📄 fetchdns.html
字号:
<a name="155" href="#155">155</a> address = InetAddress.getByName(dnsName);<a name="156" href="#156">156</a> } <strong>catch</strong> (UnknownHostException e1) {<a name="157" href="#157">157</a> address = <strong>null</strong>;<a name="158" href="#158">158</a> }<a name="159" href="#159">159</a> <strong>if</strong> (address != <strong>null</strong>) {<a name="160" href="#160">160</a> targetHost.setIP(address, DEFAULT_TTL_FOR_NON_DNS_RESOLVES);<a name="161" href="#161">161</a> curi.setFetchStatus(S_GETBYNAME_SUCCESS);<a name="162" href="#162">162</a> <strong>if</strong> (logger.isLoggable(Level.FINE)) {<a name="163" href="#163">163</a> logger.fine(<span class="string">"Found address for "</span> + dnsName +<a name="164" href="#164">164</a> <span class="string">" using native dns."</span>);<a name="165" href="#165">165</a> }<a name="166" href="#166">166</a> } <strong>else</strong> {<a name="167" href="#167">167</a> <strong>if</strong> (logger.isLoggable(Level.FINE)) {<a name="168" href="#168">168</a> logger.fine(<span class="string">"Failed find of address for "</span> + dnsName +<a name="169" href="#169">169</a> <span class="string">" using native dns."</span>);<a name="170" href="#170">170</a> }<a name="171" href="#171">171</a> setUnresolvable(curi, targetHost);<a name="172" href="#172">172</a> }<a name="173" href="#173">173</a> } <strong>else</strong> {<a name="174" href="#174">174</a> setUnresolvable(curi, targetHost);<a name="175" href="#175">175</a> }<a name="176" href="#176">176</a> }<a name="177" href="#177">177</a> curi.putLong(A_FETCH_COMPLETED_TIME, System.currentTimeMillis());<a name="178" href="#178">178</a> }<a name="179" href="#179">179</a> <a name="180" href="#180">180</a> <strong>protected</strong> <strong>void</strong> storeDNSRecord(<strong>final</strong> <a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi, <strong>final</strong> String dnsName,<a name="181" href="#181">181</a> <strong>final</strong> <a href="../../../../org/archive/crawler/datamodel/CrawlHost.html">CrawlHost</a> targetHost, <strong>final</strong> Record[] rrecordSet) {<a name="182" href="#182">182</a> <em class="comment">// Get TTL and IP info from the first A record (there may be</em><a name="183" href="#183">183</a> <em class="comment">// multiple, e.g. www.washington.edu) then update the CrawlServer</em><a name="184" href="#184">184</a> ARecord arecord = getFirstARecord(rrecordSet);<a name="185" href="#185">185</a> <strong>if</strong> (arecord == <strong>null</strong>) {<a name="186" href="#186">186</a> <strong>throw</strong> <strong>new</strong> NullPointerException(<span class="string">"Got null arecord for "</span> +<a name="187" href="#187">187</a> dnsName);<a name="188" href="#188">188</a> }<a name="189" href="#189">189</a> targetHost.setIP(arecord.getAddress(), arecord.getTTL());<a name="190" href="#190">190</a> <strong>try</strong> {<a name="191" href="#191">191</a> recordDNS(curi, rrecordSet);<a name="192" href="#192">192</a> curi.setFetchStatus(S_DNS_SUCCESS);<a name="193" href="#193">193</a> curi.putString(A_DNS_SERVER_IP_LABEL, FindServer.server());<a name="194" href="#194">194</a> } <strong>catch</strong> (IOException e) {<a name="195" href="#195">195</a> logger.log(Level.SEVERE, <span class="string">"Failed store of DNS Record for "</span> +<a name="196" href="#196">196</a> curi.toString(), e);<a name="197" href="#197">197</a> setUnresolvable(curi, targetHost);<a name="198" href="#198">198</a> }<a name="199" href="#199">199</a> }<a name="200" href="#200">200</a> <a name="201" href="#201">201</a> <strong>protected</strong> <strong>boolean</strong> isQuadAddress(<strong>final</strong> <a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi, <strong>final</strong> String dnsName,<a name="202" href="#202">202</a> <strong>final</strong> <a href="../../../../org/archive/crawler/datamodel/CrawlHost.html">CrawlHost</a> targetHost) {<a name="203" href="#203">203</a> <strong>boolean</strong> result = false;<a name="204" href="#204">204</a> Matcher matcher = InetAddressUtil.IPV4_QUADS.matcher(dnsName);<a name="205" href="#205">205</a> <em class="comment">// If