arcreader.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="427" href="#427">427</a> <em>     */</em>
<a name="428" href="#428">428</a>     <strong>public</strong> <strong>boolean</strong> isParseHttpHeaders() {
<a name="429" href="#429">429</a>         <strong>return</strong> <strong>this</strong>.parseHttpHeaders;
<a name="430" href="#430">430</a>     }
<a name="431" href="#431">431</a>     
<a name="432" href="#432">432</a>     <em>/**<em>*</em></em>
<a name="433" href="#433">433</a> <em>     * @param parse The parseHttpHeaders to set.</em>
<a name="434" href="#434">434</a> <em>     */</em>
<a name="435" href="#435">435</a>     <strong>public</strong> <strong>void</strong> setParseHttpHeaders(<strong>boolean</strong> parse) {
<a name="436" href="#436">436</a>         <strong>this</strong>.parseHttpHeaders = parse;
<a name="437" href="#437">437</a>     }
<a name="438" href="#438">438</a>     
<a name="439" href="#439">439</a> 	<strong>public</strong> String getFileExtension() {
<a name="440" href="#440">440</a> 		<strong>return</strong> ARC_FILE_EXTENSION;
<a name="441" href="#441">441</a> 	}
<a name="442" href="#442">442</a> 	
<a name="443" href="#443">443</a> 	<strong>public</strong> String getDotFileExtension() {
<a name="444" href="#444">444</a> 		<strong>return</strong> DOT_ARC_FILE_EXTENSION;
<a name="445" href="#445">445</a> 	}
<a name="446" href="#446">446</a> 	
<a name="447" href="#447">447</a> 	<strong>protected</strong> <strong>boolean</strong> output(<strong>final</strong> String format) 
<a name="448" href="#448">448</a> 	throws IOException, java.text.ParseException {
<a name="449" href="#449">449</a> 		<strong>boolean</strong> result = <strong>super</strong>.output(format);
<a name="450" href="#450">450</a> 		<strong>if</strong>(!result &amp;&amp; format.equals(NOHEAD)) {
<a name="451" href="#451">451</a> 			<strong>throw</strong> <strong>new</strong> IOException(format +
<a name="452" href="#452">452</a> 				<span class="string">" only supported for single Records"</span>);
<a name="453" href="#453">453</a> 		}
<a name="454" href="#454">454</a> 		<strong>return</strong> result;
<a name="455" href="#455">455</a> 	}
<a name="456" href="#456">456</a>     
<a name="457" href="#457">457</a>     <strong>protected</strong> <strong>boolean</strong> outputRecord(<strong>final</strong> String format)
<a name="458" href="#458">458</a>     throws IOException {
<a name="459" href="#459">459</a>     	<strong>boolean</strong> result = <strong>super</strong>.outputRecord(format);
<a name="460" href="#460">460</a>     	<strong>if</strong>(!result &amp;&amp; format.equals(NOHEAD)) {
<a name="461" href="#461">461</a>             <em class="comment">// No point digesting if dumping content.</em>
<a name="462" href="#462">462</a>             setDigest(false);
<a name="463" href="#463">463</a>             <a href="../../../../org/archive/io/arc/ARCRecord.html">ARCRecord</a> r = (ARCRecord)get();
<a name="464" href="#464">464</a>             r.skipHttpHeader();
<a name="465" href="#465">465</a>             r.dump();
<a name="466" href="#466">466</a>             result = <strong>true</strong>;
<a name="467" href="#467">467</a>         }
<a name="468" href="#468">468</a>         <strong>return</strong> result;
<a name="469" href="#469">469</a>     }
<a name="470" href="#470">470</a> 
<a name="471" href="#471">471</a>     <strong>public</strong> <strong>void</strong> dump(<strong>final</strong> <strong>boolean</strong> compress)
<a name="472" href="#472">472</a>     throws IOException, java.text.ParseException {
<a name="473" href="#473">473</a>         <em class="comment">// No point digesting if we're doing a dump.</em>
<a name="474" href="#474">474</a>         setDigest(false);
<a name="475" href="#475">475</a>         <strong>boolean</strong> firstRecord = <strong>true</strong>;
<a name="476" href="#476">476</a>         <a href="../../../../org/archive/io/arc/ARCWriter.html">ARCWriter</a> writer = <strong>null</strong>;
<a name="477" href="#477">477</a>         <strong>for</strong> (Iterator&lt;ArchiveRecord> ii = iterator(); ii.hasNext();) {
