uuri.html

一个开源的网页爬虫一个开源的网页爬虫一个开源的网页爬虫一个开源的网页爬虫一个开源的网页爬虫一个开源的网页爬虫

<a name="306" href="#306">306</a>         <strong>if</strong> (<strong>this</strong>.cachedAuthorityMinusUserinfo != <strong>null</strong>
<a name="307" href="#307">307</a>                 &amp;&amp; <strong>this</strong>.cachedHost != <strong>null</strong>
<a name="308" href="#308">308</a>                 &amp;&amp; <strong>this</strong>.cachedHost.length() ==
<a name="309" href="#309">309</a>                     <strong>this</strong>.cachedAuthorityMinusUserinfo.length()) {
<a name="310" href="#310">310</a>             <em class="comment">// lengths can only be identical if contents</em>
<a name="311" href="#311">311</a>             <em class="comment">// are identical; use only one instance</em>
<a name="312" href="#312">312</a>             <strong>this</strong>.cachedAuthorityMinusUserinfo = <strong>this</strong>.cachedHost;
<a name="313" href="#313">313</a>         }
<a name="314" href="#314">314</a>     }
<a name="315" href="#315">315</a> 
<a name="316" href="#316">316</a>     <em>/**<em>*</em></em>
<a name="317" href="#317">317</a> <em>     * Return the referenced host in the UURI, if any, also extracting the </em>
<a name="318" href="#318">318</a> <em>     * host of a DNS-lookup URI where necessary. </em>
<a name="319" href="#319">319</a> <em>     * </em>
<a name="320" href="#320">320</a> <em>     * @return the target or topic host of the URI</em>
<a name="321" href="#321">321</a> <em>     * @throws URIException</em>
<a name="322" href="#322">322</a> <em>     */</em>
<a name="323" href="#323">323</a>     <strong>public</strong> String getReferencedHost() throws URIException {
<a name="324" href="#324">324</a>         String referencedHost = <strong>this</strong>.getHost();
<a name="325" href="#325">325</a>         <strong>if</strong>(referencedHost==<strong>null</strong> &amp;&amp; <strong>this</strong>.getScheme().equals(<span class="string">"dns"</span>)) {
<a name="326" href="#326">326</a>             <em class="comment">// extract target domain of DNS lookup</em>
<a name="327" href="#327">327</a>             String possibleHost = <strong>this</strong>.getCurrentHierPath();
<a name="328" href="#328">328</a>             <strong>if</strong>(possibleHost != <strong>null</strong> &amp;&amp; possibleHost.matches(<span class="string">"[-_&#47;&#47;w&#47;&#47;.:]+"</span>)) {
<a name="329" href="#329">329</a>                 referencedHost = possibleHost;
<a name="330" href="#330">330</a>             }
<a name="331" href="#331">331</a>         }
<a name="332" href="#332">332</a>         <strong>return</strong> referencedHost;
<a name="333" href="#333">333</a>     }
<a name="334" href="#334">334</a> 
<a name="335" href="#335">335</a>     <em>/**<em>*</em></em>
<a name="336" href="#336">336</a> <em>     * @return Return the 'SURT' format of this UURI</em>
<a name="337" href="#337">337</a> <em>     */</em>
<a name="338" href="#338">338</a>     <strong>public</strong> String getSurtForm() {
<a name="339" href="#339">339</a>         <strong>if</strong> (surtForm == <strong>null</strong>) {
<a name="340" href="#340">340</a>             surtForm = SURT.fromURI(<strong>this</strong>.toString());
<a name="341" href="#341">341</a>         }
<a name="342" href="#342">342</a>         <strong>return</strong> surtForm;
<a name="343" href="#343">343</a>     }
<a name="344" href="#344">344</a>     
<a name="345" href="#345">345</a>     <em>/**<em>*</em></em>
<a name="346" href="#346">346</a> <em>     * Return the authority minus userinfo (if any).</em>
<a name="347" href="#347">347</a> <em>     * </em>
<a name="348" href="#348">348</a> <em>     * If no userinfo present, just returns the authority.</em>
<a name="349" href="#349">349</a> <em>     * </em>
<a name="350" href="#350">350</a> <em>     * @return The authority stripped of any userinfo if present.</em>
<a name="351" href="#351">351</a> <em>     * @throws URIException</em>
<a name="352" href="#352">352</a> <em>     */</em>
<a name="353" href="#353">353</a> 	<strong>public</strong> String getAuthorityMinusUserinfo()
<a name="354" href="#354">354</a>     throws URIException {
