archivereader.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="538" href="#538">538</a>             <strong>for</strong> (<strong>int</strong> i = MAX_ALLOWED_RECOVERABLES; i > 0 &amp;&amp;
<a name="539" href="#539">539</a>                     result == <strong>null</strong>; i--) {
<a name="540" href="#540">540</a>                 ioe = <strong>null</strong>;
<a name="541" href="#541">541</a>                 <strong>try</strong> {
<a name="542" href="#542">542</a>                     result = innerNext();
<a name="543" href="#543">543</a>                 } <strong>catch</strong> (RecoverableIOException e) {
<a name="544" href="#544">544</a>                     ioe = e;
<a name="545" href="#545">545</a>                     getLogger().warning(e.getMessage());
<a name="546" href="#546">546</a>                     <strong>if</strong> (hasNext()) {
<a name="547" href="#547">547</a>                         <strong>continue</strong>;
<a name="548" href="#548">548</a>                     }
<a name="549" href="#549">549</a>                     <em class="comment">// No records left.  Throw exception rather than</em>
<a name="550" href="#550">550</a>                     <em class="comment">// return null.  The caller is expecting to get</em>
<a name="551" href="#551">551</a>                     <em class="comment">// back a record since they've just called</em>
<a name="552" href="#552">552</a>                     <em class="comment">// hasNext.</em>
<a name="553" href="#553">553</a>                     <strong>break</strong>;
<a name="554" href="#554">554</a>                 }
<a name="555" href="#555">555</a>             }
<a name="556" href="#556">556</a>             <strong>if</strong> (ioe != <strong>null</strong>) {
<a name="557" href="#557">557</a>                 <em class="comment">// Then we did MAX_ALLOWED_RECOVERABLES retries.  Throw</em>
<a name="558" href="#558">558</a>                 <em class="comment">// the recoverable ioe wrapped in a RuntimeException so</em>
<a name="559" href="#559">559</a>                 <em class="comment">// it goes out pass checks for IOE.</em>
<a name="560" href="#560">560</a>                 <strong>throw</strong> <strong>new</strong> RuntimeException(<span class="string">"Retried "</span> +
<a name="561" href="#561">561</a>                     MAX_ALLOWED_RECOVERABLES + <span class="string">" times in a row"</span>, ioe);
<a name="562" href="#562">562</a>             }
<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>         <strong>protected</strong> <a href="../../../org/archive/io/ArchiveRecord.html">ArchiveRecord</a> innerNext() throws IOException {
<a name="567" href="#567">567</a>             <strong>return</strong> get(((RepositionableStream)getInputStream()).position());
<a name="568" href="#568">568</a>         }
<a name="569" href="#569">569</a>         
<a name="570" href="#570">570</a>         <strong>public</strong> <strong>void</strong> remove() {
<a name="571" href="#571">571</a>             <strong>throw</strong> <strong>new</strong> UnsupportedOperationException();
<a name="572" href="#572">572</a>         }
<a name="573" href="#573">573</a>     }
<a name="574" href="#574">574</a>     
<a name="575" href="#575">575</a>     <strong>protected</strong> <strong>static</strong> String stripExtension(<strong>final</strong> String name,
<a name="576" href="#576">576</a>     		<strong>final</strong> String ext) {
<a name="577" href="#577">577</a>         <strong>return</strong> (!name.endsWith(ext))? name:
<a name="578" href="#578">578</a>             name.substring(0, name.length() - ext.length());
<a name="579" href="#579">579</a>     }
<a name="580" href="#580">580</a>     
<a name="581" href="#581">581</a>     <em>/**<em>*</em></em>
<a name="582" href="#582">582</a> <em>     * @return short name of Archive file.</em>
<a name="583" href="#583">583</a> <em>     */</em>
<a name="584" href="#584">584</a>     <strong>public</strong> String getStrippedFileName() {
<a name="585" href="#585">585</a>     	<strong>return</strong> getStrippedFileName((<strong>new</strong> File(getReaderIdentifier())).getName(),
<a name="586" href="#586">586</a>     		getDotFileExtension());
<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>     * @param name Name of ARCFile.</em>
<a name="591" href="#591">591</a> <em>     * @param dotFileExtension '.arc' or '.warc', etc.</em>
<a name="592" href="#592">592</a> <em>     * @return short name of Archive file.</em>
<a name="593" href="#593">593</a> <em>     */</em>
<a name="594" href="#594">594</a>     <strong>public</strong> <strong>static</strong> String getStrippedFileName(String name,
<a name="595" href="#595">595</a>     		<strong>final</strong> String dotFileExtension) {
