processor.html

来自「网络爬虫开源代码」· HTML 代码 · 共 291 行 · 第 1/2 页

<a name="138" href="#138">138</a> <em>     * @throws InterruptedException</em>
<a name="139" href="#139">139</a> <em>     */</em>
<a name="140" href="#140">140</a>     <strong>protected</strong> <strong>void</strong> innerProcess(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi)
<a name="141" href="#141">141</a>     throws InterruptedException {
<a name="142" href="#142">142</a>         <em class="comment">// by default do nothing</em>
<a name="143" href="#143">143</a>     }
<a name="144" href="#144">144</a> 
<a name="145" href="#145">145</a>     <em>/**<em>*</em></em>
<a name="146" href="#146">146</a> <em>     * Classes subclassing this one should override this method to perform</em>
<a name="147" href="#147">147</a> <em>     * processor specific actions.</em>
<a name="148" href="#148">148</a> <em>     * &lt;p></em>
<a name="149" href="#149">149</a> <em>     *</em>
<a name="150" href="#150">150</a> <em>     * This method is garanteed to be called after the crawl is set up, but</em>
<a name="151" href="#151">151</a> <em>     * before any URI-processing has occured.</em>
<a name="152" href="#152">152</a> <em>     */</em>
<a name="153" href="#153">153</a>     <strong>protected</strong> <strong>void</strong> initialTasks () {
<a name="154" href="#154">154</a>         <em class="comment">// by default do nothing</em>
<a name="155" href="#155">155</a>     }
<a name="156" href="#156">156</a> 
<a name="157" href="#157">157</a>     <em>/**<em>*</em></em>
<a name="158" href="#158">158</a> <em>     * Classes subclassing this one should override this method to perform</em>
<a name="159" href="#159">159</a> <em>     * processor specific actions.</em>
<a name="160" href="#160">160</a> <em>     *</em>
<a name="161" href="#161">161</a> <em>     */</em>
<a name="162" href="#162">162</a>     <strong>protected</strong> <strong>void</strong> finalTasks () {
<a name="163" href="#163">163</a>         <em class="comment">// by default do nothing</em>
<a name="164" href="#164">164</a>     }
<a name="165" href="#165">165</a> 
<a name="166" href="#166">166</a>     <strong>protected</strong> <a href="../../../../org/archive/crawler/deciderules/DecideRule.html">DecideRule</a> getDecideRule(Object o) {
<a name="167" href="#167">167</a>         <strong>try</strong> {
<a name="168" href="#168">168</a>             <strong>return</strong> (DecideRule)getAttribute(o, attrDecideRules);
<a name="169" href="#169">169</a>         } <strong>catch</strong> (AttributeNotFoundException e) {
<a name="170" href="#170">170</a>             <strong>throw</strong> <strong>new</strong> RuntimeException(e);
<a name="171" href="#171">171</a>         }
<a name="172" href="#172">172</a>     }
<a name="173" href="#173">173</a> 
<a name="174" href="#174">174</a>     <strong>protected</strong> <strong>boolean</strong> rulesAccept(Object o) {
<a name="175" href="#175">175</a>         <strong>return</strong> rulesAccept(getDecideRule(o),o);
<a name="176" href="#176">176</a>     }
<a name="177" href="#177">177</a> 
<a name="178" href="#178">178</a>     <strong>protected</strong> <strong>boolean</strong> rulesAccept(<a href="../../../../org/archive/crawler/deciderules/DecideRule.html">DecideRule</a> rule, Object o) {
<a name="179" href="#179">179</a>         <strong>return</strong> rule.decisionFor(o) != DecideRule.REJECT;
<a name="180" href="#180">180</a>     }
<a name="181" href="#181">181</a>     <em>/**<em>*</em></em>
<a name="182" href="#182">182</a> <em>     * Returns the next processor for the given CrawlURI in the processor chain.</em>
<a name="183" href="#183">183</a> <em>     * @param curi The CrawlURI that we want to find the next processor for.</em>
<a name="184" href="#184">184</a> <em>     * @return The next processor for the given CrawlURI in the processor chain.</em>
