persistprocessor.html
来自「网络爬虫开源代码」· HTML 代码 · 共 292 行 · 第 1/2 页
HTML
292 行
<a name="138" href="#138">138</a> System.out.println(<span class="string">"Arguments: "</span>);<a name="139" href="#139">139</a> System.out.println(<span class="string">" source [target]"</span>);<a name="140" href="#140">140</a> System.out.println(<a name="141" href="#141">141</a> <span class="string">"...where source is either a txtser log file or BDB env dir"</span>);<a name="142" href="#142">142</a> System.out.println(<a name="143" href="#143">143</a> <span class="string">"and target, if present, is a BDB env dir. "</span>);<a name="144" href="#144">144</a> <strong>return</strong>;<a name="145" href="#145">145</a> }<a name="146" href="#146">146</a> <a name="147" href="#147">147</a> }<a name="148" href="#148">148</a> <a name="149" href="#149">149</a> <em>/**<em>*</em></em><a name="150" href="#150">150</a> <em> * Move the history information in the first argument (either the path </em><a name="151" href="#151">151</a> <em> * to a log or to an environment containing a uri_history database) to </em><a name="152" href="#152">152</a> <em> * the environment in the second environment (path; environment will </em><a name="153" href="#153">153</a> <em> * be created if it dow not already exist). </em><a name="154" href="#154">154</a> <em> * </em><a name="155" href="#155">155</a> <em> * @param args command-line arguments</em><a name="156" href="#156">156</a> <em> * @throws DatabaseException</em><a name="157" href="#157">157</a> <em> * @throws FileNotFoundException</em><a name="158" href="#158">158</a> <em> * @throws UnsupportedEncodingException</em><a name="159" href="#159">159</a> <em> * @throws IOException</em><a name="160" href="#160">160</a> <em> */</em><a name="161" href="#161">161</a> <strong>private</strong> <strong>static</strong> <strong>void</strong> main2args(String[] args) throws DatabaseException, FileNotFoundException, UnsupportedEncodingException, IOException {<a name="162" href="#162">162</a> File source = <strong>new</strong> File(args[0]);<a name="163" href="#163">163</a> File env = <strong>new</strong> File(args[1]);<a name="164" href="#164">164</a> <strong>if</strong>(!env.exists()) {<a name="165" href="#165">165</a> env.mkdirs();<a name="166" href="#166">166</a> }<a name="167" href="#167">167</a> <a name="168" href="#168">168</a> <em class="comment">// setup target environment</em><a name="169" href="#169">169</a> <a href="../../../../../org/archive/util/bdbje/EnhancedEnvironment.html">EnhancedEnvironment</a> targetEnv = setupEnvironment(env);<a name="170" href="#170">170</a> StoredClassCatalog classCatalog = targetEnv.getClassCatalog();<a name="171" href="#171">171</a> Database historyDB = targetEnv.openDatabase(<a name="172" href="#172">172</a> <strong>null</strong>,URI_HISTORY_DBNAME,historyDatabaseConfig());<a name="173" href="#173">173</a> StoredSortedMap historyMap = <strong>new</strong> StoredSortedMap(historyDB,<a name="174" href="#174">174</a> <strong>new</strong> StringBinding(), <strong>new</strong> SerialBinding(classCatalog,<a name="175" href="#175">175</a> AList.<strong>class</strong>), <strong>true</strong>);<a name="176" href="#176">176</a> <a name="177" href="#177">177</a> <strong>int</strong> count = 0;<a name="178" href="#178">178</a> <a name="179" href="#179">179</a> <strong>if</strong>(source.isFile()) {<a name="180" href="#180">180</a> <em class="comment">// scan log, writing to database</em><a name="181" href="#181">181</a> BufferedReader br = CrawlerJournal.getBufferedReader(source);<a name="182" href="#182">182</a> Iterator iter = <strong>new</strong> <a href="../../../../../org/archive/util/iterator/LineReadingIterator.html">LineReadingIterator</a>(br);<a name="183" href="#183">183</a> <strong>while</strong>(iter.hasNext()) {<a