complextype.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="286" href="#286">286</a>     }
<a name="287" href="#287">287</a> 
<a name="288" href="#288">288</a>     <em>/**<em>* Get the absolute name of this ComplexType.</em></em>
<a name="289" href="#289">289</a> <em>     *</em>
<a name="290" href="#290">290</a> <em>     * The absolute name is like a file path with the name of the element</em>
<a name="291" href="#291">291</a> <em>     * prepended by all the parents names separated by slashes.</em>
<a name="292" href="#292">292</a> <em>     * @return Absolute name.</em>
<a name="293" href="#293">293</a> <em>     */</em>
<a name="294" href="#294">294</a>     <strong>public</strong> String getAbsoluteName() {
<a name="295" href="#295">295</a>         <strong>return</strong> absoluteName;
<a name="296" href="#296">296</a>     }
<a name="297" href="#297">297</a> 
<a name="298" href="#298">298</a>     <em>/**<em>*</em></em>
<a name="299" href="#299">299</a> <em>     * Get settings object valid for a URI.</em>
<a name="300" href="#300">300</a> <em>     * &lt;p/></em>
<a name="301" href="#301">301</a> <em>     * This method takes an object,</em>
<a name="302" href="#302">302</a> <em>     * try to convert it into a {@link CrawlURI} and then tries to get the</em>
<a name="303" href="#303">303</a> <em>     * settings object from it. If this fails, then the global settings object</em>
<a name="304" href="#304">304</a> <em>     * is returned.</em>
<a name="305" href="#305">305</a> <em>     * &lt;p/></em>
<a name="306" href="#306">306</a> <em>     * If the requested attribute is not set on this settings</em>
<a name="307" href="#307">307</a> <em>     * object it tries its parent until it gets a settings object where this</em>
<a name="308" href="#308">308</a> <em>     * attribute is set is found. If nothing is found, global settings is</em>
<a name="309" href="#309">309</a> <em>     * returned.</em>
<a name="310" href="#310">310</a> <em>     *</em>
<a name="311" href="#311">311</a> <em>     * @param o possible {@link CrawlURI}.</em>
<a name="312" href="#312">312</a> <em>     * @param attributeName the attribute that should have a value set on the</em>
<a name="313" href="#313">313</a> <em>     *            returned settings object.</em>
<a name="314" href="#314">314</a> <em>     * @return the settings object valid for the URI.</em>
<a name="315" href="#315">315</a> <em>     */</em>
<a name="316" href="#316">316</a>     Context getSettingsFromObject(Object o, String attributeName) {
<a name="317" href="#317">317</a>         Context context;
<a name="318" href="#318">318</a>         <strong>if</strong> (o == <strong>null</strong>) {
<a name="319" href="#319">319</a>             context = <strong>null</strong>;
<a name="320" href="#320">320</a>         } <strong>else</strong> <strong>if</strong> (o instanceof Context) {
<a name="321" href="#321">321</a>             context = (Context) o;
<a name="322" href="#322">322</a>         } <strong>else</strong> <strong>if</strong> (o instanceof CrawlerSettings) {
<a name="323" href="#323">323</a>             context = <strong>new</strong> Context((CrawlerSettings) o, <strong>null</strong>);
<a name="324" href="#324">324</a>         } <strong>else</strong> <strong>if</strong> (o instanceof UURI || o instanceof CandidateURI) {
<a name="325" href="#325">325</a>             <em class="comment">// Try to get settings for URI that has no references to a</em>
<a name="326" href="#326">326</a>             <em class="comment">// CrawlServer [SIC - CrawlURI may have CrawlServer -gjm]</em>
<a name="327" href="#327">327</a>             context = <strong>new</strong> Context();
<a name="328" href="#328">328</a>             context.uri = (o instanceof CandidateURI)?
<a name="329" href="#329">329</a>                 ((<a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a>) o).getUURI(): (UURI)o;
<a name="330" href="#330">330</a>             <strong>try</strong> {
<a name="331" href="#331">331</a>                context.settings = getSettingsHandler().
