deciderulesequencetest.html

来自「网络爬虫开源代码」· HTML 代码 · 共 510 行 · 第 1/4 页

HTML
510
字号
<a name="120" href="#120">120</a>         addDecideRule(<strong>new</strong> <a href="../../../../org/archive/crawler/deciderules/DecideRule.html">DecideRule</a>(<span class="string">"PASS4"</span>));<a name="121" href="#121">121</a>         addDecideRule(<strong>new</strong> <a href="../../../../org/archive/crawler/deciderules/RejectDecideRule.html">RejectDecideRule</a>(<span class="string">"REJECT3"</span>));<a name="122" href="#122">122</a>         Object decision = <strong>this</strong>.rule.decisionFor(<span class="string">"test"</span>);<a name="123" href="#123">123</a>         assertTrue(<span class="string">"Expect REJECT but got "</span> + decision,<a name="124" href="#124">124</a>             decision == DecideRule.REJECT);<a name="125" href="#125">125</a>     }<a name="126" href="#126">126</a>         <a name="127" href="#127">127</a>     <strong>public</strong> <strong>void</strong> testRegex()<a name="128" href="#128">128</a>     throws InvalidAttributeValueException, AttributeNotFoundException,<a name="129" href="#129">129</a>     MBeanException, ReflectionException {<a name="130" href="#130">130</a>         <strong>final</strong> String regexName = <span class="string">"REGEX"</span>;<a name="131" href="#131">131</a>         <a href="../../../../org/archive/crawler/deciderules/DecideRule.html">DecideRule</a> r = addDecideRule(<strong>new</strong> <a href="../../../../org/archive/crawler/deciderules/MatchesRegExpDecideRule.html">MatchesRegExpDecideRule</a>(regexName));<a name="132" href="#132">132</a>         <em class="comment">// Set regex to be match anything that ends in archive.org.</em><a name="133" href="#133">133</a>         r.setAttribute(<strong>new</strong> Attribute(MatchesRegExpDecideRule.ATTR_REGEXP,<a name="134" href="#134">134</a>             <span class="string">"^.*&#47;&#47;.archive&#47;&#47;.org"</span>));<a name="135" href="#135">135</a>         Object decision = <strong>this</strong>.rule.decisionFor(<span class="string">"http://google.com"</span>);<a name="136" href="#136">136</a>         assertTrue(<span class="string">"Expect PASS but got "</span> + decision,<a name="137" href="#137">137</a>             decision == DecideRule.PASS);<a name="138" href="#138">138</a>         decision = <strong>this</strong>.rule.decisionFor(<span class="string">"http://archive.org"</span>);<a name="139" href="#139">139</a>         assertTrue(<span class="string">"Expect PASS but got "</span> + decision,<a name="140" href="#140">140</a>             decision == DecideRule.PASS);<a name="141" href="#141">141</a>         decision = <strong>this</strong>.rule.decisionFor(<span class="string">"http://www.archive.org"</span>);<a name="142" href="#142">142</a>         assertTrue(<span class="string">"Expect ACCEPT but got "</span> + decision,<a name="143" href="#143">143</a>             decision == DecideRule.ACCEPT);<a name="144" href="#144">144</a>     }<a name="145" href="#145">145</a>     <a name="146" href="#146">146</a>     <strong>public</strong> <strong>void</strong> testNotRegex()<a name="147" href="#147">147</a>     throws InvalidAttributeValueException, AttributeNotFoundException,<a name="148" href="#148">148</a>     MBeanException, ReflectionException {<a name="149" href="#149">149</a>         <strong>final</strong> String regexName = <span class="string">"NOT_REGEX"</span>;<a name="150" href="#150">150</a>         <a href="../../../../org/archive/crawler/deciderules/DecideRule.html">DecideRule</a> r = addDecideRule(<strong>new</strong> <a href="../../../../org/archive/crawler/deciderules/NotMatchesRegExpDecideRule.html">NotMatchesRegExpDecideRule</a>(regexName));<a name="151" href="#151">151</a>         <em class="comment">// Set regex to be match anything that ends in archive.org.</em><a name="152" href="#152">152</a>         r.setAttribute(<strong>new</strong> Attribute(MatchesRegExpDecideRule.ATTR_REGEXP,<a name="153" href="#153">153</a>             <span class="string">"^.