<a name="478" href="#478">478</a>             <a href="../../../../org/archive/io/arc/ARCRecord.html">ARCRecord</a> r = (ARCRecord)ii.next();
<a name="479" href="#479">479</a>             <em class="comment">// We're to dump the arc on stdout.</em>
<a name="480" href="#480">480</a>             <em class="comment">// Get the first record's data if any.</em>
<a name="481" href="#481">481</a>             <a href="../../../../org/archive/io/arc/ARCRecordMetaData.html">ARCRecordMetaData</a> meta = r.getMetaData();
<a name="482" href="#482">482</a>             <strong>if</strong> (firstRecord) {
<a name="483" href="#483">483</a>                 firstRecord = false;
<a name="484" href="#484">484</a>                 <em class="comment">// Get an ARCWriter.</em>
<a name="485" href="#485">485</a>                 ByteArrayOutputStream baos =
<a name="486" href="#486">486</a>                     <strong>new</strong> ByteArrayOutputStream(r.available());
<a name="487" href="#487">487</a>                 <em class="comment">// This is slow but done only once at top of ARC.</em>
<a name="488" href="#488">488</a>                 <strong>while</strong> (r.available() > 0) {
<a name="489" href="#489">489</a>                     baos.write(r.read());
<a name="490" href="#490">490</a>                 }
<a name="491" href="#491">491</a>                 List&lt;String> listOfMetadata = <strong>new</strong> ArrayList&lt;String>();
<a name="492" href="#492">492</a>                 listOfMetadata.add(baos.toString(WriterPoolMember.UTF8));
<a name="493" href="#493">493</a>                 <em class="comment">// Assume getArc returns full path to file.  ARCWriter</em>
<a name="494" href="#494">494</a>                 <em class="comment">// or new File will complain if it is otherwise.</em>
<a name="495" href="#495">495</a>                 writer = <strong>new</strong> <a href="../../../../org/archive/io/arc/ARCWriter.html">ARCWriter</a>(<strong>new</strong> AtomicInteger(), System.out,
<a name="496" href="#496">496</a>                     <strong>new</strong> File(meta.getArc()),
<a name="497" href="#497">497</a>                     compress, meta.getDate(), listOfMetadata);
<a name="498" href="#498">498</a>                 <strong>continue</strong>;
<a name="499" href="#499">499</a>             }
<a name="500" href="#500">500</a>             
<a name="501" href="#501">501</a>             writer.write(meta.getUrl(), meta.getMimetype(), meta.getIp(),
<a name="502" href="#502">502</a>                 ArchiveUtils.parse14DigitDate(meta.getDate()).getTime(),
<a name="503" href="#503">503</a>                 (<strong>int</strong>)meta.getLength(), r);
<a name="504" href="#504">504</a>         }
<a name="505" href="#505">505</a>         <em class="comment">// System.out.println(System.currentTimeMillis() - start);</em>
<a name="506" href="#506">506</a>     }
<a name="507" href="#507">507</a>     
<a name="508" href="#508">508</a>     <em class="comment">// Static methods follow.</em>
<a name="509" href="#509">509</a> 
<a name="510" href="#510">510</a>     <em>/**<em>*</em></em>
<a name="511" href="#511">511</a> <em>     *</em>
<a name="512" href="#512">512</a> <em>     * @param formatter Help formatter instance.</em>
<a name="513" href="#513">513</a> <em>     * @param options Usage options.</em>
<a name="514" href="#514">514</a> <em>     * @param exitCode Exit code.</em>
<a name="515" href="#515">515</a> <em>     */</em>
<a name="516" href="#516">516</a>     <strong>private</strong> <strong>static</strong> <strong>void</strong> usage(HelpFormatter formatter, Options options,
<a name="517" href="#517">517</a>             <strong>int</strong> exitCode) {
<a name="518" href="#518">518</a>         formatter.printHelp(<span class="string">"java org.archive.io.arc.ARCReader"</span> +
<a name="519" href="#519">519</a>             <span class="string">" [--digest=true|false] &#47;&#47;\n"</span> +
<a name="520" href="#520">520</a>             <span class="string">" [--format=cdx|cdxfile|dump|gzipdump|nohead]"</span> +
