writerpoolprocessor.html

来自「网络爬虫开源代码」· HTML 代码 · 共 732 行 · 第 1/5 页

HTML
732
字号
<a name="286" href="#286">286</a>             <strong>return</strong>;<a name="287" href="#287">287</a>         }<a name="288" href="#288">288</a>         <strong>if</strong> (max &lt;= <strong>this</strong>.totalBytesWritten) {<a name="289" href="#289">289</a>             getController().requestCrawlStop(<span class="string">"Finished - Maximum bytes ("</span> +<a name="290" href="#290">290</a>                 Long.toString(max) + <span class="string">") written"</span>);<a name="291" href="#291">291</a>         }<a name="292" href="#292">292</a>     }<a name="293" href="#293">293</a>     <a name="294" href="#294">294</a>     <em>/**<em>*</em></em><a name="295" href="#295">295</a> <em>     * Whether the given CrawlURI should be written to archive files. </em><a name="296" href="#296">296</a> <em>     * Annotates CrawlURI with a reason for any negative answer. </em><a name="297" href="#297">297</a> <em>     * </em><a name="298" href="#298">298</a> <em>     * @param curi CrawlURI</em><a name="299" href="#299">299</a> <em>     * @return true if URI should be written; false otherwise</em><a name="300" href="#300">300</a> <em>     */</em><a name="301" href="#301">301</a>     <strong>protected</strong> <strong>boolean</strong> shouldWrite(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) {<a name="302" href="#302">302</a>         <em class="comment">// check for duplicate content write suppression</em><a name="303" href="#303">303</a>         <strong>if</strong>(((Boolean)getUncheckedAttribute(curi, ATTR_SKIP_IDENTICAL_DIGESTS)) <a name="304" href="#304">304</a>             &amp;&amp; IdenticalDigestDecideRule.hasIdenticalDigest(curi)) {<a name="305" href="#305">305</a>             curi.addAnnotation(ANNOTATION_UNWRITTEN + <span class="string">":identicalDigest"</span>);<a name="306" href="#306">306</a>             <strong>return</strong> false; <a name="307" href="#307">307</a>         }<a name="308" href="#308">308</a>         String scheme = curi.getUURI().getScheme().toLowerCase();<a name="309" href="#309">309</a>         <em class="comment">// TODO: possibly move this sort of isSuccess() test into CrawlURI</em><a name="310" href="#310">310</a>         <strong>boolean</strong> retVal; <a name="311" href="#311">311</a>         <strong>if</strong> (scheme.equals(<span class="string">"dns"</span>)) {<a name="312" href="#312">312</a>             retVal = curi.getFetchStatus() == S_DNS_SUCCESS;<a name="313" href="#313">313</a>         } <strong>else</strong> <strong>if</strong> (scheme.equals(<span class="string">"http"</span>) || scheme.equals(<span class="string">"https"</span>)) {<a name="314" href="#314">314</a>             retVal = curi.getFetchStatus() > 0 &amp;&amp; curi.isHttpTransaction();<a name="315" href="#315">315</a>         } <strong>else</strong> <strong>if</strong> (scheme.equals(<span class="string">"ftp"</span>)) {<a name="316" href="#316">316</a>             retVal = curi.getFetchStatus() == 200;<a name="317" href="#317">317</a>         } <strong>else</strong> {<a name="318" href="#318">318</a>             <em class="comment">// unsupported scheme</em><a name="319" href="#319">319</a>             curi.addAnnotation(ANNOTATION_UNWRITTEN + <span class="string">":scheme"</span>);<a name="320" href="#320">320</a>             <strong>return</strong> false; <a name="321" href="#321">321</a>         }<a name="322" href="#322">322</a>         <strong>if</strong> (retVal == false) {<a name="323" href="#323">323</a>             <em class="comment">// status not deserving writing</em><a name="324" href="#324">324</a>             curi.addAnnotation(ANNOTATION_UNWRITTEN + <span class="string">":status"</span>);<a name="325" href="#325">325</a>             <strong>return</strong> false; <a name="326" href="#326">326</a>         }<a name="327" href="#327">327</a>         <strong>return</strong> <strong>true</strong>; <a name="328" href="#328">328</a>     }<a name="329" href="#329">329</a>     <a name="330" href="#330">330</a>     <em>/**<em>*</em></em><a name="331" href="#331">331</a> <em>     * Return IP address of given URI suitable for recording (as in a</em><a name="332" href="#332">332</a> <em>     * classic ARC 5-field header line).