surtprefixset.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="132" href="#132">132</a>                     <strong>new</strong> <a href="../../../org/archive/util/iterator/LineReadingIterator.html">LineReadingIterator</a>(reader),
<a name="133" href="#133">133</a>                     RegexpLineIterator.COMMENT_LINE,
<a name="134" href="#134">134</a>                     RegexpLineIterator.NONWHITESPACE_ENTRY_TRAILING_COMMENT,
<a name="135" href="#135">135</a>                     RegexpLineIterator.ENTRY);
<a name="136" href="#136">136</a> 
<a name="137" href="#137">137</a>         <strong>while</strong> (iter.hasNext()) {
<a name="138" href="#138">138</a>             s = (String) iter.next();
<a name="139" href="#139">139</a>             <em class="comment">// s is a URI (or even fragmentary hostname), not a SURT</em>
<a name="140" href="#140">140</a>             addFromPlain(s);
<a name="141" href="#141">141</a>         }
<a name="142" href="#142">142</a>     }
<a name="143" href="#143">143</a> 
<a name="144" href="#144">144</a>     <em>/**<em>*</em></em>
<a name="145" href="#145">145</a> <em>     * Import SURT prefixes from a reader with mixed URI and SURT prefix</em>
<a name="146" href="#146">146</a> <em>     * format. </em>
<a name="147" href="#147">147</a> <em>     * </em>
<a name="148" href="#148">148</a> <em>     * @param r  the reader to import the prefixes from</em>
<a name="149" href="#149">149</a> <em>     * @param deduceFromSeeds   true to also import SURT prefixes implied</em>
<a name="150" href="#150">150</a> <em>     *                          from normal URIs/hostname seeds</em>
<a name="151" href="#151">151</a> <em>     */</em>
<a name="152" href="#152">152</a>     <strong>public</strong> <strong>void</strong> importFromMixed(Reader r, <strong>boolean</strong> deduceFromSeeds) {
<a name="153" href="#153">153</a>         BufferedReader reader = <strong>new</strong> BufferedReader(r);
<a name="154" href="#154">154</a>         String s;
<a name="155" href="#155">155</a>         
<a name="156" href="#156">156</a>         Iterator iter = 
<a name="157" href="#157">157</a>             <strong>new</strong> <a href="../../../org/archive/util/iterator/RegexpLineIterator.html">RegexpLineIterator</a>(
<a name="158" href="#158">158</a>                     <strong>new</strong> <a href="../../../org/archive/util/iterator/LineReadingIterator.html">LineReadingIterator</a>(reader),
<a name="159" href="#159">159</a>                     RegexpLineIterator.COMMENT_LINE,
<a name="160" href="#160">160</a>                     RegexpLineIterator.NONWHITESPACE_ENTRY_TRAILING_COMMENT,
<a name="161" href="#161">161</a>                     RegexpLineIterator.ENTRY);
<a name="162" href="#162">162</a> 
<a name="163" href="#163">163</a>         <strong>while</strong> (iter.hasNext()) {
<a name="164" href="#164">164</a>             s = (String) iter.next();
<a name="165" href="#165">165</a>             <strong>if</strong>(s.startsWith(SURT_PREFIX_DIRECTIVE)) {
<a name="166" href="#166">166</a>                 <em class="comment">// it's specifically a SURT prefix line</em>
<a name="167" href="#167">167</a>                 String u = s.substring(SURT_PREFIX_DIRECTIVE.length()).trim();
<a name="168" href="#168">168</a>                 <strong>if</strong>(u.indexOf(<span class="string">"("</span>)>0) {
<a name="169" href="#169">169</a>                     <em class="comment">// formal SURT prefix; toLowerCase just in case</em>
<a name="170" href="#170">170</a>                     add(u.toLowerCase());
<a name="171" href="#171">171</a>                 } <strong>else</strong> {
<a name="172" href="#172">172</a>                     <em class="comment">// hostname/normal form URI from which </em>
<a name="173" href="#173">173</a>                     <em class="comment">// to deduce SURT prefix</em>
<a name="174" href="#174">174</a>                     addFromPlain(u);
