experimentalwarcwriter.html

来自「网络爬虫开源代码」· HTML 代码 · 共 454 行 · 第 1/3 页

<a name="144" href="#144">144</a>     }
<a name="145" href="#145">145</a>     
<a name="146" href="#146">146</a>     <strong>protected</strong> String checkHeaderValue(<strong>final</strong> String value)
<a name="147" href="#147">147</a>     throws IOException {
<a name="148" href="#148">148</a>         <strong>for</strong> (<strong>int</strong> i = 0; i &lt; value.length(); i++) {
<a name="149" href="#149">149</a>         	<strong>final</strong> <strong>char</strong> c = value.<strong>char</strong>At(i);
<a name="150" href="#150">150</a>         	baseCharacterCheck(c, value);
<a name="151" href="#151">151</a>         	<strong>if</strong> (Character.isWhitespace(c)) {
<a name="152" href="#152">152</a>                 <strong>throw</strong> <strong>new</strong> IOException(<span class="string">"Contains disallowed white space 0x"</span> +
<a name="153" href="#153">153</a>                     Integer.toHexString(c) + <span class="string">": "</span> + value);
<a name="154" href="#154">154</a>         	}
<a name="155" href="#155">155</a>         }
<a name="156" href="#156">156</a>         <strong>return</strong> value;
<a name="157" href="#157">157</a>     }
<a name="158" href="#158">158</a>     
<a name="159" href="#159">159</a>     <strong>protected</strong> String checkHeaderLineMimetypeParameter(<strong>final</strong> String parameter)
<a name="160" href="#160">160</a>     throws IOException {
<a name="161" href="#161">161</a>     	StringBuilder sb = <strong>new</strong> StringBuilder(parameter.length());
<a name="162" href="#162">162</a>     	<strong>boolean</strong> wasWhitespace = false;
<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>char</strong> c = parameter.<strong>char</strong>At(i);
<a name="165" href="#165">165</a>         	<strong>if</strong> (Character.isWhitespace(c)) {
<a name="166" href="#166">166</a>         		<em class="comment">// Map all to ' ' and collapse multiples into one.</em>
<a name="167" href="#167">167</a>         		<em class="comment">// TODO: Make sure white space occurs in legal location --</em>
<a name="168" href="#168">168</a>         		<em class="comment">// before parameter or inside quoted-string.</em>
<a name="169" href="#169">169</a>         		<strong>if</strong> (wasWhitespace) {
<a name="170" href="#170">170</a>         			<strong>continue</strong>;
<a name="171" href="#171">171</a>         		}
<a name="172" href="#172">172</a>         		wasWhitespace = <strong>true</strong>;
<a name="173" href="#173">173</a>         		c = ' ';
<a name="174" href="#174">174</a>         	} <strong>else</strong> {
<a name="175" href="#175">175</a>         		wasWhitespace = false;
<a name="176" href="#176">176</a>         		baseCharacterCheck(c, parameter);
<a name="177" href="#177">177</a>         	}
<a name="178" href="#178">178</a>         	sb.append(c);
<a name="179" href="#179">179</a>         }
<a name="180" href="#180">180</a>         
<a name="181" href="#181">181</a>         <strong>return</strong> sb.toString();
<a name="182" href="#182">182</a>     }
<a name="183" href="#183">183</a> 
<a name="184" href="#184">184</a>     <strong>protected</strong> String createRecordHeader(<strong>final</strong> String type,
<a name="185" href="#185">185</a>     		<strong>final</strong> String url, <strong>final</strong> String create14DigitDate,
<a name="186" href="#186">186</a>     		<strong>final</strong> String mimetype, <strong>final</strong> URI recordId,
<a name="187" href="#187">187</a>     		<strong>final</strong> <a href="../../../../org/archive/util/anvl/ANVLRecord.html">ANVLRecord</a> xtraHeaders, <strong>final</strong> <strong>long</strong> contentLength)
<a name="188" href="#188">188</a>     throws IOException {
<a name="189" href="#189">189</a>     	<strong>final</strong> StringBuilder sb =
<a name="190" href="#190">190</a>     		<strong>new</strong> StringBuilder(2048<em class="comment">/*<em class="comment">A SWAG: TODO: Do analysis.*/</em>);</em>
<a name="191" href="#191">191</a>     	sb.append(WARC_ID).append(CRLF);
<a name="192" href="#192">192</a>         sb.append(HEADER_KEY_TYPE).append(COLON_SPACE).append(type).
