📄 crawlmapper.html
字号:
<a name="159" href="#159">159</a> 0,<a name="160" href="#160">160</a> ((Integer) getUncheckedAttribute(<strong>null</strong>,<a name="161" href="#161">161</a> ATTR_ROTATION_DIGITS)).intValue());<a name="162" href="#162">162</a> <strong>if</strong>(!nowGeneration.equals(logGeneration)) {<a name="163" href="#163">163</a> updateGeneration(nowGeneration);<a name="164" href="#164">164</a> }<a name="165" href="#165">165</a> <a name="166" href="#166">166</a> <strong>if</strong> (curi.getFetchStatus() == 0<a name="167" href="#167">167</a> && ((Boolean) getUncheckedAttribute(<strong>null</strong>, ATTR_CHECK_URI))<a name="168" href="#168">168</a> .booleanValue()) {<a name="169" href="#169">169</a> <em class="comment">// apply mapping to the CrawlURI itself</em><a name="170" href="#170">170</a> String target = map(curi);<a name="171" href="#171">171</a> <strong>if</strong>(!localName.equals(target)) {<a name="172" href="#172">172</a> <em class="comment">// CrawlURI is mapped to somewhere other than here</em><a name="173" href="#173">173</a> curi.setFetchStatus(S_BLOCKED_BY_CUSTOM_PROCESSOR);<a name="174" href="#174">174</a> curi.addAnnotation(<span class="string">"to:"</span>+target);<a name="175" href="#175">175</a> curi.skipToProcessorChain(getController().<a name="176" href="#176">176</a> getPostprocessorChain());<a name="177" href="#177">177</a> divertLog(curi,target);<a name="178" href="#178">178</a> } <strong>else</strong> {<a name="179" href="#179">179</a> <em class="comment">// localName means keep locally; do nothing</em><a name="180" href="#180">180</a> }<a name="181" href="#181">181</a> }<a name="182" href="#182">182</a> <a name="183" href="#183">183</a> <strong>if</strong> (curi.getOutLinks().size() > 0 && <a name="184" href="#184">184</a> ((Boolean) getUncheckedAttribute(<strong>null</strong>, ATTR_CHECK_OUTLINKS))<a name="185" href="#185">185</a> .booleanValue()) {<a name="186" href="#186">186</a> <em class="comment">// consider outlinks for mapping</em><a name="187" href="#187">187</a> Iterator iter = curi.getOutLinks().iterator(); <a name="188" href="#188">188</a> <strong>while</strong>(iter.hasNext()) {<a name="189" href="#189">189</a> Object next = iter.next();<a name="190" href="#190">190</a> <strong>if</strong>(! (next instanceof CandidateURI)) {<a name="191" href="#191">191</a> <strong>continue</strong>;<a name="192" href="#192">192</a> }<a name="193" href="#193">193</a> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a> cauri = (CandidateURI)next; <a name="194" href="#194">194</a> <a name="195" href="#195">195</a> <strong>if</strong> (decideToMapOutlink(cauri)) {<a name="196" href="#196">196</a> <em class="comment">// apply mapping to the CandidateURI</em><a name="197" href="#197">197</a> String target = map(cauri);<a name="198" href="#198">198</a> <strong>if</strong>(!localName.equals(target)) {<a name="199" href="#199">199</a> <em class="comment">// CandidateURI is mapped to somewhere other than here</em><a name="200" href="#200">200</a> iter.remove();<a name="201" href="#201">201</a> divertLog(cauri,target);<a name="202" href="#202">202</a> } <strong>else</strong> {<a name="203" href="#203">203</a> <em class="comment">// localName means keep locally; do nothing</em><a name="204" href="#204">204</a> }<a name="205" href="#205">205</a> }<a name="206" href="#206">206</a> }<a name="207" href="#207">207</a> }<a name="208" href="#208">208</a> }<a name="209" href="#209">209</a> <a name="210" href="#210">210</a> <strong>protected</strong> <strong>boolean</strong> decideToMapOutlink(<a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a> cauri) {<a name="211" href="#211">211</a> <strong>boolean</strong> rejected = getMapOutlinkDecideRule(cauri).decisionFor(cauri)<a name="212" href="#212">212</a> .equals(DecideRule.REJECT);<a name="213" href="#213">213</a> <strong>return</strong> !rejected;<a name="214" href="#214">214</a> }<a name="215" href="#215">215</a> <a