xmlfixerreader.html

解析RSS工具源码

<a name="240" href="#240">240</a>                             loop = false;
<a name="241" href="#241">241</a>                         }
<a name="242" href="#242">242</a>                     }
<a name="243" href="#243">243</a>                     <strong>else</strong> {
<a name="244" href="#244">244</a>                         loop = false;
<a name="245" href="#245">245</a>                     }
<a name="246" href="#246">246</a>                     <strong>break</strong>;
<a name="247" href="#247">247</a>                 <strong>case</strong> 1: <em class="comment">// reading entity from stream</em>
<a name="248" href="#248">248</a>                     c = in.read();
<a name="249" href="#249">249</a>                     <strong>if</strong> (c>-1) {
<a name="250" href="#250">250</a>                         <strong>if</strong> (c==';') {
<a name="251" href="#251">251</a>                             _buffer.append((<strong>char</strong>)c);
<a name="252" href="#252">252</a>                             _state = 2;
<a name="253" href="#253">253</a>                             loop = <strong>true</strong>;
<a name="254" href="#254">254</a>                         }
<a name="255" href="#255">255</a>                         <strong>else</strong>
<a name="256" href="#256">256</a>                         <strong>if</strong> ((c>='a' &amp;&amp; c&lt;='z') || (c>='A' &amp;&amp; c&lt;='Z') || (c=='#') || (c>='0' &amp;&amp; c&lt;='9')) {
<a name="257" href="#257">257</a>                             _buffer.append((<strong>char</strong>)c);
<a name="258" href="#258">258</a>                             loop = <strong>true</strong>;
<a name="259" href="#259">259</a>                         }
<a name="260" href="#260">260</a>                         <strong>else</strong> {
<a name="261" href="#261">261</a>                             _buffer.append((<strong>char</strong>)c);
<a name="262" href="#262">262</a>                             _state = 3;
<a name="263" href="#263">263</a>                             loop = <strong>true</strong>;
<a name="264" href="#264">264</a>                         }
<a name="265" href="#265">265</a>                     }
<a name="266" href="#266">266</a>                     <strong>else</strong> {
<a name="267" href="#267">267</a>                         _state = 3;
<a name="268" href="#268">268</a>                         loop = <strong>true</strong>;
<a name="269" href="#269">269</a>                     }
<a name="270" href="#270">270</a>                     <strong>break</strong>;
<a name="271" href="#271">271</a>                 <strong>case</strong> 2: <em class="comment">// replacing entity</em>
<a name="272" href="#272">272</a>                     c = 0;
<a name="273" href="#273">273</a>                     String literalEntity = _buffer.toString();
<a name="274" href="#274">274</a>                     String codedEntity = (String) CODED_ENTITIES.get(literalEntity);
<a name="275" href="#275">275</a>                     <strong>if</strong> (codedEntity!=<strong>null</strong>) {
<a name="276" href="#276">276</a>                         _buffer.setLength(0);
<a name="277" href="#277">277</a>                         _buffer.append(codedEntity);
<a name="278" href="#278">278</a>                     } <em class="comment">// else we leave what was in the stream</em>
<a name="279" href="#279">279</a>                     _state = 3;
<a name="280" href="#280">280</a>                     loop = <strong>true</strong>;
<a name="281" href="#281">281</a>                     <strong>break</strong>;
<a name="282" href="#282">282</a>                 <strong>case</strong> 3: <em class="comment">// consuming buffer</em>
<a name="283" href="#283">283</a>                     <strong>if</strong> (_bufferPos&lt;_buffer.length()) {
<a name="284" href="#284">284</a>                         c = _buffer.charAt(_bufferPos++);
<a name="285" href="#285">285</a>                         loop = false;
<a name="286" href="#286">286</a>                     }
<a name="287" href="#287">287</a>                     <strong>else</strong> {
<a name="288" href="#288">288</a>                         c = 0;
<a name="289" href="#289">289</a>                         _state = 0;
<a name="290" href="#290">290</a>                         loop = <strong>true</strong>;
<a name="291" href="#291">291</a>                     }
<a name="292" href="#292">292</a>                     <strong>break</strong>;
<a name="293" href="#293">293</a>                  <strong>default</strong>:
<a name="294" href="#294">294</a>                     <strong>throw</strong> <strong>new</strong> IOException(<span class="string">"It shouldn't happen"</span>);
<a name="295" href="#295">295</a>             }
<a name="296" href="#296">296</a>         } <strong>while</strong> (loop);
<a name="297" href="#297">297</a>         <strong>return</strong> c;
<a name="298" href="#298">298</a>     }
<a name="299" href="#299">299</a> 
<a name="300" href="#300">300</a>     <strong>public</strong> <strong>int</strong> read(<strong>char</strong>[] buffer,<strong>int</strong> offset,<strong>int</strong> len) throws IOException {
