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="376" href="#376">376</a> ScopePlusOneDecideRule.DOMAIN));<a name="377" href="#377">377</a> <a name="378" href="#378">378</a> decision = <strong>this</strong>.rule.decisionFor(candidate);<a name="379" href="#379">379</a> assertTrue(<span class="string">"Domain: URI Expect "</span> + DecideRule.ACCEPT + <span class="string">" for "</span> +<a name="380" href="#380">380</a> candidate + <span class="string">" but got "</span> + decision, decision == DecideRule.ACCEPT); <a name="381" href="#381">381</a> decision = <strong>this</strong>.rule.decisionFor(plusOne);<a name="382" href="#382">382</a> assertTrue(<span class="string">"Domain: PlusOne Expect "</span> + DecideRule.ACCEPT + <span class="string">" for "</span> +<a name="383" href="#383">383</a> plusOne + <span class="string">" with via "</span> + plusOne.flattenVia() + <span class="string">" but got "</span> +<a name="384" href="#384">384</a> decision, decision == DecideRule.ACCEPT);<a name="385" href="#385">385</a> decision = <strong>this</strong>.rule.decisionFor(plusTwo);<a name="386" href="#386">386</a> assertTrue(<span class="string">"Domain: PlusTwo Expect "</span> + DecideRule.ACCEPT + <span class="string">" for "</span> +<a name="387" href="#387">387</a> plusTwo + <span class="string">" with via "</span> + plusTwo.flattenVia() + <span class="string">" but got "</span> +<a name="388" href="#388">388</a> decision, decision == DecideRule.ACCEPT); <a name="389" href="#389">389</a> <a href="../../../../org/archive/net/UURI.html">UURI</a> uuriThree = UURIFactory.getInstance(<span class="string">"http://sloan.org"</span>);<a name="390" href="#390">390</a> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a> plusThree = <strong>new</strong> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a>(uuriThree);<a name="391" href="#391">391</a> plusThree.setVia(uuriTwo);<a name="392" href="#392">392</a> decision = <strong>this</strong>.rule.decisionFor(plusThree);<a name="393" href="#393">393</a> assertTrue(<span class="string">"Domain: PlusThree Expect "</span> + DecideRule.ACCEPT + <span class="string">" for "</span> +<a name="394" href="#394">394</a> plusThree + <span class="string">" with via "</span> + plusThree.flattenVia() + <span class="string">" but got "</span> +<a name="395" href="#395">395</a> decision, decision == DecideRule.ACCEPT); <a name="396" href="#396">396</a> <a href="../../../../org/archive/net/UURI.html">UURI</a> uuriFour = UURIFactory.getInstance(<span class="string">"http://example.com"</span>);<a name="397" href="#397">397</a> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a> plusFour = <strong>new</strong> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a>(uuriFour);<a name="398" href="#398">398</a> plusFour.setVia(uuriThree);<a name="399" href="#399">399</a> decision = <strong>this</strong>.rule.decisionFor(plusFour); <a name="400" href="#400">400</a> assertTrue(<span class="string">"Domain: PlusFour Expect "</span> + DecideRule.PASS + <span class="string">" for "</span> +<a name="401" href="#401">401</a> plusFour + <span class="string">" with via "</span> + plusFour.flattenVia() + <span class="string">" but got "</span> +<a name="402" href="#402">402</a> decision, decision == DecideRule.PASS); <a name="403" href="#403">403</a> } <a name="404" href="#404">404</a> <a name="405" href="#405">405</a> <strong>public</strong> <strong>void</strong> testFilter()<a name="406" href="#406">406</a> throws InvalidAttributeValueException, URIException, AttributeNotFoundException, MBeanException, ReflectionException {<a name="407" href="#407">407</a> FilterDecideRule dr = <strong>new</strong> <a href="../../../../org/archive/crawler/deciderules/FilterDecideRule.html">FilterDecideRule</a>(<a name="408" href="#408">408</a> <span class="string">"FilterDecideRule(ContentTypeRegExpFilter)"</span>);<a name="409" href="#409">409</a> addDecideRule(dr);<a name="410" href="#410">410</a> StringBuffer baseUri = <strong>new</strong> StringBuffer();<a name="411" href="#411">411</a> <a href="../../../../org/archive/net/UURI.html">UURI</a> uuri = UURIFactory.getInstance(<span class="string">"http://example.com/foo"</span>);<a name="412" href="#412">412</a> <a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi = <strong>new</strong> <a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a>(uuri);<a name="413" href="#413">413</a> curi.setContentType(<span class="string">"text/html"</span>);<a name="414" href="#414">414</a> Object decision = <strong>this</strong>.rule.decisionFor(curi);<a name="415" href="#415">415</a> <em class="comment">// default for unconfigured