experimentalv10warcwriterprocessor.html

来自「网络爬虫开源代码」· HTML 代码 · 共 336 行 · 第 1/2 页

HTML
336
字号
<a name="160" href="#160">160</a>                     <em class="comment">// TODO: This is digest for content -- doesn't include</em><a name="161" href="#161">161</a>                     <em class="comment">// response headers.</em><a name="162" href="#162">162</a>                     r.addLabelValue(NAMED_FIELD_CHECKSUM_LABEL,<a name="163" href="#163">163</a>                         curi.getContentDigestSchemeString());<a name="164" href="#164">164</a>                 }<a name="165" href="#165">165</a>                 r.addLabelValue(NAMED_FIELD_IP_LABEL, getHostAddress(curi));<a name="166" href="#166">166</a>                 URI rid = writeResponse(w, timestamp, HTTP_RESPONSE_MIMETYPE,<a name="167" href="#167">167</a>                 	baseid, curi, r);<a name="168" href="#168">168</a>                 r = <strong>new</strong> <a href="../../../../org/archive/util/anvl/ANVLRecord.html">ANVLRecord</a>(1);<a name="169" href="#169">169</a>                 r.addLabelValue(NAMED_FIELD_RELATED_LABEL, rid.toString());<a name="170" href="#170">170</a>                 writeRequest(w, timestamp, HTTP_REQUEST_MIMETYPE,<a name="171" href="#171">171</a>                 	baseid, curi, r);<a name="172" href="#172">172</a>                 writeMetadata(w, timestamp, baseid, curi, r);<a name="173" href="#173">173</a>             } <strong>else</strong> <strong>if</strong> (lowerCaseScheme.equals(<span class="string">"dns"</span>)) {<a name="174" href="#174">174</a>                 String ip = curi.getString(A_DNS_SERVER_IP_LABEL);<a name="175" href="#175">175</a>                 <a href="../../../../org/archive/util/anvl/ANVLRecord.html">ANVLRecord</a> r = <strong>null</strong>;<a name="176" href="#176">176</a>                 <strong>if</strong> (ip != <strong>null</strong> &amp;&amp; ip.length() > 0) {<a name="177" href="#177">177</a>                 	r = <strong>new</strong> <a href="../../../../org/archive/util/anvl/ANVLRecord.html">ANVLRecord</a>();<a name="178" href="#178">178</a>                     r.addLabelValue(NAMED_FIELD_IP_LABEL, ip);<a name="179" href="#179">179</a>                 }<a name="180" href="#180">180</a>                 writeResponse(w, timestamp, curi.getContentType(), baseid,<a name="181" href="#181">181</a>                     curi, r);<a name="182" href="#182">182</a>             } <strong>else</strong> {<a name="183" href="#183">183</a>                 logger.warning(<span class="string">"No handler for scheme "</span> + lowerCaseScheme);<a name="184" href="#184">184</a>             }<a name="185" href="#185">185</a>         } <strong>catch</strong> (IOException e) {<a name="186" href="#186">186</a>             <em class="comment">// Invalidate this file (It gets a '.invalid' suffix).</em><a name="187" href="#187">187</a>             getPool().invalidateFile(writer);<a name="188" href="#188">188</a>             <em class="comment">// Set the writer to null otherwise the pool accounting</em><a name="189" href="#189">189</a>             <em class="comment">// of how many active writers gets skewed if we subsequently</em><a name="190" href="#190">190</a>             <em class="comment">// do a returnWriter call on this object in the finally block.</em><a name="191" href="#191">191</a>             writer = <strong>null</strong>;<a name="192" href="#192">192</a>             <strong>throw</strong> e;<a name="193" href="#193">193</a>         } <strong>finally</strong> {<a name="194" href="#194">194</a>             <strong>if</strong> (writer != <strong>null</strong>) {<a name="195" href="#195">195</a>             	setTotalBytesWritten(getTotalBytesWritten() +<a name="196" href="#196">196</a>             	     (writer.getPosition() - position));<a name="197" href="#197">197</a>                 getPool().returnFile(writer);<a name="198" href="#198">198</a>             }<a name="199" href="#199">199</a>         }<a name="200" href="#200">200</a>         checkBytesWritten();<a name="201" href="#201">201</a>     }<a name="202" href="#202">202</a>     <a name="203" href="#203">203</a>     <strong>protected</strong> URI writeRequest(<strong>final</strong> <a href="../../../../org/archive/io/warc/ExperimentalWARCWriter.html">ExperimentalWARCWriter</a> w,<a name="204" href="#204">204</a>             <strong>final</strong> String timestamp, <strong>final</strong> String mimetype,<a name="205" href="#205">205</a>             <strong>final</strong> URI baseid, <strong>final</strong> <a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi,<a name="206" href="#206">206</a>             <strong>final</strong> <a href="../../../../org/archive/util/anvl/ANVLRecord.html">ANVLRecord</a> namedFields) <a name="207" href="#207">207</a>     throws IOException {<a name="208" href="#208">208</a>         <strong>final</strong> URI uid = qualifyRecordID(baseid, TYPE, REQUEST);<a name="209" href="#209">209</a>         <a href="../../../../org/archive/io/ReplayInputStream.html">ReplayInputStream</a> ris =<a name="210" href="#210">210</a>             curi.getHttpRecorder().getRecordedOutput().getReplayInputStream();<a name="211" href="#211">211</a>         <strong>try</strong> {<a name="212" href="#212">212</a>             w.writeRequestRecord(curi.toString(), timestamp, mimetype, uid,<a name="213" href="#213">213</a>                 namedFields, ris,<a name="214" href="#214">214</a>                 curi.getHttpRecorder().getRecordedOutput().getSize());<a name="215" href="#215">215</a>         } <strong>finally</strong> {<a name="216" href="#216">216</a>             <strong>if</strong> (ris != <strong>null</strong>) {<a name="217" href="#217">217</a>                 ris.close();<a name="218" href="#218">218</a>             }<a name="219" href="#219">219</a>         }<a name="220" href="#220">220</a>         <strong>return</strong> uid;<a name="221" href="#221">221</a>     }<a name="222" href="#222">222</a>     <a name="223" href="#223">223</a>     <strong>protected</strong> URI writeResponse(<strong>final</strong> <a href="../../../../org/archive/io/warc/ExperimentalWARCWriter.html">ExperimentalWARCWriter</a> w,<a name="224" href="#224">224</a>             <strong>final</strong> String timestamp, <strong>final</strong> String mimetype,<a name="225" href="#225">225</a>             <strong>final</strong> URI baseid, <strong>final</strong> <a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi,<a name="226" href="#226">226</a>             <strong>final</strong> <a href="../../../../org/archive/util/anvl/ANVLRecord.html">ANVLRecord</a> namedFields) <a name="227" href="#227">227</a>     throws IOException {<a name="228" href="#228">228</a>         <a href="../../../../org/archive/io/ReplayInputStream.html">ReplayInputStream</a> ris =<a name="229" href="#229">229</a>             curi.getHttpRecorder().getRecordedInput().getReplayInputStream();<a name="230" href="#230">230</a>         <strong>try</strong> {<a name="231" href="#231">231</a>             w.writeResponseRecord(curi.toString(), timestamp, mimetype, baseid,<a name="232" href="#232">232</a>                 namedFields, ris,<a name="233" href="#233">233</a>                 curi.getHttpRecorder().getRecordedInput().getSize());<a name="234" href="#234">234</a>         } <strong>finally</strong> {<a name="235" href="#235">235</a>             <strong>if</strong> (ris != <strong>null</strong>) {<a name="236" href="#236">236</a>                 ris.close();<a name="237" href="#237">237</a>             }<a name="238" href="#238">238</a>         }<a name="239" href="#239">239</a>         <strong>return</strong> baseid;<a name="240" href="#240">240</a>     }<a name="241" href="#241">241</a>     <a name="242" href="#242">242</a>     <strong>protected</strong> URI writeMetadata(<strong>final</strong> <a href="../../../../org/archive/io/warc/ExperimentalWARCWriter.html">ExperimentalWARCWriter</a> w,<a name="243" href="#243">243</a>             <strong>final</strong> String timestamp,<a name="244" href="#244">244</a>             <strong>final</strong> URI baseid, <strong>final</strong> <a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi,<a name="245" href="#245">245</a>             <strong>final</strong> <a href="../../../../org/archive/util/anvl/ANVLRecord.html">ANVLRecord</a> namedFields) <a name="246" href="#246">246</a>     throws IOException {<a name="247" href="#247">247</a>         <strong>final</strong> URI uid = qualifyRecordID(baseid, TYPE, METADATA);<a name="248" href="#248">248</a>         <em class="comment">// Get some metadata from the curi.</em><a name="249" href="#249">249</a>         <em class="comment">// TODO: Get all curi metadata.</em><a name="250" href="#250">250</a>         <a href="../../../../org/archive/util/anvl/ANVLRecord.html">ANVLRecord</a> r = <strong>new</strong> <a href="../../../../org/archive/util/anvl/ANVLRecord.html">ANVLRecord</a>();<a name="251" href="#251">251</a>         <strong>if</strong> (curi.isSeed()) {<a name="252" href="#252">252</a>             r.addLabel(<span class="string">"seed"</span>);<a name="253" href="#253">253</a>         } <strong>else</strong> {<a name="254" href="#254">254</a>         	<strong>if</strong> (curi.forceFetch()) {<a name="255" href="#255">255</a>         		r.addLabel(<span class="string">"force-fetch"</span>);<a name="256" href="#256">256</a>         	}<a name="257" href="#257">257</a>             r.addLabelValue(<span class="string">"via"</span>, curi.flattenVia());<a name="258" href="#258">258</a>             r.addLabelValue(<span class="string">"pathFromSeed"</span>, curi.getPathFromSeed());<a name="259" href="#259">259</a>         }<a name="260" href="#260">260</a>         Collection&lt;Link> links = curi.getOutLinks();<a name="261" href="#261">261</a>         <strong>if</strong> (links != <strong>null</strong> &amp;&amp; links.size() > 0) {<a name="262" href="#262">262</a>             <strong>for</strong> (Link link: links) {<a name="263" href="#263">263</a>                 r.addLabelValue(<span class="string">"outlink"</span>, link.toString());<a name="264" href="#264">264</a>             }<a name="265" href="#265">265</a>         }<a name="266" href="#266">266</a>         <strong>if</strong> (curi.isTruncatedFetch()) {<a name="267" href="#267">267</a>             String value = curi.isTimeTruncatedFetch()?<a name="268" href="#268">268</a>                     NAMED_FIELD_TRUNCATED_VALUE_TIME:<a name="269" href="#269">269</a>                 curi.isLengthTruncatedFetch()?<a name="270" href="#270">270</a>                         NAMED_FIELD_TRUNCATED_VALUE_LEN:<a name="271" href="#271">271</a>                 curi.isHeaderTruncatedFetch()?<a name="272" href="#272">272</a>                         NAMED_FIELD_TRUNCATED_VALUE_HEAD:<a name="273" href="#273">273</a>                 NAMED_FIELD_TRUNCATED_VALUE_UNSPECIFIED;<a name="274" href="#274">274</a>                       <a name="275" href="#275">275</a>             r.addLabelValue(NAMED_FIELD_TRUNCATED, value);<a name="276" href="#276">276</a>         }<a name="277" href="#277">277</a>         <a name="278" href="#278">278</a>         <em class="comment">// TODO: Other curi fields to write to metadata.</em><a name="279" href="#279">279</a>         <em class="comment">// </em><a name="280" href="#280">280</a>         <em class="comment">// Credentials</em><a name="281" href="#281">281</a>         <em class="comment">// </em><a name="282" href="#282">282</a>         <em class="comment">// fetch-began-time: 1154569278774</em><a name="283" href="#283">283</a>         <em class="comment">// fetch-completed-time: 1154569281816</em><a name="284" href="#284">284</a>         <em class="comment">//</em><a name="285" href="#285">285</a>         <em class="comment">// Annotations.</em><a name="286" href="#286">286</a>         <a name="287" href="#287">287</a>         byte [] b = r.getUTF8Bytes();<a name="288" href="#288">288</a>         w.writeMetadataRecord(curi.toString(), timestamp, ANVLRecord.MIMETYPE,<a name="289" href="#289">289</a>             uid, namedFields, <strong>new</strong> ByteArrayInputStream(b), b.length);<a name="290" href="#290">290</a>         <strong>return</strong> uid;<a name="291" href="#291">291</a>     }<a name="292" href="#292">292</a>     <a name="293" href="#293">293</a>     <strong>protected</strong> URI getRecordID() throws IOException {<a name="294" href="#294">294</a>         URI result;<a name="295" href="#295">295</a>         <strong>try</strong> {<a name="296" href="#296">296</a>             result = GeneratorFactory.getFactory().getRecordID();<a name="297" href="#297">297</a>         } <strong>catch</strong> (URISyntaxException e) {<a name="298" href="#298">298</a>             <strong>throw</strong> <strong>new</strong> IOException(e.toString());<a name="299" href="#299">299</a>         }<a name="300" href="#300">300</a>         <strong>return</strong> result;<a name="301" href="#301">301</a>     }<a name="302" href="#302">302</a>     <a name="303" href="#303">303</a>     <strong>protected</strong> URI qualifyRecordID(<strong>final</strong> URI base, <strong>final</strong> String key,<a name="304" href="#304">304</a>             <strong>final</strong> String value)<a name="305" href="#305">305</a>     throws IOException {<a name="306" href="#306">306</a>         URI result;<a name="307" href="#307">307</a>         Map&lt;String, String> qualifiers = <strong>new</strong> HashMap&lt;String, String>(1);<a name="308" href="#308">308</a>         qualifiers.put(key, value);<a name="309" href="#309">309</a>         <strong>try</strong> {<a name="310" href="#310">310</a>             result = GeneratorFactory.getFactory().<a name="311" href="#311">311</a>                 qualifyRecordID(base, qualifiers);<a name="312" href="#312">312</a>         } <strong>catch</strong> (URISyntaxException e) {<a name="313" href="#313">313</a>             <strong>throw</strong> <strong>new</strong> IOException(e.toString());<a name="314" href="#314">314</a>         }<a name="315" href="#315">315</a>         <strong>return</strong> result;<a name="316" href="#316">316</a>     }  <a name="317" href="#317">317</a> <a name="318" href="#318">318</a>     <strong>public</strong> List getMetadata() {<a name="319" href="#319">319</a>         <em class="comment">// TODO: As ANVL?</em><a name="320" href="#320">320</a>         <strong>return</strong> <strong>null</strong>;<a name="321" href="#321">321</a>     }<a name="322" href="#322">322</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?