<a name="596" href="#596">596</a>     	name = stripExtension(name,
<a name="597" href="#597">597</a>     		ArchiveFileConstants.DOT_COMPRESSED_FILE_EXTENSION);
<a name="598" href="#598">598</a>     	<strong>return</strong> stripExtension(name, dotFileExtension);
<a name="599" href="#599">599</a>     }
<a name="600" href="#600">600</a>     
<a name="601" href="#601">601</a>     <em>/**<em>*</em></em>
<a name="602" href="#602">602</a> <em>     * @param value Value to test.</em>
<a name="603" href="#603">603</a> <em>     * @return True if value is 'true', else false.</em>
<a name="604" href="#604">604</a> <em>     */</em>
<a name="605" href="#605">605</a>     <strong>protected</strong> <strong>static</strong> <strong>boolean</strong> getTrueOrFalse(<strong>final</strong> String value) {
<a name="606" href="#606">606</a>     	<strong>if</strong> (value == <strong>null</strong> || value.length() &lt;= 0) {
<a name="607" href="#607">607</a>     		<strong>return</strong> false;
<a name="608" href="#608">608</a>     	}
<a name="609" href="#609">609</a>         <strong>return</strong> Boolean.TRUE.toString().equals(value.toLowerCase());
<a name="610" href="#610">610</a>     }
<a name="611" href="#611">611</a>     
<a name="612" href="#612">612</a>     <em>/**<em>*</em></em>
<a name="613" href="#613">613</a> <em>     * @param format Format to use outputting.</em>
<a name="614" href="#614">614</a> <em>     * @throws IOException</em>
<a name="615" href="#615">615</a> <em>     * @throws java.text.ParseException</em>
<a name="616" href="#616">616</a> <em>     * @return True if handled.</em>
<a name="617" href="#617">617</a> <em>     */</em>
<a name="618" href="#618">618</a>     <strong>protected</strong> <strong>boolean</strong> output(<strong>final</strong> String format)
<a name="619" href="#619">619</a>     throws IOException, java.text.ParseException {
<a name="620" href="#620">620</a>     	<strong>boolean</strong> result = <strong>true</strong>;
<a name="621" href="#621">621</a>         <em class="comment">// long start = System.currentTimeMillis();</em>
<a name="622" href="#622">622</a>     	
<a name="623" href="#623">623</a>         <em class="comment">// Write output as pseudo-CDX file.  See</em>
<a name="624" href="#624">624</a>         <em class="comment">// http://www.archive.org/web/researcher/cdx_legend.php</em>
<a name="625" href="#625">625</a>         <em class="comment">// and http://www.archive.org/web/researcher/example_cdx.php.</em>
<a name="626" href="#626">626</a>         <em class="comment">// Hash is hard-coded straight SHA-1 hash of content.</em>
<a name="627" href="#627">627</a>         <strong>if</strong> (format.equals(DUMP)) {
<a name="628" href="#628">628</a>         	<em class="comment">// No point digesting dumping.</em>
<a name="629" href="#629">629</a>         	setDigest(false);
<a name="630" href="#630">630</a>             dump(false);
<a name="631" href="#631">631</a>         } <strong>else</strong> <strong>if</strong> (format.equals(GZIP_DUMP)) {
<a name="632" href="#632">632</a>         	<em class="comment">// No point digesting dumping.</em>
<a name="633" href="#633">633</a>         	setDigest(false);
<a name="634" href="#634">634</a>             dump(<strong>true</strong>);
<a name="635" href="#635">635</a>         } <strong>else</strong> <strong>if</strong> (format.equals(CDX)) {
<a name="636" href="#636">636</a>         	cdxOutput(false);   
<a name="637" href="#637">637</a>         } <strong>else</strong> <strong>if</strong> (format.equals(CDX_FILE)) {
<a name="638" href="#638">638</a>             cdxOutput(<strong>true</strong>);
<a name="639" href="#639">639</a>         } <strong>else</strong> {
<a name="640" href="#640">640</a>         	result = false;
<a name="641" href="#641">641</a>         }	
<a name="642" href="#642">642</a>         <strong>return</strong> result;
<a name="643" href="#643">643</a>     }
<a name="644" href="#644">644</a>     
<a name="645" href="#645">645</a>     <strong>protected</strong> <strong>void</strong> cdxOutput(<strong>boolean</strong> toFile)
<a name="646" href="#646">646</a>     throws IOException {
<a name="647" href="#647">647</a>         BufferedWriter cdxWriter = <strong>null</strong>;
<a name="648" href="#648">648</a>         <strong>if</strong> (toFile) {
<a name="649" href="#649">649</a>             String cdxFilename = stripExtension(getReaderIdentifier(),
<a name="650" href="#650">650</a>                 DOT_COMPRESSED_FILE_EXTENSION);
<a name="651" href="#651">651</a>             cdxFilename = stripExtension(cdxFilename, getDotFileExtension());
<a name="652" href="#652">652</a>             cdxFilename += ('.' + CDX);