</em><a name="333" href="#333">333</a> <em>     * </em><a name="334" href="#334">334</a> <em>     * @param curi CrawlURI</em><a name="335" href="#335">335</a> <em>     * @return String of IP address</em><a name="336" href="#336">336</a> <em>     */</em><a name="337" href="#337">337</a>     <strong>protected</strong> String getHostAddress(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) {<a name="338" href="#338">338</a>         <em class="comment">// special handling for DNS URIs: want address of DNS server</em><a name="339" href="#339">339</a>         <strong>if</strong>(curi.getUURI().getScheme().toLowerCase().equals(<span class="string">"dns"</span>)) {<a name="340" href="#340">340</a>             <strong>return</strong> curi.getString(A_DNS_SERVER_IP_LABEL);<a name="341" href="#341">341</a>         }<a name="342" href="#342">342</a>         <em class="comment">// otherwise, host referenced in URI</em><a name="343" href="#343">343</a>         <a href="../../../../org/archive/crawler/datamodel/CrawlHost.html">CrawlHost</a> h = getController().getServerCache().getHostFor(curi);<a name="344" href="#344">344</a>         <strong>if</strong> (h == <strong>null</strong>) {<a name="345" href="#345">345</a>             <strong>throw</strong> <strong>new</strong> NullPointerException(<span class="string">"Crawlhost is null for "</span> +<a name="346" href="#346">346</a>                 curi + <span class="string">" "</span> + curi.getVia());<a name="347" href="#347">347</a>         }<a name="348" href="#348">348</a>         InetAddress a = h.getIP();<a name="349" href="#349">349</a>         <strong>if</strong> (a == <strong>null</strong>) {<a name="350" href="#350">350</a>             <strong>throw</strong> <strong>new</strong> NullPointerException(<span class="string">"Address is null for "</span> +<a name="351" href="#351">351</a>                 curi + <span class="string">" "</span> + curi.getVia() + <span class="string">". Address "</span> +<a name="352" href="#352">352</a>                 ((h.getIpFetched() == CrawlHost.IP_NEVER_LOOKED_UP)?<a name="353" href="#353">353</a>                      <span class="string">"was never looked up."</span>:<a name="354" href="#354">354</a>                      (System.currentTimeMillis() - h.getIpFetched()) +<a name="355" href="#355">355</a>                          <span class="string">" ms ago."</span>));<a name="356" href="#356">356</a>         }<a name="357" href="#357">357</a>         <strong>return</strong> h.getIP().getHostAddress();<a name="358" href="#358">358</a>     }<a name="359" href="#359">359</a>     <a name="360" href="#360">360</a>     <em>/**<em>*</em></em><a name="361" href="#361">361</a> <em>     * Version of getAttributes that catches and logs exceptions</em><a name="362" href="#362">362</a> <em>     * and returns null if failure to fetch the attribute.</em><a name="363" href="#363">363</a> <em>     * @param name Attribute name.</em><a name="364" href="#364">364</a> <em>     * @return Attribute or null.</em><a name="365" href="#365">365</a> <em>     */</em><a name="366" href="#366">366</a>     <strong>public</strong> Object getAttributeUnchecked(String name) {<a name="367" href="#367">367</a>         Object result = <strong>null</strong>;<a name="368" href="#368">368</a>         <strong>try</strong> {<a name="369" href="#369">369</a>             result = <strong>super</strong>.getAttribute(name);<a name="370" href="#370">370</a>         } <strong>catch</strong> (AttributeNotFoundException e) {<a name="371" href="#371">371</a>             logger.warning(e.getLocalizedMessage());<a name="372" href="#372">372</a>         } <strong>catch</strong> (MBeanException e) {<a name="373" href="#373">373</a>             logger.warning(e.getLocalizedMessage());<a name="374" href="#374">374</a>         } <strong>catch</strong> (ReflectionException e) {<a name="375" href="#375">375</a>             logger.warning(e.getLocalizedMessage());<a name="376" href="#376">376</a>         }<a name="377" href="#377">377</a>         <strong>return</strong> result;<a name="378" href="#378">378</a>     }<a name="379" href="#379">379</a> <a name="380" href="#380">380</a>    <em>/**<em>*</em></em><a name="381" href="#381">381</a> <em>    * Max size we want files to be (bytes).