crawluri.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="596" href="#596">596</a>             classname.substring(index + 1): classname);
<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>     * Add an annotation: an abbrieviated indication of something special</em>
<a name="601" href="#601">601</a> <em>     * about this URI that need not be present in every crawl.log line,</em>
<a name="602" href="#602">602</a> <em>     * but should be noted for future reference. </em>
<a name="603" href="#603">603</a> <em>     *</em>
<a name="604" href="#604">604</a> <em>     * @param annotation the annotation to add; should not contain </em>
<a name="605" href="#605">605</a> <em>     * whitespace or a comma</em>
<a name="606" href="#606">606</a> <em>     */</em>
<a name="607" href="#607">607</a>     <strong>public</strong> <strong>void</strong> addAnnotation(String annotation) {
<a name="608" href="#608">608</a>         String annotations;
<a name="609" href="#609">609</a>         <strong>if</strong>(containsKey(A_ANNOTATIONS)) {
<a name="610" href="#610">610</a>             annotations = getString(A_ANNOTATIONS);
<a name="611" href="#611">611</a>             annotations += <span class="string">","</span>+annotation;
<a name="612" href="#612">612</a>         } <strong>else</strong> {
<a name="613" href="#613">613</a>             annotations = annotation;
<a name="614" href="#614">614</a>         }
<a name="615" href="#615">615</a> 
<a name="616" href="#616">616</a>         putString(A_ANNOTATIONS,annotations);
<a name="617" href="#617">617</a>     }
<a name="618" href="#618">618</a>     
<a name="619" href="#619">619</a>     <em>/**<em>*</em></em>
<a name="620" href="#620">620</a> <em>     * TODO: Implement truncation using booleans rather than as this</em>
<a name="621" href="#621">621</a> <em>     * ugly String parse.</em>
<a name="622" href="#622">622</a> <em>     * @return True if fetch was truncated.</em>
<a name="623" href="#623">623</a> <em>     */</em>
<a name="624" href="#624">624</a>     <strong>public</strong> <strong>boolean</strong> isTruncatedFetch() {
<a name="625" href="#625">625</a>         <strong>return</strong> annotationContains(TRUNC_SUFFIX);
<a name="626" href="#626">626</a>     }
<a name="627" href="#627">627</a>     
<a name="628" href="#628">628</a>     <strong>public</strong> <strong>boolean</strong> isLengthTruncatedFetch() {
<a name="629" href="#629">629</a>         <strong>return</strong> annotationContains(LENGTH_TRUNC);
<a name="630" href="#630">630</a>     }
<a name="631" href="#631">631</a>     
<a name="632" href="#632">632</a>     <strong>public</strong> <strong>boolean</strong> isTimeTruncatedFetch() {
<a name="633" href="#633">633</a>         <strong>return</strong> annotationContains(TIMER_TRUNC);
<a name="634" href="#634">634</a>     }
<a name="635" href="#635">635</a>     
<a name="636" href="#636">636</a>     <strong>public</strong> <strong>boolean</strong> isHeaderTruncatedFetch() {
<a name="637" href="#637">637</a>         <strong>return</strong> annotationContains(HEADER_TRUNC);
<a name="638" href="#638">638</a>     }
<a name="639" href="#639">639</a>     
<a name="640" href="#640">640</a>     <strong>protected</strong> <strong>boolean</strong> annotationContains(<strong>final</strong> String str2Find) {
<a name="641" href="#641">641</a>         <strong>boolean</strong> result = false;
<a name="642" href="#642">642</a>         <strong>if</strong> (!containsKey(A_ANNOTATIONS)) {
<a name="643" href="#643">643</a>             <strong>return</strong> result;
<a name="644" href="#644">644</a>         }
<a name="645" href="#645">645</a>         String annotations = getString(A_ANNOTATIONS);
<a name="646" href="#646">646</a>         <strong>if</strong> (annotations != <strong>null</strong> &amp;&amp; annotations.length() > 0) {
<a name="647" href="#647">647</a>             result = annotations.indexOf(str2Find) >= 0;
<a name="648" href="#648">648</a>         }
<a name="649" href="#649">649</a>         <strong>return</strong> result;
<a name="650" href="#650">650</a>     }
<a name="651" href="#651">651</a> 
<a name="652" href="#652">652</a>     <em>/**<em>*</em></em>
<a name="653" href="#653">653</a> <em>     * Get the annotations set for this uri.</em>
<a name="654" href="#654">654</a> <em>     *</em>
<a name="655" href="#655">655</a> <em>     * @return the annotations set for this uri.</em>
<a name="656" href="#656">656</a> <em>     */</em>
<a name="657" href="#657">657</a>     <strong>public</strong> String getAnnotations() {
<a name="658" href="#658">658</a>         <strong>return</strong> (containsKey(A_ANNOTATIONS))?