<a name="193" href="#193">193</a>             append(CRLF);
<a name="194" href="#194">194</a>         <em class="comment">// Do not write a subject-uri if not one present.</em>
<a name="195" href="#195">195</a>         <strong>if</strong> (url != <strong>null</strong> &amp;&amp; url.length() > 0) {
<a name="196" href="#196">196</a>             sb.append(HEADER_KEY_URI).append(COLON_SPACE).
<a name="197" href="#197">197</a>                 append(checkHeaderValue(url)).append(CRLF);
<a name="198" href="#198">198</a>         }
<a name="199" href="#199">199</a>         sb.append(HEADER_KEY_DATE).append(COLON_SPACE).
<a name="200" href="#200">200</a>             append(create14DigitDate).append(CRLF);
<a name="201" href="#201">201</a>         <strong>if</strong> (xtraHeaders != <strong>null</strong>) {
<a name="202" href="#202">202</a>             <strong>for</strong> (<strong>final</strong> Iterator i = xtraHeaders.iterator(); i.hasNext();) {
<a name="203" href="#203">203</a>                 sb.append(i.next()).append(CRLF);
<a name="204" href="#204">204</a>             }
<a name="205" href="#205">205</a>         }
<a name="206" href="#206">206</a> 
<a name="207" href="#207">207</a>         <em class="comment">// TODO: Is MIME Version needed.</em>
<a name="208" href="#208">208</a>         sb.append(MIME_VERSION).append(CRLF);
<a name="209" href="#209">209</a>         sb.append(CONTENT_ID).append(COLON_SPACE).append('&lt;').
<a name="210" href="#210">210</a>             append(recordId.toString()).append('>').append(CRLF);
<a name="211" href="#211">211</a>         <strong>if</strong> (contentLength > 0) {
<a name="212" href="#212">212</a>             sb.append(CONTENT_TYPE).append(COLON_SPACE).append(
<a name="213" href="#213">213</a>                 checkHeaderLineMimetypeParameter(mimetype)).append(CRLF);
<a name="214" href="#214">214</a>         }
<a name="215" href="#215">215</a>         sb.append(CONTENT_LENGTH).append(COLON_SPACE).
<a name="216" href="#216">216</a>             append(Long.toString(contentLength)).append(CRLF);
<a name="217" href="#217">217</a>     	
<a name="218" href="#218">218</a>     	<strong>return</strong> sb.toString();
<a name="219" href="#219">219</a>     }
<a name="220" href="#220">220</a> 
<a name="221" href="#221">221</a>     <strong>protected</strong> <strong>void</strong> writeRecord(<strong>final</strong> String type, <strong>final</strong> String url,
<a name="222" href="#222">222</a>     		<strong>final</strong> String create14DigitDate, <strong>final</strong> String mimetype,
<a name="223" href="#223">223</a>     		<strong>final</strong> URI recordId, <a href="../../../../org/archive/util/anvl/ANVLRecord.html">ANVLRecord</a> xtraHeaders,
<a name="224" href="#224">224</a>             <strong>final</strong> InputStream contentStream, <strong>final</strong> <strong>long</strong> contentLength)
<a name="225" href="#225">225</a>     throws IOException {
<a name="226" href="#226">226</a>     	<strong>if</strong> (!TYPES_LIST.contains(type)) {
<a name="227" href="#227">227</a>     		<strong>throw</strong> <strong>new</strong> IllegalArgumentException(<span class="string">"Unknown record type: "</span> + type);
<a name="228" href="#228">228</a>     	}
<a name="229" href="#229">229</a>     	<strong>if</strong> (contentLength == 0 &amp;&amp;
<a name="230" href="#230">230</a>                 (xtraHeaders == <strong>null</strong> || xtraHeaders.size() &lt;= 0)) {
<a name="231" href="#231">231</a>     		<strong>throw</strong> <strong>new</strong> IllegalArgumentException(<span class="string">"Cannot write record "</span> +
<a name="232" href="#232">232</a>     		    <span class="string">"of content-length zero and base headers only."</span>);
<a name="233" href="#233">233</a>     	}
<a name="234" href="#234">234</a>     	
<a name="235" href="#235">235</a>         preWriteRecordTasks();
<a name="236" href="#236">236</a>         <strong>try</strong> {
<a name="237" href="#237">237</a>             <strong>final</strong> String header = createRecordHeader(type, url,
<a name="238" href="#238">238</a>             	create14DigitDate, mimetype, recordId, xtraHeaders,
<a name="239" href="#239">239</a>             	contentLength);
<a name="240" href="#240">240</a>             <em class="comment">// TODO: Revisit endcoding of header.</em>
