⭐ 欢迎来到虫虫下载站! | 📦 资源下载 📁 资源专辑 ℹ️ 关于我们
⭐ 虫虫下载站

📄 crawljobhandler.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
💻 HTML
📖 第 1 页 / 共 5 页
字号:
<a name="292" href="#292">292</a>             <strong>this</strong>.completedCrawlJobs.add(cjob);<a name="293" href="#293">293</a>         } <strong>else</strong> <strong>if</strong>( cjob.getStatus().equals(CrawlJob.STATUS_PENDING) ) {<a name="294" href="#294">294</a>             <em class="comment">// Was a pending job.</em><a name="295" href="#295">295</a>             <strong>this</strong>.pendingCrawlJobs.add(cjob);<a name="296" href="#296">296</a>         } <strong>else</strong> <strong>if</strong>( cjob.getStatus().equals(CrawlJob.STATUS_CREATED)<a name="297" href="#297">297</a>                 || cjob.getStatus().equals(CrawlJob.STATUS_DELETED) ) {<a name="298" href="#298">298</a>             <em class="comment">// Ignore for now. TODO: Add to 'recycle bin'</em><a name="299" href="#299">299</a>         } <strong>else</strong> {<a name="300" href="#300">300</a>             <em class="comment">// Must have been completed.</em><a name="301" href="#301">301</a>             <strong>this</strong>.completedCrawlJobs.add(cjob);<a name="302" href="#302">302</a>         }<a name="303" href="#303">303</a>     }<a name="304" href="#304">304</a> <a name="305" href="#305">305</a>     <em>/**<em>*</em></em><a name="306" href="#306">306</a> <em>     * Looks in conf dir for a profiles dir.</em><a name="307" href="#307">307</a> <em>     * @return the directory where profiles are stored else null if none</em><a name="308" href="#308">308</a> <em>     * available</em><a name="309" href="#309">309</a> <em>     * @throws IOException</em><a name="310" href="#310">310</a> <em>     */</em><a name="311" href="#311">311</a>     <strong>private</strong> File getProfilesDirectory() throws IOException {<a name="312" href="#312">312</a>         <strong>return</strong> (Heritrix.getConfdir(false) == <strong>null</strong>)? <strong>null</strong>:<a name="313" href="#313">313</a>             <strong>new</strong> File(Heritrix.getConfdir().getAbsolutePath(),<a name="314" href="#314">314</a>                 PROFILES_DIR_NAME);<a name="315" href="#315">315</a>     }<a name="316" href="#316">316</a> <a name="317" href="#317">317</a>     <em>/**<em>*</em></em><a name="318" href="#318">318</a> <em>     * Loads the default profile and all other profiles found on disk.</em><a name="319" href="#319">319</a> <em>     */</em><a name="320" href="#320">320</a>     <strong>private</strong> <strong>void</strong> loadProfiles() {<a name="321" href="#321">321</a>         <strong>boolean</strong> loadedDefault = false;<a name="322" href="#322">322</a>         File profileDir = <strong>null</strong>;<a name="323" href="#323">323</a> 		<strong>try</strong> {<a name="324" href="#324">324</a> 			profileDir = getProfilesDirectory();<a name="325" href="#325">325</a> 		} <strong>catch</strong> (IOException e) {<a name="326" href="#326">326</a> 			e.printStackTrace();<a name="327" href="#327">327</a> 		}<a name="328" href="#328">328</a> 		<strong>if</strong> (profileDir != <strong>null</strong>) {<a name="329" href="#329">329</a>             File[] ps = profileDir.listFiles();<a name="330" href="#330">330</a>             <strong>if</strong> (ps != <strong>null</strong> &amp;&amp; ps.length > 0) {<a name="331" href="#331">331</a>                 <strong>for</strong> (<strong>int</strong> i = 0; i &lt; ps.length; i++) {<a name="332" href="#332">332</a>                     File f = ps[i];<a name="333" href="#333">333</a>                     <strong>if</strong> (f.isDirectory()) {<a name="334" href="#334">334</a>                         <em class="comment">// Each directory in the profiles directory should</em><a name="335" href="#335">335</a>                         <em class="comment">// contain the file order.xml.</em><a name="336" href="#336">336</a>                         File profile = <strong>new</strong> File(f, ORDER_FILE_NAME);<a name="337" href="#337">337</a>                         <strong>if</strong> (profile.canRead()) {<a name="338" href="#338">338</a>                             <strong>boolean</strong> b = loadProfile(profile);<a name="339" href="#339">339</a>                             <strong>if</strong> (b) {<a name="340" href="#340">340</a>                                 loadedDefault = b;<a name="341" href="#341">341</a>                             }<a name="342" href="#342">342</a>                         }<a name="343" href="#343">343</a>                     }<a name="344" href="#344">344</a>                 }<a name="345" href="#345">345</a>             }<a name="346" href="#346">346</a>         }<a name="347" href="#347">347</a>         <em class="comment">// Now add in the default profile.  