<a name="653" href="#653">653</a>             cdxWriter = <strong>new</strong> BufferedWriter(<strong>new</strong> FileWriter(cdxFilename));
<a name="654" href="#654">654</a>         }
<a name="655" href="#655">655</a>         
<a name="656" href="#656">656</a>         String header = <span class="string">"CDX b e a m s c "</span> + ((isCompressed()) ? <span class="string">"V"</span> : <span class="string">"v"</span>)
<a name="657" href="#657">657</a>             + <span class="string">" n g"</span>;
<a name="658" href="#658">658</a>         <strong>if</strong> (toFile) {
<a name="659" href="#659">659</a>             cdxWriter.write(header);
<a name="660" href="#660">660</a>             cdxWriter.newLine();
<a name="661" href="#661">661</a>         } <strong>else</strong> {
<a name="662" href="#662">662</a>             System.out.println(header);
<a name="663" href="#663">663</a>         }
<a name="664" href="#664">664</a>         
<a name="665" href="#665">665</a>         String strippedFileName = getStrippedFileName();
<a name="666" href="#666">666</a>         <strong>try</strong> {
<a name="667" href="#667">667</a>             <strong>for</strong> (Iterator&lt;ArchiveRecord> ii = iterator(); ii.hasNext();) {
<a name="668" href="#668">668</a>             	ArchiveRecord r = ii.next();
<a name="669" href="#669">669</a>                 <strong>if</strong> (toFile) {
<a name="670" href="#670">670</a>                     cdxWriter.write(r.outputCdx(strippedFileName));
<a name="671" href="#671">671</a>                     cdxWriter.newLine();
<a name="672" href="#672">672</a>                 } <strong>else</strong> {
<a name="673" href="#673">673</a>                     System.out.println(r.outputCdx(strippedFileName));
<a name="674" href="#674">674</a>                 }
<a name="675" href="#675">675</a>             }
<a name="676" href="#676">676</a>         } <strong>finally</strong> {
<a name="677" href="#677">677</a>             <strong>if</strong> (toFile) {
<a name="678" href="#678">678</a>                 cdxWriter.close();
<a name="679" href="#679">679</a>             }
<a name="680" href="#680">680</a>         }
<a name="681" href="#681">681</a>     }
<a name="682" href="#682">682</a>     
<a name="683" href="#683">683</a>     <em>/**<em>*</em></em>
<a name="684" href="#684">684</a> <em>     * Output passed record using passed format specifier.</em>
<a name="685" href="#685">685</a> <em>     * @param format What format to use outputting.</em>
<a name="686" href="#686">686</a> <em>     * @throws IOException</em>
<a name="687" href="#687">687</a> <em>     * @return True if handled.</em>
<a name="688" href="#688">688</a> <em>     */</em>
<a name="689" href="#689">689</a>     <strong>protected</strong> <strong>boolean</strong> outputRecord(<strong>final</strong> String format)
<a name="690" href="#690">690</a>     throws IOException {
<a name="691" href="#691">691</a>     	<strong>boolean</strong> result = <strong>true</strong>;
<a name="692" href="#692">692</a>         <strong>if</strong> (format.equals(CDX)) {
<a name="693" href="#693">693</a>             System.out.println(get().outputCdx(getStrippedFileName()));
<a name="694" href="#694">694</a>         } <strong>else</strong> <strong>if</strong>(format.equals(ArchiveFileConstants.DUMP)) {
<a name="695" href="#695">695</a>             <em class="comment">// No point digesting if dumping content.</em>
<a name="696" href="#696">696</a>             setDigest(false);
<a name="697" href="#697">697</a>             get().dump();
<a name="698" href="#698">698</a>         } <strong>else</strong> {
<a name="699" href="#699">699</a>         	result = false;
<a name="700" href="#700">700</a>         }
<a name="701" href="#701">701</a>         <strong>return</strong> result;
<a name="702" href="#702">702</a>     }
<a name="703" href="#703">703</a> 
<a name="704" href="#704">704</a>     <em>/**<em>*</em></em>
<a name="705" href="#705">705</a> <em>     * Dump this file on STDOUT</em>
<a name="706" href="#706">706</a> <em>     * @throws compress True if dumped output is compressed.</em>
<a name="707" href="#707">707</a> <em>     * @throws IOException</em>
<a name="708" href="#708">708</a> <em>     * @throws java.text.ParseException</em>
<a name="709" href="#709">709</a> <em>     */</em>
<a name="710" href="#710">710</a>     <strong>public</strong> <strong>abstract</strong> <strong>void</strong> dump(<strong>final</strong> <strong>boolean</strong> compress)
<a name="711" href="#711">711</a>     throws IOException, java.text.ParseException;
<a name="712" href="#712">712</a> }
</pre>
<hr/><div id="footer">This page was automatically generated by <a href="http://maven.apache.org/">Maven</a></div></body>
</html>