📄 xmlsettingshandler.html
字号:
<a name="282" href="#282">282</a> getResourceAsStream(f.getPath());<a name="283" href="#283">283</a> }<a name="284" href="#284">284</a> } <strong>else</strong> {<a name="285" href="#285">285</a> is = <strong>new</strong> FileInputStream(f);<a name="286" href="#286">286</a> }<a name="287" href="#287">287</a> <strong>if</strong> (is != <strong>null</strong>) {<a name="288" href="#288">288</a> XMLReader parser = SAXParserFactory.newInstance()<a name="289" href="#289">289</a> .newSAXParser().getXMLReader();<a name="290" href="#290">290</a> InputStream file = <strong>new</strong> BufferedInputStream(is);<a name="291" href="#291">291</a> parser.setContentHandler(<strong>new</strong> <a href="../../../../org/archive/crawler/settings/CrawlSettingsSAXHandler.html">CrawlSettingsSAXHandler</a>(settings));<a name="292" href="#292">292</a> InputSource source = <strong>new</strong> InputSource(file);<a name="293" href="#293">293</a> source.setSystemId(f.toURL().toExternalForm());<a name="294" href="#294">294</a> parser.parse(source);<a name="295" href="#295">295</a> result = settings;<a name="296" href="#296">296</a> }<a name="297" href="#297">297</a> } <strong>catch</strong> (SAXParseException e) {<a name="298" href="#298">298</a> logger.warning(e.getMessage() + <span class="string">" in '"</span> + e.getSystemId()<a name="299" href="#299">299</a> + <span class="string">"', line: "</span> + e.getLineNumber() + <span class="string">", column: "</span><a name="300" href="#300">300</a> + e.getColumnNumber());<a name="301" href="#301">301</a> } <strong>catch</strong> (SAXException e) {<a name="302" href="#302">302</a> logger.warning(e.getMessage() + <span class="string">": "</span><a name="303" href="#303">303</a> + e.getException().getMessage());<a name="304" href="#304">304</a> } <strong>catch</strong> (ParserConfigurationException e) {<a name="305" href="#305">305</a> logger.warning(e.getMessage() + <span class="string">": "</span><a name="306" href="#306">306</a> + e.getCause().getMessage());<a name="307" href="#307">307</a> } <strong>catch</strong> (FactoryConfigurationError e) {<a name="308" href="#308">308</a> logger.warning(e.getMessage() + <span class="string">": "</span><a name="309" href="#309">309</a> + e.getException().getMessage());<a name="310" href="#310">310</a> } <strong>catch</strong> (IOException e) {<a name="311" href="#311">311</a> logger.warning(<span class="string">"Could not access file '"</span><a name="312" href="#312">312</a> + f.getAbsolutePath() + <span class="string">"': "</span> + e.getMessage());<a name="313" href="#313">313</a> }<a name="314" href="#314">314</a> <strong>return</strong> result;<a name="315" href="#315">315</a> }<a name="316" href="#316">316</a> <a name="317" href="#317">317</a> <strong>protected</strong> <strong>final</strong> <a href="../../../../org/archive/crawler/settings/CrawlerSettings.html">CrawlerSettings</a> readSettingsObject(<a href="../../../../org/archive/crawler/settings/CrawlerSettings.html">CrawlerSettings</a> settings) {<a name="318" href="#318">318</a> File filename = settingsToFilename(settings);<a name="319" href="#319">319</a> <strong>return</strong> readSettingsObject(settings, filename);<a name="320" href="#320">320</a> }<a name="321" href="#321">321</a> <a name="322" href="#322">322</a> <em>/**<em>* Get the <code>File</code> object pointing to the order file.</em></em><a name="323" href="#323">323</a> <em> *</em><a name="324" href="#324">324</a> <em> * @return File object for the order file.</em><a name="325" href="#325">325</a> <em> */</em><a name="326" href="#326">326</a> <strong>public</strong> File getOrderFile() {<a name="327" href="#327">327</a> <strong>return</strong> orderFile;<a name="328" href="#328">328</a> }<a name="329" href="#329">329</a> <a name="330" href="#330">330</a> <em>/**<em>* Creates a replica of the settings file structure in another directory</em></em><a name="331" href="#331">331</a> <em> * (fully recursive, includes all per host settings). The SettingsHandler</em><a name="332" href="#332">332</a> <em> * will then refer to the new files.