experimentalwarcwriter.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="148" href="#148">148</a>     	writeWarcinfoRecord(filename);
<a name="149" href="#149">149</a>         <strong>return</strong> filename;
<a name="150" href="#150">150</a>     }
<a name="151" href="#151">151</a>     
<a name="152" href="#152">152</a>     <strong>protected</strong> <strong>void</strong> baseCharacterCheck(<strong>final</strong> <strong>char</strong> c, <strong>final</strong> String parameter)
<a name="153" href="#153">153</a>     throws IOException {
<a name="154" href="#154">154</a>         <em class="comment">// TODO: Too strict?  UNICODE control characters?</em>
<a name="155" href="#155">155</a>         <strong>if</strong> (Character.isISOControl(c) || !Character.isValidCodePoint(c)) {
<a name="156" href="#156">156</a>             <strong>throw</strong> <strong>new</strong> IOException(<span class="string">"Contains illegal character 0x"</span> +
<a name="157" href="#157">157</a>                 Integer.toHexString(c) + <span class="string">": "</span> + parameter);
<a name="158" href="#158">158</a>         }
<a name="159" href="#159">159</a>     }
<a name="160" href="#160">160</a>     
<a name="161" href="#161">161</a>     <strong>protected</strong> String checkHeaderLineParameters(<strong>final</strong> String parameter)
<a name="162" href="#162">162</a>     throws IOException {
<a name="163" href="#163">163</a>         <strong>for</strong> (<strong>int</strong> i = 0; i &lt; parameter.length(); i++) {
<a name="164" href="#164">164</a>         	<strong>final</strong> <strong>char</strong> c = parameter.<strong>char</strong>At(i);
<a name="165" href="#165">165</a>         	baseCharacterCheck(c, parameter);
<a name="166" href="#166">166</a>         	<strong>if</strong> (Character.isWhitespace(c)) {
<a name="167" href="#167">167</a>                 <strong>throw</strong> <strong>new</strong> IOException(<span class="string">"Contains disallowed white space 0x"</span> +
<a name="168" href="#168">168</a>                     Integer.toHexString(c) + <span class="string">": "</span> + parameter);
<a name="169" href="#169">169</a>         	}
<a name="170" href="#170">170</a>         }
<a name="171" href="#171">171</a>         <strong>return</strong> parameter;
<a name="172" href="#172">172</a>     }
<a name="173" href="#173">173</a>     
<a name="174" href="#174">174</a>     <strong>protected</strong> String checkHeaderLineMimetypeParameter(<strong>final</strong> String parameter)
<a name="175" href="#175">175</a>     throws IOException {
<a name="176" href="#176">176</a>     	StringBuilder sb = <strong>new</strong> StringBuilder(parameter.length());
<a name="177" href="#177">177</a>     	<strong>boolean</strong> wasWhitespace = false;
<a name="178" href="#178">178</a>         <strong>for</strong> (<strong>int</strong> i = 0; i &lt; parameter.length(); i++) {
<a name="179" href="#179">179</a>         	<strong>char</strong> c = parameter.<strong>char</strong>At(i);
<a name="180" href="#180">180</a>         	<strong>if</strong> (Character.isWhitespace(c)) {
<a name="181" href="#181">181</a>         		<em class="comment">// Map all to ' ' and collapse multiples into one.</em>
<a name="182" href="#182">182</a>         		<em class="comment">// TODO: Make sure white space occurs in legal location --</em>
<a name="183" href="#183">183</a>         		<em class="comment">// before parameter or inside quoted-string.</em>
<a name="184" href="#184">184</a>         		<strong>if</strong> (wasWhitespace) {
<a name="185" href="#185">185</a>         			<strong>continue</strong>;
<a name="186" href="#186">186</a>         		}
<a name="187" href="#187">187</a>         		wasWhitespace = <strong>true</strong>;
<a name="188" href="#188">188</a>         		c = ' ';
<a name="189" href="#189">189</a>         	} <strong>else</strong> {
<a name="190" href="#190">190</a>         		wasWhitespace = false;
<a name="191" href="#191">191</a>         		baseCharacterCheck(c, parameter);
<a name="192" href="#192">192</a>         	}
<a name="193" href="#193">193</a>         	sb.append(c);
<a name="194" href="#194">194</a>         }
<a name="195" href="#195">195</a>         
<a name="196" href="#196">196</a>         <strong>return</strong> sb.toString();
<a name="197" href="#197">197</a>     }
<a name="198" href="#198">198</a> 
<a name="199" href="#199">199</a>     <strong>protected</strong> byte [] createRecordHeaderline(<strong>final</strong> String type,
<a name="200" href="#200">200</a>     		<strong>final</strong> String url, <strong>final</strong> String create14DigitDate,