Its on the CLASSPATH and needs</em><a name="348" href="#348">348</a>         <em class="comment">// special handling.  Don't add if already a default present.</em><a name="349" href="#349">349</a>         String parent = File.separator + PROFILES_DIR_NAME + File.separator;<a name="350" href="#350">350</a>         <strong>if</strong> (!loadedDefault) {<a name="351" href="#351">351</a>             loadProfile(<strong>new</strong> File(parent + DEFAULT_PROFILE, ORDER_FILE_NAME));<a name="352" href="#352">352</a>         }<a name="353" href="#353">353</a>         <em class="comment">// Look to see if a default profile system property has been</em><a name="354" href="#354">354</a>         <em class="comment">// supplied. If so, use it instead.</em><a name="355" href="#355">355</a>         <em class="comment">// TODO: Try and read default profile from some permanent storage.</em><a name="356" href="#356">356</a>         defaultProfile = DEFAULT_PROFILE;<a name="357" href="#357">357</a>     }<a name="358" href="#358">358</a>     <a name="359" href="#359">359</a>     <em>/**<em>*</em></em><a name="360" href="#360">360</a> <em>     * Load one profile.</em><a name="361" href="#361">361</a> <em>     * @param profile Profile to load.</em><a name="362" href="#362">362</a> <em>     * @return True if loaded profile was the default profile.</em><a name="363" href="#363">363</a> <em>     */</em><a name="364" href="#364">364</a>     <strong>protected</strong> <strong>boolean</strong> loadProfile(File profile) {<a name="365" href="#365">365</a>         <strong>boolean</strong> loadedDefault = false;<a name="366" href="#366">366</a>         <em class="comment">// Ok, got the order file for this profile.</em><a name="367" href="#367">367</a>         <strong>try</strong> {<a name="368" href="#368">368</a>             <em class="comment">// The directory name denotes the profiles UID and name.</em><a name="369" href="#369">369</a>             <a href="../../../../org/archive/crawler/settings/XMLSettingsHandler.html">XMLSettingsHandler</a> newSettingsHandler =<a name="370" href="#370">370</a>                 <strong>new</strong> <a href="../../../../org/archive/crawler/settings/XMLSettingsHandler.html">XMLSettingsHandler</a>(profile);<a name="371" href="#371">371</a>             <a href="../../../../org/archive/crawler/admin/CrawlJobErrorHandler.html">CrawlJobErrorHandler</a> cjseh =<a name="372" href="#372">372</a>                 <strong>new</strong> <a href="../../../../org/archive/crawler/admin/CrawlJobErrorHandler.html">CrawlJobErrorHandler</a>(Level.SEVERE);<a name="373" href="#373">373</a>             newSettingsHandler.<a name="374" href="#374">374</a>                 setErrorReportingLevel(cjseh.getLevel());<a name="375" href="#375">375</a>             newSettingsHandler.initialize();<a name="376" href="#376">376</a>             addProfile(<strong>new</strong> <a href="../../../../org/archive/crawler/admin/CrawlJob.html">CrawlJob</a>(profile.getParentFile().getName(),<a name="377" href="#377">377</a>                 newSettingsHandler, cjseh));<a name="378" href="#378">378</a>             loadedDefault = profile.getParentFile().getName().<a name="379" href="#379">379</a>                 equals(DEFAULT_PROFILE);<a name="380" href="#380">380</a>         } <strong>catch</strong> (InvalidAttributeValueException e) {<a name="381" href="#381">381</a>             System.err.println(<span class="string">"Failed to load profile '"</span> +<a name="382" href="#382">382</a>                     profile.getParentFile().getName() +<a name="383" href="#383">383</a>                     <span class="string">"'. InvalidAttributeValueException."</span>);<a name="384" href="#384">384</a>         }<a name="385" href="#385">385</a>         <strong>return</strong> loadedDefault;<a name="386" href="#386">386</a>     }<a name="387" href="#387">387</a> <a name="388" href="#388">388</a>     <em>/**<em>*</em></em><a name="389" href="#389">389</a> <em>     * Add a new profile</em><a name="390" href="#390">390</a> <em>     * @param profile The new profile</em><a name="391" href="#391">391</a> <em>     */</em><a name="392" href="#392">392</a>     <strong>public</strong> <strong>synchronized</strong> <strong>void</strong> addProfile(<a href="../../../../org/archive/crawler/admin/CrawlJob.html">CrawlJob</a> profile){<a name="393" href="#393">393</a>         profileJobs.add(profile);<a name="394" href="#394">394</a>     }<a name="395" href="#395">395</a>     <a name="396" href="#396">396</a>     <strong>public</strong> <strong>synchronized</strong> <strong>void</strong> deleteProfile(<a href="../../../../org/archive/crawler/admin/CrawlJob.html">CrawlJob</a> cj) throws IOException {<a name="397" href="#397">397</a>         File d = getProfilesDirectory();<a name="398" href="#398">398</a>         File p = <strong>new</strong> File(d, cj.getJobName());<a name="399" href="#399">399</a>         <strong>if</strong> (!p.exists()) {<a name="400" href="#400">400</a>             <strong>throw</strong> <strong>new</strong> IOException(<span class="string">"No profile named "</span> + cj.getJobName() +<a name="401" href="#401">401</a>                 <span class="string">" at "</span> + d.getAbsolutePath());<a name="402" href="#402">402</a>         }<a name="403" href="#403">403</a>         FileUtils.deleteDir(p);<a name="404" href="#404">404</a>         <strong>this</strong>.profileJobs.remove(cj);<a name="405" href="#405">405</a>     }<a name="406" href="#406">406</a> <a name="407" href="#407">407</a>     <em>/**<em>*</em></em><a name="408" href="#408">408</a> <em>     * Returns a List of all known profiles.</em><a name="409" href="#409">409</a> <em>     * @return a List of all known profiles.</em><a name="410" href="#410">410</a> <em>     */</em><a name="411" href="#411">411</a>     <strong>public</strong> <strong>synchronized</strong> List getProfiles(){<a name="412" href="#412">412</a>         ArrayList tmp = <strong>new</strong> ArrayList(profileJobs.size());<a name="413" href="#413">413</a>         tmp.addAll(profileJobs);<a name="414" href="#414">414</a>         <strong>return</strong> tmp;<a name="415" href="#415">415</a>     }<a name="416" href="#416">416</a> <a name="417" href="#417">417</a>     <em>/**<em>*</em></em><a name="418" href="#418">418</a> <em>     * Submit a job to the handler. Job will be scheduled for crawling. At</em><a name="419" href="#419">419</a> <em>     * present it will not take the job's priority into consideration.</em><a name="420" href="#420">420</a> <em>     *</em><a name="421" href="#421">421</a> <em>     * @param job A new job for the handler</em><a name="422" href="#422">422</a> <em>     * @return CrawlJob that was added or null.</em><a name="423" href="#423">423</a> <em>     */</em><a name="424" href="#424">424</a>     <strong>public</strong> <a href="../../../../org/archive/crawler/admin/CrawlJob.html">CrawlJob</a> addJob(<a href="../../../../org/archive/crawler/admin/CrawlJob.html">CrawlJob</a> job) {<a name="425" href="#425">425</a>         <strong>if</strong>(job.isProfile()){<a name="426" href="#426">426</a>             <strong>return</strong> <strong>null</strong>;     <em class="comment">// Can't crawl profiles.</em><a name="427" href="#427">427</a>         }<a name="428" href="#428">428</a>         job.setStatus(CrawlJob.STATUS_PENDING);<a name="429" href="#429">429</a>         <strong>if</strong>(job.isNew()){<a name="430" href="#430">430</a>             <em class="comment">// Are adding the new job to the pending queue.</em><a name="431" href="#431">431</a>             <strong>this</strong>.newJob = <strong>null</strong>;<a name="432" href="#432">432</a>             job.setNew(false);<a name="433" href="#433">433</a>         }<a name="434" href="#434">434</a>         <strong>this</strong>.pendingCrawlJobs.add(job);<a name="435" href="#435">435</a>         <strong>if</strong>(isCrawling() == false &amp;&amp; isRunning()) {<a name="436" href="#436">436</a>             <em class="comment">// Start crawling</em><a name="437" href="#437">437</a>             startNextJob();<a name="438" href="#438">438</a>         }<a name="439" href="#439">439</a>         <strong>return</strong> job;<a name="440" href="#440">440</a>     }<a name="441" href="#441">441</a> 

⌨️ 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 + -