surtprefixset.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="272" href="#272">272</a>         s = s.replaceAll(<span class="string">"^(.*//.*/)[^/]*"</span>,<span class="string">"$1"</span>);
<a name="273" href="#273">273</a>         <em class="comment">// Strip trailing ")", if present and NO path (no 3rd slash).</em>
<a name="274" href="#274">274</a>         <strong>if</strong> (!s.endsWith(<span class="string">"/"</span>)) {
<a name="275" href="#275">275</a>             s = s.replaceAll(<span class="string">"^(.*)&#47;&#47;)"</span>,<span class="string">"$1"</span>);
<a name="276" href="#276">276</a>         }
<a name="277" href="#277">277</a>         <strong>return</strong> s;
<a name="278" href="#278">278</a>     }
<a name="279" href="#279">279</a> 
<a name="280" href="#280">280</a>     <em>/**<em>*</em></em>
<a name="281" href="#281">281</a> <em>     * Calculate the SURT form URI to use as a candidate against prefixes</em>
<a name="282" href="#282">282</a> <em>     * from the given Object (CandidateURI or UURI)</em>
<a name="283" href="#283">283</a> <em>     * </em>
<a name="284" href="#284">284</a> <em>     * @param object CandidateURI or UURI</em>
<a name="285" href="#285">285</a> <em>     * @return SURT form of URI for evaluation, or null if unavailable</em>
<a name="286" href="#286">286</a> <em>     */</em>
<a name="287" href="#287">287</a>     <strong>public</strong> <strong>static</strong> String getCandidateSurt(Object object) {
<a name="288" href="#288">288</a>         <a href="../../../org/archive/net/UURI.html">UURI</a> u = UURI.from(object);
<a name="289" href="#289">289</a>         <strong>if</strong> (u == <strong>null</strong>) {
<a name="290" href="#290">290</a>             <strong>return</strong> <strong>null</strong>;
<a name="291" href="#291">291</a>         }
<a name="292" href="#292">292</a>         String candidateSurt = u.getSurtForm();
<a name="293" href="#293">293</a>         <em class="comment">// also want to treat https as http</em>
<a name="294" href="#294">294</a>         candidateSurt = coerceFromHttpsForComparison(candidateSurt);
<a name="295" href="#295">295</a>         <strong>return</strong> candidateSurt;
<a name="296" href="#296">296</a>     }
<a name="297" href="#297">297</a>     <em>/**<em>*</em></em>
<a name="298" href="#298">298</a> <em>     * @param fw</em>
<a name="299" href="#299">299</a> <em>     * @throws IOException</em>
<a name="300" href="#300">300</a> <em>     */</em>
<a name="301" href="#301">301</a>     <strong>public</strong> <strong>void</strong> exportTo(FileWriter fw) throws IOException {
<a name="302" href="#302">302</a>         Iterator iter = <strong>this</strong>.iterator();
<a name="303" href="#303">303</a>         <strong>while</strong>(iter.hasNext()) {
<a name="304" href="#304">304</a>             fw.write((String)iter.next() + <span class="string">"\n"</span>);
<a name="305" href="#305">305</a>         }
<a name="306" href="#306">306</a>     }
<a name="307" href="#307">307</a> 
<a name="308" href="#308">308</a>     <em>/**<em>*</em></em>
<a name="309" href="#309">309</a> <em>     * Changes all prefixes so that they enforce an exact host. For</em>
<a name="310" href="#310">310</a> <em>     * prefixes that already include a ')', this means discarding </em>
<a name="311" href="#311">311</a> <em>     * anything after ')' (path info). For prefixes that don't include</em>
<a name="312" href="#312">312</a> <em>     * a ')' -- domain prefixes open to subdomains -- add the closing</em>
<a name="313" href="#313">313</a> <em>     * ')' (or ",)").  </em>
<a name="314" href="#314">314</a> <em>     */</em>
<a name="315" href="#315">315</a>     <strong>public</strong> <strong>void</strong> convertAllPrefixesToHosts() {
<a name="316" href="#316">316</a>         <a href="../../../org/archive/util/SurtPrefixSet.html">SurtPrefixSet</a> iterCopy = (SurtPrefixSet) <strong>this</strong>.clone();
