statisticssummary.html

用JAVA编写的,在做实验的时候留下来的,本来想删的,但是传上来,大家分享吧

<a name="312" href="#312">312</a>     <strong>public</strong> String getProcessedDocsPerSec() {
<a name="313" href="#313">313</a>     	<strong>return</strong> processedDocsPerSec;
<a name="314" href="#314">314</a>     }
<a name="315" href="#315">315</a>     
<a name="316" href="#316">316</a>     <strong>public</strong> String getBandwidthKbytesPerSec() {
<a name="317" href="#317">317</a>     	<strong>return</strong> bandwidthKbytesPerSec;
<a name="318" href="#318">318</a>     }
<a name="319" href="#319">319</a>     
<a name="320" href="#320">320</a>     <strong>public</strong> String getTotalDataWritten() {
<a name="321" href="#321">321</a>     	<strong>return</strong> totalDataWritten;
<a name="322" href="#322">322</a>     }
<a name="323" href="#323">323</a> 
<a name="324" href="#324">324</a>     <em>/**<em>*</em></em>
<a name="325" href="#325">325</a> <em>     * Sort the entries of the given HashMap in descending order by their</em>
<a name="326" href="#326">326</a> <em>     * values, which must be longs wrapped with &lt;code>LongWrapper&lt;/code>.</em>
<a name="327" href="#327">327</a> <em>     * &lt;p></em>
<a name="328" href="#328">328</a> <em>     * Elements are sorted by value from largest to smallest. Equal values are</em>
<a name="329" href="#329">329</a> <em>     * sorted in an arbitrary, but consistent manner by their keys. Only items</em>
<a name="330" href="#330">330</a> <em>     * with identical value and key are considered equal.</em>
<a name="331" href="#331">331</a> <em>     *</em>
<a name="332" href="#332">332</a> <em>     * If the passed-in map requires access to be synchronized, the caller</em>
<a name="333" href="#333">333</a> <em>     * should ensure this synchronization. </em>
<a name="334" href="#334">334</a> <em>     * </em>
<a name="335" href="#335">335</a> <em>     * @param mapOfLongWrapperValues</em>
<a name="336" href="#336">336</a> <em>     *            Assumes values are wrapped with LongWrapper.</em>
<a name="337" href="#337">337</a> <em>     * @return a sorted set containing the same elements as the map.</em>
<a name="338" href="#338">338</a> <em>     */</em>
<a name="339" href="#339">339</a>     <strong>public</strong> TreeMap getReverseSortedCopy(<strong>final</strong> Map mapOfLongWrapperValues) {
<a name="340" href="#340">340</a>         TreeMap sortedMap = <strong>new</strong> TreeMap(<strong>new</strong> Comparator() {
<a name="341" href="#341">341</a>             <strong>public</strong> <strong>int</strong> compare(Object e1, Object e2) {
<a name="342" href="#342">342</a>                 <strong>long</strong> firstVal = ((LongWrapper)mapOfLongWrapperValues.get(e1)).
<a name="343" href="#343">343</a>                     longValue;
<a name="344" href="#344">344</a>                 <strong>long</strong> secondVal = ((LongWrapper)mapOfLongWrapperValues.get(e2)).
<a name="345" href="#345">345</a>                     longValue;
<a name="346" href="#346">346</a>                 <strong>if</strong> (firstVal &lt; secondVal) {
<a name="347" href="#347">347</a>                     <strong>return</strong> 1;
<a name="348" href="#348">348</a>                 }
<a name="349" href="#349">349</a>                 <strong>if</strong> (secondVal &lt; firstVal) {
<a name="350" href="#350">350</a>                     <strong>return</strong> -1;
<a name="351" href="#351">351</a>                 }
<a name="352" href="#352">352</a>                 <em class="comment">// If the values are the same, sort by keys.</em>
<a name="353" href="#353">353</a>                 <strong>return</strong> ((String)e1).compareTo((String)e2);
<a name="354" href="#354">354</a>             }
<a name="355" href="#355">355</a>         });
<a name="356" href="#356">356</a>         <strong>try</strong> {
<a name="357" href="#357">357</a>             sortedMap.putAll(mapOfLongWrapperValues);
<a name="358" href="#358">358</a>         } <strong>catch</strong> (UnsupportedOperationException e) {
<a name="359" href="#359">359</a>             Iterator i = mapOfLongWrapperValues.keySet().iterator();
<a name="360" href="#360">360</a>             <strong>for</strong> (;i.hasNext();) {
<a name="361" href="#361">361</a>                 <em class="comment">// Ok. Try doing it the slow way then.</em>