*&#47;&#47;.archive&#47;&#47;.org"</span>));<a name="154" href="#154">154</a>         Object decision = <strong>this</strong>.rule.decisionFor(<span class="string">"http://google.com"</span>);<a name="155" href="#155">155</a>         assertTrue(<span class="string">"Expect ACCEPT but got "</span> + decision,<a name="156" href="#156">156</a>             decision == DecideRule.ACCEPT);<a name="157" href="#157">157</a>         decision = <strong>this</strong>.rule.decisionFor(<span class="string">"http://www.archive.org"</span>);<a name="158" href="#158">158</a>         assertTrue(<span class="string">"Expect PASS but got "</span> + decision,<a name="159" href="#159">159</a>             decision == DecideRule.PASS);<a name="160" href="#160">160</a>     }<a name="161" href="#161">161</a>     <a name="162" href="#162">162</a>     <a name="163" href="#163">163</a>     <strong>public</strong> <strong>void</strong> testPrerequisite()<a name="164" href="#164">164</a>     throws InvalidAttributeValueException, URIException {<a name="165" href="#165">165</a>         addDecideRule(<strong>new</strong> <a href="../../../../org/archive/crawler/deciderules/PrerequisiteAcceptDecideRule.html">PrerequisiteAcceptDecideRule</a>(<span class="string">"PREREQUISITE"</span>));<a name="166" href="#166">166</a>         <a href="../../../../org/archive/net/UURI.html">UURI</a> uuri = UURIFactory.getInstance(<span class="string">"http://archive.org"</span>);<a name="167" href="#167">167</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="168" href="#168">168</a>         Object decision = <strong>this</strong>.rule.decisionFor(candidate);<a name="169" href="#169">169</a>         assertTrue(<span class="string">"Expect PASS but got "</span> + decision,<a name="170" href="#170">170</a>             decision == DecideRule.PASS);<a name="171" href="#171">171</a>         candidate = <strong>new</strong> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a>(uuri, <span class="string">"LLP"</span>, <strong>null</strong>, <strong>null</strong>);<a name="172" href="#172">172</a>         decision = <strong>this</strong>.rule.decisionFor(candidate);<a name="173" href="#173">173</a>         assertTrue(<span class="string">"Expect ACCEPT but got "</span> + decision,<a name="174" href="#174">174</a>             decision == DecideRule.ACCEPT);<a name="175" href="#175">175</a>     }<a name="176" href="#176">176</a>     <a name="177" href="#177">177</a>     <strong>public</strong> <strong>void</strong> testHops()<a name="178" href="#178">178</a>     throws InvalidAttributeValueException, URIException {<a name="179" href="#179">179</a>         addDecideRule(<strong>new</strong> <a href="../../../../org/archive/crawler/deciderules/TooManyHopsDecideRule.html">TooManyHopsDecideRule</a>(<span class="string">"HOPS"</span>));<a name="180" href="#180">180</a>         testHopLimit(TooManyHopsDecideRule.DEFAULT_MAX_HOPS.intValue(), 'L',<a name="181" href="#181">181</a>             DecideRule.PASS, DecideRule.REJECT);<a name="182" href="#182">182</a>     }<a name="183" href="#183">183</a>     <a name="184" href="#184">184</a>     <strong>public</strong> <strong>void</strong> testTransclusion()<a name="185" href="#185">185</a>     throws InvalidAttributeValueException, URIException {<a name="186" href="#186">186</a>         addDecideRule(<strong>new</strong> <a href="../../../../org/archive/crawler/deciderules/TransclusionDecideRule.html">TransclusionDecideRule</a>(<span class="string">"TRANSCLUSION"</span>));<a name="187" href="#187">187</a>         <strong>final</strong> <strong>int</strong> max =<a name="188" href="#188">188</a>             TransclusionDecideRule.DEFAULT_MAX_TRANS_HOPS.intValue();<a name="189" href="#189">189</a>         <strong>final</strong> <strong>char</strong> pathExpansion = 'E';<a name="190" href="#190">190</a>         <a href="../../../../org/archive/net/UURI.html">UURI</a> uuri = UURIFactory.getInstance(<span class="string">"http://archive.org"</span>);<a name="191" href="#191">191</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="192" href="#192">192</a>         Object decision = <strong>this</strong>.rule.decisionFor(candidate);<a name="193" href="#193">193</a>         assertTrue(<span class="string">"Expect "</span> + DecideRule.PASS + <span class="string">" but got "</span> + decision,<a name="194" href="#194">194</a>             decision == DecideRule.PASS);<a name="195" href="#195">195</a>         StringBuffer path = <strong>new</strong> StringBuffer(max);<a name="196" href="#196">196</a>         <strong>for</strong> (<strong>int</strong> i = 0; i &lt; (max - 1); i++) {<a name="197" href="#197">197</a>             path.append(pathExpansion);<a name="198" href="#198">198</a>         }<a name="199" href="#199">199</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="200" href="#200">200</a>         decision = <strong>this</strong>.rule.decisionFor(candidate);<a name="201" href="#201">201</a>         assertTrue(<span class="string">"Expect "</span> + DecideRule.ACCEPT + <span class="string">" but got "</span> + decision,<a name="202" href="#202">202</a>             decision == DecideRule.ACCEPT);<a name="203" href="#203">203</a>         