uurifactorytest.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="270" href="#270">270</a>     throws URIException {
<a name="271" href="#271">271</a>         <a href="../../../org/archive/net/UURI.html">UURI</a> uuri = UURIFactory.getInstance(original);
<a name="272" href="#272">272</a>         assertEquals(original, uuri.toString());
<a name="273" href="#273">273</a>         <strong>return</strong> uuri;
<a name="274" href="#274">274</a>     }
<a name="275" href="#275">275</a>     
<a name="276" href="#276">276</a> 	<strong>public</strong> <strong>final</strong> <strong>void</strong> testTrimSpaceNBSP() throws URIException {
<a name="277" href="#277">277</a> 		<strong>final</strong> String uri = <span class="string">"   http://archive.org/DIR WITH SPACES/"</span> +
<a name="278" href="#278">278</a> 		UURIFactory.NBSP + <span class="string">"home.html    "</span> + UURIFactory.NBSP + <span class="string">"   "</span>;
<a name="279" href="#279">279</a> 		<strong>final</strong> String tgtUri =
<a name="280" href="#280">280</a> 			<span class="string">"http://archive.org/DIR%20WITH%20SPACES/%20home.html"</span>;
<a name="281" href="#281">281</a> 		UURI uuri = UURIFactory.getInstance(uri);
<a name="282" href="#282">282</a> 		assertTrue(<span class="string">"Not equal "</span> + uuri.toString(),
<a name="283" href="#283">283</a> 				uuri.toString().equals(tgtUri));
<a name="284" href="#284">284</a> 	}
<a name="285" href="#285">285</a> 	
<a name="286" href="#286">286</a> 	<em>/**<em>*</em></em>
<a name="287" href="#287">287</a> <em>	 * Test space plus encoding ([ 1010966 ] crawl.log has URIs with spaces in them).</em>
<a name="288" href="#288">288</a> <em>	 * See &lt;a href="<a href="http://sourceforge.net/tracker/index.php?func=detail&amp;aid=1010966&amp;group_id=73833&amp;atid=539099" target="alexandria_uri">http://sourceforge.net/tracker/index.php?func=detail&amp;aid=1010966&amp;group_id=73833&amp;atid=539099</a>">[ 1010966 ] crawl.log has URIs with spaces in them&lt;/a>.</em>
<a name="289" href="#289">289</a> <em>	 * @throws URIException</em>
<a name="290" href="#290">290</a> <em>	 */</em>
<a name="291" href="#291">291</a> 	<strong>public</strong> <strong>final</strong> <strong>void</strong> testSpaceDoubleEncoding() throws URIException {
<a name="292" href="#292">292</a> 		<strong>final</strong> String uri = <span class="string">"http://www.brook.edu/i.html? %20taxonomy=Politics"</span>;
<a name="293" href="#293">293</a> 		<strong>final</strong> String encodedUri =
<a name="294" href="#294">294</a> 			<span class="string">"http://www.brook.edu/i.html?%20%20taxonomy=Politics"</span>;
<a name="295" href="#295">295</a> 		UURI uuri = UURIFactory.getInstance(uri, <span class="string">"ISO-8859-1"</span>);
<a name="296" href="#296">296</a> 		assertTrue(<span class="string">"Not equal "</span> + uuri.toString(),
<a name="297" href="#297">297</a> 				uuri.toString().equals(encodedUri));
<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>	 * Test for doubly-encoded sequences.</em>
<a name="302" href="#302">302</a> <em>	 * See &lt;a href="https://sourceforge.net/tracker/index.php?func=detail&amp;aid=966219&amp;group_id=73833&amp;atid=539099">[ 966219 ] UURI doubly-encodes %XX sequences&lt;/a>.</em>
<a name="303" href="#303">303</a> <em>	 * @throws URIException</em>
<a name="304" href="#304">304</a> <em>	 */</em>
<a name="305" href="#305">305</a> 	<strong>public</strong> <strong>final</strong> <strong>void</strong> testDoubleEncoding() throws URIException {
<a name="306" href="#306">306</a> 		<strong>final</strong> <strong>char</strong> ae = '\u00E6';
<a name="307" href="#307">307</a> 		<strong>final</strong> String uri = <span class="string">"http://archive.org/DIR WITH SPACES/home"</span> +
<a name="308" href="#308">308</a> 		    ae + <span class="string">".html"</span>;
<a name="309" href="#309">309</a> 		<strong>final</strong> String encodedUri =
<a name="310" href="#310">310</a> 			<span class="string">"http://archive.org/DIR%20WITH%20SPACES/home%E6.html"</span>;
<a name="311" href="#311">311</a> 		UURI uuri = UURIFactory.getInstance(uri, <span class="string">"ISO-8859-1"</span>);
<a name="312" href="#312">312</a> 		assertEquals(<span class="string">"single encoding"</span>, encodedUri, uuri.toString());
<a name="313" href="#313">313</a> 		<em class="comment">// Dbl-encodes.</em>
<a name="314" href="#314">314</a> 		uuri = UURIFactory.getInstance(uuri.toString(), <span class="string">"ISO-8859-1"</span>);
<a name="315" href="#315">315</a> 		uuri = UURIFactory.getInstance(uuri.toString(), <span class="string">"ISO-8859-1"</span>);
