warcrecord.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="230" href="#230">230</a>             <strong>public</strong> String toString() {
<a name="231" href="#231">231</a>                 <strong>return</strong> <strong>this</strong>.fields.toString();
<a name="232" href="#232">232</a>             }
<a name="233" href="#233">233</a>     	};
<a name="234" href="#234">234</a>     }
<a name="235" href="#235">235</a>     
<a name="236" href="#236">236</a>     <strong>protected</strong> <strong>int</strong> parseHeaderLine(<strong>final</strong> InputStream in,
<a name="237" href="#237">237</a>             <strong>final</strong> Map&lt;Object, Object> fields, <strong>final</strong> <strong>boolean</strong> strict) 
<a name="238" href="#238">238</a>     throws IOException {
<a name="239" href="#239">239</a>         byte [] line = readLine(in, strict);
<a name="240" href="#240">240</a>         <strong>if</strong> (line.length &lt;= 2) {
<a name="241" href="#241">241</a>             <strong>throw</strong> <strong>new</strong> IOException(<span class="string">"No Header Line found"</span>);
<a name="242" href="#242">242</a>         }
<a name="243" href="#243">243</a>         <em class="comment">// Strip the CRLF.</em>
<a name="244" href="#244">244</a>         String headerLine = <strong>new</strong> String(line, 0, line.length - 2,
<a name="245" href="#245">245</a>             HEADER_LINE_ENCODING);
<a name="246" href="#246">246</a>         Matcher m = HEADER_LINE.matcher(headerLine);
<a name="247" href="#247">247</a>         <strong>if</strong> (!m.matches()) {
<a name="248" href="#248">248</a>             <strong>throw</strong> <strong>new</strong> IOException(<span class="string">"Failed parse of Header Line: "</span> +
<a name="249" href="#249">249</a>                 headerLine);
<a name="250" href="#250">250</a>         }
<a name="251" href="#251">251</a>         <strong>for</strong> (<strong>int</strong> i = 0; i &lt; HEADER_FIELD_KEYS.length; i++) {
<a name="252" href="#252">252</a>             <strong>if</strong> (i == 1) {
<a name="253" href="#253">253</a>                 <em class="comment">// Do length of Record as a Long.</em>
<a name="254" href="#254">254</a>                 fields.put(HEADER_FIELD_KEYS[i],
<a name="255" href="#255">255</a>                     Long.parseLong(m.group(i + 1)));
<a name="256" href="#256">256</a>                 <strong>continue</strong>;
<a name="257" href="#257">257</a>             }
<a name="258" href="#258">258</a>             fields.put(HEADER_FIELD_KEYS[i], m.group(i + 1));
<a name="259" href="#259">259</a>         }
<a name="260" href="#260">260</a>         
<a name="261" href="#261">261</a>         <strong>return</strong> line.length;
<a name="262" href="#262">262</a>     }
<a name="263" href="#263">263</a> 
<a name="264" href="#264">264</a>     <em>/**<em>*</em></em>
<a name="265" href="#265">265</a> <em>     * Read a line.</em>
<a name="266" href="#266">266</a> <em>     * A 'line' in this context ends in CRLF and contains ascii-only and no</em>
<a name="267" href="#267">267</a> <em>     * control-characters.</em>
<a name="268" href="#268">268</a> <em>     * @param in InputStream to read.</em>
<a name="269" href="#269">269</a> <em>     * @param strict Strict parsing (If false, we'll eat whitespace before the</em>
<a name="270" href="#270">270</a> <em>     * record.</em>
<a name="271" href="#271">271</a> <em>     * @return All bytes in line including terminating CRLF.</em>
<a name="272" href="#272">272</a> <em>     * @throws IOException</em>
<a name="273" href="#273">273</a> <em>     */</em>
<a name="274" href="#274">274</a>     <strong>protected</strong> byte [] readLine(<strong>final</strong> InputStream in, <strong>final</strong> <strong>boolean</strong> strict) 
<a name="275" href="#275">275</a>     throws IOException {
<a name="276" href="#276">276</a>         <strong>boolean</strong> done = false;
<a name="277" href="#277">277</a>         <strong>boolean</strong> recordStart = strict;
<a name="278" href="#278">278</a>         <strong>int</strong> read = 0;
<a name="279" href="#279">279</a>         ByteArrayOutputStream baos = <strong>new</strong> ByteArrayOutputStream(1024 <em class="comment">/*<em class="comment">SWAG*/</em>);</em>
<a name="280" href="#280">280</a>         <strong>for</strong> (<strong>int</strong> c  = -1, previousCharacter; !done;) {
<a name="281" href="#281">281</a>             <strong>if</strong> (read++ >= MAX_LINE_LENGTH) {
<a name="282" href="#282">282</a>                 <strong>throw</strong> <strong>new</strong> IOException(<span class="string">"Read "</span> + MAX_LINE_LENGTH +
<a name="283" href="#283">283</a>                     <span class="string">" bytes without finding CRLF"</span>);
<a name="284" href="#284">284</a>             }
<a name="285" href="#285">285</a>             previousCharacter = c;
<a name="286" href="#286">286</a>             c = in.read();
<a name="287" href="#287">287</a>             <strong>if</strong> (c == -1) {
