archiveutils.html

来自「网络爬虫开源代码」· HTML 代码 · 共 757 行 · 第 1/5 页

<a name="144" href="#144">144</a>     
<a name="145" href="#145">145</a>     <em>/**<em>*</em></em>
<a name="146" href="#146">146</a> <em>     * Utility function for creating log timestamps, in</em>
<a name="147" href="#147">147</a> <em>     * W3C/ISO8601 format, assuming UTC. </em>
<a name="148" href="#148">148</a> <em>     * </em>
<a name="149" href="#149">149</a> <em>     * Format is yyyy-MM-dd'T'HH:mm:ss.SSS'Z'</em>
<a name="150" href="#150">150</a> <em>     * @param date Date to format.</em>
<a name="151" href="#151">151</a> <em>     * </em>
<a name="152" href="#152">152</a> <em>     * @return the date stamp</em>
<a name="153" href="#153">153</a> <em>     */</em>
<a name="154" href="#154">154</a>     <strong>public</strong> <strong>static</strong> String getLog17Date(<strong>long</strong> date){
<a name="155" href="#155">155</a>         <strong>return</strong> TIMESTAMP17ISO8601Z.get().format(<strong>new</strong> Date(date));
<a name="156" href="#156">156</a>     }
<a name="157" href="#157">157</a>     
<a name="158" href="#158">158</a>     <em>/**<em>*</em></em>
<a name="159" href="#159">159</a> <em>     * Utility function for creating log timestamps, in</em>
<a name="160" href="#160">160</a> <em>     * W3C/ISO8601 format, assuming UTC. Use current time. </em>
<a name="161" href="#161">161</a> <em>     * </em>
<a name="162" href="#162">162</a> <em>     * Format is yyyy-MM-dd'T'HH:mm:ss'Z'</em>
<a name="163" href="#163">163</a> <em>     * </em>
<a name="164" href="#164">164</a> <em>     * @return the date stamp</em>
<a name="165" href="#165">165</a> <em>     */</em>
<a name="166" href="#166">166</a>     <strong>public</strong> <strong>static</strong> String getLog14Date(){
<a name="167" href="#167">167</a>         <strong>return</strong> TIMESTAMP14ISO8601Z.get().format(<strong>new</strong> Date());
<a name="168" href="#168">168</a>     }
<a name="169" href="#169">169</a>     
<a name="170" href="#170">170</a>     <em>/**<em>*</em></em>
<a name="171" href="#171">171</a> <em>     * Utility function for creating log timestamps, in</em>
<a name="172" href="#172">172</a> <em>     * W3C/ISO8601 format, assuming UTC. </em>
<a name="173" href="#173">173</a> <em>     * </em>
<a name="174" href="#174">174</a> <em>     * Format is yyyy-MM-dd'T'HH:mm:ss'Z'</em>
<a name="175" href="#175">175</a> <em>     * @param date long timestamp to format.</em>
<a name="176" href="#176">176</a> <em>     * </em>
<a name="177" href="#177">177</a> <em>     * @return the date stamp</em>
<a name="178" href="#178">178</a> <em>     */</em>
<a name="179" href="#179">179</a>     <strong>public</strong> <strong>static</strong> String getLog14Date(<strong>long</strong> date){
<a name="180" href="#180">180</a>         <strong>return</strong> TIMESTAMP14ISO8601Z.get().format(<strong>new</strong> Date(date));
<a name="181" href="#181">181</a>     }
<a name="182" href="#182">182</a>     
<a name="183" href="#183">183</a>     <em>/**<em>*</em></em>
<a name="184" href="#184">184</a> <em>     * Utility function for creating log timestamps, in</em>
<a name="185" href="#185">185</a> <em>     * W3C/ISO8601 format, assuming UTC. </em>
<a name="186" href="#186">186</a> <em>     * </em>
<a name="187" href="#187">187</a> <em>     * Format is yyyy-MM-dd'T'HH:mm:ss'Z'</em>
<a name="188" href="#188">188</a> <em>     * @param date Date to format.</em>
<a name="189" href="#189">189</a> <em>     * </em>
<a name="190" href="#190">190</a> <em>     * @return the date stamp</em>
<a name="191" href="#191">191</a> <em>     */</em>
<a name="192" href="#192">192</a>     <strong>public</strong> <strong>static</strong> String getLog14Date(Date date){
<a name="193" href="#193">193</a>         <strong>return</strong> TIMESTAMP14ISO8601Z.get().format(date);
<a name="194" href="#194">194</a>     }
<a name="195" href="#195">195</a>     
<a name="196" href="#196">196</a>     <em>/**<em>*</em></em>
<a name="197" href="#197">197</a> <em>     * Utility function for creating arc-style date stamps</em>
<a name="198" href="#198">198</a> <em>     * in the format yyyyMMddHHmmssSSS.</em>
<a name="199" href="#199">199</a> <em>     * Date stamps are in the UTC time zone</em>
<a name="200" href="#200">200</a> <em>     *</em>
<a name="201" href="#201">201</a> <em>     * @param date milliseconds since epoc</em>
<a name="202" href="#202">202</a> <em>     * @return the date stamp</em>
<a name="203" href="#203">203</a> <em>     */</em>
<a name="204" href="#204">204</a>     <strong>public</strong> <strong>static</strong> String get17DigitDate(<strong>long</strong> date){