it's an ip no need to do a lookup</em><a name="206" href="#206">206</a> <strong>if</strong> (matcher == <strong>null</strong> || !matcher.matches()) {<a name="207" href="#207">207</a> <strong>return</strong> result;<a name="208" href="#208">208</a> }<a name="209" href="#209">209</a> <a name="210" href="#210">210</a> result = <strong>true</strong>;<a name="211" href="#211">211</a> <em class="comment">// Ideally this branch would never be reached: no CrawlURI</em><a name="212" href="#212">212</a> <em class="comment">// would be created for numerical IPs</em><a name="213" href="#213">213</a> <strong>if</strong> (logger.isLoggable(Level.WARNING)) {<a name="214" href="#214">214</a> logger.warning(<span class="string">"Unnecessary DNS CrawlURI created: "</span> + curi);<a name="215" href="#215">215</a> }<a name="216" href="#216">216</a> <strong>try</strong> {<a name="217" href="#217">217</a> targetHost.setIP(InetAddress.getByAddress(dnsName, <strong>new</strong> byte[] {<a name="218" href="#218">218</a> (byte) (<strong>new</strong> Integer(matcher.group(1)).intValue()),<a name="219" href="#219">219</a> (byte) (<strong>new</strong> Integer(matcher.group(2)).intValue()),<a name="220" href="#220">220</a> (byte) (<strong>new</strong> Integer(matcher.group(3)).intValue()),<a name="221" href="#221">221</a> (byte) (<strong>new</strong> Integer(matcher.group(4)).intValue()) }),<a name="222" href="#222">222</a> CrawlHost.IP_NEVER_EXPIRES); <em class="comment">// Never expire numeric IPs</em><a name="223" href="#223">223</a> curi.setFetchStatus(S_DNS_SUCCESS);<a name="224" href="#224">224</a> } <strong>catch</strong> (UnknownHostException e) {<a name="225" href="#225">225</a> logger.log(Level.SEVERE, <span class="string">"Should never be "</span> + e.getMessage(), e);<a name="226" href="#226">226</a> setUnresolvable(curi, targetHost);<a name="227" href="#227">227</a> }<a name="228" href="#228">228</a> <strong>return</strong> result;<a name="229" href="#229">229</a> }<a name="230" href="#230">230</a> <a name="231" href="#231">231</a> <strong>protected</strong> <strong>void</strong> recordDNS(<strong>final</strong> <a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi, <strong>final</strong> Record[] rrecordSet)<a name="232" href="#232">232</a> throws IOException {<a name="233" href="#233">233</a> <strong>final</strong> byte[] dnsRecord =<a name="234" href="#234">234</a> getDNSRecord(curi.getLong(A_FETCH_BEGAN_TIME), rrecordSet);<a name="235" href="#235">235</a> HttpRecorder rec = HttpRecorder.getHttpRecorder();<a name="236" href="#236">236</a> <em class="comment">// Shall we get a digest on the content downloaded?</em><a name="237" href="#237">237</a> <strong>boolean</strong> sha1Content = ((Boolean)getUncheckedAttribute(curi,<a name="238" href="#238">238</a> FetchHTTP.ATTR_SHA1_CONTENT)).booleanValue();<a name="239" href="#239">239</a> <strong>if</strong>(sha1Content) {<a name="240" href="#240">240</a> rec.getRecordedInput().setSha1Digest();<a name="241" href="#241">241</a> } <strong>else</strong> {<a name="242" href="#242">242</a> rec.getRecordedInput().setDigest(<strong>null</strong>);<a name="243" href="#243">243</a> }<a name="244" href="#244">244</a> curi.setHttpRecorder(rec);<a name="245" href="#245">245</a> InputStream is = curi.getHttpRecorder().inputWrap(<a name="246" href="#246">246</a> <strong>new</strong> ByteArrayInputStream(dnsRecord));<a name="247" href="#247">247</a> <em class="comment">// Reading from the wrapped stream, behind the scenes, will write</em><a name="248" href="#248">248</a> <em class="comment">// files into scratch space</em><a name="249" href="#249">249</a> <strong>try</strong> {<a name="250" href="#250">250</a> <strong>while</strong> (is.read(<strong>this</strong>.reusableBuffer) != -1) {<a name="251" href="#251">251</a> <strong>continue</strong>;<a name="252" href="#252">252</a> }<a name="253" href="#253">253</a> } <strong>finally</strong> {<a name="254" href="#254">254</a> is.close();<a name="255" href="#255">255</a> rec.closeRecorders();<a name="256" href="#256">256</a> }<a name="257" href="#257">257</a> curi.setContentSize(dnsRecord.length);<a