<a name="250" href="#250">250</a>     throws IOException {<a name="251" href="#251">251</a>         read.length(0);<a name="252" href="#252">252</a>         <strong>int</strong> c = is.read();<a name="253" href="#253">253</a>         <strong>while</strong>((c!=-1)&amp;&amp;c!='\n'&amp;&amp;c!='\r') {<a name="254" href="#254">254</a>             read.append((<strong>char</strong>)c);<a name="255" href="#255">255</a>             c = is.read();<a name="256" href="#256">256</a>         }<a name="257" href="#257">257</a>         <strong>if</strong>(c==-1 &amp;&amp; read.length()==0) {<a name="258" href="#258">258</a>             <em class="comment">// EOF and none read; return false</em><a name="259" href="#259">259</a>             <strong>return</strong> false;<a name="260" href="#260">260</a>         }<a name="261" href="#261">261</a>         <strong>if</strong>(c=='\n') {<a name="262" href="#262">262</a>             <em class="comment">// consume LF following CR, if present</em><a name="263" href="#263">263</a>             is.mark(1);<a name="264" href="#264">264</a>             <strong>if</strong>(is.read()!='\r') {<a name="265" href="#265">265</a>                 is.reset();<a name="266" href="#266">266</a>             }<a name="267" href="#267">267</a>         }<a name="268" href="#268">268</a>         <em class="comment">// a line (possibly blank) was read</em><a name="269" href="#269">269</a>         <strong>return</strong> <strong>true</strong>;<a name="270" href="#270">270</a>     }<a name="271" href="#271">271</a> <a name="272" href="#272">272</a>     <em>/**<em>*</em></em><a name="273" href="#273">273</a> <em>     * Import all ADDs from given recovery log into the frontier's queues</em><a name="274" href="#274">274</a> <em>     * (excepting those the frontier drops as already having been included)</em><a name="275" href="#275">275</a> <em>     * </em><a name="276" href="#276">276</a> <em>     * @param source recovery log file to use</em><a name="277" href="#277">277</a> <em>     * @param frontier frontier to update</em><a name="278" href="#278">278</a> <em>     * @param lines total lines noted in recovery log earlier</em><a name="279" href="#279">279</a> <em>     * @param enough latch signalling 'enough' URIs queued to begin crawling</em><a name="280" href="#280">280</a> <em>     */</em><a name="281" href="#281">281</a>     <strong>private</strong> <strong>static</strong> <strong>void</strong> importQueuesFromLog(File source, <a href="../../../../org/archive/crawler/framework/Frontier.html">Frontier</a> frontier,<a name="282" href="#282">282</a>             <strong>int</strong> lines, CountDownLatch enough) {<a name="283" href="#283">283</a>         BufferedInputStream is;<a name="284" href="#284">284</a>         <em class="comment">// create MutableString of good starting size (will grow if necessary)</em><a name="285" href="#285">285</a>         MutableString read = <strong>new</strong> MutableString(UURI.MAX_URL_LENGTH);<a name="286" href="#286">286</a>         <strong>long</strong> queuedAtStart = frontier.queuedUriCount();<a name="287" href="#287">287</a>         <strong>long</strong> queuedDuringRecovery = 0;<a name="288" href="#288">288</a>         <strong>int</strong> qLines = 0;<a name="289" href="#289">289</a>         <a name="290" href="#290">290</a>         <strong>try</strong> {<a name="291" href="#291">291</a>             <em class="comment">// Scan log for all 'F+' lines: if not alreadyIncluded, schedule for</em><a name="292" href="#292">292</a>             <em class="comment">// visitation</em><a name="293" href="#293">293</a>             is = getBufferedInput(source);<a name="294" href="#294">294</a>             <strong>try</strong> {<a name="295" href="#295">295</a>                 <strong>while</strong> (readLine(is,read)) {<a name="296" href="#296">296</a>                     qLines++;<a name="297" href="#297">297</a>                     <strong>if</strong> (read.startsWith(F_ADD)) {<a name="298" href="#298">298</a>                         <a href="../../../../org/archive/net/UURI.html">UURI</a> u;<a name="299" href="#299">299</a>                         CharSequence args[] = splitOnSpaceRuns(read);<a name="300" href="#300">300</a>                         <strong>try</strong> {<a name="301" href="#301">301</a>                             u = UURIFactory.getInstance(args[1].toString());<a name="302" href="#302">302</a>                             String pathFromSeed = (args.length > 2)?<a name="303" href="#303">303</a>                                 args[2].toString() : <span class="string">""</span>;<a name="304" href="#304">304</a>                             <a href="../../../../org/archive/net/UURI.html">UURI</a> via = (args.length > 3)?<a name="305" href="#305">305</a>                                 UURIFactory.getInstance(args[3].toString()):<a name="306" href="#306">306</a>                                 <strong>null</strong>;<a name="307" href="#307">307</a>                             String viaContext = (args.length > 4)?