<a name="362" href="#362">362</a>                 Object key = i.next();
<a name="363" href="#363">363</a>                 sortedMap.put(key, mapOfLongWrapperValues.get(key));
<a name="364" href="#364">364</a>             }
<a name="365" href="#365">365</a>         }
<a name="366" href="#366">366</a>         <strong>return</strong> sortedMap;
<a name="367" href="#367">367</a>     }
<a name="368" href="#368">368</a>      
<a name="369" href="#369">369</a>     <em>/**<em>*</em></em>
<a name="370" href="#370">370</a> <em>     * Get the number of hosts with a particular TLD.</em>
<a name="371" href="#371">371</a> <em>     * @param tld</em>
<a name="372" href="#372">372</a> <em>     * 				top-level domain name</em>
<a name="373" href="#373">373</a> <em>     * @return		Total crawled hosts</em>
<a name="374" href="#374">374</a> <em>     */</em>
<a name="375" href="#375">375</a>     <strong>public</strong> <strong>long</strong> getHostsPerTld(String tld) {
<a name="376" href="#376">376</a>     	LongWrapper lw = (LongWrapper)tldHostDistribution.get(tld);
<a name="377" href="#377">377</a>     	<strong>return</strong> (lw == <strong>null</strong> ? 0 : lw.longValue);
<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>     * Read status code distribution from responsecode-report.txt.</em>
<a name="382" href="#382">382</a> <em>     * DNS and HTTP status codes are separated when read.</em>
<a name="383" href="#383">383</a> <em>     * @return True if we found some stats.</em>
<a name="384" href="#384">384</a> <em>     */</em>
<a name="385" href="#385">385</a>     <strong>private</strong> <strong>boolean</strong> calculateStatusCodeDistribution() {
<a name="386" href="#386">386</a>     	<em class="comment">// Read from responsecode-report.txt</em>
<a name="387" href="#387">387</a>     	File f = <strong>new</strong> File(cjob.getDirectory(), <span class="string">"responsecode-report.txt"</span>);
<a name="388" href="#388">388</a>     	<strong>if</strong> (!f.exists()) {
<a name="389" href="#389">389</a>     		<strong>return</strong> false;
<a name="390" href="#390">390</a>     	}
<a name="391" href="#391">391</a>     	BufferedReader br = <strong>null</strong>;
<a name="392" href="#392">392</a>     	<strong>try</strong> {
<a name="393" href="#393">393</a> 	    	FileReader reader = <strong>new</strong> FileReader(f);
<a name="394" href="#394">394</a> 	    	br = <strong>new</strong> BufferedReader(reader);
<a name="395" href="#395">395</a> 	    	String line = br.readLine();  <em class="comment">// Ignore heading</em>
<a name="396" href="#396">396</a> 	    	line = br.readLine();
<a name="397" href="#397">397</a> 	    	<strong>while</strong> (line != <strong>null</strong>) {  	  
<a name="398" href="#398">398</a> 	    	  <em class="comment">// Get status code and # urls which are seperated by a space</em>
<a name="399" href="#399">399</a> 	    	  
<a name="400" href="#400">400</a> 	    	  String[] items = line.split(<span class="string">" "</span>);
<a name="401" href="#401">401</a> 	    	  <strong>if</strong> (items.length &lt; 2) {
<a name="402" href="#402">402</a> 	    		  logger.log(Level.WARNING,
<a name="403" href="#403">403</a>                           <span class="string">"Unexpected formatting on line ["</span> + line + <span class="string">"]"</span>);
<a name="404" href="#404">404</a> 	    	  }
<a name="405" href="#405">405</a> 	    	  <strong>else</strong> {
<a name="406" href="#406">406</a> 	    		  <em class="comment">// See if DNS or HTTP status code</em>
<a name="407" href="#407">407</a> 	    		  <strong>if</strong> (items[0].length() &lt; 3) {
<a name="408" href="#408">408</a> 	    			  <em class="comment">// DNS status code</em>
<a name="409" href="#409">409</a> 	    			  <strong>long</strong> total = Long.parseLong(items[1]);
<a name="410" href="#410">410</a> 	    			  dnsStatusCodeDistribution.put(items[0], 
<a name="411" href="#411">411</a> 	    					  <strong>new</strong> LongWrapper(total));
<a name="412" href="#412">412</a> 	    			  totalDnsStatusCodeDocuments += total;
<a name="413" href="#413">413</a> 	    		  }
<a name="414" href="#414">414</a> 	    		  <strong>else</strong> {
<a name="415" href="#415">415</a> 	    			  <em class="comment">// HTTP status code</em>
<a name="416" href="#416">416</a> 	    			  <strong>long</strong> total = Long.parseLong(items[1]);