<a name="332" href="#332">332</a>                    getSettings(context.uri.getReferencedHost(),
<a name="333" href="#333">333</a>                        context.uri);
<a name="334" href="#334">334</a>             }
<a name="335" href="#335">335</a>             <strong>catch</strong> (URIException e1) {
<a name="336" href="#336">336</a>                 logger.severe(<span class="string">"Failed to get host"</span>);
<a name="337" href="#337">337</a>             }
<a name="338" href="#338">338</a> 
<a name="339" href="#339">339</a>             <strong>if</strong> (attributeName != <strong>null</strong>) {
<a name="340" href="#340">340</a>                 <strong>try</strong> {
<a name="341" href="#341">341</a>                     context.settings =
<a name="342" href="#342">342</a>                         getDataContainerRecursive(context, attributeName).
<a name="343" href="#343">343</a>                             getSettings();
<a name="344" href="#344">344</a>                 } <strong>catch</strong> (AttributeNotFoundException e) {
<a name="345" href="#345">345</a>                     <em class="comment">// Nothing found, globals will be used</em>
<a name="346" href="#346">346</a>                 }
<a name="347" href="#347">347</a>             }
<a name="348" href="#348">348</a>         } <strong>else</strong> {
<a name="349" href="#349">349</a>             logger.warning(<span class="string">"Unknown object type: "</span> +
<a name="350" href="#350">350</a>                 o.getClass().getName());
<a name="351" href="#351">351</a>             context = <strong>null</strong>;
<a name="352" href="#352">352</a>         }
<a name="353" href="#353">353</a> 
<a name="354" href="#354">354</a>         <em class="comment">// if settings could not be resolved use globals.</em>
<a name="355" href="#355">355</a>         <strong>if</strong> (context == <strong>null</strong>) {
<a name="356" href="#356">356</a>             context = <strong>new</strong> Context(globalSettings(), <strong>null</strong>);
<a name="357" href="#357">357</a>         }
<a name="358" href="#358">358</a>         <strong>return</strong> context;
<a name="359" href="#359">359</a>     }
<a name="360" href="#360">360</a> 
<a name="361" href="#361">361</a>     <em>/**<em>* Get settings object valid for a URI.</em></em>
<a name="362" href="#362">362</a> <em>    *</em>
<a name="363" href="#363">363</a> <em>    * This method takes an object, try to convert it into a {@link CrawlURI}</em>
<a name="364" href="#364">364</a> <em>    * and then tries to get the settings object from it. If this fails, then</em>
<a name="365" href="#365">365</a> <em>    * the global settings object is returned.</em>
<a name="366" href="#366">366</a> <em>    *</em>
<a name="367" href="#367">367</a> <em>    * @param o possible {@link CrawlURI}.</em>
<a name="368" href="#368">368</a> <em>    * @return the settings object valid for the URI.</em>
<a name="369" href="#369">369</a> <em>    */</em>
<a name="370" href="#370">370</a>     Context getSettingsFromObject(Object o) {
<a name="371" href="#371">371</a>         <strong>return</strong> getSettingsFromObject(o, <strong>null</strong>);
<a name="372" href="#372">372</a>     }
<a name="373" href="#373">373</a> 
<a name="374" href="#374">374</a>     <em>/**<em>* Returns true if an element is overridden for this settings object.</em></em>
<a name="375" href="#375">375</a> <em>     *</em>
<a name="376" href="#376">376</a> <em>     * @param settings the settings object to investigate.</em>
<a name="377" href="#377">377</a> <em>     * @param name the name of the element to check.</em>
<a name="378" href="#378">378</a> <em>     * @return true if element is overridden for this settings object, false</em>
<a name="379" href="#379">379</a> <em>     *              if not set here or is first defined here.</em>
<a name="380" href="#380">380</a> <em>     * @throws AttributeNotFoundException if element doesn't exist.</em>
<a name="381" href="#381">381</a> <em>     */</em>
<a name="382" href="#382">382</a>     <strong>public</strong> <strong>boolean</strong> isOverridden(<a href="../../../../org/archive/crawler/settings/CrawlerSettings.html">CrawlerSettings</a> settings, String name)
<a name="383" href="#383">383</a>             throws AttributeNotFoundException {