<a name="175" href="#175">175</a>                 }
<a name="176" href="#176">176</a>                 
<a name="177" href="#177">177</a>                 <strong>continue</strong>; 
<a name="178" href="#178">178</a>             } <strong>else</strong> {
<a name="179" href="#179">179</a>                 <strong>if</strong>(deduceFromSeeds) {
<a name="180" href="#180">180</a>                     <em class="comment">// also deducing 'implied' SURT prefixes </em>
<a name="181" href="#181">181</a>                     <em class="comment">// from normal URIs/hostname seeds</em>
<a name="182" href="#182">182</a>                     addFromPlain(s);
<a name="183" href="#183">183</a>                 }
<a name="184" href="#184">184</a>             }
<a name="185" href="#185">185</a>         }
<a name="186" href="#186">186</a>     }
<a name="187" href="#187">187</a>     
<a name="188" href="#188">188</a>     <em>/**<em>*</em></em>
<a name="189" href="#189">189</a> <em>     * Given a plain URI or hostname, deduce an implied SURT prefix from</em>
<a name="190" href="#190">190</a> <em>     * it and add to active prefixes. </em>
<a name="191" href="#191">191</a> <em>     * </em>
<a name="192" href="#192">192</a> <em>     * @param u String of URI or hostname</em>
<a name="193" href="#193">193</a> <em>     */</em>
<a name="194" href="#194">194</a>     <strong>private</strong> <strong>void</strong> addFromPlain(String u) {
<a name="195" href="#195">195</a>         u = prefixFromPlain(u);
<a name="196" href="#196">196</a>         add(u);
<a name="197" href="#197">197</a>     }
<a name="198" href="#198">198</a> 
<a name="199" href="#199">199</a>     <em>/**<em>*</em></em>
<a name="200" href="#200">200</a> <em>     * Given a plain URI or hostname/hostname+path, deduce an implied SURT </em>
<a name="201" href="#201">201</a> <em>     * prefix from it. Results may be unpredictable on strings that cannot</em>
<a name="202" href="#202">202</a> <em>     * be interpreted as URIs. </em>
<a name="203" href="#203">203</a> <em>     * </em>
<a name="204" href="#204">204</a> <em>     * UURI 'fixup' is applied to the URI that is built. </em>
<a name="205" href="#205">205</a> <em>     *</em>
<a name="206" href="#206">206</a> <em>     * @param u URI or almost-URI to consider</em>
<a name="207" href="#207">207</a> <em>     * @return implied SURT prefix form</em>
<a name="208" href="#208">208</a> <em>     */</em>
<a name="209" href="#209">209</a>     <strong>public</strong> <strong>static</strong> String prefixFromPlain(String u) {
<a name="210" href="#210">210</a>         u = ArchiveUtils.addImpliedHttpIfNecessary(u);
<a name="211" href="#211">211</a>         u = coerceFromHttpsForComparison(u);
<a name="212" href="#212">212</a>         <strong>boolean</strong> trailingSlash = u.endsWith(<span class="string">"/"</span>);
<a name="213" href="#213">213</a>         <em class="comment">// ensure all typical UURI cleanup (incl. IDN-punycoding) is done</em>
<a name="214" href="#214">214</a>         <strong>try</strong> {
<a name="215" href="#215">215</a>             u = UURIFactory.getInstance(u).toString();
<a name="216" href="#216">216</a>         } <strong>catch</strong> (URIException e) {
<a name="217" href="#217">217</a>             e.printStackTrace();
<a name="218" href="#218">218</a>             <em class="comment">// allow to continue with original string uri</em>
<a name="219" href="#219">219</a>         }
<a name="220" href="#220">220</a>         <em class="comment">// except: don't let UURI-fixup add a trailing slash</em>
<a name="221" href="#221">221</a>         <em class="comment">// if it wasn't already there (presence or absence of</em>
<a name="222" href="#222">222</a>         <em class="comment">// such slash has special meaning specifying implied</em>
<a name="223" href="#223">223</a>         <em class="comment">// SURT prefixes)</em>
<a name="224" href="#224">224</a>         <strong>if</strong>(!trailingSlash &amp;&amp; u.endsWith(<span class="string">"/"</span>)) {
<a name="225" href="#225">225</a>             u = u.substring(0,u.length()-1);
<a name="226" href="#226">226</a>         }
<a name="227" href="#227">227</a>         <em class="comment">// convert to full SURT</em>
<a name="228" href="#228">228</a>         u = SURT.fromURI(u);
<a name="229" href="#229">229</a>         <em class="comment">// truncate to implied prefix</em>
<a name="230" href="#230">230</a>         u = SurtPrefixSet.asPrefix(u);
<a name="231" href="#231">231</a>         <strong>return</strong> u;
<a name="232" href="#232">232</a>     }
<a name="233" href="#233">233</a> 
<a name="234" href="#234">234</a>     <em>/**<em>*</em></em>
<a name="235" href="#235">235</a> <em>     * For SURT comparisons -- prefixes or candidates being checked against</em>
<a name="236" href="#236">236</a> <em>     * those prefixes -- we treat https URIs as if they were http.</em>
<a name="237" href="#237">237</a> <em>     * </em>
<a name="238" href="#238">238</a> <em>     * @param u string to coerce if it has https scheme</em>
<a name="239" href="#239">239</a> <em>     * @return string converted to http scheme, or original if not necessary</em>
<a name="240" href="#240">240</a> <em>     */</em>
<a name="241" href="#241">241</a>     <strong>private</strong> <strong>static</strong> String coerceFromHttpsForComparison(String u) {
<a name="242" href="#242">242</a>         <strong>if</strong> (u.startsWith(<span class="string">"https://"</span>)) {
<a name="243" href="#243">243</a>             u = <span class="string">"http"</span> + u.substring(<span class="string">"https"</span>.length());
<a name="244" href="#244">244</a>         }
<a name="245" href="#245">245</a>         <strong>return</strong> u;
<a name="246" href="#246">246</a>     }
<a name="247" href="#247">247</a> 
<a name="248" href="#248">248</a>     <em>/**<em>*</em></em>
<a name="249" href="#249">249</a> <em>     * Utility method for truncating a SURT that came from a </em>
<a name="250" href="#250">250</a> <em>     * full URI (as a seed, for example) into a prefix</em>
<a name="251" href="#251">251</a> <em>     * for determining inclusion.</em>
<a name="252" href="#252">252</a> <em>     * </em>
<a name="253" href="#253">253</a> <em>     * This involves: </em>
<a name="254" href="#254">254</a> <em>     * &lt;pre></em>
<a name="255" href="#255">255</a> <em>     *    (1) removing the last path component, if any</em>
<a name="256" href="#256">256</a> <em>     *        (anything after the last '/', if there are</em>
<a name="257" href="#257">257</a> <em>     *        at least 3 '/'s)</em>
<a name="258" href="#258">258</a> <em>     *    (2) removing a trailing ')', if present, opening</em>
<a name="259" href="#259">259</a> <em>     *        the possibility of proper subdomains. (This</em>
<a name="260" href="#260">260</a> <em>     *        means that the presence or absence of a</em>
<a name="261" href="#261">261</a> <em>     *        trailing '/' after a hostname in a seed list</em>
<a name="262" href="#262">262</a> <em>     *        is significant for the how the SURT prefix is </em>
<a name="263" href="#263">263</a> <em>     *        created, even though it is not signficant for </em>
<a name="264" href="#264">264</a> <em>     *        the URI's treatment as a seed.)</em>
<a name="265" href="#265">265</a> <em>     * &lt;/pre></em>
<a name="266" href="#266">266</a> <em>     *</em>
<a name="267" href="#267">267</a> <em>     * @param s String to work on.</em>
<a name="268" href="#268">268</a> <em>     * @return As prefix.</em>
<a name="269" href="#269">269</a> <em>     */</em>
<a name="270" href="#270">270</a>     <strong>private</strong> <strong>static</strong> String asPrefix(String s) {
<a name="271" href="#271">271</a>         <em class="comment">// Strip last path-segment, if more than 3 slashes</em>