xmlsettingshandler.html

来自「网络爬虫开源代码」· HTML 代码 · 共 568 行 · 第 1/4 页

HTML
568
字号
<a name="418" href="#418">418</a>                 }<a name="419" href="#419">419</a>             }<a name="420" href="#420">420</a>         }<a name="421" href="#421">421</a>         <strong>return</strong> confirmedSubDomains;<a name="422" href="#422">422</a>     }<a name="423" href="#423">423</a> <a name="424" href="#424">424</a>     <em>/**<em>*</em></em><a name="425" href="#425">425</a> <em>     * Checks if a file is a a 'per host' override or if it's a directory if it</em><a name="426" href="#426">426</a> <em>     * or it's subdirectories  contains a 'per host' override file.</em><a name="427" href="#427">427</a> <em>     * @param f The file or directory to check</em><a name="428" href="#428">428</a> <em>     * @return True if the file is an override or it's a directory that contains</em><a name="429" href="#429">429</a> <em>     *         such a file.</em><a name="430" href="#430">430</a> <em>     */</em><a name="431" href="#431">431</a>     <strong>private</strong> <strong>boolean</strong> isOverride(File f){<a name="432" href="#432">432</a>         <strong>if</strong>(f.isDirectory()){<a name="433" href="#433">433</a>             <em class="comment">// Have a directory, check it's contents.</em><a name="434" href="#434">434</a>             File[] subs = f.listFiles();<a name="435" href="#435">435</a>             <strong>for</strong>(<strong>int</strong> i=0 ; i &lt; subs.length ; i++){<a name="436" href="#436">436</a>                 <strong>if</strong>(isOverride(subs[i])){<a name="437" href="#437">437</a>                     <em class="comment">// Found one. Can stop looking.</em><a name="438" href="#438">438</a>                     <strong>return</strong> <strong>true</strong>;<a name="439" href="#439">439</a>                 }<a name="440" href="#440">440</a>             }<a name="441" href="#441">441</a>         } <strong>else</strong> <strong>if</strong> (f.getName().equals(<a name="442" href="#442">442</a>                 settingsFilename + <span class="string">"."</span> + settingsFilenameSuffix)) {<a name="443" href="#443">443</a>             <em class="comment">// This is an override file (or sure looks like one in any case).</em><a name="444" href="#444">444</a>             <strong>return</strong> <strong>true</strong>;<a name="445" href="#445">445</a>         }<a name="446" href="#446">446</a>         <em class="comment">// Didn't find an override.</em><a name="447" href="#447">447</a>         <strong>return</strong> false;<a name="448" href="#448">448</a>     }<a name="449" href="#449">449</a> <a name="450" href="#450">450</a>     <em>/**<em>* Delete a settings object from persistent storage.</em></em><a name="451" href="#451">451</a> <em>     *</em><a name="452" href="#452">452</a> <em>     * Deletes the file represented by the submitted settings object. All empty</em><a name="453" href="#453">453</a> <em>     * directories that are parents to the files path are also deleted.</em><a name="454" href="#454">454</a> <em>     *</em><a name="455" href="#455">455</a> <em>     * @param settings the settings object to delete.</em><a name="456" href="#456">456</a> <em>     */</em><a name="457" href="#457">457</a>     <strong>public</strong> <strong>void</strong> deleteSettingsObject(<a href="../../../../org/archive/crawler/settings/CrawlerSettings.html">CrawlerSettings</a> settings) {<a name="458" href="#458">458</a>         <strong>super</strong>.deleteSettingsObject(settings);<a name="459" href="#459">459</a>         File settingsDirectory = getSettingsDirectory();<a name="460" href="#460">460</a>         File settingsFile = settingsToFilename(settings);<a name="461" href="#461">461</a> <a name="462" href="#462">462</a>         settingsFile.delete();<a name="463" href="#463">463</a>         settingsFile = settingsFile.getParentFile();<a name="464" href="#464">464</a>         <strong>while</strong> (settingsFile.isDirectory() &amp;&amp; settingsFile.list().length == 0<a name="465" href="#465">465</a>                 &amp;&amp; !settingsFile.equals(settingsDirectory)) {<a name="466" href="#466">466</a>             settingsFile.delete();<a name="467" href="#467">467</a>             settingsFile = settingsFile.getParentFile();<a name="468" href="#468">468</a>         }<a name="469" href="#469">469</a>     }<a name="470" href="#470">470</a> <a name="471" href="#471">471</a>     <em class="comment">/*<em class="comment"> (non-Javadoc)</em></em><a name="472" href="#472">472</a> <em class="comment">     * @see org.archive.crawler.settings.SettingsHandler#getListOfAllFiles()</em><a name="473" href="#473">473</a> <em class="comment">     */</em><a name="474" href="#474">474</a>     <strong>public</strong> List&lt;String> getListOfAllFiles() {<a name="475" href="#475">475</a>         ArrayList&lt;String> list = <strong>new</strong> ArrayList&lt;String>();<a name="476" href="#476">476</a>         <em class="comment">// Add CrawlOrder.</em><a name="477" href="#477">477</a>         list.add(getOrderFile().getAbsolutePath());<a name="478" href="#478">478</a>         <em class="comment">// Iterate through the entire override hierarchy</em><a name="479" href="#479">479</a>         <strong>if</strong> (getSettingsDirectory().exists()) {<a name="480" href="#480">480</a>             recursiveFindFiles(getSettingsDirectory(),list);<a name="481" href="#481">481</a>         }<a name="482" href="#482">482</a>         <em class="comment">// Get files used by settings modules.</em><a name="483" href="#483">483</a>         recursiveFindSecondaryFiles(getOrder(),list);<a name="484" href="#484">484</a>         <strong>return</strong> list;<a name="485" href="#485">485</a>     }<a name="486" href="#486">486</a> <a name="487" href="#487">487</a>     <em>/**<em>*</em></em><a name="488" href="#488">488</a> <em>     * Add any files being used by any of the Modules making up the settings to</em><a name="489" href="#489">489</a> <em>     * the list.</em><a name="490" href="#490">490</a> <em>     *</em><a name="491" href="#491">491</a> <em>     * @param mbean A ModuleType to interrogate for files. Any child modules</em><a name="492" href="#492">492</a> <em>     *           will be recursively interrogated.</em><a name="493" href="#493">493</a> <em>     * @param list The list to add found files to.</em><a name="494" href="#494">494</a> <em>     */</em><a name="495" href="#495">495</a>     <strong>private</strong> <strong>void</strong> recursiveFindSecondaryFiles(<a href="../../../../org/archive/crawler/settings/ComplexType.html">ComplexType</a> mbean, <a name="496" href="#496">496</a>             ArrayList&lt;String> list) {<a name="497" href="#497">497</a>         MBeanInfo info = mbean.getMBeanInfo();<a name="498" href="#498">498</a>         MBeanAttributeInfo[] a = info.getAttributes();<a name="499" href="#499">499</a>         <em class="comment">// Interrogate the current module</em><a name="500" href="#500">500</a>         <strong>if</strong>(mbean instanceof <a href="../../../../org/archive/crawler/settings/ModuleType.html">ModuleType</a>){<a name="501" href="#501">501</a>             ((<a href="../../../../org/archive/crawler/settings/ModuleType.html">ModuleType</a>)mbean).listUsedFiles(list);<a name="502" href="#502">502</a>         }<a name="503" href="#503">503</a> <a name="504" href="#504">504</a>         <em class="comment">// Recursively interrogate all sub modules that are of ModuleType</em><a name="505" href="#505">505</a>         <strong>for</strong>(<strong>int</strong> n=0; n&lt;a.length; n++) {<a name="506" href="#506">506</a>             <strong>if</strong>(a[n] == <strong>null</strong>) {<a name="507" href="#507">507</a>                 <em class="comment">// Error null attribute.