<a name="384" href="#384">384</a>         settings = settings == <strong>null</strong> ? globalSettings() : settings;
<a name="385" href="#385">385</a>         <a href="../../../../org/archive/crawler/settings/DataContainer.html">DataContainer</a> data = settings.getData(<strong>this</strong>);
<a name="386" href="#386">386</a>         <strong>if</strong> (data == <strong>null</strong> || !data.containsKey(name)) {
<a name="387" href="#387">387</a>             <strong>return</strong> false;
<a name="388" href="#388">388</a>         }
<a name="389" href="#389">389</a> 
<a name="390" href="#390">390</a>         <em class="comment">// Try to find attribute, will throw an exception if not found.</em>
<a name="391" href="#391">391</a>         Context context = <strong>new</strong> Context(settings.getParent(), <strong>null</strong>);
<a name="392" href="#392">392</a>         getDataContainerRecursive(context, name);
<a name="393" href="#393">393</a>         <strong>return</strong> <strong>true</strong>;
<a name="394" href="#394">394</a>     }
<a name="395" href="#395">395</a> 
<a name="396" href="#396">396</a>     <em>/**<em>* Obtain the value of a specific attribute from the crawl order.</em></em>
<a name="397" href="#397">397</a> <em>     *</em>
<a name="398" href="#398">398</a> <em>     * If the attribute doesn't exist in the crawl order, the default</em>
<a name="399" href="#399">399</a> <em>     * value will be returned.</em>
<a name="400" href="#400">400</a> <em>     *</em>
<a name="401" href="#401">401</a> <em>     * @param name the name of the attribute to be retrieved.</em>
<a name="402" href="#402">402</a> <em>     * @return The value of the attribute retrieved.</em>
<a name="403" href="#403">403</a> <em>     * @throws AttributeNotFoundException</em>
<a name="404" href="#404">404</a> <em>     * @throws MBeanException</em>
<a name="405" href="#405">405</a> <em>     * @throws ReflectionException</em>
<a name="406" href="#406">406</a> <em>     */</em>
<a name="407" href="#407">407</a>     <strong>public</strong> Object getAttribute(String name)
<a name="408" href="#408">408</a>         throws AttributeNotFoundException, MBeanException, ReflectionException {
<a name="409" href="#409">409</a>         <strong>return</strong> getAttribute(<strong>null</strong>, name);
<a name="410" href="#410">410</a>     }
<a name="411" href="#411">411</a> 
<a name="412" href="#412">412</a>     <em>/**<em>* Obtain the value of a specific attribute that is valid for a</em></em>
<a name="413" href="#413">413</a> <em>     * specific CrawlURI.</em>
<a name="414" href="#414">414</a> <em>     *</em>
<a name="415" href="#415">415</a> <em>     * This method will try to get the attribute from the host settings</em>
<a name="416" href="#416">416</a> <em>     * valid for the CrawlURI. If it is not found it will traverse the</em>
<a name="417" href="#417">417</a> <em>     * settings up to the order and as a last resort deliver the default</em>
<a name="418" href="#418">418</a> <em>     * value. This is also the case if the CrawlURI is null or if the CrawlURI</em>
<a name="419" href="#419">419</a> <em>     * hasn't been assigned a CrawlServer.</em>
<a name="420" href="#420">420</a> <em>     *</em>
<a name="421" href="#421">421</a> <em>     * @param name the name of the attribute to be retrieved.</em>
<a name="422" href="#422">422</a> <em>     * @param uri the CrawlURI that this attribute should be valid for.</em>
<a name="423" href="#423">423</a> <em>     * @return The value of the attribute retrieved.</em>
<a name="424" href="#424">424</a> <em>     * @see #getAttribute(Object settings, String name)</em>
<a name="425" href="#425">425</a> <em>     * @throws AttributeNotFoundException</em>
<a name="426" href="#426">426</a> <em>     */</em>
<a name="427" href="#427">427</a>     <strong>public</strong> Object getAttribute(String name, <a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> uri)
<a name="428" href="#428">428</a>         throws AttributeNotFoundException {
<a name="429" href="#429">429</a>         <strong>return</strong> getAttribute(uri, name);
<a name="430" href="#430">430</a>     }
<a name="431" href="#431">431</a>     
<a name="432" href="#432">432</a>     <em>/**<em>*</em></em>