<a name="355" href="#355">355</a>         <strong>if</strong> (<strong>this</strong>.cachedAuthorityMinusUserinfo == <strong>null</strong>) {
<a name="356" href="#356">356</a>             String tmp = getAuthority();
<a name="357" href="#357">357</a>             <strong>if</strong> (tmp != <strong>null</strong> &amp;&amp; tmp.length() > 0) {
<a name="358" href="#358">358</a>             	<strong>int</strong> index = tmp.indexOf('@');
<a name="359" href="#359">359</a>                 <strong>if</strong> (index >= 0 &amp;&amp; index &lt; tmp.length()) {
<a name="360" href="#360">360</a>                     tmp = tmp.substring(index + 1);
<a name="361" href="#361">361</a>                 }
<a name="362" href="#362">362</a>             }
<a name="363" href="#363">363</a>             <strong>this</strong>.cachedAuthorityMinusUserinfo = tmp;
<a name="364" href="#364">364</a>             coalesceHostAuthorityStrings();
<a name="365" href="#365">365</a>         }
<a name="366" href="#366">366</a>         <strong>return</strong> <strong>this</strong>.cachedAuthorityMinusUserinfo;
<a name="367" href="#367">367</a> 	}
<a name="368" href="#368">368</a> 
<a name="369" href="#369">369</a>     <em class="comment">/*<em class="comment"> (non-Javadoc)</em></em>
<a name="370" href="#370">370</a> <em class="comment">     * @see java.lang.CharSequence#length()</em>
<a name="371" href="#371">371</a> <em class="comment">     */</em>
<a name="372" href="#372">372</a>     <strong>public</strong> <strong>int</strong> length() {
<a name="373" href="#373">373</a>         <strong>return</strong> getEscapedURI().length();
<a name="374" href="#374">374</a>     }
<a name="375" href="#375">375</a> 
<a name="376" href="#376">376</a>     <em class="comment">/*<em class="comment"> (non-Javadoc)</em></em>
<a name="377" href="#377">377</a> <em class="comment">     * @see java.lang.CharSequence#charAt(int)</em>
<a name="378" href="#378">378</a> <em class="comment">     */</em>
<a name="379" href="#379">379</a>     <strong>public</strong> <strong>char</strong> <strong>char</strong>At(<strong>int</strong> index) {
<a name="380" href="#380">380</a>         <strong>return</strong> getEscapedURI().charAt(index);
<a name="381" href="#381">381</a>     }
<a name="382" href="#382">382</a> 
<a name="383" href="#383">383</a>     <em class="comment">/*<em class="comment"> (non-Javadoc)</em></em>
<a name="384" href="#384">384</a> <em class="comment">     * @see java.lang.CharSequence#subSequence(int, int)</em>
<a name="385" href="#385">385</a> <em class="comment">     */</em>
<a name="386" href="#386">386</a>     <strong>public</strong> CharSequence subSequence(<strong>int</strong> start, <strong>int</strong> end) {
<a name="387" href="#387">387</a>         <strong>return</strong> getEscapedURI().subSequence(start,end);
<a name="388" href="#388">388</a>     }
<a name="389" href="#389">389</a> 
<a name="390" href="#390">390</a>     <em class="comment">/*<em class="comment"> (non-Javadoc)</em></em>
<a name="391" href="#391">391</a> <em class="comment">     * @see java.lang.Comparable#compareTo(java.lang.Object)</em>
<a name="392" href="#392">392</a> <em class="comment">     */</em>
<a name="393" href="#393">393</a>     <strong>public</strong> <strong>int</strong> compareTo(Object arg0) {
<a name="394" href="#394">394</a>         <strong>return</strong> getEscapedURI().compareTo(arg0.toString());
<a name="395" href="#395">395</a>     }
<a name="396" href="#396">396</a>     
<a name="397" href="#397">397</a>     <em>/**<em>*</em></em>
<a name="398" href="#398">398</a> <em>     * Convenience method for finding the UURI inside an</em>
<a name="399" href="#399">399</a> <em>     * Object likely to have (or be/imply) one.</em>
<a name="400" href="#400">400</a> <em>     * </em>
<a name="401" href="#401">401</a> <em>     * @param o Object that is, has, or implies a UURI</em>
<a name="402" href="#402">402</a> <em>     * @return the UURI found, or null if none</em>
<a name="403" href="#403">403</a> <em>     */</em>
<a name="404" href="#404">404</a>     <strong>public</strong> <strong>static</strong> <a href="../../../org/archive/net/UURI.html">UURI</a> from(Object o) {