<a name="288" href="#288">288</a>                 <strong>throw</strong> <strong>new</strong> IOException(<span class="string">"End-Of-Stream before CRLF:\n"</span> +
<a name="289" href="#289">289</a>                     <strong>new</strong> String(baos.toByteArray()));
<a name="290" href="#290">290</a>             }
<a name="291" href="#291">291</a>             <strong>if</strong> (isLF((<strong>char</strong>)c) &amp;&amp; isCR((<strong>char</strong>)previousCharacter)) {
<a name="292" href="#292">292</a>                 done = <strong>true</strong>;
<a name="293" href="#293">293</a>             } <strong>else</strong> <strong>if</strong> (!recordStart &amp;&amp; Character.isWhitespace(c)) {
<a name="294" href="#294">294</a>                 <em class="comment">// Skip any whitespace at start.</em>
<a name="295" href="#295">295</a>                 <strong>continue</strong>;
<a name="296" href="#296">296</a>             } <strong>else</strong> {
<a name="297" href="#297">297</a>                 <strong>if</strong> (isCR((<strong>char</strong>)previousCharacter)) {
<a name="298" href="#298">298</a>                     <em class="comment">// If previous character was a CR and this character is not</em>
<a name="299" href="#299">299</a>                     <em class="comment">// a LF, we tested above, thats illegal.</em>
<a name="300" href="#300">300</a>                     <strong>throw</strong> <strong>new</strong> IOException(<span class="string">"CR in middle of Header:\n"</span> +
<a name="301" href="#301">301</a>                         <strong>new</strong> String(baos.toByteArray()));
<a name="302" href="#302">302</a>                 }
<a name="303" href="#303">303</a>                 
<a name="304" href="#304">304</a>                 <em class="comment">// Not whitespace so start record if we haven't already.</em>
<a name="305" href="#305">305</a>                 <strong>if</strong> (!recordStart) {
<a name="306" href="#306">306</a>                     recordStart = <strong>true</strong>;
<a name="307" href="#307">307</a>                 }
<a name="308" href="#308">308</a>             }
<a name="309" href="#309">309</a>             baos.write(c);
<a name="310" href="#310">310</a>         }
<a name="311" href="#311">311</a>         <strong>return</strong> baos.toByteArray();
<a name="312" href="#312">312</a>     }
<a name="313" href="#313">313</a>  
<a name="314" href="#314">314</a>     <strong>protected</strong> <strong>void</strong> parseNamedFields(<strong>final</strong> InputStream in,
<a name="315" href="#315">315</a>         <strong>final</strong> Map&lt;Object, Object> fields) 
<a name="316" href="#316">316</a>     throws IOException {
<a name="317" href="#317">317</a>         <a href="../../../../org/archive/util/anvl/ANVLRecord.html">ANVLRecord</a> r = ANVLRecord.load(in);
<a name="318" href="#318">318</a>         fields.putAll(r.asMap());
<a name="319" href="#319">319</a>     }
<a name="320" href="#320">320</a>     
<a name="321" href="#321">321</a>     <strong>public</strong> <strong>static</strong> <strong>boolean</strong> isCROrLF(<strong>final</strong> <strong>char</strong> c) {
<a name="322" href="#322">322</a>         <strong>return</strong> isCR(c) || isLF(c);
<a name="323" href="#323">323</a>     }
<a name="324" href="#324">324</a>     
<a name="325" href="#325">325</a>     <strong>public</strong> <strong>static</strong> <strong>boolean</strong> isCR(<strong>final</strong> <strong>char</strong> c) {
<a name="326" href="#326">326</a>         <strong>return</strong> c == CRLF.charAt(0);
<a name="327" href="#327">327</a>     }
<a name="328" href="#328">328</a>     
<a name="329" href="#329">329</a>     <strong>public</strong> <strong>static</strong> <strong>boolean</strong> isLF(<strong>final</strong> <strong>char</strong> c) {
<a name="330" href="#330">330</a>         <strong>return</strong> c == CRLF.charAt(1);
<a name="331" href="#331">331</a>     }
<a name="332" href="#332">332</a>     
<a name="333" href="#333">333</a>     
<a name="334" href="#334">334</a>     @Override
<a name="335" href="#335">335</a>     <strong>protected</strong> String getMimetype4Cdx(<a href="../../../../org/archive/io/ArchiveRecordHeader.html">ArchiveRecordHeader</a> h) {
<a name="336" href="#336">336</a>         <strong>final</strong> String m = <strong>super</strong>.getMimetype4Cdx(h);
<a name="337" href="#337">337</a>         <em class="comment">// Mimetypes can have spaces in WARCs.  Emitting for CDX, just</em>
<a name="338" href="#338">338</a>         <em class="comment">// squash them for now.  Later, quote them since squashing spaces won't</em>
<a name="339" href="#339">339</a>         <em class="comment">// work for params that have quoted-string values.</em>
<a name="340" href="#340">340</a>         Matcher matcher = WHITESPACE.matcher(m);
<a name="341" href="#341">341</a>         <strong>return</strong> matcher.replaceAll(<span class="string">""</span>);
<a name="342" href="#342">342</a>     }
<a name="343" href="#343">343</a> }
</pre>
<hr/><div id="footer">This page was automatically generated by <a href="http://maven.apache.org/">Maven</a></div></body>
</html>