<a name="521" href="#521">521</a>             <span class="string">" [--offset=#] &#47;&#47;\n[--strict] [--parse] ARC_FILE|ARC_URL"</span>,
<a name="522" href="#522">522</a>                 options);
<a name="523" href="#523">523</a>         System.exit(exitCode);
<a name="524" href="#524">524</a>     }
<a name="525" href="#525">525</a> 
<a name="526" href="#526">526</a>     <em>/**<em>*</em></em>
<a name="527" href="#527">527</a> <em>     * Write out the arcfile.</em>
<a name="528" href="#528">528</a> <em>     * </em>
<a name="529" href="#529">529</a> <em>     * @param reader</em>
<a name="530" href="#530">530</a> <em>     * @param format Format to use outputting.</em>
<a name="531" href="#531">531</a> <em>     * @throws IOException</em>
<a name="532" href="#532">532</a> <em>     * @throws java.text.ParseException</em>
<a name="533" href="#533">533</a> <em>     */</em>
<a name="534" href="#534">534</a>     <strong>protected</strong> <strong>static</strong> <strong>void</strong> output(<a href="../../../../org/archive/io/arc/ARCReader.html">ARCReader</a> reader, String format)
<a name="535" href="#535">535</a>     throws IOException, java.text.ParseException {
<a name="536" href="#536">536</a>     	<strong>if</strong> (!reader.output(format)) {
<a name="537" href="#537">537</a>             <strong>throw</strong> <strong>new</strong> IOException(<span class="string">"Unsupported format: "</span> + format);
<a name="538" href="#538">538</a>     	}
<a name="539" href="#539">539</a>     }
<a name="540" href="#540">540</a>     
<a name="541" href="#541">541</a>     
<a name="542" href="#542">542</a>     <em>/**<em>*</em></em>
<a name="543" href="#543">543</a> <em>     * Output passed record using passed format specifier.</em>
<a name="544" href="#544">544</a> <em>     * @param r ARCReader instance to output.</em>
<a name="545" href="#545">545</a> <em>     * @param format What format to use outputting.</em>
<a name="546" href="#546">546</a> <em>     * @throws IOException</em>
<a name="547" href="#547">547</a> <em>     */</em>
<a name="548" href="#548">548</a>     <strong>protected</strong> <strong>static</strong> <strong>void</strong> outputRecord(<strong>final</strong> <a href="../../../../org/archive/io/arc/ARCReader.html">ARCReader</a> r, <strong>final</strong> String format)
<a name="549" href="#549">549</a>     throws IOException {
<a name="550" href="#550">550</a>     	<strong>if</strong> (!r.outputRecord(format)) {
<a name="551" href="#551">551</a>             <strong>throw</strong> <strong>new</strong> IOException(<span class="string">"Unsupported format"</span> +
<a name="552" href="#552">552</a>                 <span class="string">" (or unsupported on a single record): "</span> + format);
<a name="553" href="#553">553</a>     	}
<a name="554" href="#554">554</a>     }
<a name="555" href="#555">555</a> 
<a name="556" href="#556">556</a>     <em>/**<em>*</em></em>
<a name="557" href="#557">557</a> <em>     * Generate a CDX index file for an ARC file.</em>
<a name="558" href="#558">558</a> <em>     *</em>
<a name="559" href="#559">559</a> <em>     * @param urlOrPath The ARC file to generate a CDX index for</em>
<a name="560" href="#560">560</a> <em>     * @throws IOException</em>
<a name="561" href="#561">561</a> <em>     * @throws java.text.ParseException</em>
<a name="562" href="#562">562</a> <em>     */</em>
<a name="563" href="#563">563</a>     <strong>public</strong> <strong>static</strong> <strong>void</strong> createCDXIndexFile(String urlOrPath)
<a name="564" href="#564">564</a>     throws IOException, java.text.ParseException {
<a name="565" href="#565">565</a>     	ARCReader r = ARCReaderFactory.get(urlOrPath);
<a name="566" href="#566">566</a>     	r.setStrict(false);
<a name="567" href="#567">567</a>     	r.setParseHttpHeaders(<strong>true</strong>);
<a name="568" href="#568">568</a>     	r.setDigest(<strong>true</strong>);
<a name="569" href="#569">569</a>     	output(r, CDX_FILE);
<a name="570" href="#570">570</a>     }
<a name="571" href="#571">571</a>