fetchftp.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="481" href="#481">481</a>     <em>/**<em>*</em></em>
<a name="482" href="#482">482</a> <em>     * Returns the &lt;code>extract.parent&lt;/code> attribute for this</em>
<a name="483" href="#483">483</a> <em>     * &lt;code>FetchFTP&lt;/code> and the given curi.</em>
<a name="484" href="#484">484</a> <em>     * </em>
<a name="485" href="#485">485</a> <em>     * @param curi  the curi whose attribute to return</em>
<a name="486" href="#486">486</a> <em>     * @return  that curi's &lt;code>extract-parent&lt;/code></em>
<a name="487" href="#487">487</a> <em>     */</em>
<a name="488" href="#488">488</a>     <strong>public</strong> <strong>boolean</strong> getExtractParent(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) {
<a name="489" href="#489">489</a>         <strong>return</strong> (Boolean)get(curi, ATTR_EXTRACT_PARENT, DEFAULT_EXTRACT_PARENT);
<a name="490" href="#490">490</a>     }
<a name="491" href="#491">491</a> 
<a name="492" href="#492">492</a> 
<a name="493" href="#493">493</a>     <em>/**<em>*</em></em>
<a name="494" href="#494">494</a> <em>     * Returns the &lt;code>timeout-seconds&lt;/code> attribute for this</em>
<a name="495" href="#495">495</a> <em>     * &lt;code>FetchFTP&lt;/code> and the given curi.</em>
<a name="496" href="#496">496</a> <em>     * </em>
<a name="497" href="#497">497</a> <em>     * @param curi   the curi whose attribute to return</em>
<a name="498" href="#498">498</a> <em>     * @return   that curi's &lt;code>timeout-seconds&lt;/code></em>
<a name="499" href="#499">499</a> <em>     */</em>
<a name="500" href="#500">500</a>     <strong>public</strong> <strong>int</strong> getTimeout(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) {
<a name="501" href="#501">501</a>         <strong>return</strong> (Integer)get(curi, ATTR_TIMEOUT, DEFAULT_TIMEOUT);
<a name="502" href="#502">502</a>     }
<a name="503" href="#503">503</a> 
<a name="504" href="#504">504</a> 
<a name="505" href="#505">505</a>     <em>/**<em>*</em></em>
<a name="506" href="#506">506</a> <em>     * Returns the &lt;code>max-length-bytes&lt;/code> attribute for this</em>
<a name="507" href="#507">507</a> <em>     * &lt;code>FetchFTP&lt;/code> and the given curi.</em>
<a name="508" href="#508">508</a> <em>     * </em>
<a name="509" href="#509">509</a> <em>     * @param curi  the curi whose attribute to return</em>
<a name="510" href="#510">510</a> <em>     * @return  that curi's &lt;code>max-length-bytes&lt;/code></em>
<a name="511" href="#511">511</a> <em>     */</em>
<a name="512" href="#512">512</a>     <strong>public</strong> <strong>long</strong> getMaxLength(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) {
<a name="513" href="#513">513</a>         <strong>return</strong> (Long)get(curi, ATTR_MAX_LENGTH, DEFAULT_MAX_LENGTH);
<a name="514" href="#514">514</a>     }
<a name="515" href="#515">515</a> 
<a name="516" href="#516">516</a> 
<a name="517" href="#517">517</a>     <em>/**<em>*</em></em>
<a name="518" href="#518">518</a> <em>     * Returns the &lt;code>fetch-bandwidth&lt;/code> attribute for this</em>
<a name="519" href="#519">519</a> <em>     * &lt;code>FetchFTP&lt;/code> and the given curi.</em>
<a name="520" href="#520">520</a> <em>     * </em>
<a name="521" href="#521">521</a> <em>     * @param curi  the curi whose attribute to return</em>
<a name="522" href="#522">522</a> <em>     * @return  that curi's &lt;code>fetch-bandwidth&lt;/code></em>
<a name="523" href="#523">523</a> <em>     */</em>
<a name="524" href="#524">524</a>     <strong>public</strong> <strong>int</strong> getFetchBandwidth(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) {
<a name="525" href="#525">525</a>         <strong>return</strong> (Integer)get(curi, ATTR_BANDWIDTH, DEFAULT_BANDWIDTH);
<a name="526" href="#526">526</a>     }
<a name="527" href="#527">527</a> 
<a name="528" href="#528">528</a> 
<a name="529" href="#529">529</a>     <em>/**<em>*</em></em>
<a name="530" href="#530">530</a> <em>     * Returns the username and password for the given URI.  This method</em>
<a name="531" href="#531">531</a> <em>     * always returns an array of length 2.  The first element in the returned</em>