String pathCopy = path.toString();<a name="204" href="#204">204</a>         path.append(pathExpansion);<a name="205" href="#205">205</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="206" href="#206">206</a>         decision = <strong>this</strong>.rule.decisionFor(candidate);<a name="207" href="#207">207</a>         assertTrue(<span class="string">"Expect "</span> + DecideRule.ACCEPT + <span class="string">" but got "</span> + decision,<a name="208" href="#208">208</a>             decision == DecideRule.ACCEPT);<a name="209" href="#209">209</a>         path.append(pathExpansion);<a name="210" href="#210">210</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="211" href="#211">211</a>         decision = <strong>this</strong>.rule.decisionFor(candidate);<a name="212" href="#212">212</a>         assertTrue(<span class="string">"Expect "</span> + DecideRule.PASS + <span class="string">" but got "</span> + decision,<a name="213" href="#213">213</a>             decision == DecideRule.PASS);<a name="214" href="#214">214</a>         candidate = <strong>new</strong> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a>(uuri, pathCopy + 'L', <strong>null</strong>, <strong>null</strong>);<a name="215" href="#215">215</a>         decision = <strong>this</strong>.rule.decisionFor(candidate);<a name="216" href="#216">216</a>         assertTrue(<span class="string">"Expect "</span> + DecideRule.PASS + <span class="string">" but got "</span> + decision,<a name="217" href="#217">217</a>             decision == DecideRule.PASS);<a name="218" href="#218">218</a>     }<a name="219" href="#219">219</a>     <a name="220" href="#220">220</a>     <strong>public</strong> <strong>void</strong> testPathologicalPath()<a name="221" href="#221">221</a>     throws InvalidAttributeValueException, URIException {<a name="222" href="#222">222</a>         addDecideRule(<strong>new</strong> <a href="../../../../org/archive/crawler/deciderules/PathologicalPathDecideRule.html">PathologicalPathDecideRule</a>(<span class="string">"PATHOLOGICAL"</span>));<a name="223" href="#223">223</a>         <strong>final</strong> <strong>int</strong> max =<a name="224" href="#224">224</a>             PathologicalPathDecideRule.DEFAULT_REPETITIONS.intValue();<a name="225" href="#225">225</a>         String uri = <span class="string">"http://archive.org/"</span>;<a name="226" href="#226">226</a>         <strong>final</strong> String segment = <span class="string">"abc/"</span>;<a name="227" href="#227">227</a>         <strong>for</strong> (<strong>int</strong> i = 1; i &lt; max; i++) {<a name="228" href="#228">228</a>             uri = uri + segment;<a name="229" href="#229">229</a>         }<a name="230" href="#230">230</a>         <strong>final</strong> String baseUri = uri;<a name="231" href="#231">231</a>         <a href="../../../../org/archive/net/UURI.html">UURI</a> uuri = UURIFactory.getInstance(uri);<a name="232" href="#232">232</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="233" href="#233">233</a>         Object decision = <strong>this</strong>.rule.decisionFor(candidate);<a name="234" href="#234">234</a>         assertTrue(<span class="string">"Expect "</span> + DecideRule.PASS + <span class="string">" but got "</span> + decision,<a name="235" href="#235">235</a>             decision == DecideRule.PASS);<a name="236" href="#236">236</a>         uuri = UURIFactory.getInstance(baseUri + segment);<a name="237" href="#237">237</a>         candidate = <strong>new</strong> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a>(uuri);<a name="238" href="#238">238</a>         decision = <strong>this</strong>.rule.decisionFor(candidate);<a name="239" href="#239">239</a>         assertTrue(<span class="string">"Expect "</span> + DecideRule.PASS + <span class="string">" but got "</span> + decision,<a name="240" href="#240">240</a>             decision == DecideRule.PASS);<a name="241" href="#241">241</a>         uuri = UURIFactory.getInstance(baseUri + segment + segment);<a name="242" href="#242">242</a>         candidate = <strong>new</strong> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a>(uuri);<a name="243" href="#243">243</a>         decision = <strong>this</strong>.rule.decisionFor(candidate);<a name="244" href="#244">244</a>         assertTrue(<span class="string">"Expect "</span> + DecideRule.REJECT + <span class="string">" but got "</span> + decision,<a name="245" href="#245">245</a>             decision == DecideRule.REJECT);<a name="246" href="#246">246</a>     }<a name="247" href="#247">247</a>     

⌨️ 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?