<a name="201" href="#201">201</a>     		<strong>final</strong> String mimetype, <strong>final</strong> URI recordId,
<a name="202" href="#202">202</a>     		<strong>final</strong> <strong>int</strong> namedFieldsLength, <strong>final</strong> <strong>long</strong> contentLength)
<a name="203" href="#203">203</a>     throws IOException {
<a name="204" href="#204">204</a>     	<strong>final</strong> StringBuilder sb =
<a name="205" href="#205">205</a>     		<strong>new</strong> StringBuilder(2048<em class="comment">/*<em class="comment">A SWAG: TODO: Do analysis.*/</em>);</em>
<a name="206" href="#206">206</a>     	sb.append(WARC_ID);
<a name="207" href="#207">207</a>     	sb.append(HEADER_FIELD_SEPARATOR);
<a name="208" href="#208">208</a>     	sb.append(PLACEHOLDER_RECORD_LENGTH_STRING);
<a name="209" href="#209">209</a>     	sb.append(HEADER_FIELD_SEPARATOR);
<a name="210" href="#210">210</a>     	sb.append(type);
<a name="211" href="#211">211</a>     	sb.append(HEADER_FIELD_SEPARATOR);
<a name="212" href="#212">212</a>     	sb.append(checkHeaderLineParameters(url));
<a name="213" href="#213">213</a>     	sb.append(HEADER_FIELD_SEPARATOR);
<a name="214" href="#214">214</a>     	sb.append(checkHeaderLineParameters(create14DigitDate));
<a name="215" href="#215">215</a>     	sb.append(HEADER_FIELD_SEPARATOR);
<a name="216" href="#216">216</a>     	<em class="comment">// 0.9 of spec. has mimetype second-to-last and recordid last on</em>
<a name="217" href="#217">217</a>     	<em class="comment">// header line.  Here we swap their positions and allow writing</em>
<a name="218" href="#218">218</a>     	<em class="comment">// of full mimetypes rather than the curtailed type we used write into</em>
<a name="219" href="#219">219</a>     	<em class="comment">// ARCs.  These two deviations to be proposed as amendments to spec 0.9.</em>
<a name="220" href="#220">220</a>     	sb.append(checkHeaderLineParameters(recordId.toString()));
<a name="221" href="#221">221</a>     	sb.append(HEADER_FIELD_SEPARATOR);
<a name="222" href="#222">222</a>     	sb.append(checkHeaderLineMimetypeParameter(mimetype));
<a name="223" href="#223">223</a>         <em class="comment">// Add terminating CRLF.</em>
<a name="224" href="#224">224</a>         sb.append(CRLF);
<a name="225" href="#225">225</a>     	
<a name="226" href="#226">226</a>     	<strong>long</strong> length = sb.length() + namedFieldsLength + contentLength;
<a name="227" href="#227">227</a>     	
<a name="228" href="#228">228</a>     	<em class="comment">// Insert length and pad out to fixed width with zero prefix to</em>
<a name="229" href="#229">229</a>         <em class="comment">// highlight 'fixed-widthness' of length.</em>
<a name="230" href="#230">230</a>     	<strong>int</strong> start = WARC_ID.length() + 1 <em class="comment">/*<em class="comment">HEADER_FIELD_SEPARATOR */</em>;</em>
<a name="231" href="#231">231</a>         <strong>int</strong> end = start + PLACEHOLDER_RECORD_LENGTH_STRING.length();
<a name="232" href="#232">232</a>     	String lenStr = RECORD_LENGTH_FORMATTER.format(length);
<a name="233" href="#233">233</a>     	sb.replace(start, end, lenStr);
<a name="234" href="#234">234</a> 
<a name="235" href="#235">235</a>         <strong>return</strong> sb.toString().getBytes(HEADER_LINE_ENCODING);
<a name="236" href="#236">236</a>     }
<a name="237" href="#237">237</a> 
<a name="238" href="#238">238</a>     <strong>protected</strong> <strong>void</strong> writeRecord(<strong>final</strong> String type, <strong>final</strong> String url,
<a name="239" href="#239">239</a>     		<strong>final</strong> String create14DigitDate, <strong>final</strong> String mimetype,
<a name="240" href="#240">240</a>     		<strong>final</strong> URI recordId, <a href="../../../../org/archive/util/anvl/ANVLRecord.html">ANVLRecord</a> namedFields,
<a name="241" href="#241">241</a>             <strong>final</strong> InputStream contentStream, <strong>final</strong> <strong>long</strong> contentLength)
<a name="242" href="#242">242</a>     throws IOException {
<a name="243" href="#243">243</a>     	<strong>if</strong> (!TYPES_LIST.contains(type)) {
<a name="244" href="#244">244</a>     		<strong>throw</strong> <strong>new</strong> IllegalArgumentException(<span class="string">"Unknown record type: "</span> + type);
<a name="245" href="#245">245</a>     	}
<a name="246" href="#246">246</a>     	<strong>if</strong> (contentLength == 0 &amp;&amp;
<a name="247" href="#247">247</a>                 (namedFields == <strong>null</strong> || namedFields.size() &lt;= 0)) {
