xmlreader.html

解析RSS工具源码

<a name="288" href="#288">288</a> <em>     * &lt;p></em>
<a name="289" href="#289">289</a> <em>     * @return charset encoding.</em>
<a name="290" href="#290">290</a> <em>     *</em>
<a name="291" href="#291">291</a> <em>     */</em>
<a name="292" href="#292">292</a>     <strong>public</strong> String getEncoding() {
<a name="293" href="#293">293</a>         <strong>return</strong> _encoding;
<a name="294" href="#294">294</a>     }
<a name="295" href="#295">295</a> 
<a name="296" href="#296">296</a>     <strong>public</strong> <strong>int</strong> read(<strong>char</strong>[] buf,<strong>int</strong> offset,<strong>int</strong> len) throws IOException {
<a name="297" href="#297">297</a>         <strong>return</strong> _reader.read(buf,offset,len);
<a name="298" href="#298">298</a>     }
<a name="299" href="#299">299</a> 
<a name="300" href="#300">300</a>     <em>/**<em>*</em></em>
<a name="301" href="#301">301</a> <em>     * Closes the XmlReader stream.</em>
<a name="302" href="#302">302</a> <em>     * &lt;p></em>
<a name="303" href="#303">303</a> <em>     * @throws IOException thrown if there was a problem closing the stream.</em>
<a name="304" href="#304">304</a> <em>     *</em>
<a name="305" href="#305">305</a> <em>     */</em>
<a name="306" href="#306">306</a>     <strong>public</strong> <strong>void</strong> close() throws IOException {
<a name="307" href="#307">307</a>         _reader.close();
<a name="308" href="#308">308</a>     }
<a name="309" href="#309">309</a> 
<a name="310" href="#310">310</a>     <strong>private</strong> <strong>void</strong> doRawStream(InputStream is,<strong>boolean</strong> lenient) throws IOException {
<a name="311" href="#311">311</a>         PushbackInputStream pis = <strong>new</strong> PushbackInputStream(is,PUSHBACK_MAX_SIZE);
<a name="312" href="#312">312</a>         String bomEnc = getBOMEncoding(pis);
<a name="313" href="#313">313</a>         String xmlGuessEnc =  getXMLGuessEncoding(pis);
<a name="314" href="#314">314</a>         String xmlEnc = getXmlProlog(pis,xmlGuessEnc);
<a name="315" href="#315">315</a>         String encoding = calculateRawEncoding(bomEnc, xmlGuessEnc, xmlEnc, pis);
<a name="316" href="#316">316</a>         prepareReader(pis,encoding);
<a name="317" href="#317">317</a>     }
<a name="318" href="#318">318</a> 
<a name="319" href="#319">319</a>     <strong>private</strong> <strong>void</strong> doHttpStream(InputStream is,String httpContentType,<strong>boolean</strong> lenient) throws IOException {
<a name="320" href="#320">320</a>         PushbackInputStream pis = <strong>new</strong> PushbackInputStream(is,PUSHBACK_MAX_SIZE);
<a name="321" href="#321">321</a>         String cTMime = getContentTypeMime(httpContentType);
<a name="322" href="#322">322</a>         String cTEnc  = getContentTypeEncoding(httpContentType);
<a name="323" href="#323">323</a>         String bomEnc = getBOMEncoding(pis);
<a name="324" href="#324">324</a>         String xmlGuessEnc =  getXMLGuessEncoding(pis);
<a name="325" href="#325">325</a>         String xmlEnc = getXmlProlog(pis,xmlGuessEnc);
<a name="326" href="#326">326</a>         String encoding = calculateHttpEncoding(cTMime, cTEnc, bomEnc, xmlGuessEnc, xmlEnc, pis,lenient);
<a name="327" href="#327">327</a>         prepareReader(pis,encoding);
<a name="328" href="#328">328</a>     }
<a name="329" href="#329">329</a> 
<a name="330" href="#330">330</a>     <strong>private</strong> <strong>void</strong> prepareReader(InputStream is,String encoding) throws IOException {
<a name="331" href="#331">331</a>         _reader = <strong>new</strong> InputStreamReader(is,encoding);
<a name="332" href="#332">332</a>         _encoding = encoding;
<a name="333" href="#333">333</a>     }
<a name="334" href="#334">334</a> 
<a name="335" href="#335">335</a>     <em class="comment">// InputStream is passed for XmlReaderException creation only</em>
<a name="336" href="#336">336</a>     <strong>private</strong> <strong>static</strong> String calculateRawEncoding(String bomEnc, String xmlGuessEnc, String xmlEnc, InputStream is) throws IOException {
<a name="337" href="#337">337</a>         String encoding;
<a name="338" href="#338">338</a>         <strong>if</strong> (bomEnc==<strong>null</strong>) {
<a name="339" href="#339">339</a>             <strong>if</strong> (xmlGuessEnc==<strong>null</strong> || xmlEnc==<strong>null</strong>) {
<a name="340" href="#340">340</a>                 encoding = UTF_8;
<a name="341" href="#341">341</a>             }
<a name="342" href="#342">342</a>             <strong>else</strong>
<a name="343" href="#343">343</a>             <strong>if</strong> (xmlEnc.equals(UTF_16) &amp;&amp; (xmlGuessEnc.equals(UTF_16BE) || xmlGuessEnc.equals(UTF_16LE))) {
<a name="344" href="#344">344</a>                 encoding = xmlGuessEnc;
<a name="345" href="#345">345</a>             }
<a name="346" href="#346">346</a>             <strong>else</strong> {
<a name="347" href="#347">347</a>                 encoding = xmlEnc;
<a name="348" href="#348">348</a>             }
<a name="349" href="#349">349</a>         }
<a name="350" href="#350">350</a>         <strong>else</strong>
<a name="351" href="#351">351</a>         <strong>if</strong> (bomEnc.equals(UTF_8)) {
<a name="352" href="#352">352</a>             <strong>if</strong> (xmlGuessEnc!=<strong>null</strong> &amp;&amp; !xmlGuessEnc.equals(UTF_8)) {
<a name="353" href="#353">353</a>                 <strong>throw</strong> <strong>new</strong> <a href="../../../../com/sun/syndication/io/XmlReaderException.html">XmlReaderException</a>(RAW_EX_1.format(<strong>new</strong> Object[]{bomEnc,xmlGuessEnc,xmlEnc}),
<a name="354" href="#354">354</a>                                              bomEnc,xmlGuessEnc,xmlEnc,is);
<a name="355" href="#355">355</a>             }
<a name="356" href="#356">356</a>             <strong>if</strong> (xmlEnc!=<strong>null</strong> &amp;&amp; !xmlEnc.equals(UTF_8)) {
<a name="357" href="#357">357</a>                 <strong>throw</strong> <strong>new</strong> <a href="../../../../com/sun/syndication/io/XmlReaderException.html">XmlReaderException</a>(RAW_EX_1.format(<strong>new</strong> Object[]{bomEnc,xmlGuessEnc,xmlEnc}),
<a name="358" href="#358">358</a>                                              bomEnc,xmlGuessEnc,xmlEnc,is);
<a name="359" href="#359">359</a>             }
<a name="360" href="#360">360</a>             encoding = UTF_8;
<a name="361" href="#361">361</a>         }
<a name="362" href="#362">362</a>         <strong>else</strong>
<a name="363" href="#363">363</a>         <strong>if</strong> (bomEnc.equals(UTF_16BE) || bomEnc.equals(UTF_16LE)) {
<a name="364" href="#364">364</a>             <strong>if</strong> (xmlGuessEnc!=<strong>null</strong> &amp;&amp; !xmlGuessEnc.equals(bomEnc)) {
<a name="365" href="#365">365</a>                 <strong>throw</strong> <strong>new</strong> IOException(RAW_EX_1.format(<strong>new</strong> Object[]{bomEnc,xmlGuessEnc,xmlEnc}));
<a name="366" href="#366">366</a>             }
<a name="367" href="#367">367</a>             <strong>if</strong> (xmlEnc!=<strong>null</strong> &amp;&amp; !xmlEnc.equals(UTF_16) &amp;&amp; !xmlEnc.equals(bomEnc)) {
<a name="368" href="#368">368</a>                 <strong>throw</strong> <strong>new</strong> <a href="../../../../com/sun/syndication/io/XmlReaderException.html">XmlReaderException</a>(RAW_EX_1.format(<strong>new</strong> Object[]{bomEnc,xmlGuessEnc,xmlEnc}),
<a name="369" href="#369">369</a>                                              bomEnc,xmlGuessEnc,xmlEnc,is);
<a name="370" href="#370">370</a>             }
<a name="371" href="#371">371</a>             encoding =bomEnc;
<a name="372" href="#372">372</a>         }
<a name="373" href="#373">373</a>         <strong>else</strong> {
<a name="374" href="#374">374</a>             <strong>throw</strong> <strong>new</strong> <a href="../../../../com/sun/syndication/io/XmlReaderException.html">XmlReaderException</a>(RAW_EX_2.format(<strong>new</strong> Object[]{bomEnc,xmlGuessEnc,xmlEnc}),
<a name="375" href="#375">375</a>                                          bomEnc,xmlGuessEnc,xmlEnc,is);
<a name="376" href="#376">376</a>         }
<a name="377" href="#377">377</a>         <strong>return</strong> encoding;
<a name="378" href="#378">378</a>     }
<a name="379" href="#379">379</a> 
<a name="380" href="#380">380</a>     <em class="comment">// InputStream is passed for XmlReaderException creation only</em>
<a name="381" href="#381">381</a>     <strong>private</strong> <strong>static</strong> String calculateHttpEncoding(String cTMime, String cTEnc, String bomEnc, String xmlGuessEnc, String xmlEnc, InputStream is,<strong>boolean</strong> lenient) throws IOException {
<a name="382" href="#382">382</a>         String encoding;
<a name="383" href="#383">383</a>         <strong>if</strong> (lenient &amp; xmlEnc!=<strong>null</strong>) {
<a name="384" href="#384">384</a>             encoding = xmlEnc;
<a name="385" href="#385">385</a>         }
<a name="386" href="#386">386</a>         <strong>else</strong> {
<a name="387" href="#387">387</a>             <strong>boolean</strong> appXml = isAppXml(cTMime);
<a name="388" href="#388">388</a>             <strong>boolean</strong> textXml = isTextXml(cTMime);
<a name="389" href="#389">389</a>             <strong>if</strong> (appXml || textXml) {
<a name="390" href="#390">390</a>                 <strong>if</strong> (cTEnc==<strong>null</strong>) {
<a name="391" href="#391">391</a>                     <strong>if</strong> (appXml) {
<a name="392" href="#392">392</a>                         encoding = calculateRawEncoding(bomEnc, xmlGuessEnc, xmlEnc, is);
<a name="393" href="#393">393</a>                     }
<a name="394" href="#394">394</a>                     <strong>else</strong> {
<a name="395" href="#395">395</a>                         encoding = US_ASCII;
<a name="396" href="#396">396</a>                     }
<a name="397" href="#397">397</a>                 }
<a name="398" href="#398">398</a>                 <strong>else</strong>
<a name="399" href="#399">399</a>                 <strong>if</strong> (bomEnc!=<strong>null</strong> &amp;&amp; (cTEnc.equals(UTF_16BE) || cTEnc.equals(UTF_16LE))) {
<a name="400" href="#400">400</a>                     <strong>throw</strong> <strong>new</strong> <a href="../../../../com/sun/syndication/io/XmlReaderException.html">XmlReaderException</a>(HTTP_EX_1.format(<strong>new</strong> Object[]{cTMime,cTEnc,bomEnc,xmlGuessEnc,xmlEnc}),
<a name="401" href="#401">401</a>                                                  cTMime,cTEnc,bomEnc,xmlGuessEnc,xmlEnc,is);
<a name="402" href="#402">402</a>                 }
<a name="403" href="#403">403</a>                 <strong>else</strong>
<a name="404" href="#404">404</a>                 <strong>if</strong> (cTEnc.equals(UTF_16)) {
<a name="405" href="#405">405</a>                     <strong>if</strong> (bomEnc!=<strong>null</strong> &amp;&amp; bomEnc.startsWith(UTF_16)) {
<a name="406" href="#406">406</a>                         encoding = bomEnc;
<a name="407" href="#407">407</a>                     }
<a name="408" href="#408">408</a>                     <strong>else</strong> {
<a name="409" href="#409">409</a>                         <strong>throw</strong> <strong>new</strong> <a href="../../../../com/sun/syndication/io/XmlReaderException.html">XmlReaderException</a>(HTTP_EX_2.format(<strong>new</strong> Object[]{cTMime,cTEnc,bomEnc,xmlGuessEnc,xmlEnc}),
<a name="410" href="#410">410</a>                                                      cTMime,cTEnc,bomEnc,xmlGuessEnc,xmlEnc,is);
<a name="411" href="#411">411</a>                     }
<a name="412" href="#412">412</a>                 }
<a name="413" href="#413">413</a>                 <strong>else</strong> {
<a name="414" href="#414">414</a>                     encoding = cTEnc;
<a name="415" href="#415">415</a>                 }
<a name="416" href="#416">416</a>             }
<a name="417" href="#417">417</a>             <strong>else</strong> {
<a name="418" href="#418">418</a>                 <strong>throw</strong> <strong>new</strong> <a href="../../../../com/sun/syndication/io/XmlReaderException.html">XmlReaderException</a>(HTTP_EX_3.format(<strong>new</strong> Object[]{cTMime,cTEnc,bomEnc,xmlGuessEnc,xmlEnc}),
<a name="419" href="#419">419</a>                                              cTMime,cTEnc,bomEnc,xmlGuessEnc,xmlEnc,is);
<a name="420" href="#420">420</a>             }
<a name="421" href="#421">421</a>         }
<a name="422" href="#422">422</a>         <strong>return</strong> encoding;
<a name="423" href="#423">423</a>     }
<a name="424" href="#424">424</a> 
<a name="425" href="#425">425</a>     <em class="comment">// returns MIME type or NULL if httpContentType is NULL</em>
<a name="426" href="#426">426</a>     <strong>private</strong> <strong>static</strong> String getContentTypeMime(String httpContentType) {
<a name="427" href="#427">427</a>         String mime = <strong>null</strong>;
<a name="428" href="#428">428</a>         <strong>if</strong> (httpContentType!=<strong>null</strong>) {
<a name="429" href="#429">429</a>             <strong>int</strong> i = httpContentType.indexOf(<span class="string">";"</span>);
<a name="430" href="#430">430</a>             mime = ((i==-1) ? httpContentType : httpContentType.substring(0,i)).trim();
<a name="431" href="#431">431</a>         }
<a name="432" href="#432">432</a>         <strong>return</strong> mime;
<a name="433" href="#433">433</a>     }
<a name="434" href="#434">434</a> 
<a name="435" href="#435">435</a>     <strong>private</strong> <strong>static</strong> <strong>final</strong> Pattern CHARSET_PATTERN = Pattern.compile(<span class="string">"charset=([.[^; ]]*)"</span>);