</em><a name="508" href="#508">508</a>             } <strong>else</strong> {<a name="509" href="#509">509</a>                 <a href="../../../../org/archive/crawler/settings/ModuleAttributeInfo.html">ModuleAttributeInfo</a> att = (ModuleAttributeInfo)a[n];<a name="510" href="#510">510</a>                 Object currentAttribute;<a name="511" href="#511">511</a>                 <strong>try</strong> {<a name="512" href="#512">512</a>                     currentAttribute = mbean.getAttribute(att.getName());<a name="513" href="#513">513</a>                     <strong>if</strong>(currentAttribute instanceof <a href="../../../../org/archive/crawler/settings/ComplexType.html">ComplexType</a>) {<a name="514" href="#514">514</a>                         recursiveFindSecondaryFiles((ComplexType)currentAttribute,list);<a name="515" href="#515">515</a>                     }<a name="516" href="#516">516</a>                 } <strong>catch</strong> (AttributeNotFoundException e) {<a name="517" href="#517">517</a>                     <em class="comment">// TODO Auto-generated catch block</em><a name="518" href="#518">518</a>                     e.printStackTrace();<a name="519" href="#519">519</a>                 } <strong>catch</strong> (MBeanException e) {<a name="520" href="#520">520</a>                     <em class="comment">// TODO Auto-generated catch block</em><a name="521" href="#521">521</a>                     e.printStackTrace();<a name="522" href="#522">522</a>                 } <strong>catch</strong> (ReflectionException e) {<a name="523" href="#523">523</a>                     <em class="comment">// TODO Auto-generated catch block</em><a name="524" href="#524">524</a>                     e.printStackTrace();<a name="525" href="#525">525</a>                 }<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>     <em>/**<em>*</em></em><a name="531" href="#531">531</a> <em>     * Starting at the specific directory this method will iterate through all</em><a name="532" href="#532">532</a> <em>     * sub directories and add each file (as absolute name, with path as a</em><a name="533" href="#533">533</a> <em>     * string) to the provided ArrayList. Any file found under the settings</em><a name="534" href="#534">534</a> <em>     * directory with the proper suffix will be considered valid and added to</em><a name="535" href="#535">535</a> <em>     * the list.</em><a name="536" href="#536">536</a> <em>     * @param dir Starting directory</em><a name="537" href="#537">537</a> <em>     * @param list The list to add to</em><a name="538" href="#538">538</a> <em>     */</em><a name="539" href="#539">539</a>     <strong>private</strong> <strong>void</strong> recursiveFindFiles(File dir, ArrayList&lt;String> list){<a name="540" href="#540">540</a>         File[] subs = dir.listFiles();<a name="541" href="#541">541</a>         <strong>if</strong> (subs != <strong>null</strong>) {<a name="542" href="#542">542</a>             <strong>for</strong>(<strong>int</strong> i=0 ; i &lt; subs.length ; i++){<a name="543" href="#543">543</a>                 <strong>if</strong>(subs[i].isDirectory()){<a name="544" href="#544">544</a>                     recursiveFindFiles(subs[i],list);<a name="545" href="#545">545</a>                 } <strong>else</strong> {<a name="546" href="#546">546</a>                     <strong>if</strong>(subs[i].getName().endsWith(settingsFilenameSuffix)){<a name="547" href="#547">547</a>                         <em class="comment">// Add it to list</em><a name="548" href="#548">548</a>                         list.add(subs[i].getAbsolutePath());<a name="549" href="#549">549</a>                     }<a name="550" href="#550">550</a>                 }<a name="551" href="#551">551</a>             }<a name="552" href="#552">552</a>         }<a name="553" href="#553">553</a>     }<a name="554" href="#554">554</a> }</pre><hr/><div id="footer">This page was automatically generated by <a href="http://maven.apache.org/">Maven</a></div></body></html>

⌨️ 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?