deciderulesequencetest.html
来自「网络爬虫开源代码」· HTML 代码 · 共 510 行 · 第 1/4 页
HTML
510 行
<a name="248" href="#248">248</a> <strong>public</strong> <strong>void</strong> testTooManyPathSegments()<a name="249" href="#249">249</a> throws InvalidAttributeValueException, URIException {<a name="250" href="#250">250</a> addDecideRule(<strong>new</strong> <a href="../../../../org/archive/crawler/deciderules/TooManyPathSegmentsDecideRule.html">TooManyPathSegmentsDecideRule</a>(<span class="string">"SEGMENTS"</span>));<a name="251" href="#251">251</a> <strong>final</strong> <strong>int</strong> max =<a name="252" href="#252">252</a> TooManyPathSegmentsDecideRule.DEFAULT_MAX_PATH_DEPTH.intValue();<a name="253" href="#253">253</a> StringBuffer baseUri = <strong>new</strong> StringBuffer(<span class="string">"http://archive.org"</span>);<a name="254" href="#254">254</a> <strong>for</strong> (<strong>int</strong> i = 0; i < max; i++) {<a name="255" href="#255">255</a> baseUri.append('/');<a name="256" href="#256">256</a> baseUri.append(Integer.toString(i + 1));<a name="257" href="#257">257</a> }<a name="258" href="#258">258</a> <a href="../../../../org/archive/net/UURI.html">UURI</a> uuri = UURIFactory.getInstance(baseUri.toString());<a name="259" href="#259">259</a> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a> candidate = <strong>new</strong> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a>(uuri);<a name="260" href="#260">260</a> Object decision = <strong>this</strong>.rule.decisionFor(candidate);<a name="261" href="#261">261</a> assertTrue(<span class="string">"Expect "</span> + DecideRule.PASS + <span class="string">" but got "</span> + decision,<a name="262" href="#262">262</a> decision == DecideRule.PASS);<a name="263" href="#263">263</a> baseUri.append(<span class="string">"/x"</span>);<a name="264" href="#264">264</a> uuri = UURIFactory.getInstance(baseUri.toString());<a name="265" href="#265">265</a> candidate = <strong>new</strong> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a>(uuri);<a name="266" href="#266">266</a> decision = <strong>this</strong>.rule.decisionFor(candidate);<a name="267" href="#267">267</a> assertTrue(<span class="string">"Expect "</span> + DecideRule.REJECT + <span class="string">" but got "</span> + decision,<a name="268" href="#268">268</a> decision == DecideRule.REJECT);<a name="269" href="#269">269</a> }<a name="270" href="#270">270</a> <a name="271" href="#271">271</a> <strong>public</strong> <strong>void</strong> testMatchesFilePattern()<a name="272" href="#272">272</a> throws InvalidAttributeValueException, URIException {<a name="273" href="#273">273</a> addDecideRule(<strong>new</strong> <a href="../../../../org/archive/crawler/deciderules/MatchesFilePatternDecideRule.html">MatchesFilePatternDecideRule</a>(<span class="string">"FILE_PATTERN"</span>));<a name="274" href="#274">274</a> StringBuffer baseUri = <strong>new</strong> StringBuffer(<span class="string">"http://archive.org/"</span>);<a name="275" href="#275">275</a> <a href="../../../../org/archive/net/UURI.html">UURI</a> uuri = UURIFactory.getInstance(baseUri.toString() + <span class="string">"ms.doc"</span>);<a name="276" href="#276">276</a> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a> candidate = <strong>new</strong> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a>(uuri);<a name="277" href="#277">277</a> Object decision = <strong>this</strong>.rule.decisionFor(candidate);<a name="278" href="#278">278</a> assertTrue(<span class="string">"Expect "</span> + DecideRule.ACCEPT + <span class="string">" but got "</span> + decision,<a name="279" href="#279">279</a> decision == DecideRule.ACCEPT);<a name="280" href="#280">280</a> uuri = UURIFactory.getInstance(baseUri.toString() + <span class="string">"index.html"</span>);<a name="281" href="#281">281</a> candidate = <strong>new</strong> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a>(uuri);<a name="282" href="#282">282</a> decision = <strong>this</strong>.rule.decisionFor(candidate);<a name="283" href="#283">283</a> assertTrue(<span class="string">"Expect "</span> + DecideRule.PASS + <span class="string">" but got "</span> + decision,<a name="284" href="#284">284</a> decision == DecideRule.PASS);<a name="285" href="#285">285</a> }<a name="286" href="#286">286</a> <a name="287" href="#287">287</a> <strong>public</strong> <strong>void</strong> testNotMatchesFilePattern()<a name="288" href="#288">288</a> throws InvalidAttributeValueException, URIException {<a name="289" href="#289">289</a> addDecideRule(<strong>new</strong> <a href="../../../../org/archive/crawler/deciderules/NotMatchesFilePatternDecideRule.html">NotMatchesFilePatternDecideRule</a>(<span class="string">"NOT_FILE_PATTERN"</span>));<a name="290" href="#290">290</a> StringBuffer baseUri = <strong>new</strong> StringBuffer(<span class="string">"http://archive.org/"</span>);<a name="291" href="#291">291</a> <a href="../../../../org/archive/net/UURI.html">UURI</a> uuri = UURIFactory.getInstance(baseUri.toString() + <span class="string">"ms.doc"</span>);<a name="292" href="#292">292</a> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a> candidate = <strong>new</strong> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a>(uuri);<a name="293" href="#293">293</a> Object decision = <strong>this</strong>.rule.decisionFor(candidate);<a name="294" href="#294">294</a> assertTrue(<span class="string">"Expect "</span> + DecideRule.PASS + <span class="string">" but got "</span> + decision,<a name="295" href="#295">295</a> decision == DecideRule.PASS);<a name="296" href="#296">296</a> uuri = UURIFactory.getInstance(baseUri.toString() + <span class="string">"index.html"</span>);<a name="297" href="#297">297</a> candidate = <strong>new</strong> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a>(uuri);<a name="298" href="#298">298</a> decision = <strong>this</strong>.rule.decisionFor(candidate);<a name="299" href="#299">299</a> assertTrue(<span class="string">"Expect "</span> + DecideRule.ACCEPT + <span class="string">" but got "</span> + decision,<a name="300" href="#300">300</a> decision == DecideRule.ACCEPT);<a name="301" href="#301">301</a> }<a name="302" href="#302">302</a> <a name="303" href="#303">303</a> <strong>protected</strong> <strong>void</strong> testHopLimit(<strong>final</strong> <strong>int</strong> max, <strong>final</strong> <strong>char</strong> pathExpansion,<a name="304" href="#304">304</a> <strong>final</strong> String defaultDecision, <strong>final</strong> String overLimitDecision)<a name="305" href="#305">305</a> throws URIException {<a name="306" href="#306">306</a> <a href="../../../../org/archive/net/UURI.html">UURI</a> uuri = UURIFactory.getInstance(<span class="string">"http://archive.org"</span>);<a name="307" href="#307">307</a> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a> candidate = <strong>new</strong> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a>(uuri);<a name="308" href="#308">308</a> Object decision = <strong>this</strong>.rule.decisionFor(candidate);<a name="309" href="#309">309</a> assertTrue(<span class="string">"Expect "</span> + defaultDecision + <span class="string">" but got "</span> + decision,<a name="310" href="#310">310</a> decision == defaultDecision);<a name="311" href="#311">311</a> StringBuffer path = <strong>new</strong> StringBuffer(max);<a name="312" href="#312">312</a> <strong>for</strong> (<strong>int</strong> i = 0; i < (max - 1); i++) {<a name="313" href="#313">313</a> path.append(pathExpansion);<a name="314" href="#314">314</a> }<a name="315" href="#315">315</a> candidate = <strong>new</strong> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a>(uuri, path.toString(), <strong>null</strong>, <strong>null</strong>);<a name="316" href="#316">316</a> decision = <strong>this</strong>.rule.decisionFor(candidate);<a name="317" href="#317">317</a> assertTrue(<span class="string">"Expect "</span> + defaultDecision + <span class="string">" but got "</span> + decision,<a name="318" href="#318">318</a> decision == defaultDecision);<a name="319" href="#319">319</a> path.append(pathExpansion);<a name="320" href="#320">320</a> candidate = <strong>new</strong> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a>(uuri, path.toString(), <strong>null</strong>, <strong>null</strong>);<a name="321" href="#321">321</a> decision = <strong>this</strong>.rule.decisionFor(candidate);<a name="322" href="#322">322</a> assertTrue(<span class="string">"Expect "</span> + defaultDecision + <span class="string">" but got "</span> + decision,<a name="323" href="#323">323</a> decision == defaultDecision);<a name="324" href="#324">324</a> path.append(pathExpansion);<a name="325" href="#325">325</a> candidate = <strong>new</strong> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a>(uuri, path.toString(), <strong>null</strong>, <strong>null</strong>);<a name="326" href="#326">326</a> decision = <strong>this</strong>.rule.decisionFor(candidate);<a name="327" href="#327">327</a> assertTrue(<span class="string">"Expect "</span> + overLimitDecision + <span class="string">" but got "</span> + decision,<a name="328" href="#328">328</a> decision == overLimitDecision); <a name="329" href="#329">329</a> }<a name="330" href="#330">330</a> <a name="331" href="#331">331</a> <strong>public</strong> <strong>void</strong> testScopePlusOne() <a name="332" href="#332">332</a> throws URIException, InvalidAttributeValueException, <a name="333" href="#333">333</a> AttributeNotFoundException, MBeanException,<a name="334" href="#334">334</a> ReflectionException {<a name="335" href="#335">335</a> <em class="comment">// first test host scope </em><a name="336" href="#336">336</a> <a href="../../../../org/archive/crawler/deciderules/ScopePlusOneDecideRule.html">ScopePlusOneDecideRule</a> t = <strong>new</strong> <a href="../../../../org/archive/crawler/deciderules/ScopePlusOneDecideRule.html">ScopePlusOneDecideRule</a>(<span class="string">"host"</span>);<a name="337" href="#337">337</a> <a href="../../../../org/archive/util/SurtPrefixSet.html">SurtPrefixSet</a> mSet = <strong>new</strong> <a href="../../../../org/archive/util/SurtPrefixSet.html">SurtPrefixSet</a>();<a name="338" href="#338">338</a> mSet.add(SurtPrefixSet.prefixFromPlain(<span class="string">"http://audio.archive.org"</span>));<a name="339" href="#339">339</a> mSet.convertAllPrefixesToHosts();<a name="340" href="#340">340</a> t.surtPrefixes = mSet;<a name="341" href="#341">341</a> <a href="../../../../org/archive/crawler/deciderules/DecideRule.html">DecideRule</a> s = addDecideRule(t);<a name="342" href="#342">342</a> s.setAttribute(<strong>new</strong> Attribute(ScopePlusOneDecideRule.ATTR_SCOPE,<a name="343" href="#343">343</a> ScopePlusOneDecideRule.HOST));<a name="344" href="#344">344</a> <a name="345" href="#345">345</a> <a name="346" href="#346">346</a> <a href="../../../../org/archive/net/UURI.html">UURI</a> uuri =<a name="347" href="#347">347</a> UURIFactory.getInstance(<span class="string">"http://audio.archive.org/examples"</span>);<a name="348" href="#348">348</a> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a> candidate = <strong>new</strong> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a>(uuri);<a name="349" href="#349">349</a> Object decision = <strong>this</strong>.rule.decisionFor(candidate);<a name="350" href="#350">350</a> assertTrue(<span class="string">"URI Expect "</span> + DecideRule.ACCEPT + <span class="string">" for "</span> + candidate +<a name="351" href="#351">351</a> <span class="string">" but got "</span> + decision, decision == DecideRule.ACCEPT); <a name="352" href="#352">352</a> <a href="../../../../org/archive/net/UURI.html">UURI</a> uuriOne = UURIFactory.getInstance(<span class="string">"http://movies.archive.org"</span>);<a name="353" href="#353">353</a> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a> plusOne = <strong>new</strong> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a>(uuriOne);<a name="354" href="#354">354</a> plusOne.setVia(uuri);<a name="355" href="#355">355</a> decision = <strong>this</strong>.rule.decisionFor(plusOne);<a name="356" href="#356">356</a> assertTrue(<span class="string">"PlusOne Expect "</span> + DecideRule.ACCEPT + <span class="string">" for "</span> + plusOne +<a name="357" href="#357">357</a> <span class="string">" with via "</span> + plusOne.flattenVia() + <span class="string">" but got "</span> + decision,<a name="358" href="#358">358</a> decision == DecideRule.ACCEPT);<a name="359" href="#359">359</a> <a href="../../../../org/archive/net/UURI.html">UURI</a> uuriTwo = UURIFactory.getInstance(<span class="string">"http://sloan.archive.org"</span>);<a name="360" href="#360">360</a> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a> plusTwo = <strong>new</strong> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a>(uuriTwo);<a name="361" href="#361">361</a> plusTwo.setVia(uuriOne);<a name="362" href="#362">362</a> decision = <strong>this</strong>.rule.decisionFor(plusTwo);<a name="363" href="#363">363</a> assertTrue(<span class="string">"PlusTwo Expect "</span> + DecideRule.PASS + <span class="string">" for "</span> + plusTwo +<a name="364" href="#364">364</a> <span class="string">" with via "</span> + plusTwo.flattenVia() + <span class="string">" but got "</span> + decision,<a name="365" href="#365">365</a> decision == DecideRule.PASS); <a name="366" href="#366">366</a> <a name="367" href="#367">367</a> <a name="368" href="#368">368</a> <em class="comment">//now test domain scope</em><a name="369" href="#369">369</a> <a href="../../../../org/archive/crawler/deciderules/ScopePlusOneDecideRule.html">ScopePlusOneDecideRule</a> u = <strong>new</strong> <a href="../../../../org/archive/crawler/deciderules/ScopePlusOneDecideRule.html">ScopePlusOneDecideRule</a>(<span class="string">"domain"</span>);<a name="370" href="#370">370</a> <a href="../../../../org/archive/util/SurtPrefixSet.html">SurtPrefixSet</a> mSet1 = <strong>new</strong> <a href="../../../../org/archive/util/SurtPrefixSet.html">SurtPrefixSet</a>();<a name="371" href="#371">371</a> mSet1.add(SurtPrefixSet.prefixFromPlain(<span class="string">"archive.org"</span>));<a name="372" href="#372">372</a> mSet1.convertAllPrefixesToDomains();<a name="373" href="#373">373</a> u.surtPrefixes = mSet1;<a name="374" href="#374">374</a> <a href="../../../../org/archive/crawler/deciderules/DecideRule.html">DecideRule</a> v = addDecideRule(u);<a name="375" href="#375">375</a> v.setAttribute(<strong>new</strong> Attribute(ScopePlusOneDecideRule.ATTR_SCOPE,
⌨️ 快捷键说明
复制代码Ctrl + C
搜索代码Ctrl + F
全屏模式F11
增大字号Ctrl + =
减小字号Ctrl + -
显示快捷键?