</em><a name="333" href="#333">333</a> <em> *</em><a name="334" href="#334">334</a> <em> * Observe that this method should only be called after the SettingsHandler</em><a name="335" href="#335">335</a> <em> * has been initialized.</em><a name="336" href="#336">336</a> <em> *</em><a name="337" href="#337">337</a> <em> * @param newOrderFileName where the new order file should be saved.</em><a name="338" href="#338">338</a> <em> * @param newSettingsDirectory the top level directory of the per host/domain</em><a name="339" href="#339">339</a> <em> * settings files.</em><a name="340" href="#340">340</a> <em> * @throws IOException</em><a name="341" href="#341">341</a> <em> */</em><a name="342" href="#342">342</a> <strong>public</strong> <strong>void</strong> copySettings(File newOrderFileName, String newSettingsDirectory)<a name="343" href="#343">343</a> throws IOException {<a name="344" href="#344">344</a> File oldSettingsDirectory = getSettingsDirectory();<a name="345" href="#345">345</a> <a name="346" href="#346">346</a> <em class="comment">// Write new orderfile and point the settingshandler to it</em><a name="347" href="#347">347</a> orderFile = newOrderFileName;<a name="348" href="#348">348</a> <strong>try</strong> {<a name="349" href="#349">349</a> getOrder().setAttribute(<a name="350" href="#350">350</a> <strong>new</strong> Attribute(<a name="351" href="#351">351</a> CrawlOrder.ATTR_SETTINGS_DIRECTORY, newSettingsDirectory));<a name="352" href="#352">352</a> } <strong>catch</strong> (Exception e) {<a name="353" href="#353">353</a> <strong>throw</strong> <strong>new</strong> IOException(<span class="string">"Could not update settings with new location: "</span><a name="354" href="#354">354</a> + e.getMessage());<a name="355" href="#355">355</a> }<a name="356" href="#356">356</a> writeSettingsObject(getSettingsObject(<strong>null</strong>));<a name="357" href="#357">357</a> <a name="358" href="#358">358</a> File newDir = getPathRelativeToWorkingDirectory(newSettingsDirectory);<a name="359" href="#359">359</a> <a name="360" href="#360">360</a> <em class="comment">// Copy the per host files if src and dest directories are different.</em><a name="361" href="#361">361</a> <strong>if</strong> (oldSettingsDirectory.compareTo(newDir) != 0) {<a name="362" href="#362">362</a> FileUtils.copyFiles(oldSettingsDirectory, newDir);<a name="363" href="#363">363</a> }<a name="364" href="#364">364</a> }<a name="365" href="#365">365</a> <a name="366" href="#366">366</a> <em>/**<em>*</em></em><a name="367" href="#367">367</a> <em> * Transforms a relative path so that it is relative to the location of the</em><a name="368" href="#368">368</a> <em> * order file. If an absolute path is given, it will be returned unchanged.<p></em><a name="369" href="#369">369</a> <em> * The location of it's order file is always considered as the 'working'</em><a name="370" href="#370">370</a> <em> * directory for any given settings.</em><a name="371" href="#371">371</a> <em> * @param path A relative path to a file (or directory)</em><a name="372" href="#372">372</a> <em> * @return The same path modified so that it is relative to the file level</em><a name="373" href="#373">373</a> <em> * location of the order file for the settings handler.