<a name="317" href="#317">317</a>         Iterator iter = iterCopy.iterator();
<a name="318" href="#318">318</a>         <strong>while</strong> (iter.hasNext()) {
<a name="319" href="#319">319</a>             String prefix = (String) iter.next();
<a name="320" href="#320">320</a>             String convPrefix = convertPrefixToHost(prefix);
<a name="321" href="#321">321</a>             <strong>if</strong>(prefix!=convPrefix) {
<a name="322" href="#322">322</a>             	<em class="comment">// if returned value not unchanged, update set</em>
<a name="323" href="#323">323</a>             	<strong>this</strong>.remove(prefix);
<a name="324" href="#324">324</a>             	<strong>this</strong>.add(convPrefix);
<a name="325" href="#325">325</a>             }
<a name="326" href="#326">326</a>         }
<a name="327" href="#327">327</a>     }
<a name="328" href="#328">328</a>     
<a name="329" href="#329">329</a>     <strong>public</strong> <strong>static</strong> String convertPrefixToHost(String prefix) {
<a name="330" href="#330">330</a>         <strong>if</strong>(prefix.endsWith(<span class="string">")"</span>)) {
<a name="331" href="#331">331</a>             <strong>return</strong> prefix; <em class="comment">// no change necessary</em>
<a name="332" href="#332">332</a>         }
<a name="333" href="#333">333</a>         <strong>if</strong>(prefix.indexOf(')')&lt;0) {
<a name="334" href="#334">334</a>             <em class="comment">// open-ended domain prefix</em>
<a name="335" href="#335">335</a>             <strong>if</strong>(!prefix.endsWith(<span class="string">","</span>)) {
<a name="336" href="#336">336</a>                 prefix += <span class="string">","</span>;
<a name="337" href="#337">337</a>             }
<a name="338" href="#338">338</a>             prefix += <span class="string">")"</span>;
<a name="339" href="#339">339</a>         } <strong>else</strong> {
<a name="340" href="#340">340</a>             <em class="comment">// prefix with excess path-info</em>
<a name="341" href="#341">341</a>             prefix = prefix.substring(0,prefix.indexOf(')')+1);
<a name="342" href="#342">342</a>         }
<a name="343" href="#343">343</a>         <strong>return</strong> prefix;
<a name="344" href="#344">344</a>     }
<a name="345" href="#345">345</a> 
<a name="346" href="#346">346</a>     <em>/**<em>*</em></em>
<a name="347" href="#347">347</a> <em>     * Changes all prefixes so that they only enforce a general</em>
<a name="348" href="#348">348</a> <em>     * domain (allowing subdomains).For prefixes that don't include</em>
<a name="349" href="#349">349</a> <em>     * a ')', no change is necessary. For others, truncate everything</em>
<a name="350" href="#350">350</a> <em>     * from the ')' onward. Additionally, truncate off "www," if it</em>
<a name="351" href="#351">351</a> <em>     * appears.</em>
<a name="352" href="#352">352</a> <em>     */</em>
<a name="353" href="#353">353</a>     <strong>public</strong> <strong>void</strong> convertAllPrefixesToDomains() {
<a name="354" href="#354">354</a>         <a href="../../../org/archive/util/SurtPrefixSet.html">SurtPrefixSet</a> iterCopy = (SurtPrefixSet) <strong>this</strong>.clone();
<a name="355" href="#355">355</a>         Iterator iter = iterCopy.iterator();
<a name="356" href="#356">356</a>         <strong>while</strong> (iter.hasNext()) {
<a name="357" href="#357">357</a>             String prefix = (String) iter.next();
<a name="358" href="#358">358</a>             String convPrefix = convertPrefixToDomain(prefix);
<a name="359" href="#359">359</a>             <strong>if</strong>(prefix!=convPrefix) {
<a name="360" href="#360">360</a>             	<em class="comment">// if returned value not unchanged, update set</em>