<a name="241" href="#241">241</a>             write(header.getBytes(WARC_HEADER_ENCODING));
<a name="242" href="#242">242</a>             
<a name="243" href="#243">243</a>             <strong>if</strong> (contentStream != <strong>null</strong> &amp;&amp; contentLength > 0) {
<a name="244" href="#244">244</a>                 <em class="comment">// Write out the header/body separator.</em>
<a name="245" href="#245">245</a>                 write(CRLF_BYTES); <em class="comment">// TODO: should this be written even for zero-length?</em>
<a name="246" href="#246">246</a>             	readToLimitFrom(contentStream, contentLength, <strong>this</strong>.readbuffer);
<a name="247" href="#247">247</a>             }
<a name="248" href="#248">248</a>             
<a name="249" href="#249">249</a>             <em class="comment">// Write out the two blank lines at end of all records.</em>
<a name="250" href="#250">250</a>             <em class="comment">// TODO: Why? Messes up skipping through file. Also not in grammar.</em>
<a name="251" href="#251">251</a>             write(CRLF_BYTES);
<a name="252" href="#252">252</a>             write(CRLF_BYTES);
<a name="253" href="#253">253</a>         } <strong>finally</strong> {
<a name="254" href="#254">254</a>             postWriteRecordTasks();
<a name="255" href="#255">255</a>         }
<a name="256" href="#256">256</a>     }
<a name="257" href="#257">257</a>     
<a name="258" href="#258">258</a>     <strong>protected</strong> URI generateRecordId(<strong>final</strong> Map&lt;String, String> qualifiers)
<a name="259" href="#259">259</a>     throws IOException {
<a name="260" href="#260">260</a>     	URI rid = <strong>null</strong>;
<a name="261" href="#261">261</a>     	<strong>try</strong> {
<a name="262" href="#262">262</a>     		rid = GeneratorFactory.getFactory().
<a name="263" href="#263">263</a>     			getQualifiedRecordID(qualifiers);
<a name="264" href="#264">264</a>     	} <strong>catch</strong> (URISyntaxException e) {
<a name="265" href="#265">265</a>     		<em class="comment">// Convert to IOE so can let it out.</em>
<a name="266" href="#266">266</a>     		<strong>throw</strong> <strong>new</strong> IOException(e.getMessage());
<a name="267" href="#267">267</a>     	}
<a name="268" href="#268">268</a>     	<strong>return</strong> rid;
<a name="269" href="#269">269</a>     }
<a name="270" href="#270">270</a>     
<a name="271" href="#271">271</a>     <strong>protected</strong> URI generateRecordId(<strong>final</strong> String key, <strong>final</strong> String value)
<a name="272" href="#272">272</a>     throws IOException {
<a name="273" href="#273">273</a>     	URI rid = <strong>null</strong>;
<a name="274" href="#274">274</a>     	<strong>try</strong> {
<a name="275" href="#275">275</a>     		rid = GeneratorFactory.getFactory().
<a name="276" href="#276">276</a>     			getQualifiedRecordID(key, value);
<a name="277" href="#277">277</a>     	} <strong>catch</strong> (URISyntaxException e) {
<a name="278" href="#278">278</a>     		<em class="comment">// Convert to IOE so can let it out.</em>
<a name="279" href="#279">279</a>     		<strong>throw</strong> <strong>new</strong> IOException(e.getMessage());
<a name="280" href="#280">280</a>     	}
<a name="281" href="#281">281</a>     	<strong>return</strong> rid;
<a name="282" href="#282">282</a>     }
<a name="283" href="#283">283</a>     
<a name="284" href="#284">284</a>     <strong>public</strong> URI writeWarcinfoRecord(String filename)
<a name="285" href="#285">285</a> 	throws IOException {
<a name="286" href="#286">286</a>     	<strong>return</strong> writeWarcinfoRecord(filename, <strong>null</strong>);
<a name="287" href="#287">287</a>     }
<a name="288" href="#288">288</a>     
<a name="289" href="#289">289</a>     <strong>public</strong> URI writeWarcinfoRecord(String filename, <strong>final</strong> String description)
<a name="290" href="#290">290</a>         	throws IOException {
<a name="291" href="#291">291</a>         <em class="comment">// Strip .open suffix if present.</em>
<a name="292" href="#292">292</a>         <strong>if</strong> (filename.endsWith(WriterPoolMember.OCCUPIED_SUFFIX)) {
<a name="293" href="#293">293</a>         	filename = filename.substring(0,
<a name="294" href="#294">294</a>         		filename.length() - WriterPoolMember.OCCUPIED_SUFFIX.length());
<a name="295" href="#295">295</a>         }