<a name="316" href="#316">316</a> 		assertEquals(<span class="string">"double encoding"</span>, encodedUri, uuri.toString());
<a name="317" href="#317">317</a> 		<em class="comment">// Do default utf-8 test.</em>
<a name="318" href="#318">318</a> 		uuri = UURIFactory.getInstance(uri);
<a name="319" href="#319">319</a> 		<strong>final</strong> String encodedUtf8Uri =
<a name="320" href="#320">320</a> 			<span class="string">"http://archive.org/DIR%20WITH%20SPACES/home%C3%A6.html"</span>;
<a name="321" href="#321">321</a> 		assertEquals(<span class="string">"Not equal utf8"</span>, encodedUtf8Uri, uuri.toString());      
<a name="322" href="#322">322</a> 		<em class="comment">// Now dbl-encode.</em>
<a name="323" href="#323">323</a> 		uuri = UURIFactory.getInstance(uuri.toString());
<a name="324" href="#324">324</a> 		uuri = UURIFactory.getInstance(uuri.toString());
<a name="325" href="#325">325</a> 		assertEquals(<span class="string">"Not equal (dbl-encoding) utf8"</span>, encodedUtf8Uri, uuri.toString());
<a name="326" href="#326">326</a> 	}
<a name="327" href="#327">327</a> 	
<a name="328" href="#328">328</a> 	<em>/**<em>*</em></em>
<a name="329" href="#329">329</a> <em>	 * Test for syntax errors stop page parsing.</em>
<a name="330" href="#330">330</a> <em>	 * @see &lt;a href="https://sourceforge.net/tracker/?func=detail&amp;aid=788219&amp;group_id=73833&amp;atid=539099">[ 788219 ] URI Syntax Errors stop page parsing&lt;/a></em>
<a name="331" href="#331">331</a> <em>	 * @throws URIException</em>
<a name="332" href="#332">332</a> <em>	 */</em>
<a name="333" href="#333">333</a> 	<strong>public</strong> <strong>final</strong> <strong>void</strong> testThreeSlashes() throws URIException {
<a name="334" href="#334">334</a> 		UURI goodURI = UURIFactory.
<a name="335" href="#335">335</a> 		getInstance(<span class="string">"http://lcweb.loc.gov/rr/goodtwo.html"</span>);
<a name="336" href="#336">336</a> 		String uuri = <span class="string">"http:///lcweb.loc.gov/rr/goodtwo.html"</span>;
<a name="337" href="#337">337</a> 		UURI rewrittenURI = UURIFactory.getInstance(uuri);
<a name="338" href="#338">338</a> 		assertTrue(<span class="string">"Not equal "</span> + goodURI + <span class="string">", "</span> + uuri,
<a name="339" href="#339">339</a> 				goodURI.toString().equals(rewrittenURI.toString()));
<a name="340" href="#340">340</a> 		uuri = <span class="string">"http:////lcweb.loc.gov/rr/goodtwo.html"</span>;
<a name="341" href="#341">341</a> 		rewrittenURI = UURIFactory.getInstance(uuri);
<a name="342" href="#342">342</a> 		assertTrue(<span class="string">"Not equal "</span> + goodURI + <span class="string">", "</span> + uuri,
<a name="343" href="#343">343</a> 				goodURI.toString().equals(rewrittenURI.toString()));
<a name="344" href="#344">344</a> 		<em class="comment">// Check https.</em>
<a name="345" href="#345">345</a> 		goodURI = UURIFactory.
<a name="346" href="#346">346</a> 		getInstance(<span class="string">"https://lcweb.loc.gov/rr/goodtwo.html"</span>);
<a name="347" href="#347">347</a> 		uuri = <span class="string">"https:////lcweb.loc.gov/rr/goodtwo.html"</span>;
<a name="348" href="#348">348</a> 		rewrittenURI = UURIFactory.getInstance(uuri);
<a name="349" href="#349">349</a> 		assertTrue(<span class="string">"Not equal "</span> + goodURI + <span class="string">", "</span> + uuri,
<a name="350" href="#350">350</a> 				goodURI.toString().equals(rewrittenURI.toString()));
<a name="351" href="#351">351</a> 	}
<a name="352" href="#352">352</a> 	
<a name="353" href="#353">353</a> 	<strong>public</strong> <strong>final</strong> <strong>void</strong> testNoScheme() {
<a name="354" href="#354">354</a> 		<strong>boolean</strong> expectedException = false;
<a name="355" href="#355">355</a> 		String uuri = <span class="string">"www.loc.gov/rr/european/egw/polishex.html"</span>;
<a name="356" href="#356">356</a> 		<strong>try</strong> {
<a name="357" href="#357">357</a> 			UURIFactory.getInstance(uuri);
<a name="358" href="#358">358</a> 		} <strong>catch</strong> (URIException e) {
<a name="359" href="#359">359</a> 			<em class="comment">// Expected exception.</em>
<a name="360" href="#360">360</a> 			expectedException = <strong>true</strong>;
<a name="361" href="#361">361</a> 		}
<a name="362" href="#362">362</a> 		assertTrue(<span class="string">"Didn't get expected exception: "</span> + uuri, 
<a name="363" href="#363">363</a> 				expectedException); 
<a name="364" href="#364">364</a> 	}
<a name="365" href="#365">365</a> 	
<a name="366" href="#366">366</a> 	<strong>public</strong> <strong>final</strong> <strong>void</strong> testRelative() throws URIException {
<a name="367" href="#367">367</a> 		UURI uuriTgt = UURIFactory.