<a name="361" href="#361">361</a>             	<strong>this</strong>.remove(prefix);
<a name="362" href="#362">362</a>             	<strong>this</strong>.add(convPrefix);
<a name="363" href="#363">363</a>             }
<a name="364" href="#364">364</a>         } 
<a name="365" href="#365">365</a>     }
<a name="366" href="#366">366</a>     
<a name="367" href="#367">367</a>     <strong>public</strong> <strong>static</strong> String convertPrefixToDomain(String prefix) {
<a name="368" href="#368">368</a>         <strong>if</strong>(prefix.indexOf(')')>=0) {
<a name="369" href="#369">369</a>             prefix = prefix.substring(0,prefix.indexOf(')'));
<a name="370" href="#370">370</a>         }
<a name="371" href="#371">371</a>         <em class="comment">// strip 'www,' when present</em>
<a name="372" href="#372">372</a>         <strong>if</strong>(prefix.endsWith(<span class="string">"www,"</span>)) {
<a name="373" href="#373">373</a>             prefix = prefix.substring(0,prefix.length()-4);
<a name="374" href="#374">374</a>         }
<a name="375" href="#375">375</a>         <strong>return</strong> prefix;
<a name="376" href="#376">376</a>     }
<a name="377" href="#377">377</a>     
<a name="378" href="#378">378</a>     <em>/**<em>*</em></em>
<a name="379" href="#379">379</a> <em>     * Allow class to be used as a command-line tool for converting </em>
<a name="380" href="#380">380</a> <em>     * URL lists (or naked host or host/path fragments implied</em>
<a name="381" href="#381">381</a> <em>     * to be HTTP URLs) to implied SURT prefix form. </em>
<a name="382" href="#382">382</a> <em>     * </em>
<a name="383" href="#383">383</a> <em>     * Read from stdin or first file argument. Writes to stdout. </em>
<a name="384" href="#384">384</a> <em>     *</em>
<a name="385" href="#385">385</a> <em>     * @param args cmd-line arguments: may include input file</em>
<a name="386" href="#386">386</a> <em>     * @throws IOException</em>
<a name="387" href="#387">387</a> <em>     */</em>
<a name="388" href="#388">388</a>     <strong>public</strong> <strong>static</strong> <strong>void</strong> main(String[] args) throws IOException {
<a name="389" href="#389">389</a>         InputStream in = args.length > 0 ? <strong>new</strong> BufferedInputStream(
<a name="390" href="#390">390</a>                 <strong>new</strong> FileInputStream(args[0])) : System.in;
<a name="391" href="#391">391</a>         PrintStream out = args.length > 1 ? <strong>new</strong> PrintStream(
<a name="392" href="#392">392</a>                 <strong>new</strong> BufferedOutputStream(<strong>new</strong> FileOutputStream(args[1])))
<a name="393" href="#393">393</a>                 : System.out;
<a name="394" href="#394">394</a>         BufferedReader br =
<a name="395" href="#395">395</a>             <strong>new</strong> BufferedReader(<strong>new</strong> InputStreamReader(in));
<a name="396" href="#396">396</a>         String line;
<a name="397" href="#397">397</a>         <strong>while</strong>((line = br.readLine())!=<strong>null</strong>) {
<a name="398" href="#398">398</a>             <strong>if</strong>(line.indexOf(<span class="string">"#"</span>)>0) line=line.substring(0,line.indexOf(<span class="string">"#"</span>));
<a name="399" href="#399">399</a>             line = line.trim();
<a name="400" href="#400">400</a>             <strong>if</strong>(line.length()==0) <strong>continue</strong>;
<a name="401" href="#401">401</a>             out.println(prefixFromPlain(line));
<a name="402" href="#402">402</a>         }
<a name="403" href="#403">403</a>         br.close();
<a name="404" href="#404">404</a>         out.close();
<a name="405" href="#405">405</a>     }
<a name="406" href="#406">406</a> }
</pre>
<hr/><div id="footer">This page was automatically generated by <a href="http://maven.apache.org/">Maven</a></div></body>
</html>