</em><a name="374" href="#374">374</a> <em> */</em><a name="375" href="#375">375</a> <strong>public</strong> File getPathRelativeToWorkingDirectory(String path) {<a name="376" href="#376">376</a> File f = <strong>new</strong> File(path);<a name="377" href="#377">377</a> <em class="comment">// If path is not absolute, set f's directory</em><a name="378" href="#378">378</a> <em class="comment">// relative to the path of the order file</em><a name="379" href="#379">379</a> <strong>if</strong> (!f.isAbsolute()) {<a name="380" href="#380">380</a> f = <strong>new</strong> File(<strong>this</strong>.getOrderFile().getParent(), path);<a name="381" href="#381">381</a> }<a name="382" href="#382">382</a> <strong>return</strong> f;<a name="383" href="#383">383</a> }<a name="384" href="#384">384</a> <a name="385" href="#385">385</a> <strong>public</strong> Collection getDomainOverrides(String rootDomain) {<a name="386" href="#386">386</a> File settingsDir = getSettingsDirectory();<a name="387" href="#387">387</a> <a name="388" href="#388">388</a> <em class="comment">//Find the right start directory.</em><a name="389" href="#389">389</a> ArrayList domains = <strong>new</strong> ArrayList();<a name="390" href="#390">390</a> <em class="comment">//First we deconstruct the rootDomain string</em><a name="391" href="#391">391</a> <strong>while</strong>(rootDomain != <strong>null</strong> && rootDomain.length()>0){<a name="392" href="#392">392</a> <strong>if</strong>(rootDomain.indexOf('.')<0){<a name="393" href="#393">393</a> <em class="comment">// Last level.</em><a name="394" href="#394">394</a> domains.add(rootDomain);<a name="395" href="#395">395</a> <strong>break</strong>; <em class="comment">//We're done.</em><a name="396" href="#396">396</a> } <strong>else</strong> {<a name="397" href="#397">397</a> <em class="comment">// Got more then one level left.</em><a name="398" href="#398">398</a> domains.add(rootDomain.substring(0,rootDomain.indexOf('.')));<a name="399" href="#399">399</a> <em class="comment">// Strip down rootDomain.</em><a name="400" href="#400">400</a> rootDomain = rootDomain.substring(rootDomain.indexOf('.')+1);<a name="401" href="#401">401</a> }<a name="402" href="#402">402</a> }<a name="403" href="#403">403</a> <em class="comment">//Build up a proper path</em><a name="404" href="#404">404</a> <em class="comment">//Since the domains are right to left, we start at the end of the array.</em><a name="405" href="#405">405</a> StringBuffer subDir = <strong>new</strong> StringBuffer();<a name="406" href="#406">406</a> <strong>for</strong>(<strong>int</strong> i=(domains.size()-1) ; i>=0 ; i--){<a name="407" href="#407">407</a> subDir.append(File.separator+domains.get(i));<a name="408" href="#408">408</a> }<a name="409" href="#409">409</a> <em class="comment">//Then we move to the approprite directory.</em><a name="410" href="#410">410</a> settingsDir = <strong>new</strong> File(settingsDir.getPath()+subDir);<a name="411" href="#411">411</a> TreeSet confirmedSubDomains = <strong>new</strong> TreeSet(<strong>new</strong> Comparator() {<a name="412" href="#412">412</a> <strong>public</strong> <strong>int</strong> compare(Object o1, Object o2) {<a name="413" href="#413">413</a> <strong>if</strong>(o1 instanceof String && o2 instanceof String){<a name="414" href="#414">414</a> <strong>return</strong> ((String)o1).compareTo(o2.toString());<a name="415" href="#415">415</a> } <strong>else</strong> {<a name="416" href="#416">416</a> <em class="comment">// We only account for strings.</em><a name="417" href="#417">417</a> <strong>return</strong> 0;<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> );<a name="422" href="#422">422</a> <strong>if</strong>(settingsDir.exists()){<a name="423" href="#423">423</a> <em class="comment">// Found our place! Search through it's subdirs.</em><a name="424" href="#424">424</a> File[] possibleSubDomains = settingsDir.listFiles();<a name="425" href="#425">425</a> <strong>for</strong> (<strong>int</strong> i = 0; i < possibleSubDomains.length; i++) {<a name="426" href="#426">426</a> <strong>if</strong> (possibleSubDomains[i].isDirectory()
⌨️ 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 + -