<a name="659" href="#659">659</a>             getString(A_ANNOTATIONS): <strong>null</strong>;
<a name="660" href="#660">660</a>     }
<a name="661" href="#661">661</a> 
<a name="662" href="#662">662</a>     <em>/**<em>*</em></em>
<a name="663" href="#663">663</a> <em>     * Get the embeded hop count.</em>
<a name="664" href="#664">664</a> <em>     *</em>
<a name="665" href="#665">665</a> <em>     * @return the embeded hop count.</em>
<a name="666" href="#666">666</a> <em>     */</em>
<a name="667" href="#667">667</a>     <strong>public</strong> <strong>int</strong> getEmbedHopCount() {
<a name="668" href="#668">668</a>         <strong>return</strong> embedHopCount;
<a name="669" href="#669">669</a>     }
<a name="670" href="#670">670</a> 
<a name="671" href="#671">671</a>     <em>/**<em>*</em></em>
<a name="672" href="#672">672</a> <em>     * Get the link hop count.</em>
<a name="673" href="#673">673</a> <em>     *</em>
<a name="674" href="#674">674</a> <em>     * @return the link hop count.</em>
<a name="675" href="#675">675</a> <em>     */</em>
<a name="676" href="#676">676</a>     <strong>public</strong> <strong>int</strong> getLinkHopCount() {
<a name="677" href="#677">677</a>         <strong>return</strong> linkHopCount;
<a name="678" href="#678">678</a>     }
<a name="679" href="#679">679</a> 
<a name="680" href="#680">680</a>     <em>/**<em>*</em></em>
<a name="681" href="#681">681</a> <em>     * Mark this uri as being a seed.</em>
<a name="682" href="#682">682</a> <em>     *</em>
<a name="683" href="#683">683</a> <em>     */</em>
<a name="684" href="#684">684</a>     <strong>public</strong> <strong>void</strong> markAsSeed() {
<a name="685" href="#685">685</a>         linkHopCount = 0;
<a name="686" href="#686">686</a>         embedHopCount = 0;
<a name="687" href="#687">687</a>     }
<a name="688" href="#688">688</a> 
<a name="689" href="#689">689</a>     <em>/**<em>*</em></em>
<a name="690" href="#690">690</a> <em>     * Get the user agent to use for crawling this URI.</em>
<a name="691" href="#691">691</a> <em>     *</em>
<a name="692" href="#692">692</a> <em>     * If null the global setting should be used.</em>
<a name="693" href="#693">693</a> <em>     *</em>
<a name="694" href="#694">694</a> <em>     * @return user agent or null</em>
<a name="695" href="#695">695</a> <em>     */</em>
<a name="696" href="#696">696</a>     <strong>public</strong> String getUserAgent() {
<a name="697" href="#697">697</a>         <strong>return</strong> userAgent;
<a name="698" href="#698">698</a>     }
<a name="699" href="#699">699</a> 
<a name="700" href="#700">700</a>     <em>/**<em>*</em></em>
<a name="701" href="#701">701</a> <em>     * Set the user agent to use when crawling this URI.</em>
<a name="702" href="#702">702</a> <em>     *</em>
<a name="703" href="#703">703</a> <em>     * If not set the global settings should be used.</em>
<a name="704" href="#704">704</a> <em>     *</em>
<a name="705" href="#705">705</a> <em>     * @param string user agent to use</em>
<a name="706" href="#706">706</a> <em>     */</em>
<a name="707" href="#707">707</a>     <strong>public</strong> <strong>void</strong> setUserAgent(String string) {
<a name="708" href="#708">708</a>         userAgent = string;
<a name="709" href="#709">709</a>     }
<a name="710" href="#710">710</a> 
<a name="711" href="#711">711</a>     <em>/**<em>*</em></em>
<a name="712" href="#712">712</a> <em>     * Set which processor should be the next processor to process this uri</em>
<a name="713" href="#713">713</a> <em>     * instead of using the default next processor.</em>
<a name="714" href="#714">714</a> <em>     *</em>
<a name="715" href="#715">715</a> <em>     * @param processorChain the processor chain to skip to.</em>
<a name="716" href="#716">716</a> <em>     * @param processor the processor in the processor chain to skip to.</em>
<a name="717" href="#717">717</a> <em>     */</em>
<a name="718" href="#718">718</a>     <strong>public</strong> <strong>void</strong> skipToProcessor(<a href="../../../../org/archive/crawler/framework/ProcessorChain.html">ProcessorChain</a> processorChain,
<a name="719" href="#719">719</a>             <a href="../../../../org/archive/crawler/framework/Processor.html">Processor</a> processor) {
<a name="720" href="#720">720</a>         setNextProcessorChain(processorChain);
<a name="721" href="#721">721</a>         setNextProcessor(processor);
<a name="722" href="#722">722</a>     }
<a name="723" href="#723">723</a> 
<a name="724" href="#724">724</a>     <em>/**<em>*</em></em>
<a name="725" href="#725">725</a> <em>     * Set which processor chain should be processing this uri next.</em>
<a name="726" href="#726">726</a> <em>     *</em>
<a name="727" href="#727">727</a> <em>     * @param processorChain the processor chain to skip to.</em>
<a name="728" href="#728">728</a> <em>     */</em>
<a name="729" href="#729">729</a>     <strong>public</strong> <strong>void</strong> skipToProcessorChain(<a href="../../../../org/archive/crawler/framework/ProcessorChain.html">ProcessorChain</a> processorChain) {
<a name="730" href="#730">730</a>         setNextProcessorChain(processorChain);
<a name="731" href="#731">731</a>         setNextProcessor(<strong>null</strong>);
<a name="732" href="#732">732</a>     }
<a name="733" href="#733">733</a> 
<a name="734" href="#734">734</a>     <em>/**<em>*</em></em>
<a name="735" href="#735">735</a> <em>     * For completed HTTP transactions, the length of the content-body.</em>
<a name="736" href="#736">736</a> <em>     *</em>
<a name="737" href="#737">737</a> <em>     * @return For completed HTTP transactions, the length of the content-body.</em>
<a name="738" href="#738">738</a> <em>     */</em>
<a name="739" href="#739">739</a>     <strong>public</strong> <strong>long</strong> getContentLength() {
<a name="740" href="#740">740</a>         <strong>if</strong> (<strong>this</strong>.contentLength &lt; 0) {
<a name="741" href="#741">741</a>             <strong>this</strong>.contentLength = (getHttpRecorder() != <strong>null</strong>)?
<a name="742" href="#742">742</a>