?? statisticssummary.html
字號:
<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>)) {<a name="472" href="#472">472</a> mimeTypeDnsDistribution.put(mime,<a name="473" href="#473">473</a> <strong>new</strong> <a href="../../../../org/archive/util/LongWrapper.html">LongWrapper</a>(total));<a name="474" href="#474">474</a> mimeTypeDnsBytes.put(mime, <strong>new</strong> LongWrapper(bytes));<a name="475" href="#475">475</a> totalDnsMimeTypeDocuments += total;<a name="476" href="#476">476</a> totalDnsMimeSize += bytes;<a name="477" href="#477">477</a> }<a name="478" href="#478">478</a> <strong>else</strong> {<a name="479" href="#479">479</a> mimeTypeDistribution.put(mime, <strong>new</strong> LongWrapper(total));<a name="480" href="#480">480</a> mimeTypeBytes.put(mime, <strong>new</strong> LongWrapper(bytes));<a name="481" href="#481">481</a> totalMimeTypeDocuments += total;<a name="482" href="#482">482</a> totalMimeSize += bytes;<a name="483" href="#483">483</a> }<a name="484" href="#484">484</a> }<a name="485" href="#485">485</a> line = br.readLine();<a name="486" href="#486">486</a> }<a name="487" href="#487">487</a> } <strong>catch</strong> (IOException e) {<a name="488" href="#488">488</a> logger.log(Level.SEVERE, <span class="string">"Reading "</span> + f.getAbsolutePath(), e);<a name="489" href="#489">489</a> } <strong>finally</strong> {<a name="490" href="#490">490</a> <strong>if</strong> (br != <strong>null</strong>) {<a name="491" href="#491">491</a> <strong>try</strong> {<a name="492" href="#492">492</a> br.close();<a name="493" href="#493">493</a> } <strong>catch</strong> (IOException e) {<a name="494" href="#494">494</a> logger.log(Level.SEVERE,<a name="495" href="#495">495</a> <span class="string">"Closing "</span> + f.getAbsolutePath(), e);<a name="496" href="#496">496</a> }<a name="497" href="#497">497</a> }<a name="498" href="#498">498</a> }<a name="499" href="#499">499</a> <strong>return</strong> <strong>true</strong>;<a name="500" href="#500">500</a> }<a name="501" href="#501">501</a> <a name="502" href="#502">502</a> <em>/**<em>*</em></em><a name="503" href="#503">503</a> <em> * Read number of URLs and total bytes for each host name from</em><a name="504" href="#504">504</a> <em> * hosts-report.txt.</em><a name="505" href="#505">505</a> <em> * Host name of "dns:" is separated from others.</em><a name="506" href="#506">506</a> <em> * @return true if stats found.</em><a name="507" href="#507">507</a> <em> */</em><a name="508" href="#508">508</a> <strong>private</strong> <strong>boolean</strong> calculateHostsDistribution() {<a name="509" href="#509">509</a> File f = <strong>new</strong> File(cjob.getDirectory(), <span class="string">"hosts-report.txt"</span>);<a name="510" href="#510">510</a> <strong>if</strong> (!f.exists()) {<a name="511" href="#511">511</a> <strong>return</strong> false;<a name="512" href="#512">512</a> }<a name="513" href="#513">513</a> BufferedReader br = <strong>null</strong>;<a name="514" href="#514">514</a> <strong>try</strong> {<a name="515" href="#515">515</a> FileReader reader = <strong>new</strong> FileReader(f);<a name="516" href="#516">516</a> br = <strong>new</strong> BufferedReader(reader);<a name="517" href="#517">517</a> String line = br.readLine(); <em class="comment">// Ignore heading</em><a name="518" href="#518">518</a> line = br.readLine();<a name="519" href="#519">519</a> <strong>while</strong> (line != <strong>null</strong>) { <a name="520" href="#520">520</a> <em class="comment">// Get num urls, num bytes, and host name (seperated by a space)</em><a name="521" href="#521">521</a> <em class="comment">// Example: 9 7468 www.blogger.com</em><a name="522" href="#522">522</a> <a name="523" href="#523">523</a> String[] items = line.split(<span class="string">" "</span>);<a name="524" href="#524">524</a> <strong>if</strong> (items.length < 3) {<a name="525" href="#525">525</a> logger.log(Level.WARNING,<a name="526" href="#526">526</a> <span class="string">"Unexpected formatting on line ["</span> + line + <span class="string">"]"</span>);<a name="527" href="#527">527</a> }<a name="528" href="#528">528</a> <strong>else</strong> {<a name="529" href="#529">529</a> <strong>long</strong> total = Long.parseLong(items[0]);<a name="530" href="#530">530</a> <strong>long</strong> bytes = Long.parseLong(items[1]);<a name="531" href="#531">531</a> String host = items[2];<a name="532" href="#532">532</a> <a name="533" href="#533">533</a> <em class="comment">// Seperate DNS reconrds from HTTP</em><a name="534" href="#534">534</a> <strong>if</strong> (host.startsWith(<span class="string">"dns:"</span>, 0)) {<a name="535" href="#535">535</a> hostsDnsDistribution.put(host, <strong>new</strong> LongWrapper(total));<a name="536" href="#536">536</a> hostsDnsBytes.put(host, <strong>new</strong> LongWrapper(bytes));<a name="537" href="#537">537</a> totalDnsHostDocuments += total;<a name="538" href="#538">538</a> totalDnsHostSize += bytes;<a name="539" href="#539">539</a> }<a name="540" href="#540">540</a> <strong>else</strong> {<a name="541" href="#541">541</a> hostsDistribution.put(host, <strong>new</strong> LongWrapper(total));<a name="542" href="#542">542</a> hostsBytes.put(host, <strong>new</strong> LongWrapper(bytes));<a name="543" href="#543">543</a> totalHostDocuments += total;<a name="544" href="#544">544</a> totalHostSize += bytes;<a name="545" href="#545">545</a> <a name="546" href="#546">546</a> <em class="comment">// Count top level domain (TLD)</em><a name="547" href="#547">547</a> String tld = host.substring(host.lastIndexOf('.')+1);<a name="548" href="#548">548</a> incrementMapCount(tldDistribution, tld, total); <a name="549" href="#549">549</a> incrementMapCount(tldBytes, tld, bytes);<a name="550" href="#550">550</a> incrementMapCount(tldHostDistribution, tld);<a name="551" href="#551">551</a> totalTldDocuments += total;<a name="552" href="#552">552</a> totalTldSize += bytes;<a name="553" href="#553">553</a> <a name="554" href="#554">554</a> totalHosts++;<a name="555" href="#555">555</a> }<a name="556" href="#556">556</a> }<a name="557" href="#557">557</a> line = br.readLine();<a name="558" href="#558">558</a> }<a name="559" href="#559">559</a> } <strong>catch</strong> (IOException e) {<a name="560" href="#560">560</a> logger.log(Level.SEVERE, <span class="string">"Reading "</span> + f.getAbsolutePath(), e);<a name="561" href="#561">561</a> } <strong>finally</strong> {<a name="562" href="#562">562</a> <strong>if</strong> (br != <strong>null</strong>) {<a name="563" href="#563">563</a> <strong>try</strong> {<a name="564" href="#564">564</a> br.close();<a name="565" href="#565">565</a> } <strong>catch</strong> (IOException e) {<a name="566" href="#566">566</a> logger.log(Level.SEVERE,<a name="567" href="#567">567</a> <span class="string">"Closing "</span> + f.getAbsolutePath(), e);<a name="568" href="#568">568</a> }<a name="569" href="#569">569</a> }<a name="570" href="#570">570</a> }<a name="571" href="#571">571</a> <strong>return</strong> <strong>true</strong>;<a name="572" href="#572">572</a> }<a name="573" href="#573">573</a> <a name="574" href="#574">574</a> <em>/**<em>*</em></em><a name="575" href="#575">575</a> <em> * Returns the accumulated number of bytes downloaded from a given host.