<a name="308" href="#308">308</a>                                     args[4].toString(): <span class="string">""</span>;<a name="309" href="#309">309</a>                             <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a> caUri = <strong>new</strong> <a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a>(u, <a name="310" href="#310">310</a>                                     pathFromSeed, via, viaContext);<a name="311" href="#311">311</a>                             frontier.schedule(caUri);<a name="312" href="#312">312</a>                             <a name="313" href="#313">313</a>                             queuedDuringRecovery =<a name="314" href="#314">314</a>                                 frontier.queuedUriCount() - queuedAtStart;<a name="315" href="#315">315</a>                             <strong>if</strong>(((queuedDuringRecovery + 1) %<a name="316" href="#316">316</a>                                     ENOUGH_TO_START_CRAWLING) == 0) {<a name="317" href="#317">317</a>                                 enough.countDown();<a name="318" href="#318">318</a>                             }<a name="319" href="#319">319</a>                         } <strong>catch</strong> (URIException e) {<a name="320" href="#320">320</a>                             e.printStackTrace();<a name="321" href="#321">321</a>                         }<a name="322" href="#322">322</a>                     }<a name="323" href="#323">323</a>                     <strong>if</strong>((qLines%PROGRESS_INTERVAL)==0) {<a name="324" href="#324">324</a>                         <em class="comment">// every 1 million lines, print progress</em><a name="325" href="#325">325</a>                         LOGGER.info(<a name="326" href="#326">326</a>                                 <span class="string">"through line "</span> <a name="327" href="#327">327</a>                                 + qLines + <span class="string">"/"</span> + lines <a name="328" href="#328">328</a>                                 + <span class="string">" queued count = "</span> +<a name="329" href="#329">329</a>                                 frontier.queuedUriCount());<a name="330" href="#330">330</a>                     }<a name="331" href="#331">331</a>                 }<a name="332" href="#332">332</a>             } <strong>catch</strong> (EOFException e) {<a name="333" href="#333">333</a>                 <em class="comment">// no problem: untidy end of recovery journal</em><a name="334" href="#334">334</a>             } <strong>finally</strong> {<a name="335" href="#335">335</a>             	    is.close(); <a name="336" href="#336">336</a>             }<a name="337" href="#337">337</a>         } <strong>catch</strong> (IOException e) {<a name="338" href="#338">338</a>             <em class="comment">// TODO Auto-generated catch block</em><a name="339" href="#339">339</a>             e.printStackTrace();<a name="340" href="#340">340</a>         }<a name="341" href="#341">341</a>         LOGGER.info(<span class="string">"finished recovering frontier from "</span>+source+<span class="string">" "</span><a name="342" href="#342">342</a>                 +qLines+<span class="string">" lines processed"</span>);<a name="343" href="#343">343</a>         enough.countDown();<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>     * Return an array of the subsequences of the passed-in sequence,</em><a name="348" href="#348">348</a> <em>     * split on space runs. </em><a name="349" href="#349">349</a> <em>     * </em><a name="350" href="#350">350</a> <em>     * @param read</em><a name="351" href="#351">351</a> <em>     * @return CharSequence.</em><a name="352" href="#352">352</a> <em>     */</em><a name="353" href="#353">353</a>     <strong>private</strong> <strong>static</strong> CharSequence[] splitOnSpaceRuns(CharSequence read) {<a name="354" href="#354">354</a>         <strong>int</strong> lastStart = 0;<a name="355" href="#355">355</a>         ArrayList&lt;CharSequence> segs = <strong>new</strong> ArrayList&lt;CharSequence>(5);<a name="356" href="#356">356</a>         <strong>int</strong> i;<a name="357" href="#357">357</a>         <strong>for</strong>(i=0;i&lt;read.length();i++) {<a name="358" href="#358">358</a>             <strong>if</strong> (read.charAt(i)==' ') {<a name="359" href="#359">359</a>                 segs.add(read.subSequence(lastStart,i));<a name="360" href="#360">360</a>                 i++;<a name="361" href="#361">361</a>                 <strong>while</strong>(i &lt; read.length() &amp;&amp; read.charAt(i)==' ') {<a name="362" href="#362">362</a>                     <em class="comment">// skip any space runs</em><a name="363" href="#363">363</a>                     i++;<a name="364" href="#364">364</a>                 }<a name="365" href="#365">365</a>                 lastStart = i;<a name="366" href="#366">366</a>             }<a name="367" href="#367">367</a>         }<a name="368" href="#368">368</a>         <strong>if</strong>(lastStart&lt;read.length()) {<a name="369" href="#369">369</a>             segs.add(read.subSequence(lastStart,i));<a name="370" href="#370">370</a>         }<a name="371" href="#371">371</a>         <strong>return</strong> (CharSequence[]) segs.toArray(<strong>new</strong> CharSequence[segs.size()]);        <a name="372" href="#372">372</a>     }<a name="373" href="#373">373</a> }</pre><hr/><div id="footer">This page was automatically generated by <a href="http://maven.apache.org/">Maven</a></div></body></html>