name="216" href="#216">216</a> <strong>protected</strong> <a href="../../../../org/archive/crawler/deciderules/DecideRule.html">DecideRule</a> getMapOutlinkDecideRule(Object o) {<a name="217" href="#217">217</a> <strong>try</strong> {<a name="218" href="#218">218</a> <strong>return</strong> (DecideRule)getAttribute(o, ATTR_MAP_OUTLINK_DECIDE_RULES);<a name="219" href="#219">219</a> } <strong>catch</strong> (AttributeNotFoundException e) {<a name="220" href="#220">220</a> <strong>throw</strong> <strong>new</strong> RuntimeException(e);<a name="221" href="#221">221</a> }<a name="222" href="#222">222</a> }<a name="223" href="#223">223</a> <a name="224" href="#224">224</a> <a name="225" href="#225">225</a> <em>/**<em>*</em></em><a name="226" href="#226">226</a> <em> * Close and mark as finished all existing diversion logs, and</em><a name="227" href="#227">227</a> <em> * arrange for new logs to use the new generation prefix.</em><a name="228" href="#228">228</a> <em> * </em><a name="229" href="#229">229</a> <em> * @param nowGeneration new generation (timestamp prefix) to use</em><a name="230" href="#230">230</a> <em> */</em><a name="231" href="#231">231</a> <strong>protected</strong> <strong>synchronized</strong> <strong>void</strong> updateGeneration(String nowGeneration) {<a name="232" href="#232">232</a> <em class="comment">// all existing logs are of a previous generation</em><a name="233" href="#233">233</a> Iterator iter = diversionLogs.values().iterator();<a name="234" href="#234">234</a> <strong>while</strong>(iter.hasNext()) {<a name="235" href="#235">235</a> FilePrintWriter writer = (FilePrintWriter) iter.next();<a name="236" href="#236">236</a> writer.close();<a name="237" href="#237">237</a> writer.getFile().renameTo(<a name="238" href="#238">238</a> <strong>new</strong> File(writer.getFile().getAbsolutePath()<a name="239" href="#239">239</a> .replaceFirst(<span class="string">"//.open$"</span>, <span class="string">".divert"</span>)));<a name="240" href="#240">240</a> }<a name="241" href="#241">241</a> diversionLogs.clear();<a name="242" href="#242">242</a> logGeneration = nowGeneration;<a name="243" href="#243">243</a> }<a name="244" href="#244">244</a> <a name="245" href="#245">245</a> <em>/**<em>*</em></em><a name="246" href="#246">246</a> <em> * Look up the crawler node name to which the given CandidateURI </em><a name="247" href="#247">247</a> <em> * should be mapped. </em><a name="248" href="#248">248</a> <em> * </em><a name="249" href="#249">249</a> <em> * @param cauri CandidateURI to consider</em><a name="250" href="#250">250</a> <em> * @return String node name which should handle URI</em><a name="251" href="#251">251</a> <em> */</em><a name="252" href="#252">252</a> <strong>protected</strong> <strong>abstract</strong> String map(<a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a> cauri);<a name="253" href="#253">253</a> <a name="254" href="#254">254</a> <a name="255" href="#255">255</a> <em>/**<em>*</em></em><a name="256" href="#256">256</a> <em> * Note the given CandidateURI in the appropriate diversion log. </em><a name="257" href="#257">257</a> <em> * </em><a name="258" href="#258">258</a> <em> * @param cauri CandidateURI to append to a diversion log</em><a name="259" href="#259">259</a> <em> * @param target String node name (log name) to receive URI</em><a name="260" href="#260">260</a> <em> */</em><a name="261" href="#261">261</a> <strong>protected</strong> <strong>synchronized</strong> <strong>void</strong> divertLog(<a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a> cauri, String target) {<a name="262" href="#262">262</a> <strong>if</strong>(recentlySeen(cauri)) {<a name="263" href="#263">263</a> <strong>return</strong>;<a name="264" href="#264">264</a> }<a name="265" href="#265">265</a> PrintWriter diversionLog = getDiversionLog(target);<a name="266" href="#266">266</a> cauri.singleLineReportTo(diversionLog);<a name="267" href="#267">267</a> diversionLog.println();<a name="268" href="#268">268</a> }<a