</em><a name="576" href="#576">576</a> <em> * @param host name of the host</em><a name="577" href="#577">577</a> <em> * @return the accumulated number of bytes downloaded from a given host</em><a name="578" href="#578">578</a> <em> */</em><a name="579" href="#579">579</a> <strong>public</strong> <strong>long</strong> getBytesPerHost(String host) { <a name="580" href="#580">580</a> <strong>long</strong> bytes = -1;<a name="581" href="#581">581</a> <a name="582" href="#582">582</a> bytes = host != <strong>null</strong> && host.startsWith(<span class="string">"dns:"</span>, 0) ? <a name="583" href="#583">583</a> ((LongWrapper)hostsDnsBytes.get(host)).longValue :<a name="584" href="#584">584</a> ((LongWrapper)hostsBytes.get(host)).longValue; <a name="585" href="#585">585</a> <a name="586" href="#586">586</a> <strong>return</strong> bytes;<a name="587" href="#587">587</a> }<a name="588" href="#588">588</a> <a name="589" href="#589">589</a> <em>/**<em>*</em></em><a name="590" href="#590">590</a> <em> * Returns the total number of bytes downloaded for a given TLD.</em><a name="591" href="#591">591</a> <em> * @param tld TLD</em><a name="592" href="#592">592</a> <em> * @return the total number of bytes downloaded for a given TLD</em><a name="593" href="#593">593</a> <em> */</em><a name="594" href="#594">594</a> <strong>public</strong> <strong>long</strong> getBytesPerTld(String tld) {<a name="595" href="#595">595</a> LongWrapper lw = (LongWrapper)tldBytes.get(tld);<a name="596" href="#596">596</a> <strong>return</strong> (lw == <strong>null</strong> ? 0 : lw.longValue);<a name="597" href="#597">597</a> }<a name="598" href="#598">598</a> <a name="599" href="#599">599</a> <em>/**<em>*</em></em><a name="600" href="#600">600</a> <em> * Returns the accumulated number of bytes from files of a given file type.</em><a name="601" href="#601">601</a> <em> * @param filetype Filetype to check.</em><a name="602" href="#602">602</a> <em> * @return the accumulated number of bytes from files of a given mime type</em><a name="603" href="#603">603</a> <em> */</em><a name="604" href="#604">604</a> <strong>public</strong> <strong>long</strong> getBytesPerMimeType(String filetype) {<a name="605" href="#605">605</a> <strong>long</strong> bytes = -1;<a name="606" href="#606">606</a> <a name="607" href="#607">607</a> <strong>if</strong> (filetype != <strong>null</strong>) { <a name="608" href="#608">608</a> <strong>if</strong> (filetype.equals(<span class="string">"text/dns"</span>)) { <a name="609" href="#609">609</a> bytes = mimeTypeDnsBytes.get(filetype) == <strong>null</strong> ? 0 :<a name="610" href="#610">610</a> ((LongWrapper)mimeTypeDnsBytes.get(filetype)).longValue;<a name="611" href="#611">611</a> }<a name="612" href="#612">612</a> <strong>else</strong> {<a name="613" href="#613">613</a> bytes = mimeTypeBytes.get(filetype) == <strong>null</strong> ? 0 :<a name="614" href="#614">614</a> ((LongWrapper)mimeTypeBytes.get(filetype)).longValue;<a name="615" href="#615">615</a> }<a name="616" href="#616">616</a> }<a name="617" href="#617">617</a> <strong>return</strong> bytes;<a name="618" href="#618">618</a> }<a name="619" href="#619">619</a> <a name="620" href="#620">620</a> <em>/**<em>*</em></em><a name="621" href="#621">621</a> <em> * Reads duration time, processed docs/sec, bandwidth, and total size</em><a name="622" href="#622">622</a> <em> * of crawl from crawl-report.txt.</em><a name="623" href="#623">623</a> <em> * @return true if stats found.</em><a name="624" href="#624">624</a> <em> */</em><a name="625" href="#625">625</a> <strong>public</strong> <strong>boolean</strong> readCrawlReport() {<a name="626" href="#626">626</a> File f = <strong>new</strong> File(cjob.getDirectory(), <span class="string">"crawl-report.txt"</span>);<a name="627" href="#627">627</a> <strong>if</strong> (!f.exists()) {
?? 快捷鍵說明
復制代碼
Ctrl + C
搜索代碼
Ctrl + F
全屏模式
F11
切換主題
Ctrl + Shift + D
顯示快捷鍵
?
增大字號
Ctrl + =
減小字號
Ctrl + -