<a name="248" href="#248">248</a>     		<strong>throw</strong> <strong>new</strong> IllegalArgumentException(<span class="string">"Cannot have a record made "</span> +
<a name="249" href="#249">249</a>     		    <span class="string">"of a Header line only (Content and Named Fields are empty)."</span>);
<a name="250" href="#250">250</a>     	}
<a name="251" href="#251">251</a>     	
<a name="252" href="#252">252</a>         preWriteRecordTasks();
<a name="253" href="#253">253</a>         <strong>try</strong> {
<a name="254" href="#254">254</a>         	<strong>if</strong> (namedFields == <strong>null</strong>) {
<a name="255" href="#255">255</a>         		<em class="comment">// Use the empty anvl record so the length of blank line on</em>
<a name="256" href="#256">256</a>         		<em class="comment">// end gets counted as part of the record length.</em>
<a name="257" href="#257">257</a>         		namedFields = ANVLRecord.EMPTY_ANVL_RECORD;
<a name="258" href="#258">258</a>         	}
<a name="259" href="#259">259</a>         	
<a name="260" href="#260">260</a>         	<em class="comment">// Serialize metadata first so we have metadata length.</em>
<a name="261" href="#261">261</a>         	<strong>final</strong> byte [] namedFieldsBlock = namedFields.getUTF8Bytes();
<a name="262" href="#262">262</a>         	<em class="comment">// Now serialize the Header line.</em>
<a name="263" href="#263">263</a>             <strong>final</strong> byte [] header = createRecordHeaderline(type, url,
<a name="264" href="#264">264</a>             	create14DigitDate, mimetype, recordId, namedFieldsBlock.length,
<a name="265" href="#265">265</a>             	contentLength);
<a name="266" href="#266">266</a>             write(header);
<a name="267" href="#267">267</a>             write(namedFieldsBlock);
<a name="268" href="#268">268</a>             <strong>if</strong> (contentStream != <strong>null</strong> &amp;&amp; contentLength > 0) {
<a name="269" href="#269">269</a>             	readFullyFrom(contentStream, contentLength, <strong>this</strong>.readbuffer);
<a name="270" href="#270">270</a>             }
<a name="271" href="#271">271</a>             
<a name="272" href="#272">272</a>             <em class="comment">// Write out the two blank lines at end of all records.</em>
<a name="273" href="#273">273</a>             <em class="comment">// TODO: Why? Messes up skipping through file. Also not in grammar.</em>
<a name="274" href="#274">274</a>             write(CRLF_BYTES);
<a name="275" href="#275">275</a>             write(CRLF_BYTES);
<a name="276" href="#276">276</a>         } <strong>finally</strong> {
<a name="277" href="#277">277</a>             postWriteRecordTasks();
<a name="278" href="#278">278</a>         }
<a name="279" href="#279">279</a>     }
<a name="280" href="#280">280</a>     
<a name="281" href="#281">281</a>     <strong>protected</strong> URI generateRecordId(<strong>final</strong> Map&lt;String, String> qualifiers)
<a name="282" href="#282">282</a>     throws IOException {
<a name="283" href="#283">283</a>     	URI rid = <strong>null</strong>;
<a name="284" href="#284">284</a>     	<strong>try</strong> {
<a name="285" href="#285">285</a>     		rid = GeneratorFactory.getFactory().
<a name="286" href="#286">286</a>     			getQualifiedRecordID(qualifiers);
<a name="287" href="#287">287</a>     	} <strong>catch</strong> (URISyntaxException e) {
<a name="288" href="#288">288</a>     		<em class="comment">// Convert to IOE so can let it out.</em>
<a name="289" href="#289">289</a>     		<strong>throw</strong> <strong>new</strong> IOException(e.getMessage());
<a name="290" href="#290">290</a>     	}
<a name="291" href="#291">291</a>     	<strong>return</strong> rid;
<a name="292" href="#292">292</a>     }
<a name="293" href="#293">293</a>     
<a name="294" href="#294">294</a>     <strong>protected</strong> URI generateRecordId(<strong>final</strong> String key, <strong>final</strong> String value)
<a name="295" href="#295">295</a>     throws IOException {
<a name="296" href="#296">296</a>     	URI rid = <strong>null</strong>;
<a name="297" href="#297">297</a>     	<strong>try</strong> {
<a name="298" href="#298">298</a>     		rid = GeneratorFactory.getFactory().
<a name="299" href="#299">299</a>     			getQualifiedRecordID(key, value);
<a name="300" href="#300">300</a>     	} <strong>catch</strong> (URISyntaxException e) {
<a name="301" href="#301">301</a>     		<em class="comment">// Convert to IOE so can let it out.</em>
<a name="302" href="#302">302</a>     		<strong>throw</strong> <strong>new</strong> IOException(e.getMessage());
<a name="303" href="#303">303</a>     	}