📄 preconditionenforcer.html
字号:
<a name="248" href="#248">248</a> <strong>if</strong> (isIpExpired(curi) && !curi.getUURI().getScheme().equals(<span class="string">"dns"</span>)) {<a name="249" href="#249">249</a> logger.fine(<span class="string">"Deferring processing of CrawlURI "</span> + curi.toString()<a name="250" href="#250">250</a> + <span class="string">" for dns lookup."</span>);<a name="251" href="#251">251</a> String preq = <span class="string">"dns:"</span> + ch.getHostName();<a name="252" href="#252">252</a> <strong>try</strong> {<a name="253" href="#253">253</a> curi.markPrerequisite(preq,<a name="254" href="#254">254</a> getController().getPostprocessorChain());<a name="255" href="#255">255</a> } <strong>catch</strong> (URIException e) {<a name="256" href="#256">256</a> <strong>throw</strong> <strong>new</strong> RuntimeException(e); <em class="comment">// shouldn't ever happen</em><a name="257" href="#257">257</a> }<a name="258" href="#258">258</a> <strong>return</strong> <strong>true</strong>;<a name="259" href="#259">259</a> }<a name="260" href="#260">260</a> <a name="261" href="#261">261</a> <em class="comment">// DNS preconditions OK</em><a name="262" href="#262">262</a> <strong>return</strong> false;<a name="263" href="#263">263</a> }<a name="264" href="#264">264</a> <a name="265" href="#265">265</a> <em>/**<em>*</em></em><a name="266" href="#266">266</a> <em> * Get the maximum time a dns-record is valid.</em><a name="267" href="#267">267</a> <em> *</em><a name="268" href="#268">268</a> <em> * @param curi the uri this time is valid for.</em><a name="269" href="#269">269</a> <em> * @return the maximum time a dns-record is valid -- in seconds -- or</em><a name="270" href="#270">270</a> <em> * negative if record's ttl should be used.</em><a name="271" href="#271">271</a> <em> */</em><a name="272" href="#272">272</a> <strong>public</strong> <strong>long</strong> getIPValidityDuration(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) {<a name="273" href="#273">273</a> Integer d;<a name="274" href="#274">274</a> <strong>try</strong> {<a name="275" href="#275">275</a> d = (Integer)getAttribute(ATTR_IP_VALIDITY_DURATION, curi);<a name="276" href="#276">276</a> } <strong>catch</strong> (AttributeNotFoundException e) {<a name="277" href="#277">277</a> d = DEFAULT_IP_VALIDITY_DURATION;<a name="278" href="#278">278</a> }<a name="279" href="#279">279</a> <a name="280" href="#280">280</a> <strong>return</strong> d.longValue();<a name="281" href="#281">281</a> }<a name="282" href="#282">282</a> <a name="283" href="#283">283</a> <em>/**<em>* Return true if ip should be looked up.</em></em><a name="284" href="#284">284</a> <em> *</em><a name="285" href="#285">285</a> <em> * @param curi the URI to check.</em><a name="286" href="#286">286</a> <em> * @return true if ip should be looked up.</em><a name="287" href="#287">287</a> <em> */</em><a name="288" href="#288">288</a> <strong>public</strong> <strong>boolean</strong> isIpExpired(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) {<a name="289" href="#289">289</a> <a href="../../../../org/archive/crawler/datamodel/CrawlHost.html">CrawlHost</a> host = getController().getServerCache().getHostFor(curi);<a name="290" href="#290">290</a> <strong>if</strong> (!host.hasBeenLookedUp()) {<a name="291" href="#291">291</a> <em class="comment">// IP has not been looked up yet.</em><a name="292" href="#292">292</a> <strong>return</strong> <strong>true</strong>;<a name="293" href="#293">293</a> }<a name="294" href="#294">294</a> <a name="295" href="#295">295</a> <strong>if</strong> (host.getIpTTL() == CrawlHost.IP_NEVER_EXPIRES) {<a name="296" href="#296">296</a> <em class="comment">// IP never expires (numeric IP)</em><a name="297" href="#297">297</a> <strong>return</strong> false;<a name="298" href="#298">298</a> }<a name="299" href="#299">299</a> <a name="300" href="#300">300</a> <strong>long</strong> duration = getIPValidityDuration(curi);<a name="301" href="#301">301</a> <strong>if</strong> (duration == 0) {<a name="302" href="#302">302</a> <em class="comment">// Never expire ip if duration is null (set by user or more likely,</em><a name="303" href="#303">303</a> <em class="comment">// set to zero in case where we tried in FetchDNS but failed).