){
<a name="581" href="#581">581</a>         <strong>return</strong> name;
<a name="582" href="#582">582</a>     }
<a name="583" href="#583">583</a> 
<a name="584" href="#584">584</a>     <em>/**<em>*</em></em>
<a name="585" href="#585">585</a> <em>     * Return the combination of given name and UID most commonly</em>
<a name="586" href="#586">586</a> <em>     * used in administrative interface.</em>
<a name="587" href="#587">587</a> <em>     *</em>
<a name="588" href="#588">588</a> <em>     * @return Job's name with UID notation</em>
<a name="589" href="#589">589</a> <em>     */</em>
<a name="590" href="#590">590</a>     <strong>public</strong> String getDisplayName() {
<a name="591" href="#591">591</a>         <strong>return</strong> getJobName()+<span class="string">" ["</span>+getUID()+<span class="string">"]"</span>;
<a name="592" href="#592">592</a>     }
<a name="593" href="#593">593</a> 
<a name="594" href="#594">594</a>     <em>/**<em>*</em></em>
<a name="595" href="#595">595</a> <em>     * Set this job's level of priority.</em>
<a name="596" href="#596">596</a> <em>     *</em>
<a name="597" href="#597">597</a> <em>     * @param priority The level of priority</em>
<a name="598" href="#598">598</a> <em>     *</em>
<a name="599" href="#599">599</a> <em>     * @see #getJobPriority()</em>
<a name="600" href="#600">600</a> <em>     * @see #PRIORITY_MINIMAL</em>
<a name="601" href="#601">601</a> <em>     * @see #PRIORITY_LOW</em>
<a name="602" href="#602">602</a> <em>     * @see #PRIORITY_AVERAGE</em>
<a name="603" href="#603">603</a> <em>     * @see #PRIORITY_HIGH</em>
<a name="604" href="#604">604</a> <em>     * @see #PRIORITY_CRITICAL</em>
<a name="605" href="#605">605</a> <em>     */</em>
<a name="606" href="#606">606</a>     <strong>public</strong> <strong>void</strong> setJobPriority(<strong>int</strong> priority) {
<a name="607" href="#607">607</a>         <strong>this</strong>.priority = priority;
<a name="608" href="#608">608</a>     }
<a name="609" href="#609">609</a> 
<a name="610" href="#610">610</a>     <em>/**<em>*</em></em>
<a name="611" href="#611">611</a> <em>     * Get this job's level of priority.</em>
<a name="612" href="#612">612</a> <em>     *</em>
<a name="613" href="#613">613</a> <em>     * @return this job's priority</em>
<a name="614" href="#614">614</a> <em>     * @see #setJobPriority(int)</em>
<a name="615" href="#615">615</a> <em>     * @see #PRIORITY_MINIMAL</em>
<a name="616" href="#616">616</a> <em>     * @see #PRIORITY_LOW</em>
<a name="617" href="#617">617</a> <em>     * @see #PRIORITY_AVERAGE</em>
<a name="618" href="#618">618</a> <em>     * @see #PRIORITY_HIGH</em>
<a name="619" href="#619">619</a> <em>     * @see #PRIORITY_CRITICAL</em>
<a name="620" href="#620">620</a> <em>     */</em>
<a name="621" href="#621">621</a>     <strong>public</strong> <strong>int</strong> getJobPriority() {
<a name="622" href="#622">622</a>         <strong>return</strong> priority;
<a name="623" href="#623">623</a>     }
<a name="624" href="#624">624</a> 
<a name="625" href="#625">625</a>     <em>/**<em>*</em></em>
<a name="626" href="#626">626</a> <em>     * Once called no changes can be made to the settings for this job.</em>
<a name="627" href="#627">627</a> <em>     * Typically this is done once a crawl is completed and further changes</em>
<a name="628" href="#628">628</a> <em>     * to the crawl order are therefor meaningless.</em>
<a name="629" href="#629">629</a> <em>     */</em>
<a name="630" href="#630">630</a>     <strong>public</strong> <strong>void</strong> setReadOnly() {
<a name="631" href="#631">631</a>         isReadOnly = <strong>true</strong>;
<a name="632" href="#632">632</a>         writeJobFile(); <em class="comment">//Save changes</em>
<a name="633" href="#633">633</a>     }
<a name="634" href="#634">634</a> 
<a name="635" href="#635">635</a>     <em>/**<em>*</em></em>
<a name="636" href="#636">636</a> <em>     * Is job read only?</em>
<a name="637" href="#637">637</a> <em>     * @return false until setReadOnly has been invoked, after that it returns true.</em>
<a name="638" href="#638">638</a> <em>     */</em>
<a name="639" href="#639">639</a>     <strong>public</strong> <strong>boolean</strong> isReadOnly(){
<a name="640" href="#640">640</a>         <strong>return</strong> isReadOnly;
<a name="641" href="#641">641</a>     }
<a name="642" href="#642">642</a> 
<a name="643" href="#643">643</a>     <em>/**<em>*</em></em>
<a name="644" href="#644">644</a> <em>     * Set the status of this CrawlJob.</em>
<a name="645" href="#645">645</a> <em>     *</em>
<a name="646" href="#646">646</a> <em>     * @param status Current status of CrawlJob</em>
<a name="647" href="#647">647</a> <em>     *         (see constants defined here beginning with STATUS)</em>
<a name="648" href="#648">648</a> <em>     */</em>
<a name="649" href="#649">649</a>     <strong>public</strong> <strong>void</strong> setStatus(String status) {
<a name="650" href="#650">650</a>         <strong>this</strong>.status = status;
<a name="651" href="#651">651</a>         writeJobFile(); <em class="comment">//Save changes</em>
<a name="652" href="#652">652</a>         <em class="comment">// TODO: If job finished, save StatisticsTracker!</em>
<a name="653" href="#653">653</a>     }
<a name="654" href="#654">654</a> 
<a name="655" href="#655">655</a>     <em>/**<em>*</em></em>
<a name="656" href="#656">656</a> <em>     * @return Status of the crawler (Used by JMX).</em>
<a name="657" href="#657">657</a> <em>     */</em>
<a name="658" href="#658">658</a>     <strong>public</strong> String getCrawlStatus() {
<a name="659" href="#659">659</a>         <strong>return</strong> <strong>this</strong>.controller != <strong>null</strong>?
<a name="660" href="#660">660</a>             <strong>this</strong>.controller.getState().toString(): <span class="string">"Illegal State"</span>;
<a name="661" href="#661">661</a>     }
<a name="662" href="#662">662</a>     
<a name="663" href="#663">663</a>     <em>/**<em>*</em></em>
<a name="664" href="#664">664</a> <em>     * Get the current status of this CrawlJob</em>
<a name="665" href="#665">665</a> <em>     *</em>
<a name="666" href="#666">666</a> <em>     * @return The current status of this CrawlJob</em>
<a name="667" href="#667">667</a> <em>     *         (see constants defined