FilterDecideRule is true from (empty)</em><a name="416" href="#416">416</a> <em class="comment">// filters, then ACCEPT because of true</em><a name="417" href="#417">417</a> assertTrue(<span class="string">"Expect "</span> + DecideRule.ACCEPT + <span class="string">" but got "</span> + decision,<a name="418" href="#418">418</a> decision == DecideRule.ACCEPT);<a name="419" href="#419">419</a> <a href="../../../../org/archive/crawler/filter/ContentTypeRegExpFilter.html">ContentTypeRegExpFilter</a> filt = <a name="420" href="#420">420</a> <strong>new</strong> <a href="../../../../org/archive/crawler/filter/ContentTypeRegExpFilter.html">ContentTypeRegExpFilter</a>(<span class="string">"ContentTypeRegExpFilter"</span>,<span class="string">"app.*"</span>);<a name="421" href="#421">421</a> dr.filters.addElement(<strong>null</strong>,filt);<a name="422" href="#422">422</a> decision = <strong>this</strong>.rule.decisionFor(curi);<a name="423" href="#423">423</a> <em class="comment">// filter should now return false, making decision REJECT</em><a name="424" href="#424">424</a> assertTrue(<span class="string">"Expect "</span> + DecideRule.REJECT + <span class="string">" but got "</span> + decision,<a name="425" href="#425">425</a> decision == DecideRule.REJECT);<a name="426" href="#426">426</a> curi.setContentType(<span class="string">"application/octet-stream"</span>);<a name="427" href="#427">427</a> decision = <strong>this</strong>.rule.decisionFor(curi);<a name="428" href="#428">428</a> <em class="comment">// filter should now return true, making decision ACCEPT</em><a name="429" href="#429">429</a> assertTrue(<span class="string">"Expect "</span> + DecideRule.ACCEPT + <span class="string">" but got "</span> + decision,<a name="430" href="#430">430</a> decision == DecideRule.ACCEPT);<a name="431" href="#431">431</a> <em class="comment">// change true answer to "PASS"; use String to simulate settings non-identity</em><a name="432" href="#432">432</a> dr.setAttribute(<strong>new</strong> Attribute(FilterDecideRule.ATTR_TRUE_DECISION,<span class="string">"PASS"</span>));<a name="433" href="#433">433</a> decision = <strong>this</strong>.rule.decisionFor(curi);<a name="434" href="#434">434</a> assertTrue(<span class="string">"Expect "</span> + DecideRule.PASS + <span class="string">" but got "</span> + decision,<a name="435" href="#435">435</a> decision == DecideRule.PASS); <a name="436" href="#436">436</a> }<a name="437" href="#437">437</a> <a name="438" href="#438">438</a> <strong>protected</strong> <a href="../../../../org/archive/crawler/deciderules/DecideRule.html">DecideRule</a> addDecideRule(<a href="../../../../org/archive/crawler/deciderules/DecideRule.html">DecideRule</a> dr)<a name="439" href="#439">439</a> throws InvalidAttributeValueException {<a name="440" href="#440">440</a> <a href="../../../../org/archive/crawler/settings/MapType.html">MapType</a> rules = <strong>this</strong>.rule.getRules(<strong>null</strong>);<a name="441" href="#441">441</a> rules.addElement(<strong>null</strong>, dr);<a name="442" href="#442">442</a> <strong>return</strong> dr;<a name="443" href="#443">443</a> }<a name="444" href="#444">444</a> <a name="445" href="#445">445</a> <strong>public</strong> <strong>void</strong> testContentTypeMatchesRegexpDecideRule() throws Exception{<a name="446" href="#446">446</a> <a href="../../../../org/archive/crawler/deciderules/ContentTypeMatchesRegExpDecideRule.html">ContentTypeMatchesRegExpDecideRule</a> dr = <strong>new</strong> <a href="../../../../org/archive/crawler/deciderules/ContentTypeMatchesRegExpDecideRule.html">ContentTypeMatchesRegExpDecideRule</a>(<span class="string">"CTMREDRtest"</span>);<a name="447" href="#447">447</a> <a href="../../../../org/archive/crawler/deciderules/DecideRule.html">DecideRule</a> v = addDecideRule(dr);<a name="448" href="#448">448</a> <a name="449" href="#449">449</a> v.setAttribute(<strong>new</strong> Attribute(MatchesRegExpDecideRule.ATTR_REGEXP,<span class="string">"text/html"</span>));<a name="450" href="#450">450</a> <a href="../../../../org/archive/net/UURI.html">UURI</a> uuri = UURIFactory.getInstance(<span class="string">"http://www.archive.org"</span>);<a name="451" href="#451">451</a> <a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> crawlUri = <strong>new</strong> <a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a>(uuri);<a name="452" href="#452">452</a> <a name="453" href="#453">453</a> <em class="comment">// no content type - let curi pass</em><a name="454" href="#454">454</a> Object decision = <strong>this</strong>.rule.decisionFor(crawlUri);<a name="455" href="#455">455</a> assertTrue(<span