archivereaderfactory.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="115" href="#115">115</a>     	}
<a name="116" href="#116">116</a>     	<strong>throw</strong> <strong>new</strong> IOException(<span class="string">"Unknown file extension (Not ARC nor WARC): "</span>
<a name="117" href="#117">117</a>     		+ f.getName());
<a name="118" href="#118">118</a>     }
<a name="119" href="#119">119</a>     
<a name="120" href="#120">120</a>     <em>/**<em>*</em></em>
<a name="121" href="#121">121</a> <em>     * Wrap a Reader around passed Stream.</em>
<a name="122" href="#122">122</a> <em>     * @param s Identifying String for this Stream used in error messages.</em>
<a name="123" href="#123">123</a> <em>     * @param is Stream.</em>
<a name="124" href="#124">124</a> <em>     * @param atFirstRecord Are we at first Record?</em>
<a name="125" href="#125">125</a> <em>     * @return ArchiveReader.</em>
<a name="126" href="#126">126</a> <em>     * @throws IOException</em>
<a name="127" href="#127">127</a> <em>     */</em>
<a name="128" href="#128">128</a>     <strong>public</strong> <strong>static</strong> <a href="../../../org/archive/io/ArchiveReader.html">ArchiveReader</a> get(<strong>final</strong> String s, <strong>final</strong> InputStream is,
<a name="129" href="#129">129</a>         <strong>final</strong> <strong>boolean</strong> atFirstRecord)
<a name="130" href="#130">130</a>     throws IOException {
<a name="131" href="#131">131</a>         <strong>return</strong> ArchiveReaderFactory.factory.getArchiveReader(s, is,
<a name="132" href="#132">132</a>         	atFirstRecord);
<a name="133" href="#133">133</a>     }
<a name="134" href="#134">134</a>     
<a name="135" href="#135">135</a>     <strong>protected</strong> <a href="../../../org/archive/io/ArchiveReader.html">ArchiveReader</a> getArchiveReader(<strong>final</strong> String id, 
<a name="136" href="#136">136</a>     		<strong>final</strong> InputStream is, <strong>final</strong> <strong>boolean</strong> atFirstRecord)
<a name="137" href="#137">137</a>     throws IOException {
<a name="138" href="#138">138</a>     	<em class="comment">// Default to compressed WARC.</em>
<a name="139" href="#139">139</a>     	<strong>return</strong> WARCReaderFactory.get(id, is, atFirstRecord);
<a name="140" href="#140">140</a>     }
<a name="141" href="#141">141</a>     
<a name="142" href="#142">142</a>     <em>/**<em>*</em></em>
<a name="143" href="#143">143</a> <em>     * Get an Archive Reader aligned at &lt;code>offset&lt;/code>.</em>
<a name="144" href="#144">144</a> <em>     * This version of get will not bring the file local but will try to</em>
<a name="145" href="#145">145</a> <em>     * stream across the net making an HTTP 1.1 Range request on remote</em>
<a name="146" href="#146">146</a> <em>     * http server (RFC1435 Section 14.35).</em>
<a name="147" href="#147">147</a> <em>     * @param u HTTP URL for an Archive file.</em>
<a name="148" href="#148">148</a> <em>     * @param offset Offset into file at which to start fetching.</em>
<a name="149" href="#149">149</a> <em>     * @return An ArchiveReader aligned at offset.</em>
<a name="150" href="#150">150</a> <em>     * @throws IOException</em>
<a name="151" href="#151">151</a> <em>     */</em>
<a name="152" href="#152">152</a>     <strong>public</strong> <strong>static</strong> <a href="../../../org/archive/io/ArchiveReader.html">ArchiveReader</a> get(<strong>final</strong> URL u, <strong>final</strong> <strong>long</strong> offset)
<a name="153" href="#153">153</a>     throws IOException {
<a name="154" href="#154">154</a>     	<strong>return</strong> ArchiveReaderFactory.factory.getArchiveReader(u, offset);
<a name="155" href="#155">155</a>     }
<a name="156" href="#156">156</a>     
<a name="157" href="#157">157</a>     <strong>protected</strong> <a href="../../../org/archive/io/ArchiveReader.html">ArchiveReader</a> getArchiveReader(<strong>final</strong> URL f, <strong>final</strong> <strong>long</strong> offset)
<a name="158" href="#158">158</a>     throws IOException {
<a name="159" href="#159">159</a>         <em class="comment">// Get URL connection.</em>
<a name="160" href="#160">160</a>         URLConnection connection = f.openConnection();
<a name="161" href="#161">161</a>         <strong>if</strong> (!(connection instanceof HttpURLConnection)) {
<a name="162" href="#162">162</a>             <strong>throw</strong> <strong>new</strong> IOException(<span class="string">"This method only handles HTTP connections."</span>);
<a name="163" href="#163">163</a>         }
<a name="164" href="#164">164</a>         addUserAgent((HttpURLConnection)connection);
<a name="165" href="#165">165</a>         <em class="comment">// Use a Range request (Assumes HTTP 1.1 on other end). If length &lt;= 0,</em>
<a name="166" href="#166">166</a>         <em class="comment">// add open-ended range header to the request.  Else, because end-byte</em>
<a name="167" href="#167">167</a>         <em class="comment">// is inclusive, subtract 1.</em>
<a name="168" href="#168">168</a>         connection.addRequestProperty(<span class="string">"Range"</span>, <span class="string">"bytes="</span> + offset + <span class="string">"-"</span>);
<a name="169" href="#169">169</a>         <em class="comment">// TODO: Get feedback on this ArchiveReader maker. If fetching single</em>
<a name="170" href="#170">170</a>         <em class="comment">// record remotely, might make sense to do a slimmed down</em>
<a name="171" href="#171">171</a>         <em class="comment">// ArchiveReader getter.</em>
<a name="172" href="#172">172</a>         <em class="comment">// Good if size 2 * inflator buffer to avoid buffer boundaries</em>
<a name="173" href="#173">173</a>         <em class="comment">// (TODO: Implement better handling across buffer boundaries).</em>
<a name="174" href="#174">174</a>         <strong>return</strong> getArchiveReader(f.toString(),
<a name="175" href="#175">175</a>             <strong>new</strong> <a href="../../../org/archive/io/RepositionableInputStream.html">RepositionableInputStream</a>(connection.getInputStream(),
<a name="176" href="#176">176</a>                 16 * 1024),
<a name="177" href="#177">177</a>             (offset == 0));
<a name="178" href="#178">178</a>     }
<a name="179" href="#179">179</a>     
<a name="180" href="#180">180</a>     <em>/**<em>*</em></em>
<a name="181" href="#181">181</a> <em>     * Get an ARCReader.</em>
<a name="182" href="#182">182</a> <em>     * Pulls the ARC local into whereever the System Property</em>
<a name="183" href="#183">183</a> <em>     * &lt;code>java.io.tmpdir&lt;/code> points. It then hands back an ARCReader that</em>
<a name="184" href="#184">184</a> <em>     * points at this local copy.  A close on this ARCReader instance will</em>
<a name="185" href="#185">185</a> <em>     * remove the local copy.</em>
<a name="186" href="#186">186</a> <em>     * @param u An URL that points at an ARC.</em>
<a name="187" href="#187">187</a> <em>     * @return An ARCReader.</em>
<a name="188" href="#188">188</a> <em>     * @throws IOException </em>
<a name="189" href="#189">189</a> <em>     */</em>
<a name="190" href="#190">190</a>     <strong>public</strong> <strong>static</strong> <a href="../../../org/archive/io/ArchiveReader.html">ArchiveReader</a> get(<strong>final</strong> URL u)
<a name="191" href="#191">191</a>     throws IOException {
<a name="192" href="#192">192</a>     	<strong>return</strong> ArchiveReaderFactory.factory.getArchiveReader(u);
<a name="193" href="#193">193</a>     }
<a name="194" href="#194">194</a>     
<a name="195" href="#195">195</a>     <strong>protected</strong> <a href="../../../org/archive/io/ArchiveReader.html">ArchiveReader</a> getArchiveReader(<strong>final</strong> URL u)
<a name="196" href="#196">196</a>     throws IOException {
<a name="197" href="#197">197</a>         <em class="comment">// If url represents a local file then return file it points to.</em>
<a name="198" href="#198">198</a>         <strong>if</strong> (u.getPath() != <strong>null</strong>) {
<a name="199" href="#199">199</a>             <em class="comment">// TODO: Add scheme check and host check.</em>
<a name="200" href="#200">200</a>             File f = <strong>new</strong> File(u.getPath());
<a name="201" href="#201">201</a>             <strong>if</strong> (f.exists()) {
<a name="202" href="#202">202</a>                 <strong>return</strong> get(f, 0);
<a name="203" href="#203">203</a>             }
<a name="204" href="#204">204</a>         }
<a name="205" href="#205">205</a>         
<a name="206" href="#206">206</a>         <strong>return</strong> makeARCLocal(u.openConnection());
<a name="207" href="#207">207</a>     }
<a name="208" href="#208">208</a>     
<a name="209" href="#209">209</a>     <strong>protected</strong> <a href="../../../org/archive/io/ArchiveReader.html">ArchiveReader</a> makeARCLocal(<strong>final</strong> URLConnection connection)
<a name="210" href="#210">210</a>     throws IOException {
<a name="211" href="#211">211</a>         File localFile = <strong>null</strong>;
<a name="212" href="#212">212</a>         <strong>if</strong> (connection instanceof HttpURLConnection) {
<a name="213" href="#213">213</a>             <em class="comment">// If http url connection, bring down the resouce local.</em>
<a name="214" href="#214">214</a>             localFile = File.createTempFile(ArchiveReader.<strong>class</strong>.getName(),
<a name="215" href="#215">215</a>             	<span class="string">".tmp"</span>, FileUtils.TMPDIR);
<a name="216" href="#216">216</a>             connection.connect();
<a name="217" href="#217">217</a>             addUserAgent((HttpURLConnection)connection);
<a name="218" href="#218">218</a>             <strong>try</strong> {
<a name="219" href="#219">219</a>                 IoUtils.readFullyToFile(connection.getInputStream(), localFile,
<a name="220" href="#220">220</a>                         <strong>new</strong> byte[16 * 1024]);
<a name="221" href="#221">221</a>             } <strong>catch</strong> (IOException ioe) {
<a name="222" href="#222">222</a>                 localFile.delete();
<a name="223" href="#223">223</a>                 <strong>throw</strong> ioe;
<a name="224" href="#224">224</a>             }
<a name="225" href="#225">225</a>         } <strong>else</strong> <strong>if</strong> (connection instanceof RsyncURLConnection) {
<a name="226" href="#226">226</a>             <em class="comment">// Then, connect and this will create a local file.</em>
<a name="227" href="#227">227</a>             <em class="comment">// See implementation of the rsync handler.</em>
<a name="228" href="#228">228</a>             connection.connect();
<a name="229" href="#229">229</a>             localFile = ((RsyncURLConnection)connection).getFile();
<a name="230" href="#230">230</a>         } <strong>else</strong> <strong>if</strong> (connection instanceof Md5URLConnection) {
<a name="231" href="#231">231</a>             <em class="comment">// Then, connect and this will create a local file.</em>
<a name="232" href="#232">232</a>             <em class="comment">// See implementation of the md5 handler.</em>
<a name="233" href="#233">233</a>             connection.connect();
<a name="234" href="#234">234</a>             localFile = ((Md5URLConnection)connection).getFile();
<a name="235" href="#235">235</a>         } <strong>else</strong> {
<a name="236" href="#236">236</a>             <strong>throw</strong> <strong>new</strong> UnsupportedOperationException(<span class="string">"No support for "</span> +
<a name="237" href="#237">237</a>                 connection);