<a name="532" href="#532">532</a> <em>     * array is the username for the URI, and the second element is the</em>
<a name="533" href="#533">533</a> <em>     * password.</em>
<a name="534" href="#534">534</a> <em>     * </em>
<a name="535" href="#535">535</a> <em>     * &lt;p>If the URI itself contains the username and password (i.e., it looks</em>
<a name="536" href="#536">536</a> <em>     * like &lt;code>ftp://username:password@host/path&lt;/code>) then that username</em>
<a name="537" href="#537">537</a> <em>     * and password are returned.</em>
<a name="538" href="#538">538</a> <em>     * </em>
<a name="539" href="#539">539</a> <em>     * &lt;p>Otherwise the settings system is probed for the &lt;code>username&lt;/code></em>
<a name="540" href="#540">540</a> <em>     * and &lt;code>password&lt;/code> attributes for this &lt;code>FTPFetch&lt;/code></em>
<a name="541" href="#541">541</a> <em>     * and the given &lt;code>curi&lt;/code> context.  The values of those </em>
<a name="542" href="#542">542</a> <em>     * attributes are then returned.</em>
<a name="543" href="#543">543</a> <em>     * </em>
<a name="544" href="#544">544</a> <em>     * @param curi  the curi whose username and password to return</em>
<a name="545" href="#545">545</a> <em>     * @return  an array containing the username and password</em>
<a name="546" href="#546">546</a> <em>     */</em>
<a name="547" href="#547">547</a>     <strong>private</strong> String[] getAuth(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) {
<a name="548" href="#548">548</a>         String[] result = <strong>new</strong> String[2];
<a name="549" href="#549">549</a>         <a href="../../../../org/archive/net/UURI.html">UURI</a> uuri = curi.getUURI();
<a name="550" href="#550">550</a>         String userinfo;
<a name="551" href="#551">551</a>         <strong>try</strong> {
<a name="552" href="#552">552</a>             userinfo = uuri.getUserinfo();
<a name="553" href="#553">553</a>         } <strong>catch</strong> (URIException e) {
<a name="554" href="#554">554</a>             assert false;
<a name="555" href="#555">555</a>             logger.finest(<span class="string">"getUserinfo raised URIException."</span>);
<a name="556" href="#556">556</a>             userinfo = <strong>null</strong>;
<a name="557" href="#557">557</a>         }
<a name="558" href="#558">558</a>         <strong>if</strong> (userinfo != <strong>null</strong>) {
<a name="559" href="#559">559</a>             <strong>int</strong> p = userinfo.indexOf(':');
<a name="560" href="#560">560</a>             <strong>if</strong> (p > 0) {
<a name="561" href="#561">561</a>                 result[0] = userinfo.substring(0,p);
<a name="562" href="#562">562</a>                 result[1] = userinfo.substring(p + 1);
<a name="563" href="#563">563</a>                 <strong>return</strong> result;
<a name="564" href="#564">564</a>             }
<a name="565" href="#565">565</a>         }
<a name="566" href="#566">566</a>         result[0] = (String)get(curi, ATTR_USERNAME, DEFAULT_USERNAME);
<a name="567" href="#567">567</a>         result[1] = (String)get(curi, ATTR_PASSWORD, DEFAULT_PASSWORD);
<a name="568" href="#568">568</a>         <strong>return</strong> result;
<a name="569" href="#569">569</a>     }
<a name="570" href="#570">570</a>     
<a name="571" href="#571">571</a>     
<a name="572" href="#572">572</a>     <em>/**<em>*</em></em>
<a name="573" href="#573">573</a> <em>     * Determines the password for the given URI.  If the URI itself contains</em>
<a name="574" href="#574">574</a> <em>     * a password, then that password is returned.  Otherwise the settings</em>
<a name="575" href="#575">575</a> <em>     * system is probed for the &lt;code>password&lt;/code> attribute, and the value</em>
<a name="576" href="#576">576</a> <em>     * for that attribute is returned.</em>
<a name="577" href="#577">577</a> <em>     * </em>
<a name="578" href="#578">578</a> <em>     * @param curi  the curi whose password to return</em>
<a name="579" href="#579">579</a> <em>     * @return  that password</em>
<a name="580" href="#580">580</a> <em>     */</em>
