atom10parser.html

解析RSS工具源码

<a name="288" href="#288">288</a>     
<a name="289" href="#289">289</a>     <strong>private</strong> String parseTextConstructToString(Element e) {
<a name="290" href="#290">290</a>         String value = <strong>null</strong>;
<a name="291" href="#291">291</a>         String type = e.getAttributeValue(<span class="string">"type"</span>);<em class="comment">//getAtomNamespace()); DONT KNOW WHY DOESN'T WORK</em>
<a name="292" href="#292">292</a>         type = (type!=<strong>null</strong>) ? type : Content.TEXT;
<a name="293" href="#293">293</a>         <strong>if</strong> (type.equals(Content.XHTML)) {
<a name="294" href="#294">294</a>             <em class="comment">// XHTML content needs special handling</em>
<a name="295" href="#295">295</a>             XMLOutputter outputter = <strong>new</strong> XMLOutputter();
<a name="296" href="#296">296</a>             List eContent = e.getContent();
<a name="297" href="#297">297</a>             Iterator i = eContent.iterator();
<a name="298" href="#298">298</a>             <strong>while</strong> (i.hasNext()) {
<a name="299" href="#299">299</a>                 org.jdom.Content c = (org.jdom.Content) i.next();
<a name="300" href="#300">300</a>                 <strong>if</strong> (c instanceof Element) {
<a name="301" href="#301">301</a>                     Element eC = (Element) c;
<a name="302" href="#302">302</a>                     <strong>if</strong> (eC.getNamespace().equals(getAtomNamespace())) {
<a name="303" href="#303">303</a>                         ((Element)c).setNamespace(Namespace.NO_NAMESPACE);
<a name="304" href="#304">304</a>                     }
<a name="305" href="#305">305</a>                 }
<a name="306" href="#306">306</a>             }
<a name="307" href="#307">307</a>             value = outputter.outputString(eContent);
<a name="308" href="#308">308</a>         } <strong>else</strong> {
<a name="309" href="#309">309</a>             <em class="comment">// Everything else comes in verbatim</em>
<a name="310" href="#310">310</a>             value = e.getText();
<a name="311" href="#311">311</a>         }
<a name="312" href="#312">312</a>         <strong>return</strong> value;
<a name="313" href="#313">313</a>     }
<a name="314" href="#314">314</a>     
<a name="315" href="#315">315</a>     <em class="comment">// List(Elements) -> List(Entries)</em>
<a name="316" href="#316">316</a>     <strong>protected</strong> List parseEntries(<a href="../../../../../com/sun/syndication/feed/atom/Feed.html">Feed</a> feed, String baseURI, List eEntries) {
<a name="317" href="#317">317</a>         List entries = <strong>new</strong> ArrayList();
<a name="318" href="#318">318</a>         <strong>for</strong> (<strong>int</strong> i=0;i&lt;eEntries.size();i++) {
<a name="319" href="#319">319</a>             entries.add(parseEntry(feed, (Element)eEntries.get(i), baseURI));
<a name="320" href="#320">320</a>         }
<a name="321" href="#321">321</a>         <strong>return</strong> (entries.size()>0) ? entries : <strong>null</strong>;
<a name="322" href="#322">322</a>     }
<a name="323" href="#323">323</a>     
<a name="324" href="#324">324</a>     <strong>protected</strong> <a href="../../../../../com/sun/syndication/feed/atom/Entry.html">Entry</a> parseEntry(<a href="../../../../../com/sun/syndication/feed/atom/Feed.html">Feed</a> feed, Element eEntry, String baseURI) {
<a name="325" href="#325">325</a>         <a href="../../../../../com/sun/syndication/feed/atom/Entry.html">Entry</a> entry = <strong>new</strong> <a href="../../../../../com/sun/syndication/feed/atom/Entry.html">Entry</a>();
<a name="326" href="#326">326</a>         
<a name="327" href="#327">327</a>         String xmlBase = eEntry.getAttributeValue(<span class="string">"base"</span>, Namespace.XML_NAMESPACE);
<a name="328" href="#328">328</a>         <strong>if</strong> (xmlBase != <strong>null</strong>) {
<a name="329" href="#329">329</a>             entry.setXmlBase(xmlBase);
<a name="330" href="#330">330</a>         }
<a name="331" href="#331">331</a>         
<a name="332" href="#332">332</a>         Element e = eEntry.getChild(<span class="string">"title"</span>,getAtomNamespace());
<a name="333" href="#333">333</a>         <strong>if</strong> (e!=<strong>null</strong>) {
<a name="334" href="#334">334</a>             <a href="../../../../../com/sun/syndication/feed/atom/Content.html">Content</a> c = <strong>new</strong> <a href="../../../../../com/sun/syndication/feed/atom/Content.html">Content</a>();
<a name="335" href="#335">335</a>             c.setValue(parseTextConstructToString(e));
<a name="336" href="#336">336</a>             c.setType(e.getAttributeValue(<span class="string">"type"</span>)); <em class="comment">//, Namespace.XML_NAMESPACE));</em>
<a name="337" href="#337">337</a>             entry.setTitleEx(c);
<a name="338" href="#338">338</a>         }
<a name="339" href="#339">339</a>         
<a name="340" href="#340">340</a>         List eList = eEntry.getChildren(<span class="string">"link"</span>,getAtomNamespace());
<a name="341" href="#341">341</a>         entry.setAlternateLinks(parseAlternateLinks(feed, entry, baseURI, eList));
<a name="342" href="#342">342</a>         entry.setOtherLinks(parseOtherLinks(feed, entry, baseURI, eList));
<a name="343" href="#343">343</a>         
<a name="344" href="#344">344</a>         eList = eEntry.getChildren(<span class="string">"author"</span>, getAtomNamespace());
<a name="345" href="#345">345</a>         <strong>if</strong> (eList.size()>0) {
<a name="346" href="#346">346</a>             entry.setAuthors(parsePersons(baseURI, eList));
<a name="347" href="#347">347</a>         }
<a name="348" href="#348">348</a>         
<a name="349" href="#349">349</a>         eList = eEntry.getChildren(<span class="string">"contributor"</span>,getAtomNamespace());
<a name="350" href="#350">350</a>         <strong>if</strong> (eList.size()>0) {
<a name="351" href="#351">351</a>             entry.setContributors(parsePersons(baseURI, eList));
<a name="352" href="#352">352</a>         }
<a name="353" href="#353">353</a>         
<a name="354" href="#354">354</a>         e = eEntry.getChild(<span class="string">"id"</span>,getAtomNamespace());
<a name="355" href="#355">355</a>         <strong>if</strong> (e!=<strong>null</strong>) {
<a name="356" href="#356">356</a>             entry.setId(e.getText());
<a name="357" href="#357">357</a>         }
<a name="358" href="#358">358</a>         
<a name="359" href="#359">359</a>         e = eEntry.getChild(<span class="string">"updated"</span>,getAtomNamespace());
<a name="360" href="#360">360</a>         <strong>if</strong> (e!=<strong>null</strong>) {
<a name="361" href="#361">361</a>             entry.setUpdated(DateParser.parseW3CDateTime(e.getText()));
<a name="362" href="#362">362</a>         }
<a name="363" href="#363">363</a>         
<a name="364" href="#364">364</a>         e = eEntry.getChild(<span class="string">"published"</span>,getAtomNamespace());
<a name="365" href="#365">365</a>         <strong>if</strong> (e!=<strong>null</strong>) {
<a name="366" href="#366">366</a>             entry.setPublished(DateParser.parseW3CDateTime(e.getText()));
<a name="367" href="#367">367</a>         }
<a name="368" href="#368">368</a>         
<a name="369" href="#369">369</a>         e = eEntry.getChild(<span class="string">"summary"</span>,getAtomNamespace());
<a name="370" href="#370">370</a>         <strong>if</strong> (e!=<strong>null</strong>) {
<a name="371" href="#371">371</a>             entry.setSummary(parseContent(e));
<a name="372" href="#372">372</a>         }
<a name="373" href="#373">373</a>         
<a name="374" href="#374">374</a>         e = eEntry.getChild(<span class="string">"content"</span>,getAtomNamespace());
<a name="375" href="#375">375</a>         <strong>if</strong> (e!=<strong>null</strong>) {
<a name="376" href="#376">376</a>             List contents = <strong>new</strong> ArrayList();
<a name="377" href="#377">377</a>             contents.add(parseContent(e));
<a name="378" href="#378">378</a>             entry.setContents(contents);
<a name="379" href="#379">379</a>         }
<a name="380" href="#380">380</a>         
<a name="381" href="#381">381</a>         e = eEntry.getChild(<span class="string">"rights"</span>,getAtomNamespace());
<a name="382" href="#382">382</a>         <strong>if</strong> (e!=<strong>null</strong>) {
<a name="383" href="#383">383</a>             entry.setRights(e.getText());
<a name="384" href="#384">384</a>         }
<a name="385" href="#385">385</a>         
<a name="386" href="#386">386</a>         List cList = eEntry.getChildren(<span class="string">"category"</span>,getAtomNamespace());
<a name="387" href="#387">387</a>         entry.setCategories(parseCategories(baseURI, cList));
<a name="388" href="#388">388</a>         
<a name="389" href="#389">389</a>         <em class="comment">// TODO: SHOULD handle Atom entry source element</em>
<a name="390" href="#390">390</a>         
<a name="391" href="#391">391</a>         entry.setModules(parseItemModules(eEntry));
<a name="392" href="#392">392</a>         
<a name="393" href="#393">393</a>         List foreignMarkup =
<a name="394" href="#394">394</a>                 extractForeignMarkup(eEntry, entry, getAtomNamespace());
<a name="395" href="#395">395</a>         <strong>if</strong> (foreignMarkup.size() > 0) {
<a name="396" href="#396">396</a>             entry.setForeignMarkup(foreignMarkup);
<a name="397" href="#397">397</a>         }
<a name="398" href="#398">398</a>         <strong>return</strong> entry;
<a name="399" href="#399">399</a>     }
<a name="400" href="#400">400</a>     
<a name="401" href="#401">401</a>     <strong>private</strong> List parseCategories(String baseURI, List eCategories) {
<a name="402" href="#402">402</a>         List cats = <strong>new</strong> ArrayList();
<a name="403" href="#403">403</a>         <strong>for</strong> (<strong>int</strong> i=0;i&lt;eCategories.size();i++) {
<a name="404" href="#404">404</a>             Element eCategory = (Element) eCategories.get(i);
<a name="405" href="#405">405</a>             cats.add(parseCategory(baseURI, eCategory));
<a name="406" href="#406">406</a>         }
<a name="407" href="#407">407</a>         <strong>return</strong> (cats.size()>0) ? cats : <strong>null</strong>;
<a name="408" href="#408">408</a>     }
<a name="409" href="#409">409</a>     
<a name="410" href="#410">410</a>     <strong>private</strong> <a href="../../../../../com/sun/syndication/feed/atom/Category.html">Category</a> parseCategory(String baseURI, Element eCategory) {
<a name="411" href="#411">411</a>         <a href="../../../../../com/sun/syndication/feed/atom/Category.html">Category</a> category = <strong>new</strong> <a href="../../../../../com/sun/syndication/feed/atom/Category.html">Category</a>();
<a name="412" href="#412">412</a>         String att = eCategory.getAttributeValue(<span class="string">"term"</span>);<em class="comment">//getAtomNamespace()); DONT KNOW WHY DOESN'T WORK</em>
<a name="413" href="#413">413</a>         <strong>if</strong> (att!=<strong>null</strong>) {
<a name="414" href="#414">414</a>             category.setTerm(att);
<a name="415" href="#415">415</a>         }
<a name="416" href="#416">416</a>         att = eCategory.getAttributeValue(<span class="string">"scheme"</span>);<em class="comment">//getAtomNamespace()); DONT KNOW WHY DOESN'T WORK</em>
<a name="417" href="#417">417</a>         <strong>if</strong> (att!=<strong>null</strong>) {
<a name="418" href="#418">418</a>             category.setScheme(resolveURI(baseURI, eCategory, att));
<a name="419" href="#419">419</a>         }
<a name="420" href="#420">420</a>         att = eCategory.getAttributeValue(<span class="string">"label"</span>);<em class="comment">//getAtomNamespace()); DONT KNOW WHY DOESN'T WORK</em>
<a name="421" href="#421">421</a>         <strong>if</strong> (att!=<strong>null</strong>) {
<a name="422" href="#422">422</a>             category.setLabel(att);
<a name="423" href="#423">423</a>         }
<a name="424" href="#424">424</a>         <strong>return</strong> category;
<a name="425" href="#425">425</a>         
<a name="426" href="#426">426</a>     }
<a name="427" href="#427">427</a>     
<a name="428" href="#428">428</a>     
<a name="429" href="#429">429</a>     <em class="comment">// Fix for issue #34 "valid IRI href attributes are stripped for atom:link"</em>
<a name="430" href="#430">430</a>     <em class="comment">// URI's that didn't start with http were being treated as relative URIs.</em>
<a name="431" href="#431">431</a>     <em class="comment">// So now consider an absolute URI to be any alpha-numeric string followed</em>
<a name="432" href="#432">432</a>     <em class="comment">// by a colon, followed by anything -- specified by this regex:</em>
<a name="433" href="#433">433</a>     <strong>static</strong> Pattern absoluteURIPattern = Pattern.compile(<span class="string">"^[a-z0-9]*:.*$"</span>);
<a name="434" href="#434">434</a>     
<a name="435" href="#435">435</a>     <strong>private</strong> <strong>boolean</strong> isAbsoluteURI(String uri) {