<a name="405" href="#405">405</a>         <a href="../../../org/archive/net/UURI.html">UURI</a> u = <strong>null</strong>;
<a name="406" href="#406">406</a>         <strong>if</strong> (o instanceof UURI) {
<a name="407" href="#407">407</a>             u = (UURI)o;
<a name="408" href="#408">408</a>         } <strong>else</strong> <strong>if</strong> (o instanceof CandidateURI) {
<a name="409" href="#409">409</a>             u = ((CandidateURI) o).getUURI();
<a name="410" href="#410">410</a>         } <strong>else</strong> <strong>if</strong> (o instanceof CharSequence) {
<a name="411" href="#411">411</a>             String s = o.toString();
<a name="412" href="#412">412</a>             <strong>try</strong> {
<a name="413" href="#413">413</a>                 u = UURIFactory.getInstance(s);
<a name="414" href="#414">414</a>             } <strong>catch</strong> (URIException e) {
<a name="415" href="#415">415</a>                 LOGGER.log(Level.FINE,<span class="string">"bad URI"</span>,e);
<a name="416" href="#416">416</a>             }
<a name="417" href="#417">417</a>         } 
<a name="418" href="#418">418</a>         <strong>return</strong> u;
<a name="419" href="#419">419</a>     }
<a name="420" href="#420">420</a>     
<a name="421" href="#421">421</a>     <em>/**<em>*</em></em>
<a name="422" href="#422">422</a> <em>     * Test if passed String has likely URI scheme prefix.</em>
<a name="423" href="#423">423</a> <em>     * @param possibleUrl URL string to examine.</em>
<a name="424" href="#424">424</a> <em>     * @return True if passed string looks like it could be an URL.</em>
<a name="425" href="#425">425</a> <em>     */</em>
<a name="426" href="#426">426</a>     <strong>public</strong> <strong>static</strong> <strong>boolean</strong> hasScheme(String possibleUrl) {
<a name="427" href="#427">427</a>         <strong>boolean</strong> result = false;
<a name="428" href="#428">428</a>         <strong>for</strong> (<strong>int</strong> i = 0; i &lt; possibleUrl.length(); i++) {
<a name="429" href="#429">429</a>             <strong>char</strong> c = possibleUrl.<strong>char</strong>At(i);
<a name="430" href="#430">430</a>             <strong>if</strong> (c == ':') {
<a name="431" href="#431">431</a>                 <strong>if</strong> (i != 0) {
<a name="432" href="#432">432</a>                     result = <strong>true</strong>;
<a name="433" href="#433">433</a>                 }
<a name="434" href="#434">434</a>                 <strong>break</strong>;
<a name="435" href="#435">435</a>             }
<a name="436" href="#436">436</a>             <strong>if</strong> (!scheme.get(c)) {
<a name="437" href="#437">437</a>                 <strong>break</strong>;
<a name="438" href="#438">438</a>             }
<a name="439" href="#439">439</a>         }
<a name="440" href="#440">440</a>         <strong>return</strong> result;
<a name="441" href="#441">441</a>     }
<a name="442" href="#442">442</a>     
<a name="443" href="#443">443</a>     <em>/**<em>*</em></em>
<a name="444" href="#444">444</a> <em>     * @param pathOrUri A file path or a URI.</em>
<a name="445" href="#445">445</a> <em>     * @return Path parsed from passed &lt;code>pathOrUri&lt;/code>.</em>
<a name="446" href="#446">446</a> <em>     * @throws URISyntaxException</em>
<a name="447" href="#447">447</a> <em>     */</em>
<a name="448" href="#448">448</a>     <strong>public</strong> <strong>static</strong> String parseFilename(<strong>final</strong> String pathOrUri)
<a name="449" href="#449">449</a>     throws URISyntaxException {
<a name="450" href="#450">450</a>         String path = pathOrUri;
<a name="451" href="#451">451</a>         <strong>if</strong> (UURI.hasScheme(pathOrUri)) {
<a name="452" href="#452">452</a>             URI url = <strong>new</strong> URI(pathOrUri);
<a name="453" href="#453">453</a>             path = url.getPath();
<a name="454" href="#454">454</a>         }
<a name="455" href="#455">455</a>         <strong>return</strong> (<strong>new</strong> File(path)).getName();
<a name="456" href="#456">456</a>     }
<a name="457" href="#457">457</a> }
</pre>
<hr/><div id="footer">This page was automatically generated by <a href="http://maven.apache.org/">Maven</a></div></body>
</html>