<a name="185" href="#185">185</a> <em>     */</em>
<a name="186" href="#186">186</a>     <strong>public</strong> <a href="../../../../org/archive/crawler/framework/Processor.html">Processor</a> getDefaultNextProcessor(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) {
<a name="187" href="#187">187</a>         <strong>return</strong> defaultNextProcessor;
<a name="188" href="#188">188</a>     }
<a name="189" href="#189">189</a> 
<a name="190" href="#190">190</a>     <em>/**<em>* Set the default next processor in the chain.</em></em>
<a name="191" href="#191">191</a> <em>     *</em>
<a name="192" href="#192">192</a> <em>     * @param nextProcessor the default next processor in the chain.</em>
<a name="193" href="#193">193</a> <em>     */</em>
<a name="194" href="#194">194</a>     <strong>public</strong> <strong>void</strong> setDefaultNextProcessor(<a href="../../../../org/archive/crawler/framework/Processor.html">Processor</a> nextProcessor) {
<a name="195" href="#195">195</a>         defaultNextProcessor = nextProcessor;
<a name="196" href="#196">196</a>     }
<a name="197" href="#197">197</a> 
<a name="198" href="#198">198</a>     <em>/**<em>* </em></em>
<a name="199" href="#199">199</a> <em>     * Get the controller object.</em>
<a name="200" href="#200">200</a> <em>     *</em>
<a name="201" href="#201">201</a> <em>     * @return the controller object.</em>
<a name="202" href="#202">202</a> <em>     */</em>
<a name="203" href="#203">203</a>     <strong>public</strong> <a href="../../../../org/archive/crawler/framework/CrawlController.html">CrawlController</a> getController() {
<a name="204" href="#204">204</a>         <strong>return</strong> getSettingsHandler().getOrder().getController();
<a name="205" href="#205">205</a>     }
<a name="206" href="#206">206</a> 
<a name="207" href="#207">207</a>     <strong>public</strong> <a href="../../../../org/archive/crawler/framework/Processor.html">Processor</a> spawn(<strong>int</strong> serialNum) {
<a name="208" href="#208">208</a>         <a href="../../../../org/archive/crawler/framework/Processor.html">Processor</a> newInst = <strong>null</strong>;
<a name="209" href="#209">209</a>         <strong>try</strong> {
<a name="210" href="#210">210</a>             Constructor co =
<a name="211" href="#211">211</a>                 getClass().getConstructor(<strong>new</strong> Class[] { String.<strong>class</strong> });
<a name="212" href="#212">212</a>             newInst =
<a name="213" href="#213">213</a>                 (<a href="../../../../org/archive/crawler/framework/Processor.html">Processor</a>) co.<strong>new</strong>Instance(<strong>new</strong> Object[] {
<a name="214" href="#214">214</a>                     getName() + serialNum
<a name="215" href="#215">215</a>                     });
<a name="216" href="#216">216</a>             getParent().setAttribute(newInst);
<a name="217" href="#217">217</a>             newInst.setTransient(<strong>true</strong>);
<a name="218" href="#218">218</a>         } <strong>catch</strong> (Exception e) {
<a name="219" href="#219">219</a>             <em class="comment">// TODO Auto-generated catch block</em>
<a name="220" href="#220">220</a>             e.printStackTrace();
<a name="221" href="#221">221</a>         }
<a name="222" href="#222">222</a>         <strong>return</strong> newInst;
<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>     * Compiles and returns a report (in human readable form) about the status</em>
<a name="227" href="#227">227</a> <em>     * of the processor.  The processor's name (of implementing class) should</em>
<a name="228" href="#228">228</a> <em>     * always be included.</em>
<a name="229" href="#229">229</a> <em>     * &lt;p></em>
<a name="230" href="#230">230</a> <em>     * Examples of stats declared would include:&lt;br></em>
<a name="231" href="#231">231</a> <em>     * * Number of CrawlURIs handled.&lt;br></em>
<a name="232" href="#232">232</a> <em>     * * Number of links extracted (for link extractors)&lt;br></em>
<a name="233" href="#233">233</a> <em>     * etc.</em>
<a name="234" href="#234">234</a> <em>     *</em>
<a name="235" href="#235">235</a> <em>     * @return A human readable report on the processor's state.</em>
<a name="236" href="#236">236</a> <em>     */</em>
<a name="237" href="#237">237</a>     <strong>public</strong> String report(){
<a name="238" href="#238">238</a>         <strong>return</strong> <span class="string">""</span>; <em class="comment">// Default behavior.</em>
<a name="239" href="#239">239</a>     }
<a name="240" href="#240">240</a>     
<a name="241" href="#241">241</a>     <em>/**<em>*</em></em>
<a name="242" href="#242">242</a> <em>     * @param curi CrawlURI to examine.</em>
<a name="243" href="#243">243</a> <em>     * @return True if content to process -- content length is > 0 </em>
<a name="244" href="#244">244</a> <em>     * -- and links have not yet been extracted.</em>
<a name="245" href="#245">245</a> <em>     */</em>
<a name="246" href="#246">246</a>     <strong>protected</strong> <strong>boolean</strong> isContentToProcess(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) {
<a name="247" href="#247">247</a>         <strong>return</strong> !curi.hasBeenLinkExtracted() &amp;&amp; curi.getContentLength() > 0;
<a name="248" href="#248">248</a>     }
<a name="249" href="#249">249</a>     
<a name="250" href="#250">250</a>     <em>/**<em>*</em></em>
<a name="251" href="#251">251</a> <em>     * @param curi CrawlURI to examine.</em>
<a name="252" href="#252">252</a> <em>     * @return True if {@link #isContentToProcess(CrawlURI)} and</em>
<a name="253" href="#253">253</a> <em>     * the CrawlURI represents a successful http transaction.</em>
<a name="254" href="#254">254</a> <em>     */</em>
<a name="255" href="#255">255</a>     <strong>protected</strong> <strong>boolean</strong> isHttpTransactionContentToProcess(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) {
<a name="256" href="#256">256</a>         <strong>return</strong> isContentToProcess(curi) &amp;&amp;
<a name="257" href="#257">257</a>             curi.isHttpTransaction() &amp;&amp;
<a name="258" href="#258">258</a>             curi.isSuccess();
<a name="259" href="#259">259</a>     }
<a name="260" href="#260">260</a>     
<a name="261" href="#261">261</a>     <em>/**<em>*</em></em>
<a name="262" href="#262">262</a> <em>     * @param contentType Found content type.</em>
<a name="263" href="#263">263</a> <em>     * @param expectedPrefix String to find at start of contenttype: e.g.</em>
<a name="264" href="#264">264</a> <em>     * &lt;code>text/html&lt;/code>.</em>
<a name="265" href="#265">265</a> <em>     * @return True if passed content-type begins with</em>
<a name="266" href="#266">266</a> <em>     * expected mimetype.</em>
<a name="267" href="#267">267</a> <em>     */</em>
<a name="268" href="#268">268</a>     <strong>protected</strong> <strong>boolean</strong> isExpectedMimeType(String contentType,
<a name="269" href="#269">269</a>             String expectedPrefix) {
<a name="270" href="#270">270</a>         <strong>return</strong> contentType != <strong>null</strong> &amp;&amp;
<a name="271" href="#271">271</a>             contentType.toLowerCase().startsWith(expectedPrefix);
<a name="272" href="#272">272</a>     }
<a name="273" href="#273">273</a> 
<a name="274" href="#274">274</a>     <strong>public</strong> <strong>void</strong> kickUpdate() {
<a name="275" href="#275">275</a>         <em class="comment">// by default do nothing</em>
<a name="276" href="#276">276</a>     }
<a name="277" href="#277">277</a> }
</pre>
<hr/><div id="footer">This page was automatically generated by <a href="http://maven.apache.org/">Maven</a></div></body>
</html>