📄 adaptiverevisithostqueuetest.html
字号:
<a name="118" href="#118">118</a> assertEquals(<span class="string">"Size of HQ should now be 1"</span>,1,hq.getSize());<a name="119" href="#119">119</a> <a name="120" href="#120">120</a> <em class="comment">/*</em><a name="121" href="#121">121</a> <em class="comment"> * Invoke next and ensure that the HQ is now busy (initial valence was</em><a name="122" href="#122">122</a> <em class="comment"> * set to 1). Also check for proper errors for a busy HQ. Such as when</em><a name="123" href="#123">123</a> <em class="comment"> * trying to reinvoke next().</em><a name="124" href="#124">124</a> <em class="comment"> *</em><a name="125" href="#125">125</a> <em class="comment"> */</em><a name="126" href="#126">126</a> <a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi = hq.next(); <em class="comment">// Should return curis[2]</em><a name="127" href="#127">127</a> assertEquals(<span class="string">"next() did not return 'top' URI"</span>,<a name="128" href="#128">128</a> curis[0].toString(),curi.toString());<a name="129" href="#129">129</a> assertTrue(<span class="string">"HQ should now be busy, is "</span> + hq.getStateByName(),<a name="130" href="#130">130</a> hq.getState()==AdaptiveRevisitHostQueue.HQSTATE_BUSY);<a name="131" href="#131">131</a> <strong>try</strong>{<a name="132" href="#132">132</a> hq.next();<a name="133" href="#133">133</a> assertTrue(<span class="string">"next() should throw an IllegalStateException if HQ "</span> +<a name="134" href="#134">134</a> <span class="string">"not ready"</span>,false);<a name="135" href="#135">135</a> } <strong>catch</strong>(IllegalStateException e){<a name="136" href="#136">136</a> <em class="comment">// This is supposed to happen.</em><a name="137" href="#137">137</a> }<a name="138" href="#138">138</a> assertEquals(<span class="string">"New top URI should be null"</span>,<a name="139" href="#139">139</a> <strong>null</strong>,hq.peek());<a name="140" href="#140">140</a> <a name="141" href="#141">141</a> hq.add(curis[1],false);<a name="142" href="#142">142</a> assertEquals(<span class="string">"Second CrawlURI should be top"</span>,curis[1].toString(),<a name="143" href="#143">143</a> hq.peek().toString());<a name="144" href="#144">144</a> assertEquals(<span class="string">"Size of HQ should now be 2"</span>,2,hq.getSize());<a name="145" href="#145">145</a> <a name="146" href="#146">146</a> <em class="comment">// Return it with next fetch time in the future.</em><a name="147" href="#147">147</a> curi.putLong(A_TIME_OF_NEXT_PROCESSING,<a name="148" href="#148">148</a> hq.peek().getLong(A_TIME_OF_NEXT_PROCESSING)<a name="149" href="#149">149</a> +100000); <em class="comment">// 100 sec behind current top.</em><a name="150" href="#150">150</a> hq.update(curi,false,0);<a name="151" href="#151">151</a> assertEquals(<span class="string">"Second CrawlURI should be still be top"</span>,<a name="152" href="#152">152</a> curis[1].toString(),hq.peek().toString());<a name="153" href="#153">153</a> assertEquals(<span class="string">"Size of HQ should still be 2"</span>,2,hq.getSize());<a name="154" href="#154">154</a> <a name="155" href="#155">155</a> hq.add(curis[2],false);<a name="156" href="#156">156</a> assertEquals(<span class="string">"Second CrawlURI should still be top"</span>,<a name="157" href="#157">157</a> curis[1].toString(), hq.peek().toString());<a name="158" href="#158">158</a> assertEquals(<span class="string">"Size of HQ should now be 3"</span>,3,hq.getSize());<a name="159" href="#159">159</a> <a name="160" href="#160">160</a> <em class="comment">/*</em><a name="161" href="#161">161</a> <em class="comment"> * If there are no URIs ready, the queue should snooze, even though no</em><a name="162" href="#162">162</a> <em class="comment"> * politeness demand has been made.</em><a name="163" href="#163">163</a> <em class="comment"> * <p></em><a name="164" href="#164">164</a> <em class="comment"> * Confirms this and that it wakes up.</em><a name="165" href="#165">165</a> <em class="comment"> */</em><a name="166" href="#166">166</a> assertTrue(<span class="string">"HQ should be snoozed, is "</span> + hq.getStateByName(),<a name="167" href="#167">167</a> hq.getState()==AdaptiveRevisitHostQueue.HQSTATE_SNOOZED);<a name="168" href="#168">168</a> <em class="comment">// Wait past wakeup time </em><a name="169" href="#169">169</a> <strong>synchronized</strong>(<strong>this</strong>){<a name="170" href="#170">170</a> wait(hq.getNextReadyTime()-System.currentTimeMillis()+100);<a name="171" href="#171">171</a> }<a name="172" href="#172">172</a> assertTrue(<span class="string">"HQ should now be ready, is "</span> + hq.getStateByName(),<a name="173" href="#173">173</a> hq.getState()==AdaptiveRevisitHostQueue.HQSTATE_READY);<a name="174" href="#174">174</a> <a name="175" href="#175">175</a> <em class="comment">/*</em><a name="176" href="#176">176</a> <em class="comment"> * Re-adds a URI with a lower ready time which should promote it to the</em><a name="177" href="#177">177</a> <em class="comment"> * top of the queue. Checks if this happens correctly.</em><a name="178" href="#178">178</a> <em class="comment"> * </em><a name="179" href="#179">179</a> <em class="comment"> * Then tests an add override which would demote it back, ensures that </em><a name="180" href="#180">180</a> <em class="comment"> * this fails as it should (i.e. URIs time of next processing remains </em><a name="181" href="#181">181</a> <em class="comment"> * unchanged).