name="258" href="#258">258</a> curi.setContentDigest(FetchHTTP.SHA1,<a name="259" href="#259">259</a> rec.getRecordedInput().getDigestValue());<a name="260" href="#260">260</a> }<a name="261" href="#261">261</a> <a name="262" href="#262">262</a> <strong>protected</strong> byte [] getDNSRecord(<strong>final</strong> <strong>long</strong> fetchStart,<a name="263" href="#263">263</a> <strong>final</strong> Record[] rrecordSet)<a name="264" href="#264">264</a> throws IOException {<a name="265" href="#265">265</a> ByteArrayOutputStream baos = <strong>new</strong> ByteArrayOutputStream();<a name="266" href="#266">266</a> <em class="comment">// Start the record with a 14-digit date per RFC 2540</em><a name="267" href="#267">267</a> byte[] fetchDate = ArchiveUtils.get14DigitDate(fetchStart).getBytes();<a name="268" href="#268">268</a> baos.write(fetchDate);<a name="269" href="#269">269</a> <em class="comment">// Don't forget the newline</em><a name="270" href="#270">270</a> baos.write(<span class="string">"\n"</span>.getBytes());<a name="271" href="#271">271</a> <strong>int</strong> recordLength = fetchDate.length + 1;<a name="272" href="#272">272</a> <strong>if</strong> (rrecordSet != <strong>null</strong>) {<a name="273" href="#273">273</a> <strong>for</strong> (<strong>int</strong> i = 0; i < rrecordSet.length; i++) {<a name="274" href="#274">274</a> byte[] record = rrecordSet[i].toString().getBytes();<a name="275" href="#275">275</a> recordLength += record.length;<a name="276" href="#276">276</a> baos.write(record);<a name="277" href="#277">277</a> <em class="comment">// Add the newline between records back in</em><a name="278" href="#278">278</a> baos.write(<span class="string">"\n"</span>.getBytes());<a name="279" href="#279">279</a> recordLength += 1;<a name="280" href="#280">280</a> }<a name="281" href="#281">281</a> }<a name="282" href="#282">282</a> <strong>return</strong> baos.toByteArray();<a name="283" href="#283">283</a> }<a name="284" href="#284">284</a> <a name="285" href="#285">285</a> <strong>protected</strong> <strong>void</strong> setUnresolvable(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi, <a href="../../../../org/archive/crawler/datamodel/CrawlHost.html">CrawlHost</a> host) {<a name="286" href="#286">286</a> host.setIP(<strong>null</strong>, 0);<a name="287" href="#287">287</a> curi.setFetchStatus(S_DOMAIN_UNRESOLVABLE); <a name="288" href="#288">288</a> }<a name="289" href="#289">289</a> <a name="290" href="#290">290</a> <strong>protected</strong> ARecord getFirstARecord(Record[] rrecordSet) {<a name="291" href="#291">291</a> ARecord arecord = <strong>null</strong>;<a name="292" href="#292">292</a> <strong>if</strong> (rrecordSet == <strong>null</strong> || rrecordSet.length == 0) {<a name="293" href="#293">293</a> <strong>if</strong> (logger.isLoggable(Level.FINEST)) {<a name="294" href="#294">294</a> logger.finest(<span class="string">"rrecordSet is null or zero length: "</span> +<a name="295" href="#295">295</a> rrecordSet);<a name="296" href="#296">296</a> }<a name="297" href="#297">297</a> <strong>return</strong> arecord;<a name="298" href="#298">298</a> }<a name="299" href="#299">299</a> <strong>for</strong> (<strong>int</strong> i = 0; i < rrecordSet.length; i++) {<a name="300" href="#300">300</a> <strong>if</strong> (rrecordSet[i].getType() != Type.A) {<a name="301" href="#301">301</a> <strong>if</strong> (logger.isLoggable(Level.FINEST)) {<a name="302" href="#302">302</a> logger.finest(<span class="string">"Record "</span> + Integer.toString(i) +<a name="303" href="#303">303</a> <span class="string">" is not A type but "</span> + rrecordSet[i].getType());<a name="304" href="#304">304</a> }<a name="305" href="#305">305</a> <strong>continue</strong>;<a name="306" href="#306">306</a> }<a name="307" href="#307">307</a> arecord = (ARecord) rrecordSet[i];<a name="308" href="#308">308</a> <strong>break</strong>;<a name="309" href="#309">309</a> }<a name="310" href="#310">310</a> <strong>return</strong> arecord;<a name="311" href="#311">311</a> }<a name="312" href="#312">312</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 + -