class="string">"URI Expect "</span> + DecideRule.PASS + <span class="string">" for "</span> + crawlUri +<a name="456" href="#456">456</a> <span class="string">" but got "</span> + decision, decision == DecideRule.PASS);<a name="457" href="#457">457</a> <a name="458" href="#458">458</a> <em class="comment">// non-matching content type - let curi pass</em><a name="459" href="#459">459</a> crawlUri.setContentType(<span class="string">"application/pdf"</span>);<a name="460" href="#460">460</a> decision = <strong>this</strong>.rule.decisionFor(crawlUri);<a name="461" href="#461">461</a> assertTrue(<span class="string">"URI Expect "</span> + DecideRule.PASS + <span class="string">" for "</span> + crawlUri +<a name="462" href="#462">462</a> <span class="string">" but got "</span> + decision, decision == DecideRule.PASS);<a name="463" href="#463">463</a> <a name="464" href="#464">464</a> <em class="comment">// matching content type - accept curi</em><a name="465" href="#465">465</a> crawlUri.setContentType(<span class="string">"text/html"</span>);<a name="466" href="#466">466</a> decision = <strong>this</strong>.rule.decisionFor(crawlUri);<a name="467" href="#467">467</a> assertTrue(<span class="string">"URI Expect "</span> + DecideRule.ACCEPT + <span class="string">" for "</span> + crawlUri +<a name="468" href="#468">468</a> <span class="string">" but got "</span> + decision, decision == DecideRule.ACCEPT);<a name="469" href="#469">469</a> }<a name="470" href="#470">470</a> <a name="471" href="#471">471</a> <strong>public</strong> <strong>void</strong> testContentTypeNotMatchesRegexpDecideRule() throws Exception{<a name="472" href="#472">472</a> <a href="../../../../org/archive/crawler/deciderules/ContentTypeNotMatchesRegExpDecideRule.html">ContentTypeNotMatchesRegExpDecideRule</a> dr = <strong>new</strong> <a href="../../../../org/archive/crawler/deciderules/ContentTypeNotMatchesRegExpDecideRule.html">ContentTypeNotMatchesRegExpDecideRule</a>(<span class="string">"CTNMREDRtest"</span>);<a name="473" href="#473">473</a> <a href="../../../../org/archive/crawler/deciderules/DecideRule.html">DecideRule</a> v = addDecideRule(dr);<a name="474" href="#474">474</a> <a name="475" href="#475">475</a> v.setAttribute(<strong>new</strong> Attribute(MatchesRegExpDecideRule.ATTR_REGEXP,<span class="string">"text/html"</span>));<a name="476" href="#476">476</a> <a href="../../../../org/archive/net/UURI.html">UURI</a> uuri = UURIFactory.getInstance(<span class="string">"http://www.archive.org"</span>);<a name="477" href="#477">477</a> <a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> crawlUri = <strong>new</strong> <a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a>(uuri);<a name="478" href="#478">478</a> <a name="479" href="#479">479</a> <em class="comment">// no content type - let curi pass</em><a name="480" href="#480">480</a> Object decision = <strong>this</strong>.rule.decisionFor(crawlUri);<a name="481" href="#481">481</a> assertTrue(<span class="string">"URI Expect "</span> + DecideRule.PASS + <span class="string">" for "</span> + crawlUri +<a name="482" href="#482">482</a> <span class="string">" but got "</span> + decision, decision == DecideRule.PASS);<a name="483" href="#483">483</a> <a name="484" href="#484">484</a> <em class="comment">// matching content type - let curi pass</em><a name="485" href="#485">485</a> crawlUri.setContentType(<span class="string">"text/html"</span>);<a name="486" href="#486">486</a> decision = <strong>this</strong>.rule.decisionFor(crawlUri);<a name="487" href="#487">487</a> assertTrue(<span class="string">"URI Expect "</span> + DecideRule.PASS + <span class="string">" for "</span> + crawlUri +<a name="488" href="#488">488</a> <span class="string">" but got "</span> + decision, decision == DecideRule.PASS);<a name="489" href="#489">489</a> <a name="490" href="#490">490</a> <em class="comment">// non-matching content type - accept curi</em><a name="491" href="#491">491</a> crawlUri.setContentType(<span class="string">"application/pdf"</span>);<a name="492" href="#492">492</a> decision = <strong>this</strong>.rule.decisionFor(crawlUri);<a name="493" href="#493">493</a> assertTrue(<span class="string">"URI Expect "</span> + DecideRule.ACCEPT + <span class="string">" for "</span> + crawlUri +<a name="494" href="#494">494</a> <span class="string">" but got "</span> + decision, decision == DecideRule.ACCEPT);<a name="495" href="#495">495</a> }<a name="496" href="#496">496</a> }</pre><hr/><div id="footer">This page was automatically generated by <a href="http://maven.apache.org/">Maven</a></div></body></html>
⌨️ 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?