</em><a name="182" href="#182">182</a> <em class="comment"> */</em><a name="183" href="#183">183</a> curis[2].putLong(<a name="184" href="#184">184</a> A_TIME_OF_NEXT_PROCESSING,<a name="185" href="#185">185</a> curis[1].getLong(A_TIME_OF_NEXT_PROCESSING)<a name="186" href="#186">186</a> -1000); <em class="comment">// 1 sec. prior to current top </em><a name="187" href="#187">187</a> hq.add(curis[2],<strong>true</strong>);<a name="188" href="#188">188</a> assertEquals(<span class="string">"Size of HQ should still be 3"</span>,hq.getSize(),3);<a name="189" href="#189">189</a> assertEquals(<span class="string">"Third CrawlURI should be now be top"</span>,<a name="190" href="#190">190</a> curis[2].toString(), hq.peek().toString());<a name="191" href="#191">191</a> curis[2].putLong(A_TIME_OF_NEXT_PROCESSING,<a name="192" href="#192">192</a> curis[1].getLong(A_TIME_OF_NEXT_PROCESSING)<a name="193" href="#193">193</a> +10000); <em class="comment">// 10 sec. later </em><a name="194" href="#194">194</a> hq.add(curis[2],<strong>true</strong>);<a name="195" href="#195">195</a> assertEquals(<span class="string">"Size of HQ should still be 3"</span>,hq.getSize(),3);<a name="196" href="#196">196</a> assertEquals(<span class="string">"Third CrawlURI should still top"</span>,<a name="197" href="#197">197</a> curis[2].toString(), hq.peek().toString());<a name="198" href="#198">198</a> <a name="199" href="#199">199</a> <a name="200" href="#200">200</a> <em class="comment">/*</em><a name="201" href="#201">201</a> <em class="comment"> * Invoke next and ensure that the HQ is now busy (initial valence was</em><a name="202" href="#202">202</a> <em class="comment"> * set to 1). Also check for proper errors for a busy HQ. Such as when</em><a name="203" href="#203">203</a> <em class="comment"> * trying to reinvoke next().</em><a name="204" href="#204">204</a> <em class="comment"> *</em><a name="205" href="#205">205</a> <em class="comment"> */</em><a name="206" href="#206">206</a> curi = hq.next(); <em class="comment">// Should return curis[2]</em><a name="207" href="#207">207</a> assertEquals(<span class="string">"next() did not return 'top' URI"</span>,<a name="208" href="#208">208</a> curis[2].toString(),curi.toString());<a name="209" href="#209">209</a> assertTrue(<span class="string">"HQ should now be busy, is "</span> + hq.getStateByName(),<a name="210" href="#210">210</a> hq.getState()==AdaptiveRevisitHostQueue.HQSTATE_BUSY);<a name="211" href="#211">211</a> <strong>try</strong>{<a name="212" href="#212">212</a> hq.next();<a name="213" href="#213">213</a> assertTrue(<span class="string">"next() should throw an IllegalStateException if HQ "</span> +<a name="214" href="#214">214</a> <span class="string">"not ready"</span>,false);<a name="215" href="#215">215</a> } <strong>catch</strong>(IllegalStateException e){<a name="216" href="#216">216</a> <em class="comment">// This is supposed to happen.</em><a name="217" href="#217">217</a> }<a name="218" href="#218">218</a> assertEquals(<span class="string">"New top URI"</span>,<a name="219" href="#219">219</a> curis[1].toString(),hq.peek().toString());<a name="220" href="#220">220</a> <a name="221" href="#221">221</a> <em class="comment">/*</em><a name="222" href="#222">222</a> <em class="comment"> * Add a URI while HQ is busy. Check if this succeeds normally.</em><a name="223" href="#223">223</a> <em class="comment"> *</em><a name="224" href="#224">224</a> <em class="comment"> */</em><a name="225" href="#225">225</a> <a name="226" href="#226">226</a> curis[3].putLong(A_TIME_OF_NEXT_PROCESSING,<a name="227" href="#227">227</a> curis[1].getLong(A_TIME_OF_NEXT_PROCESSING) <a name="228" href="#228">228</a> - 1); <em class="comment">// 1 msec. ahead of current top (order [2] 3 1 0) </em><a name="229" href="#229">229</a> hq.add(curis[3],false);<a name="230" href="#230">230</a> assertEquals(<span class="string">"Size of HQ should now be 4"</span>,4,hq.getSize());<a name="231" href="#231">231</a> <a name="232" href="#232">232</a> <a name="233" href="#233">233</a> <em class="comment">/*</em><a name="234" href="#234">234</a> <em class="comment"> * Invoke update, first with an invalid URI (not the one issued by </em><a name="235" href="#235">235</a> <em class="comment"> * next() earlier), this should fail. Then with the correct one, this </em><a name="236" href="#236">236</a> <em class="comment"> * should succeed. Then finally test update again with an invalid URI </em><a name="237" href="#237">237</a> <em class="comment"> * (i.e. when no HQ has no outstanding URIs, that should fail.</em><a name="238" href="#238">238</a> <em class="comment"> * </em><a name="239" href="#239">239</a> <em class="comment"> * At each step, proper checks are made of state and that methods give </em><a name="240" href="#240">240</a> <em class="comment"> * appropriate errors.</em><a name="241" href="#241">241</a> <em class="comment"> * </em><a name="242" href="#242">242</a> <em class="comment"> * Updated URI is given low time of next processing to put it 'in front'</em><a name="243" href="#243">243</a> <em class="comment"> */</em>
⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -