piecetable.html

来自「网络爬虫开源代码」· HTML 代码 · 共 298 行 · 第 1/2 页

<a name="141" href="#141">141</a>     <em>/**<em>*</em></em>
<a name="142" href="#142">142</a> <em>     * Wraps the raw table stream.  This is used to create the charPos and</em>
<a name="143" href="#143">143</a> <em>     * filePos streams.  The streams that this method returns are "safe",</em>
<a name="144" href="#144">144</a> <em>     * meaning that the charPos and filePos position() fields never clobber</em>
<a name="145" href="#145">145</a> <em>     * each other.  They are buffered, meaning that up to &lt;i>n&lt;/i> elements</em>
<a name="146" href="#146">146</a> <em>     * can be read before the disk is accessed again.  And they are "origined",</em>
<a name="147" href="#147">147</a> <em>     * meaning result.position(0) actually positions the stream at the </em>
<a name="148" href="#148">148</a> <em>     * beginning of the piece table array, not the beginning of the file.</em>
<a name="149" href="#149">149</a> <em>     * </em>
<a name="150" href="#150">150</a> <em>     * @param input   the stream to wrap</em>
<a name="151" href="#151">151</a> <em>     * @param pos     the origin for the returned stream</em>
<a name="152" href="#152">152</a> <em>     * @param cache   the number of bytes for the returned stream to buffer</em>
<a name="153" href="#153">153</a> <em>     * @return   the wrapped stream</em>
<a name="154" href="#154">154</a> <em>     * @throws IOException  if an IO error occurs</em>
<a name="155" href="#155">155</a> <em>     */</em>
<a name="156" href="#156">156</a>     <strong>private</strong> <a href="../../../../org/archive/io/SeekInputStream.html">SeekInputStream</a> wrap(<a href="../../../../org/archive/io/SeekInputStream.html">SeekInputStream</a> input, <strong>long</strong> pos, <strong>int</strong> cache) 
<a name="157" href="#157">157</a>     throws IOException {
<a name="158" href="#158">158</a>         input.position(pos);
<a name="159" href="#159">159</a>         <a href="../../../../org/archive/io/SeekInputStream.html">SeekInputStream</a> r = <strong>new</strong> <a href="../../../../org/archive/io/SafeSeekInputStream.html">SafeSeekInputStream</a>(input);
<a name="160" href="#160">160</a>         r = <strong>new</strong> <a href="../../../../org/archive/io/OriginSeekInputStream.html">OriginSeekInputStream</a>(r, pos);
<a name="161" href="#161">161</a>         r = <strong>new</strong> <a href="../../../../org/archive/io/BufferedSeekInputStream.html">BufferedSeekInputStream</a>(r, cache);
<a name="162" href="#162">162</a>         <strong>return</strong> r;
<a name="163" href="#163">163</a>     }
<a name="164" href="#164">164</a>     
<a name="165" href="#165">165</a>     
<a name="166" href="#166">166</a>     <em>/**<em>*</em></em>
<a name="167" href="#167">167</a> <em>     * Skips over any property information that may precede a piece table.</em>
<a name="168" href="#168">168</a> <em>     * These property structures contain stylesheet information that applies</em>
<a name="169" href="#169">169</a> <em>     * to the piece table.  Since we're only interested in the text itself,</em>
<a name="170" href="#170">170</a> <em>     * we just ignore this property stuff.  (I suppose a third buffered</em>
<a name="171" href="#171">171</a> <em>     * stream could be used to add style information to {@link Piece}, but</em>
<a name="172" href="#172">172</a> <em>     * we don't need it.)</em>
<a name="173" href="#173">173</a> <em>     * </em>
<a name="174" href="#174">174</a> <em>     * @param input  the input stream containing the piece table</em>
<a name="175" href="#175">175</a> <em>     * @throws IOException  if an IO error occurs</em>
<a name="176" href="#176">176</a> <em>     */</em>
<a name="177" href="#177">177</a>     <strong>private</strong> <strong>static</strong> <strong>void</strong> skipProperties(<a href="../../../../org/archive/io/SeekInputStream.html">SeekInputStream</a> input) throws IOException {
<a name="178" href="#178">178</a>         <strong>int</strong> tag = input.read();
<a name="179" href="#179">179</a>         <strong>while</strong> (tag == 1) {
<a name="180" href="#180">180</a>             <strong>int</strong> size = Endian.littleChar(input);
<a name="181" href="#181">181</a>             <strong>while</strong> (size > 0) {
<a name="182" href="#182">182</a>                 size -= input.skip(size);
<a name="183" href="#183">183</a>             }
<a name="184" href="#184">184</a>             tag = input.read();
<a name="185" href="#185">185</a>         }
<a name="186" href="#186">186</a>         <strong>if</strong> (tag != 2) {
<a name="187" href="#187">187</a>             <strong>throw</strong> <strong>new</strong> IllegalStateException();
<a name="188" href="#188">188</a>         }
<a name="189" href="#189">189</a>     }
<a name="190" href="#190">190</a> 
<a name="191" href="#191">191</a> 
<a name="192" href="#192">192</a>     <em>/**<em>*</em></em>
<a name="193" href="#193">193</a> <em>     * Returns the maximum character position.  Put another way, returns the</em>
<a name="194" href="#194">194</a> <em>     * total number of characters in the document.</em>
<a name="195" href="#195">195</a> <em>     * </em>
<a name="196" href="#196">196</a> <em>     * @return  the maximum character position</em>
<a name="197" href="#197">197</a> <em>     */</em>
<a name="198" href="#198">198</a>     <strong>public</strong> <strong>int</strong> getMaxCharPos() {
<a name="199" href="#199">199</a>         <strong>return</strong> maxCharPos;
<a name="200" href="#200">200</a>     }
<a name="201" href="#201">201</a> 
<a name="202" href="#202">202</a> 
<a name="203" href="#203">203</a>     <em>/**<em>*</em></em>
<a name="204" href="#204">204</a> <em>     * Returns the next piece in the piece table.</em>
<a name="205" href="#205">205</a> <em>     * </em>
<a name="206" href="#206">206</a> <em>     * @return  the next piece in the piece table, or null if there is no </em>
<a name="207" href="#207">207</a> <em>     *   next piece</em>
<a name="208" href="#208">208</a> <em>     * @throws IOException  if an IO error occurs</em>
<a name="209" href="#209">209</a> <em>     */</em>
<a name="210" href="#210">210</a>     <strong>public</strong> <a href="../../../../org/archive/util/ms/Piece.html">Piece</a> next() throws IOException {
<a name="211" href="#211">211</a>         <strong>if</strong> (current >= count) {
<a name="212" href="#212">212</a>             currentPiece = <strong>null</strong>;
<a name="213" href="#213">213</a>             <strong>return</strong> <strong>null</strong>;
<a name="214" href="#214">214</a>         }
<a name="215" href="#215">215</a>                 
<a name="216" href="#216">216</a>         <strong>int</strong> cp;
<a name="217" href="#217">217</a>         <strong>if</strong> (current == count - 1) {
<a name="218" href="#218">218</a>             cp = maxCharPos;
<a name="219" href="#219">219</a>         } <strong>else</strong> {
<a name="220" href="#220">220</a>             charPos.position(current * 4);
<a name="221" href="#221">221</a>             cp = Endian.littleInt(charPos);
<a name="222" href="#222">222</a>         }
<a name="223" href="#223">223</a>         filePos.position(current * 8);
<a name="224" href="#224">224</a>         <strong>int</strong> encoded = Endian.littleInt(filePos);
<a name="225" href="#225">225</a> 
<a name="226" href="#226">226</a>         <strong>if</strong> (LOGGER.isLoggable(Level.FINEST)) {
<a name="227" href="#227">227</a>             StringBuffer sb = <strong>new</strong> StringBuffer(Integer.toBinaryString(encoded));
<a name="228" href="#228">228</a>             <strong>while</strong> (sb.length() &lt; 32) {
<a name="229" href="#229">229</a>                 sb.insert(0, '0');
<a name="230" href="#230">230</a>             }
<a name="231" href="#231">231</a>             LOGGER.finest(<span class="string">"Encoded offset: "</span> + sb.toString());
<a name="232" href="#232">232</a>         }
<a name="233" href="#233">233</a>         
<a name="234" href="#234">234</a>         current++;
<a name="235" href="#235">235</a> 
<a name="236" href="#236">236</a>         <strong>int</strong> start;
<a name="237" href="#237">237</a>         <strong>if</strong> (currentPiece == <strong>null</strong>) {
<a name="238" href="#238">238</a>             start = 0;
<a name="239" href="#239">239</a>         } <strong>else</strong> {
<a name="240" href="#240">240</a>             start = currentPiece.getCharPosLimit();
<a name="241" href="#241">241</a>         }
<a name="242" href="#242">242</a>         <strong>if</strong> ((encoded &amp; CP1252_INDICATOR) == 0) {
<a name="243" href="#243">243</a>             <a href="../../../../org/archive/util/ms/Piece.html">Piece</a> piece = <strong>new</strong> <a href="../../../../org/archive/util/ms/Piece.html">Piece</a>(encoded, start, cp, <strong>true</strong>);
<a name="244" href="#244">244</a>             currentPiece = piece;
<a name="245" href="#245">245</a>             <strong>return</strong> piece;
<a name="246" href="#246">246</a>         } <strong>else</strong> {
<a name="247" href="#247">247</a>             <strong>int</strong> filePos = (encoded &amp; CP1252_MASK) / 2;
<a name="248" href="#248">248</a>             <a href="../../../../org/archive/util/ms/Piece.html">Piece</a> piece = <strong>new</strong> <a href="../../../../org/archive/util/ms/Piece.html">Piece</a>(filePos, start, cp, false);
<a name="249" href="#249">249</a>             currentPiece = piece;
<a name="250" href="#250">250</a>             <strong>return</strong> piece;
<a name="251" href="#251">251</a>         }
<a name="252" href="#252">252</a>     }
<a name="253" href="#253">253</a> 
<a name="254" href="#254">254</a>     
<a name="255" href="#255">255</a>     <em>/**<em>*</em></em>
<a name="256" href="#256">256</a> <em>     * Returns the piece containing the given character position.</em>
<a name="257" href="#257">257</a> <em>     * </em>
<a name="258" href="#258">258</a> <em>     * @param charPos   the character position whose piece to return</em>
<a name="259" href="#259">259</a> <em>     * @return   that piece, or null if no such piece exists (if charPos </em>
<a name="260" href="#260">260</a> <em>     *   is greater than getMaxCharPos())</em>
<a name="261" href="#261">261</a> <em>     * @throws IOException   if an IO error occurs</em>
<a name="262" href="#262">262</a> <em>     */</em>
<a name="263" href="#263">263</a>     <strong>public</strong> <a href="../../../../org/archive/util/ms/Piece.html">Piece</a> pieceFor(<strong>int</strong> charPos) throws IOException {
<a name="264" href="#264">264</a>         <strong>if</strong> (currentPiece.contains(charPos)) {
<a name="265" href="#265">265</a>             <strong>return</strong> currentPiece;
<a name="266" href="#266">266</a>         }
<a name="267" href="#267">267</a>      
<a name="268" href="#268">268</a>         <em class="comment">// FIXME: Use binary search to find piece index</em>
<a name="269" href="#269">269</a>         
<a name="270" href="#270">270</a>         current = 0;
<a name="271" href="#271">271</a>         currentPiece = <strong>null</strong>;
<a name="272" href="#272">272</a>         next();
<a name="273" href="#273">273</a>         
<a name="274" href="#274">274</a>         <strong>while</strong> (currentPiece != <strong>null</strong>) {
<a name="275" href="#275">275</a>             <strong>if</strong> (currentPiece.contains(charPos)) {
<a name="276" href="#276">276</a>                 <strong>return</strong> currentPiece;
<a name="277" href="#277">277</a>             }
<a name="278" href="#278">278</a>             next();
<a name="279" href="#279">279</a>         }
<a name="280" href="#280">280</a>         
<a name="281" href="#281">281</a>         <strong>return</strong> <strong>null</strong>;
<a name="282" href="#282">282</a>     }
<a name="283" href="#283">283</a> 
<a name="284" href="#284">284</a> }
</pre>
<hr/><div id="footer">This page was automatically generated by <a href="http://maven.apache.org/">Maven</a></div></body>
</html>