<a name="368" href="#368">368</a> 		getInstance(<span class="string">"http://archive.org:83/home.html"</span>);
<a name="369" href="#369">369</a> 		UURI uri = UURIFactory.
<a name="370" href="#370">370</a> 		getInstance(<span class="string">"http://archive.org:83/one/two/three.html"</span>);
<a name="371" href="#371">371</a> 		UURI uuri = UURIFactory.
<a name="372" href="#372">372</a> 		getInstance(uri, <span class="string">"/home.html"</span>);
<a name="373" href="#373">373</a> 		assertTrue(<span class="string">"Not equal"</span>,
<a name="374" href="#374">374</a> 				uuriTgt.toString().equals(uuri.toString()));
<a name="375" href="#375">375</a> 	}
<a name="376" href="#376">376</a> 	
<a name="377" href="#377">377</a> 	<em>/**<em>*</em></em>
<a name="378" href="#378">378</a> <em>	 * Test that an empty uuri does the right thing -- that we get back the</em>
<a name="379" href="#379">379</a> <em>	 * base.</em>
<a name="380" href="#380">380</a> <em>	 *</em>
<a name="381" href="#381">381</a> <em>	 * @throws URIException</em>
<a name="382" href="#382">382</a> <em>	 */</em>
<a name="383" href="#383">383</a> 	<strong>public</strong> <strong>final</strong> <strong>void</strong> testRelativeEmpty() throws URIException {
<a name="384" href="#384">384</a> 		UURI uuriTgt = UURIFactory.
<a name="385" href="#385">385</a> 		getInstance(<span class="string">"http://archive.org:83/one/two/three.html"</span>);
<a name="386" href="#386">386</a> 		UURI uri = UURIFactory.
<a name="387" href="#387">387</a> 		getInstance(<span class="string">"http://archive.org:83/one/two/three.html"</span>);
<a name="388" href="#388">388</a> 		UURI uuri = UURIFactory.
<a name="389" href="#389">389</a> 		getInstance(uri, <span class="string">""</span>);
<a name="390" href="#390">390</a> 		assertTrue(<span class="string">"Empty length don't work"</span>,
<a name="391" href="#391">391</a> 				uuriTgt.toString().equals(uuri.toString()));
<a name="392" href="#392">392</a> 	}
<a name="393" href="#393">393</a> 	
<a name="394" href="#394">394</a> 	<strong>public</strong> <strong>final</strong> <strong>void</strong> testAbsolute() throws URIException {
<a name="395" href="#395">395</a> 		UURI uuriTgt = UURIFactory.
<a name="396" href="#396">396</a> 		getInstance(<span class="string">"http://archive.org:83/home.html"</span>);
<a name="397" href="#397">397</a> 		UURI uri = UURIFactory.
<a name="398" href="#398">398</a> 		getInstance(<span class="string">"http://archive.org:83/one/two/three.html"</span>);
<a name="399" href="#399">399</a> 		UURI uuri = UURIFactory.
<a name="400" href="#400">400</a> 		getInstance(uri, <span class="string">"http://archive.org:83/home.html"</span>);
<a name="401" href="#401">401</a> 		assertTrue(<span class="string">"Not equal"</span>,
<a name="402" href="#402">402</a> 				uuriTgt.toString().equals(uuri.toString()));
<a name="403" href="#403">403</a> 	}
<a name="404" href="#404">404</a> 	
<a name="405" href="#405">405</a> 	<em>/**<em>*</em></em>
<a name="406" href="#406">406</a> <em>	 * Test for [ 962892 ] UURI accepting/creating unUsable URIs (bad hosts).</em>
<a name="407" href="#407">407</a> <em>	 * @see &lt;a href="https://sourceforge.net/tracker/?func=detail&amp;atid=539099&amp;aid=962892&amp;group_id=73833">[ 962892 ] UURI accepting/creating unUsable URIs (bad hosts)&lt;/a></em>
<a name="408" href="#408">408</a> <em>	 */</em>