name="269" href="#269">269</a> <a name="270" href="#270">270</a> <em>/**<em>*</em></em><a name="271" href="#271">271</a> <em> * Consult the cache to determine if the given URI</em><a name="272" href="#272">272</a> <em> * has been recently seen -- entering it if not. </em><a name="273" href="#273">273</a> <em> * </em><a name="274" href="#274">274</a> <em> * @param cauri CandidateURI to test</em><a name="275" href="#275">275</a> <em> * @return true if URI was already in the cache; false otherwise </em><a name="276" href="#276">276</a> <em> */</em><a name="277" href="#277">277</a> <strong>private</strong> <strong>boolean</strong> recentlySeen(<a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a> cauri) {<a name="278" href="#278">278</a> <strong>long</strong> fp = FPGenerator.std64.fp(cauri.getURIString());<a name="279" href="#279">279</a> <strong>return</strong> ! cache.add(fp);<a name="280" href="#280">280</a> }<a name="281" href="#281">281</a> <a name="282" href="#282">282</a> <em>/**<em>*</em></em><a name="283" href="#283">283</a> <em> * Get the diversion log for a given target crawler node node. </em><a name="284" href="#284">284</a> <em> * </em><a name="285" href="#285">285</a> <em> * @param target crawler node name of requested log</em><a name="286" href="#286">286</a> <em> * @return PrintWriter open on an appropriately-named </em><a name="287" href="#287">287</a> <em> * log file</em><a name="288" href="#288">288</a> <em> */</em><a name="289" href="#289">289</a> <strong>protected</strong> PrintWriter getDiversionLog(String target) {<a name="290" href="#290">290</a> FilePrintWriter writer = (FilePrintWriter) diversionLogs.get(target);<a name="291" href="#291">291</a> <strong>if</strong>(writer == <strong>null</strong>) {<a name="292" href="#292">292</a> String divertDirPath = (String) getUncheckedAttribute(<strong>null</strong>,ATTR_DIVERSION_DIR);<a name="293" href="#293">293</a> File divertDir = <strong>new</strong> File(divertDirPath);<a name="294" href="#294">294</a> <strong>if</strong> (!divertDir.isAbsolute()) {<a name="295" href="#295">295</a> divertDir = <strong>new</strong> File(getSettingsHandler().getOrder()<a name="296" href="#296">296</a> .getController().getDisk(), divertDirPath);<a name="297" href="#297">297</a> }<a name="298" href="#298">298</a> divertDir.mkdirs();<a name="299" href="#299">299</a> File divertLog = <a name="300" href="#300">300</a> <strong>new</strong> File(divertDir,<a name="301" href="#301">301</a> logGeneration+<span class="string">"-"</span>+localName+<span class="string">"-to-"</span>+target+<span class="string">".open"</span>);<a name="302" href="#302">302</a> <strong>try</strong> {<a name="303" href="#303">303</a> writer = <strong>new</strong> FilePrintWriter(divertLog);<a name="304" href="#304">304</a> } <strong>catch</strong> (FileNotFoundException e) {<a name="305" href="#305">305</a> <em class="comment">// TODO Auto-generated catch block</em><a name="306" href="#306">306</a> e.printStackTrace();<a name="307" href="#307">307</a> <strong>throw</strong> <strong>new</strong> RuntimeException(e);<a name="308" href="#308">308</a> }<a name="309" href="#309">309</a> diversionLogs.put(target,writer);<a name="310" href="#310">310</a> } <a name="311" href="#311">311</a> <strong>return</strong> writer;<a name="312" href="#312">312</a> }<a name="313" href="#313">313</a> <a name="314" href="#314">314</a> <strong>protected</strong> <strong>void</strong> initialTasks() {<a name="315" href="#315">315</a> <strong>super</strong>.initialTasks();<a name="316" href="#316">316</a> localName = (String) getUncheckedAttribute(<strong>null</strong>, ATTR_LOCAL_NAME);<a name="317" href="#317">317</a> cache = <strong>new</strong> <a href="../../../../org/archive/util/fingerprint/ArrayLongFPCache.html">ArrayLongFPCache</a>();<a name="318" href="#318">318</a> }<a name="319" href="#319">319</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
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -