📄 experimentalwarcwriterprocessor.html
字号:
<a name="152" href="#152">152</a> <strong>if</strong> (lowerCaseScheme.startsWith(<span class="string">"http"</span>)) {<a name="153" href="#153">153</a> <em class="comment">// Add named fields for ip, checksum, and relate the metadata</em><a name="154" href="#154">154</a> <em class="comment">// and request to the resource field.</em><a name="155" href="#155">155</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="156" href="#156">156</a> <strong>if</strong> (curi.getContentDigest() != <strong>null</strong>) {<a name="157" href="#157">157</a> <em class="comment">// TODO: This is digest for content -- doesn't include</em><a name="158" href="#158">158</a> <em class="comment">// response headers.</em><a name="159" href="#159">159</a> r.addLabelValue(NAMED_FIELD_CHECKSUM_LABEL,<a name="160" href="#160">160</a> curi.getContentDigestSchemeString());<a name="161" href="#161">161</a> }<a name="162" href="#162">162</a> r.addLabelValue(NAMED_FIELD_IP_LABEL, getHostAddress(curi));<a name="163" href="#163">163</a> URI rid = writeResponse(w, timestamp, HTTP_RESPONSE_MIMETYPE,<a name="164" href="#164">164</a> baseid, curi, r);<a name="165" href="#165">165</a> r = <strong>new</strong> <a href="../../../../org/archive/util/anvl/ANVLRecord.html">ANVLRecord</a>(1);<a name="166" href="#166">166</a> r.addLabelValue(NAMED_FIELD_RELATED_LABEL, rid.toString());<a name="167" href="#167">167</a> writeRequest(w, timestamp, HTTP_REQUEST_MIMETYPE,<a name="168" href="#168">168</a> baseid, curi, r);<a name="169" href="#169">169</a> writeMetadata(w, timestamp, baseid, curi, r);<a name="170" href="#170">170</a> } <strong>else</strong> <strong>if</strong> (lowerCaseScheme.equals(<span class="string">"dns"</span>)) {<a name="171" href="#171">171</a> String ip = curi.getString(A_DNS_SERVER_IP_LABEL);<a name="172" href="#172">172</a> <a href="../../../../org/archive/util/anvl/ANVLRecord.html">ANVLRecord</a> r = <strong>null</strong>;<a name="173" href="#173">173</a> <strong>if</strong> (ip != <strong>null</strong> && ip.length() > 0) {<a name="174" href="#174">174</a> r = <strong>new</strong> <a href="../../../../org/archive/util/anvl/ANVLRecord.html">ANVLRecord</a>();<a name="175" href="#175">175</a> r.addLabelValue(NAMED_FIELD_IP_LABEL, ip);<a name="176" href="#176">176</a> }<a name="177" href="#177">177</a> writeResponse(w, timestamp, curi.getContentType(), baseid,<a name="178" href="#178">178</a> curi, r);<a name="179" href="#179">179</a> } <strong>else</strong> {<a name="180" href="#180">180</a> logger.warning(<span class="string">"No handler for scheme "</span> + lowerCaseScheme);<a name="181" href="#181">181</a> }<a name="182" href="#182">182</a> } <strong>catch</strong> (IOException e) {<a name="183" href="#183">183</a> <em class="comment">// Invalidate this file (It gets a '.invalid' suffix).</em><a name="184" href="#184">184</a> getPool().invalidateFile(writer);<a name="185" href="#185">185</a> <em class="comment">// Set the writer to null otherwise the pool accounting</em><a name="186" href="#186">186</a> <em class="comment">// of how many active writers gets skewed if we subsequently</em><a name="187" href="#187">187</a> <em class="comment">// do a returnWriter call on this object in the finally block.</em><a name="188" href="#188">188</a> writer = <strong>null</strong>;<a name="189" href="#189">189</a> <strong>throw</strong> e;<a name="190" href="#190">190</a> } <strong>finally</strong> {<a name="191" href="#191">191</a> <strong>if</strong> (writer != <strong>null</strong>) {<a name="192" href="#192">192</a> setTotalBytesWritten(getTotalBytesWritten() +<a name="193" href="#193">193</a> (writer.getPosition() - position));<a name="194" href="#194">194</a> getPool().returnFile(writer);<a name="195" href="#195">195</a> }<a name="196" href="#196">196</a> }<a name="197" href="#197">197</a> checkBytesWritten();<a name="198" href="#198">198</a> }<a name="199" href="#199">199</a> <a name="200" href="#200">200</a> <strong>protected</strong> URI writeRequest(<strong>final</strong> <a href="../../../../org/archive/io/warc/ExperimentalWARCWriter.html">ExperimentalWARCWriter</a> w,<a name="201" href="#201">201</a> <strong>final</strong> String timestamp, <strong>final</strong> String mimetype,<a name="202" href="#202">202</a> <strong>final</strong> URI baseid, <strong>final</strong> <a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi,<a name="203" href="#203">203</a> <strong>final</strong> <a href="../../../../org/archive/util/anvl/ANVLRecord.html">ANVLRecord</a> namedFields) <a name="204" href="#204">204</a> throws IOException {<a name="205" href="#205">205</a> <strong>final</strong> URI uid = qualifyRecordID(baseid, TYPE, REQUEST);<a name="206" href="#206">206</a> w.writeRequestRecord(curi.toString(), timestamp, mimetype, uid,<a name="207" href="#207">207</a> namedFields,<a name="208" href="#208">208</a> curi.getHttpRecorder().getRecordedOutput().getReplayInputStream(),<a name="209" href="#209">209</a> curi.getHttpRecorder().getRecordedOutput().getSize());<a name="210" href="#210">210</a> <strong>return</strong> uid;<a name="211" href="#211">211</a> }<a name="212" href="#212">212</a> <a name="213" href="#213">213</a> <strong>protected</strong> URI writeResponse(<strong>final</strong> <a href="../../../../org/archive/io/warc/ExperimentalWARCWriter.html">ExperimentalWARCWriter</a> w,<a name="214" href="#214">214</a> <strong>final</strong> String timestamp, <strong>final</strong> String mimetype,<a name="215" href="#215">215</a> <strong>final</strong> URI baseid, <strong>final</strong> <a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi,<a name="216" href="#216">216</a> <strong>final</strong> <a href="../../../../org/archive/util/anvl/ANVLRecord.html">ANVLRecord</a> namedFields) <a name="217" href="#217">217</a> throws IOException {<a name="218" href="#218">218</a> w.writeResponseRecord(curi.toString(), timestamp, mimetype, baseid,<a name="219" href="#219">219</a> namedFields,<a name="220" href="#220">220</a> curi.getHttpRecorder().getRecordedInput().getReplayInputStream(),<a name="221" href="#221">221</a> curi.getHttpRecorder().getRecordedInput().getSize());<a name="222" href="#222">222</a> <strong>return</strong> baseid;<a name="223" href="#223">223</a> }<a name="224" href="#224">224</a> <a name="225" href="#225">225</a> <strong>protected</strong> URI writeMetadata(<strong>final</strong> <a href="../../../../org/archive/io/warc/ExperimentalWARCWriter.html">ExperimentalWARCWriter</a> w,<a name="226" href="#226">226</a> <strong>final</strong> String timestamp,<a name="227" href="#227">227</a> <strong>final</strong> URI baseid, <strong>final</strong> <a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi,<a name="228" href="#228">228</a> <strong>final</strong> <a href="../../../../org/archive/util/anvl/ANVLRecord.html">ANVLRecord</a> namedFields) <a name="229" href="#229">229</a> throws IOException {<a name="230" href="#230">230</a> <strong>final</strong> URI uid = qualifyRecordID(baseid, TYPE, METADATA);<a name="231" href="#231">231</a> <em class="comment">// Get some metadata from the curi.</em><a name="232" href="#232">232</a> <em class="comment">// TODO: Get all curi metadata.</em><a name="233" href="#233">233</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="234" href="#234">234</a> <strong>if</strong> (curi.isSeed()) {<a name="235" href="#235">235</a> r.addLabel(<span class="string">"seed"</span>);<a name="236" href="#236">236</a> } <strong>else</strong> {<a name="237" href="#237">237</a> <strong>if</strong> (curi.forceFetch()) {<a name="238" href="#238">238</a> r.addLabel(<span class="string">"force-fetch"</span>);<a name="239" href="#239">239</a> }<a name="240" href="#240">240</a> r.addLabelValue(<span class="string">"via"</span>, curi.flattenVia());<a name="241" href="#241">241</a> r.addLabelValue(<span class="string">"pathFromSeed"</span>, curi.getPathFromSeed());<a name="242" href="#242">242</a> }<a name="243" href="#243">243</a> Collection links = curi.getOutLinks();<a name="244" href="#244">244</a> <strong>if</strong> (links != <strong>null</strong> || links.size() > 0) {<a name="245" href="#245">245</a> <strong>for</strong> (<strong>final</strong> Iterator i = links.iterator(); i.hasNext();) {<a name="246" href="#246">246</a> r.addLabelValue(<span class="string">"outlink"</span>, ((Link)i.next()).toString());<a name="247" href="#247">247</a> }<a name="248" href="#248">248</a> }<a name="249" href="#249">249</a> <strong>if</strong> (curi.isTruncatedFetch()) {<a name="250" href="#250">250</a> String value = curi.isTimeTruncatedFetch()?<a name="251" href="#251">251</a> NAMED_FIELD_TRUNCATED_VALUE_TIME:<a name="252" href="#252">252</a> curi.isLengthTruncatedFetch()?<a name="253" href="#253">253</a> NAMED_FIELD_TRUNCATED_VALUE_LEN:<a name="254" href="#254">254</a> curi.isHeaderTruncatedFetch()?<a name="255" href="#255">255</a> NAMED_FIELD_TRUNCATED_VALUE_HEAD:<a name="256" href="#256">256</a> NAMED_FIELD_TRUNCATED_VALUE_UNSPECIFIED;<a name="257" href="#257">257</a> <a name="258" href="#258">258</a> r.addLabelValue(NAMED_FIELD_TRUNCATED, value);<a name="259" href="#259">259</a> }<a name="260" href="#260">260</a> <a name="261" href="#261">261</a> <em class="comment">// TODO: Other curi fields to write to metadata.</em><a name="262" href="#262">262</a> <em class="comment">// </em><a name="263" href="#263">263</a> <em class="comment">// Credentials</em><a name="264" href="#264">264</a> <em class="comment">// </em><a name="265" href="#265">265</a> <em class="comment">// fetch-began-time: 1154569278774</em><a name="266" href="#266">266</a> <em class="comment">// fetch-completed-time: 1154569281816</em><a name="267" href="#267">267</a> <em class="comment">//</em><a name="268" href="#268">268</a> <em class="comment">// Annotations.</em><a name="269" href="#269">269</a> <a name="270" href="#270">270</a> byte [] b = r.getUTF8Bytes();<a name="271" href="#271">271</a> w.writeMetadataRecord(curi.toString(), timestamp, ANVLRecord.MIMETYPE,<a name="272" href="#272">272</a> uid, namedFields, <strong>new</strong> ByteArrayInputStream(b), b.length);<a name="273" href="#273">273</a> <strong>return</strong> uid;<a name="274" href="#274">274</a> }<a name="275" href="#275">275</a> <a name="276" href="#276">276</a> <strong>protected</strong> URI getRecordID() throws IOException {<a name="277" href="#277">277</a> URI result;<a name="278" href="#278">278</a> <strong>try</strong> {<a name="279" href="#279">279</a> result = GeneratorFactory.getFactory().getRecordID();<a name="280" href="#280">280</a> } <strong>catch</strong> (URISyntaxException e) {<a name="281" href="#281">281</a> <strong>throw</strong> <strong>new</strong> IOException(e.toString());<a name="282" href="#282">282</a> }<a name="283" href="#283">283</a> <strong>return</strong> result;<a name="284" href="#284">284</a> }<a name="285" href="#285">285</a> <a name="286" href="#286">286</a> <strong>protected</strong> URI qualifyRecordID(<strong>final</strong> URI base, <strong>final</strong> String key,<a name="287" href="#287">287</a> <strong>final</strong> String value)<a name="288" href="#288">288</a> throws IOException {<a name="289" href="#289">289</a> URI result;<a name="290" href="#290">290</a> Map<String, String> qualifiers = <strong>new</strong> HashMap<String, String>(1);<a name="291" href="#291">291</a> qualifiers.put(key, value);<a name="292" href="#292">292</a> <strong>try</strong> {<a name="293" href="#293">293</a> result = GeneratorFactory.getFactory().<a name="294" href="#294">294</a> qualifyRecordID(base, qualifiers);<a name="295" href="#295">295</a> } <strong>catch</strong> (URISyntaxException e) {<a name="296" href="#296">296</a> <strong>throw</strong> <strong>new</strong> IOException(e.toString());<a name="297" href="#297">297</a> }<a name="298" href="#298">298</a> <strong>return</strong> result;<a name="299" href="#299">299</a> } <a name="300" href="#300">300</a> <a name="301" href="#301">301</a> <strong>public</strong> List getMetadata() {<a name="302" href="#302">302</a> <em class="comment">// TODO: As ANVL?</em><a name="303" href="#303">303</a> <strong>return</strong> <strong>null</strong>;<a name="304" href="#304">304</a> }<a name="305" href="#305">305</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 + -