name="184" href="#184">184</a> String line = (String) iter.next(); <a name="185" href="#185">185</a> String[] splits = line.split(<span class="string">" "</span>);<a name="186" href="#186">186</a> historyMap.put(<a name="187" href="#187">187</a> splits[0], <a name="188" href="#188">188</a> IoUtils.deserializeFromByteArray(<a name="189" href="#189">189</a> Base64.decodeBase64(splits[1].getBytes(<span class="string">"UTF8"</span>))));<a name="190" href="#190">190</a> count++;<a name="191" href="#191">191</a> }<a name="192" href="#192">192</a> br.close();<a name="193" href="#193">193</a> } <strong>else</strong> {<a name="194" href="#194">194</a> <em class="comment">// open the source env history DB, copying entries to target env</em><a name="195" href="#195">195</a> <a href="../../../../../org/archive/util/bdbje/EnhancedEnvironment.html">EnhancedEnvironment</a> sourceEnv = setupEnvironment(source);<a name="196" href="#196">196</a> StoredClassCatalog sourceClassCatalog = sourceEnv.getClassCatalog();<a name="197" href="#197">197</a> Database sourceHistoryDB = sourceEnv.openDatabase(<a name="198" href="#198">198</a> <strong>null</strong>,URI_HISTORY_DBNAME,historyDatabaseConfig());<a name="199" href="#199">199</a> StoredSortedMap sourceHistoryMap = <strong>new</strong> StoredSortedMap(sourceHistoryDB,<a name="200" href="#200">200</a> <strong>new</strong> StringBinding(), <strong>new</strong> SerialBinding(sourceClassCatalog,<a name="201" href="#201">201</a> AList.<strong>class</strong>), <strong>true</strong>);<a name="202" href="#202">202</a> Iterator iter = sourceHistoryMap.entrySet().iterator();<a name="203" href="#203">203</a> <strong>while</strong>(iter.hasNext()) {<a name="204" href="#204">204</a> Entry item = (Entry) iter.next(); <a name="205" href="#205">205</a> historyMap.put(item.getKey(), item.getValue());<a name="206" href="#206">206</a> count++;<a name="207" href="#207">207</a> }<a name="208" href="#208">208</a> StoredIterator.close(iter);<a name="209" href="#209">209</a> sourceHistoryDB.close();<a name="210" href="#210">210</a> sourceEnv.close();<a name="211" href="#211">211</a> }<a name="212" href="#212">212</a> <a name="213" href="#213">213</a> <em class="comment">// cleanup</em><a name="214" href="#214">214</a> historyDB.sync();<a name="215" href="#215">215</a> historyDB.close();<a name="216" href="#216">216</a> targetEnv.close();<a name="217" href="#217">217</a> System.out.println(count+<span class="string">" records imported from "</span>+source+<span class="string">" to BDB env "</span>+env);<a name="218" href="#218">218</a> }<a name="219" href="#219">219</a> <a name="220" href="#220">220</a> <em>/**<em>*</em></em><a name="221" href="#221">221</a> <em> * Dump the contents of the argument (path to a persist log) to stdout</em><a name="222" href="#222">222</a> <em> * in a slightly more readable format. </em><a name="223" href="#223">223</a> <em> * </em><a name="224" href="#224">224</a> <em> * @param args command-line arguments</em><a name="225" href="#225">225</a> <em> * @throws DatabaseException</em><a name="226" href="#226">226</a> <em> * @throws FileNotFoundException</em><a name="227" href="#227">227</a> <em> * @throws UnsupportedEncodingException</em><a name="228" href="#228">228</a> <em> * @throws IOException</em><a name="229" href="#229">229</a> <em> */</em><a name="230" href="#230">230</a> <strong>private</strong> <strong>static</strong> <strong>void</strong> main1arg(String[] args) throws DatabaseException, FileNotFoundException, UnsupportedEncodingException, IOException {<a name="231" href="#231">231</a> File source = <strong>new</strong> File(args[0]);<a name="232" href="#232">232</a> <a name="233" href="#233">233</a> <strong>int</strong> count = 0;<a name="234" href="#234">234</a> <a name="235" href="#235">235</a> <strong>if</strong>(source.isFile()) {<a