</em><a name="304" href="#304">304</a> <strong>return</strong> false;<a name="305" href="#305">305</a> }<a name="306" href="#306">306</a> <a name="307" href="#307">307</a> <em class="comment">// catch old "default" -1 settings that are now problematic,</em><a name="308" href="#308">308</a> <em class="comment">// convert to new minimum</em><a name="309" href="#309">309</a> <strong>if</strong> (duration <= 0) {<a name="310" href="#310">310</a> duration = DEFAULT_IP_VALIDITY_DURATION.intValue();<a name="311" href="#311">311</a> }<a name="312" href="#312">312</a> <a name="313" href="#313">313</a> <strong>long</strong> ttl = host.getIpTTL();<a name="314" href="#314">314</a> <strong>if</strong> (ttl > duration) {<a name="315" href="#315">315</a> <em class="comment">// Use the larger of the operator-set minimum duration </em><a name="316" href="#316">316</a> <em class="comment">// or the DNS record TTL</em><a name="317" href="#317">317</a> duration = ttl;<a name="318" href="#318">318</a> }<a name="319" href="#319">319</a> <a name="320" href="#320">320</a> <em class="comment">// Duration and ttl are in seconds. Convert to millis.</em><a name="321" href="#321">321</a> <strong>if</strong> (duration > 0) {<a name="322" href="#322">322</a> duration *= 1000;<a name="323" href="#323">323</a> }<a name="324" href="#324">324</a> <a name="325" href="#325">325</a> <strong>return</strong> (duration + host.getIpFetched()) < System.currentTimeMillis();<a name="326" href="#326">326</a> }<a name="327" href="#327">327</a> <a name="328" href="#328">328</a> <em>/**<em>* Get the maximum time a robots.txt is valid.</em></em><a name="329" href="#329">329</a> <em> *</em><a name="330" href="#330">330</a> <em> * @param curi</em><a name="331" href="#331">331</a> <em> * @return the time a robots.txt is valid in milliseconds.</em><a name="332" href="#332">332</a> <em> */</em><a name="333" href="#333">333</a> <strong>public</strong> <strong>long</strong> getRobotsValidityDuration(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) {<a name="334" href="#334">334</a> Integer d;<a name="335" href="#335">335</a> <strong>try</strong> {<a name="336" href="#336">336</a> d = (Integer) getAttribute(ATTR_ROBOTS_VALIDITY_DURATION, curi);<a name="337" href="#337">337</a> } <strong>catch</strong> (AttributeNotFoundException e) {<a name="338" href="#338">338</a> <em class="comment">// This should never happen, but if it does, return default</em><a name="339" href="#339">339</a> logger.severe(e.getLocalizedMessage());<a name="340" href="#340">340</a> d = DEFAULT_ROBOTS_VALIDITY_DURATION;<a name="341" href="#341">341</a> }<a name="342" href="#342">342</a> <em class="comment">// convert from seconds to milliseconds</em><a name="343" href="#343">343</a> <strong>return</strong> d.longValue() * 1000;<a name="344" href="#344">344</a> }<a name="345" href="#345">345</a> <a name="346" href="#346">346</a> <em>/**<em>*</em></em><a name="347" href="#347">347</a> <em> * Is the robots policy expired.</em><a name="348" href="#348">348</a> <em> *</em><a name="349" href="#349">349</a> <em> * This method will also return true if we haven't tried to get the</em><a name="350" href="#350">350</a> <em> * robots.txt for this server.</em><a name="351" href="#351">351</a> <em> *</em><a name="352" href="#352">352</a> <em> * @param curi</em><a name="353" href="#353">353</a> <em> * @return true if the robots policy is expired.</em><a name="354" href="#354">354</a> <em> */</em><a name="355" href="#355">355</a> <strong>public</strong> <strong>boolean</strong> isRobotsExpired(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) {<a name="356" href="#356">356</a> <a href="../../../../org/archive/crawler/datamodel/CrawlServer.html">CrawlServer</a> server =<a name="357" href="#357">357</a> getController().getServerCache().getServerFor(curi);<a name="358" href="#358">358</a> <strong>long</strong> robotsFetched = server.getRobotsFetchedTime();<a name="359" href="#359">359</a> <strong>if</strong> (robotsFetched == CrawlServer.ROBOTS_NOT_FETCHED) {<a name="360" href="#360">360</a> <em class="comment">// Have not attempted to fetch robots</em><a name="361" href="#361">361</a> <strong>return</strong> <strong>true</strong>;<a name="362" href="#362">362</a> }<a name="363" href="#363">363</a> <strong>long</strong> duration = getRobotsValidityDuration(curi);<a name="364" href="#364">364</a> <strong>if</strong> (duration == 0) {<a name="365" href="#365">365</a> <em class="comment">// When zero, robots should be valid forever</em><a name="366" href="#366">366</a> <strong>return</strong> false;<a name="367" href="#367">367</a> }<a name="368" href="#368">368</a> <strong>if</strong> (robotsFetched + duration < System.currentTimeMillis()) {<a name="369" href="#369">369</a> <em class="comment">// Robots is still valid</em><a name="370" href="#370">370</a> <strong>return</strong> <strong>true</strong>;<a name="371" href="#371">371</a> }<a name="372" href="#372">372</a> <strong>return</strong> false;<a name="373" href="#373">373</a> }<a name="374" href="#374">374</a> <a name="375" href="#375">375</a> <em>/**<em>*</em></em>
⌨️ 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 + -