</em><a name="382" href="#382">382</a> <em>    *</em><a name="383" href="#383">383</a> <em>    * Default is ARCConstants.DEFAULT_MAX_ARC_FILE_SIZE.  Note that ARC</em><a name="384" href="#384">384</a> <em>    * files will usually be bigger than maxSize; they'll be maxSize + length</em><a name="385" href="#385">385</a> <em>    * to next boundary.</em><a name="386" href="#386">386</a> <em>    * @return ARC maximum size.</em><a name="387" href="#387">387</a> <em>    */</em><a name="388" href="#388">388</a>     <strong>public</strong> <strong>long</strong> getMaxSize() {<a name="389" href="#389">389</a>         Object obj = getAttributeUnchecked(ATTR_MAX_SIZE_BYTES);<a name="390" href="#390">390</a>         <strong>return</strong> (obj == <strong>null</strong>)? DEFAULT_MAX_FILE_SIZE: ((Long)obj).longValue();<a name="391" href="#391">391</a>     }<a name="392" href="#392">392</a> <a name="393" href="#393">393</a>     <strong>public</strong> String getPrefix() {<a name="394" href="#394">394</a>         Object obj = getAttributeUnchecked(ATTR_PREFIX);<a name="395" href="#395">395</a>         <strong>return</strong> (obj == <strong>null</strong>)? WriterPoolMember.DEFAULT_PREFIX: (String)obj;<a name="396" href="#396">396</a>     }<a name="397" href="#397">397</a> <a name="398" href="#398">398</a>     <strong>public</strong> List&lt;File> getOutputDirs() {<a name="399" href="#399">399</a>         Object obj = getAttributeUnchecked(ATTR_PATH);<a name="400" href="#400">400</a>         List list = (obj == <strong>null</strong>)? Arrays.asList(DEFAULT_PATH): (StringList)obj;<a name="401" href="#401">401</a>         ArrayList&lt;File> results = <strong>new</strong> ArrayList&lt;File>();<a name="402" href="#402">402</a>         <strong>for</strong> (Iterator i = list.iterator(); i.hasNext();) {<a name="403" href="#403">403</a>             String path = (String)i.next();<a name="404" href="#404">404</a>             File f = <strong>new</strong> File(path);<a name="405" href="#405">405</a>             <strong>if</strong> (!f.isAbsolute()) {<a name="406" href="#406">406</a>                 f = <strong>new</strong> File(getController().getDisk(), path);<a name="407" href="#407">407</a>             }<a name="408" href="#408">408</a>             <strong>if</strong> (!f.exists()) {<a name="409" href="#409">409</a>                 <strong>try</strong> {<a name="410" href="#410">410</a>                     f.mkdirs();<a name="411" href="#411">411</a>                 } <strong>catch</strong> (Exception e) {<a name="412" href="#412">412</a>                     e.printStackTrace();<a name="413" href="#413">413</a>                     <strong>continue</strong>;<a name="414" href="#414">414</a>                 }<a name="415" href="#415">415</a>             }<a name="416" href="#416">416</a>             results.add(f);<a name="417" href="#417">417</a>         }<a name="418" href="#418">418</a>         <strong>return</strong> results;<a name="419" href="#419">419</a>     }<a name="420" href="#420">420</a>     <a name="421" href="#421">421</a>     <strong>public</strong> <strong>boolean</strong> isCompressed() {<a name="422" href="#422">422</a>         Object obj = getAttributeUnchecked(ATTR_COMPRESS);<a name="423" href="#423">423</a>         <strong>return</strong> (obj == <strong>null</strong>)? DEFAULT_COMPRESS:<a name="424" href="#424">424</a>             ((Boolean)obj).booleanValue();<a name="425" href="#425">425</a>     }<a name="426" href="#426">426</a> <a name="427" href="#427">427</a>     <em>/**<em>*</em></em><a name="428" href="#428">428</a> <em>     * @return Returns the poolMaximumActive.</em><a name="429" href="#429">429</a> <em>     */</em><a name="430" href="#430">430</a>     <strong>public</strong> <strong>int</strong> getPoolMaximumActive() {<a name="431" href="#431">431</a>         Object obj = getAttributeUnchecked(ATTR_POOL_MAX_ACTIVE);<a name="432" href="#432">432</a>         <strong>return</strong> (obj == <strong>null</strong>)? WriterPool.DEFAULT_MAX_ACTIVE:

⌨️ 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?