name="236" href="#236">236</a> <em class="comment">// scan log, writing to database</em><a name="237" href="#237">237</a> BufferedReader br = CrawlerJournal.getBufferedReader(source);<a name="238" href="#238">238</a> Iterator iter = <strong>new</strong> <a href="../../../../../org/archive/util/iterator/LineReadingIterator.html">LineReadingIterator</a>(br);<a name="239" href="#239">239</a> <strong>while</strong>(iter.hasNext()) {<a name="240" href="#240">240</a> String line = (String) iter.next(); <a name="241" href="#241">241</a> String[] splits = line.split(<span class="string">" "</span>);<a name="242" href="#242">242</a> <a href="../../../../../st/ata/util/AList.html">AList</a> alist = (AList)IoUtils.deserializeFromByteArray(<a name="243" href="#243">243</a> Base64.decodeBase64(splits[1].getBytes(<span class="string">"UTF8"</span>)));<a name="244" href="#244">244</a> System.out.println(<a name="245" href="#245">245</a> splits[0] + <span class="string">" "</span> + alist.toPrettyString());<a name="246" href="#246">246</a> count++;<a name="247" href="#247">247</a> }<a name="248" href="#248">248</a> br.close();<a name="249" href="#249">249</a> } <strong>else</strong> {<a name="250" href="#250">250</a> <em class="comment">// open the source env history DB, copying entries to target env</em><a name="251" href="#251">251</a> <a href="../../../../../org/archive/util/bdbje/EnhancedEnvironment.html">EnhancedEnvironment</a> sourceEnv = setupEnvironment(source);<a name="252" href="#252">252</a> StoredClassCatalog sourceClassCatalog = sourceEnv.getClassCatalog();<a name="253" href="#253">253</a> Database sourceHistoryDB = sourceEnv.openDatabase(<a name="254" href="#254">254</a> <strong>null</strong>,URI_HISTORY_DBNAME,historyDatabaseConfig());<a name="255" href="#255">255</a> StoredSortedMap sourceHistoryMap = <strong>new</strong> StoredSortedMap(sourceHistoryDB,<a name="256" href="#256">256</a> <strong>new</strong> StringBinding(), <strong>new</strong> SerialBinding(sourceClassCatalog,<a name="257" href="#257">257</a> AList.<strong>class</strong>), <strong>true</strong>);<a name="258" href="#258">258</a> Iterator iter = sourceHistoryMap.entrySet().iterator();<a name="259" href="#259">259</a> <strong>while</strong>(iter.hasNext()) {<a name="260" href="#260">260</a> Entry item = (Entry) iter.next(); <a name="261" href="#261">261</a> <a href="../../../../../st/ata/util/AList.html">AList</a> alist = (AList)item.getValue();<a name="262" href="#262">262</a> System.out.println(item.getKey() + <span class="string">" "</span> + alist.toPrettyString());<a name="263" href="#263">263</a> count++;<a name="264" href="#264">264</a> }<a name="265" href="#265">265</a> StoredIterator.close(iter);<a name="266" href="#266">266</a> sourceHistoryDB.close();<a name="267" href="#267">267</a> sourceEnv.close();<a name="268" href="#268">268</a> }<a name="269" href="#269">269</a> <a name="270" href="#270">270</a> System.out.println(count+<span class="string">" records dumped from "</span>+source);<a name="271" href="#271">271</a> }<a name="272" href="#272">272</a> <a name="273" href="#273">273</a> <strong>private</strong> <strong>static</strong> <a href="../../../../../org/archive/util/bdbje/EnhancedEnvironment.html">EnhancedEnvironment</a> setupEnvironment(File env) throws DatabaseException {<a name="274" href="#274">274</a> EnvironmentConfig envConfig = <strong>new</strong> EnvironmentConfig();<a name="275" href="#275">275</a> envConfig.setAllowCreate(<strong>true</strong>);<a name="276" href="#276">276</a> <strong>return</strong> <strong>new</strong> <a href="../../../../../org/archive/util/bdbje/EnhancedEnvironment.html">EnhancedEnvironment</a>(env, envConfig);<a name="277" href="#277">277</a> }<a name="278" href="#278">278</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?