<a name="301" href="#301">301</a>         <strong>int</strong> charsRead = 0;
<a name="302" href="#302">302</a>         <strong>int</strong> c = read();
<a name="303" href="#303">303</a>         <strong>if</strong> (c==-1) {
<a name="304" href="#304">304</a>             <strong>return</strong> -1;
<a name="305" href="#305">305</a>         }
<a name="306" href="#306">306</a>         buffer[offset+(<strong>char</strong>sRead++)] = (<strong>char</strong>) c;
<a name="307" href="#307">307</a>         <strong>while</strong> (charsRead&lt;len &amp;&amp; (c=read())>-1) {
<a name="308" href="#308">308</a>             buffer[offset+(<strong>char</strong>sRead++)] = (<strong>char</strong>) c;
<a name="309" href="#309">309</a>         }
<a name="310" href="#310">310</a>         <strong>return</strong> charsRead;
<a name="311" href="#311">311</a>     }
<a name="312" href="#312">312</a> 
<a name="313" href="#313">313</a>     <strong>public</strong> <strong>long</strong> skip(<strong>long</strong> n) throws IOException {
<a name="314" href="#314">314</a>         <strong>if</strong> (n==0) {
<a name="315" href="#315">315</a>             <strong>return</strong> 0;
<a name="316" href="#316">316</a>         }
<a name="317" href="#317">317</a>         <strong>else</strong>
<a name="318" href="#318">318</a>         <strong>if</strong> (n&lt;0) {
<a name="319" href="#319">319</a>             <strong>throw</strong> <strong>new</strong> IllegalArgumentException(<span class="string">"'n' cannot be negative"</span>);
<a name="320" href="#320">320</a>         }
<a name="321" href="#321">321</a>         <strong>int</strong> c = read();
<a name="322" href="#322">322</a>         <strong>long</strong> counter = 1;
<a name="323" href="#323">323</a>         <strong>while</strong> (c>-1 &amp;&amp; counter&lt;n) {
<a name="324" href="#324">324</a>             c = read();
<a name="325" href="#325">325</a>             counter++;
<a name="326" href="#326">326</a>         }
<a name="327" href="#327">327</a>         <strong>return</strong> counter;
<a name="328" href="#328">328</a>     }
<a name="329" href="#329">329</a> 
<a name="330" href="#330">330</a>     <strong>public</strong> <strong>boolean</strong> ready() throws IOException {
<a name="331" href="#331">331</a>         <strong>return</strong> (_state!=0) || in.ready();
<a name="332" href="#332">332</a>     }
<a name="333" href="#333">333</a> 
<a name="334" href="#334">334</a>     <strong>public</strong> <strong>boolean</strong> markSupported() {
<a name="335" href="#335">335</a>         <strong>return</strong> false;
<a name="336" href="#336">336</a>     }
<a name="337" href="#337">337</a> 
<a name="338" href="#338">338</a>     <strong>public</strong> <strong>void</strong> mark(<strong>int</strong> readAheadLimit) throws IOException {
<a name="339" href="#339">339</a>         <strong>throw</strong> <strong>new</strong> IOException(<span class="string">"Stream does not support mark"</span>);
<a name="340" href="#340">340</a>     }
<a name="341" href="#341">341</a> 
<a name="342" href="#342">342</a>     <strong>public</strong> <strong>void</strong> reset() throws IOException {
<a name="343" href="#343">343</a>         <strong>throw</strong> <strong>new</strong> IOException(<span class="string">"Stream does not support mark"</span>);
<a name="344" href="#344">344</a>     }
<a name="345" href="#345">345</a> 
<a name="346" href="#346">346</a>     <strong>public</strong> <strong>void</strong> close() throws IOException {
<a name="347" href="#347">347</a>         in.close();
<a name="348" href="#348">348</a>     }
<a name="349" href="#349">349</a> 
<a name="350" href="#350">350</a>     <strong>private</strong> <strong>static</strong> Map CODED_ENTITIES = <strong>new</strong> HashMap();
<a name="351" href="#351">351</a> 
<a name="352" href="#352">352</a>     <strong>static</strong> {
<a name="353" href="#353">353</a>         <em class="comment">// note: refer to Character entity references in HTML 4</em>
<a name="354" href="#354">354</a>         <em class="comment">// at http://www.w3.org/TR/REC-html40/sgml/entities.html</em>
<a name="355" href="#355">355</a>             	
<a name="356" href="#356">356</a>     	<em class="comment">// Character entity set.</em>
<a name="357" href="#357">357</a>     	<em class="comment">// HTMLlat1 "-//W3C//ENTITIES Latin 1//EN//HTML"</em>
<a name="358" href="#358">358</a>     
<a name="359" href="#359">359</a>     	CODED_ENTITIES.put(<span class="string">"&amp;nbsp;"</span>,  <span class="string">"&amp;#160;"</span>);
<a name="360" href="#360">360</a>         CODED_ENTITIES.put(<span class="string">"&amp;iexcl;"</span>, <span class="string">"&amp;#161;"</span>);
<a name="361" href="#361">361</a>         CODED_ENTITIES.put(<span class="string">"&amp;cent;"</span>,  <span class="string">"&amp;#162;"</span>);
<a name="362" href="#362">362</a>         CODED_ENTITIES.put(<span class="string">"&amp;pound;"</span>, <span class="string">"&amp;#163;"</span>);
<a name="363" href="#363">363</a>         CODED_ENTITIES.put(<span class="string">"&amp;curren;"</span>,<span class="string">"&amp;#164;"</span>);