<a name="417" href="#417">417</a> 	    			  statusCodeDistribution.put(items[0], 
<a name="418" href="#418">418</a> 	    					  <strong>new</strong> LongWrapper(total));
<a name="419" href="#419">419</a> 	    			  totalStatusCodeDocuments += total;
<a name="420" href="#420">420</a> 	    		  }
<a name="421" href="#421">421</a> 	    	  }
<a name="422" href="#422">422</a> 	    	  line = br.readLine();
<a name="423" href="#423">423</a> 	    	}
<a name="424" href="#424">424</a>     	} <strong>catch</strong> (IOException e) {
<a name="425" href="#425">425</a>     		logger.log(Level.SEVERE, <span class="string">"Unable to read "</span> + f.getAbsolutePath(),
<a name="426" href="#426">426</a>     			e);
<a name="427" href="#427">427</a>     	} <strong>finally</strong> {
<a name="428" href="#428">428</a>     		<strong>if</strong> (br != <strong>null</strong>) {
<a name="429" href="#429">429</a>     			<strong>try</strong> {
<a name="430" href="#430">430</a> 					br.close();
<a name="431" href="#431">431</a> 				} <strong>catch</strong> (IOException e) {
<a name="432" href="#432">432</a> 					logger.log(Level.SEVERE,
<a name="433" href="#433">433</a> 						<span class="string">"Closing "</span> + f.getAbsolutePath(), e);
<a name="434" href="#434">434</a> 				}
<a name="435" href="#435">435</a>     		}
<a name="436" href="#436">436</a>     	}
<a name="437" href="#437">437</a>     	<strong>return</strong> <strong>true</strong>;
<a name="438" href="#438">438</a>     }
<a name="439" href="#439">439</a>     
<a name="440" href="#440">440</a>     <em>/**<em>*</em></em>
<a name="441" href="#441">441</a> <em>     * Read MIME type data from mimetype-report.txt.</em>
<a name="442" href="#442">442</a> <em>     * MIME type of text/dns is separated from other MIME types.</em>
<a name="443" href="#443">443</a> <em>     * @return True if we found some stats.</em>
<a name="444" href="#444">444</a> <em>     */</em>
<a name="445" href="#445">445</a>     <strong>private</strong> <strong>boolean</strong> calculateMimeTypeDistribution() {    	
<a name="446" href="#446">446</a>     	File f = <strong>new</strong> File(cjob.getDirectory(), <span class="string">"mimetype-report.txt"</span>);
<a name="447" href="#447">447</a>     	<strong>if</strong> (!f.exists()) {
<a name="448" href="#448">448</a>     		<strong>return</strong> false;
<a name="449" href="#449">449</a>     	}
<a name="450" href="#450">450</a>     	BufferedReader br = <strong>null</strong>;
<a name="451" href="#451">451</a>     	<strong>try</strong> {
<a name="452" href="#452">452</a> 	    	FileReader reader = <strong>new</strong> FileReader(f);
<a name="453" href="#453">453</a> 	    	br = <strong>new</strong> BufferedReader(reader);
<a name="454" href="#454">454</a> 	    	String line = br.readLine();  <em class="comment">// Ignore heading</em>
<a name="455" href="#455">455</a> 	    	line = br.readLine();
<a name="456" href="#456">456</a> 	    	<strong>while</strong> (line != <strong>null</strong>) {	    			    	  
<a name="457" href="#457">457</a> 	    		<em class="comment">// Get num urls, num bytes, and MIME type (seperated by a space)</em>
<a name="458" href="#458">458</a> 	    		<em class="comment">// Example: 12 134279 text/html</em>
<a name="459" href="#459">459</a>   
<a name="460" href="#460">460</a> 	    		String[] items = line.split(<span class="string">" "</span>);
<a name="461" href="#461">461</a> 	    		<strong>if</strong> (items.length &lt; 3) {
<a name="462" href="#462">462</a> 	    			logger.log(Level.WARNING,
<a name="463" href="#463">463</a>                             <span class="string">"Unexpected formatting on line ["</span> + line + <span class="string">"]"</span>);
<a name="464" href="#464">464</a> 	    		}
<a name="465" href="#465">465</a> 	    		<strong>else</strong> {
<a name="466" href="#466">466</a> 	    			<strong>long</strong> total = Long.parseLong(items[0]);
<a name="467" href="#467">467</a> 	    			<strong>long</strong> bytes = Long.parseLong(items[1]);
<a name="468" href="#468">468</a> 	    			String mime = items[2];
<a name="469" href="#469">469</a> 
<a name="470" href="#470">470</a> 	    			<em class="comment">// Seperate DNS reconrds from HTTP</em>
<a name="471" href="#471">471</a> 	    			<strong>if</strong> (mime.equalsIgnoreCase(<span class="string">"text/dns"</span>)) {