<a name="581" href="#581">581</a>     <strong>public</strong> String determinePassword(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) {
<a name="582" href="#582">582</a>         <strong>return</strong> (String)get(curi, ATTR_PASSWORD, DEFAULT_PASSWORD);
<a name="583" href="#583">583</a>     }
<a name="584" href="#584">584</a> 
<a name="585" href="#585">585</a> 
<a name="586" href="#586">586</a>     <em>/**<em>*</em></em>
<a name="587" href="#587">587</a> <em>     * Quietly closes the given socket.</em>
<a name="588" href="#588">588</a> <em>     * </em>
<a name="589" href="#589">589</a> <em>     * @param socket  the socket to close</em>
<a name="590" href="#590">590</a> <em>     */</em>
<a name="591" href="#591">591</a>     <strong>private</strong> <strong>static</strong> <strong>void</strong> close(Socket socket) {
<a name="592" href="#592">592</a>         <strong>try</strong> {
<a name="593" href="#593">593</a>             socket.close();
<a name="594" href="#594">594</a>         } <strong>catch</strong> (IOException e) {
<a name="595" href="#595">595</a>             logger.log(Level.WARNING, <span class="string">"IO error closing socket."</span>, e);
<a name="596" href="#596">596</a>         }
<a name="597" href="#597">597</a>     }
<a name="598" href="#598">598</a> 
<a name="599" href="#599">599</a> 
<a name="600" href="#600">600</a>     <em>/**<em>*</em></em>
<a name="601" href="#601">601</a> <em>     * Quietly closes the given sequence.</em>
<a name="602" href="#602">602</a> <em>     * If an IOException is raised, this method logs it as a warning.</em>
<a name="603" href="#603">603</a> <em>     * </em>
<a name="604" href="#604">604</a> <em>     * @param seq  the sequence to close</em>
<a name="605" href="#605">605</a> <em>     */</em>
<a name="606" href="#606">606</a>     <strong>private</strong> <strong>static</strong> <strong>void</strong> close(<a href="../../../../org/archive/io/ReplayCharSequence.html">ReplayCharSequence</a> seq) {
<a name="607" href="#607">607</a>         <strong>if</strong> (seq == <strong>null</strong>) {
<a name="608" href="#608">608</a>             <strong>return</strong>;
<a name="609" href="#609">609</a>         }
<a name="610" href="#610">610</a>         <strong>try</strong> {
<a name="611" href="#611">611</a>             seq.close();
<a name="612" href="#612">612</a>         } <strong>catch</strong> (IOException e) {
<a name="613" href="#613">613</a>             logger.log(Level.WARNING, <span class="string">"IO error closing ReplayCharSequence."</span>, 
<a name="614" href="#614">614</a>              e);
<a name="615" href="#615">615</a>         }
<a name="616" href="#616">616</a>     }
<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>     * Quietly disconnects from the given FTP client.</em>
<a name="621" href="#621">621</a> <em>     * If an IOException is raised, this method logs it as a warning.</em>
<a name="622" href="#622">622</a> <em>     * </em>
<a name="623" href="#623">623</a> <em>     * @param client  the client to disconnect</em>
<a name="624" href="#624">624</a> <em>     */</em>
<a name="625" href="#625">625</a>     <strong>private</strong> <strong>static</strong> <strong>void</strong> disconnect(<a href="../../../../org/archive/net/ClientFTP.html">ClientFTP</a> client) {
<a name="626" href="#626">626</a>         <strong>if</strong> (client.isConnected()) <strong>try</strong> {
<a name="627" href="#627">627</a>             client.disconnect();
<a name="628" href="#628">628</a>         } <strong>catch</strong> (IOException e) {
<a name="629" href="#629">629</a>             <strong>if</strong> (logger.isLoggable(Level.WARNING)) {
<a name="630" href="#630">630</a>                 logger.warning(<span class="string">"Could not disconnect from FTP client: "</span> 
<a name="631" href="#631">631</a>                  + e.getMessage());
<a name="632" href="#632">632</a>             }
<a name="633" href="#633">633</a>         }        
<a name="634" href="#634">634</a>     }
<a name="635" href="#635">635</a> 
<a name="636" href="#636">636</a> 
<a name="637" href="#637">637</a> }
</pre>
<hr/><div id="footer">This page was automatically generated by <a href="http://maven.apache.org/">Maven</a></div></body>
</html>