<a name="205" href="#205">205</a>         <strong>return</strong> TIMESTAMP17.get().format(<strong>new</strong> Date(date));
<a name="206" href="#206">206</a>     }
<a name="207" href="#207">207</a>     
<a name="208" href="#208">208</a>     <strong>public</strong> <strong>static</strong> String get17DigitDate(Date date){
<a name="209" href="#209">209</a>         <strong>return</strong> TIMESTAMP17.get().format(date);
<a name="210" href="#210">210</a>     }
<a name="211" href="#211">211</a> 
<a name="212" href="#212">212</a>     <em>/**<em>*</em></em>
<a name="213" href="#213">213</a> <em>     * Utility function for creating arc-style date stamps</em>
<a name="214" href="#214">214</a> <em>     * in the format yyyyMMddHHmmss.</em>
<a name="215" href="#215">215</a> <em>     * Date stamps are in the UTC time zone</em>
<a name="216" href="#216">216</a> <em>     *</em>
<a name="217" href="#217">217</a> <em>     * @param date milliseconds since epoc</em>
<a name="218" href="#218">218</a> <em>     * @return the date stamp</em>
<a name="219" href="#219">219</a> <em>     */</em>
<a name="220" href="#220">220</a>     <strong>public</strong> <strong>static</strong> String get14DigitDate(<strong>long</strong> date){
<a name="221" href="#221">221</a>         <strong>return</strong> TIMESTAMP14.get().format(<strong>new</strong> Date(date));
<a name="222" href="#222">222</a>     }
<a name="223" href="#223">223</a> 
<a name="224" href="#224">224</a>     <strong>public</strong> <strong>static</strong> String get14DigitDate(Date d) {
<a name="225" href="#225">225</a>         <strong>return</strong> TIMESTAMP14.get().format(d);
<a name="226" href="#226">226</a>     }
<a name="227" href="#227">227</a> 
<a name="228" href="#228">228</a>     <em>/**<em>*</em></em>
<a name="229" href="#229">229</a> <em>     * Utility function for creating arc-style date stamps</em>
<a name="230" href="#230">230</a> <em>     * in the format yyyyMMddHHmm.</em>
<a name="231" href="#231">231</a> <em>     * Date stamps are in the UTC time zone</em>
<a name="232" href="#232">232</a> <em>     *</em>
<a name="233" href="#233">233</a> <em>     * @param date milliseconds since epoc</em>
<a name="234" href="#234">234</a> <em>     * @return the date stamp</em>
<a name="235" href="#235">235</a> <em>     */</em>
<a name="236" href="#236">236</a>     <strong>public</strong> <strong>static</strong> String get12DigitDate(<strong>long</strong> date){
<a name="237" href="#237">237</a>         <strong>return</strong> TIMESTAMP12.get().format(<strong>new</strong> Date(date));
<a name="238" href="#238">238</a>     }
<a name="239" href="#239">239</a>     
<a name="240" href="#240">240</a>     <strong>public</strong> <strong>static</strong> String get12DigitDate(Date d) {
<a name="241" href="#241">241</a>         <strong>return</strong> TIMESTAMP12.get().format(d);
<a name="242" href="#242">242</a>     }
<a name="243" href="#243">243</a>     
<a name="244" href="#244">244</a>     <em>/**<em>*</em></em>
<a name="245" href="#245">245</a> <em>     * Parses an ARC-style date.  If passed String is &lt; 12 characters in length,</em>
<a name="246" href="#246">246</a> <em>     * we pad.  At a minimum, String should contain a year (>=4 characters).</em>
<a name="247" href="#247">247</a> <em>     * Parse will also fail if day or month are incompletely specified.  Depends</em>
<a name="248" href="#248">248</a> <em>     * on the above getXXDigitDate methods.</em>
<a name="249" href="#249">249</a> <em>     * @param A 4-17 digit date in ARC style (&lt;code>yyyy&lt;/code> to</em>
<a name="250" href="#250">250</a> <em>     * &lt;code>yyyyMMddHHmmssSSS&lt;/code>) formatting.  </em>
<a name="251" href="#251">251</a> <em>     * @return A Date object representing the passed String. </em>
<a name="252" href="#252">252</a> <em>     * @throws ParseException</em>
<a name="253" href="#253">253</a> <em>     */</em>
<a name="254" href="#254">254</a>     <strong>public</strong> <strong>static</strong> Date getDate(String d) throws ParseException {
<a name="255" href="#255">255</a>         Date date = <strong>null</strong>;
<a name="256" href="#256">256</a>         <strong>if</strong> (d == <strong>null</strong>) {
<a name="257" href="#257">257</a>             <strong>throw</strong> <strong>new</strong> IllegalArgumentException(<span class="string">"Passed date is null"</span>);
<a name="258" href="#258">258</a>         }
<a name="259" href="#259">259</a>         <strong>switch</strong> (d.length()) {
<a name="260" href="#260">260</a>         <strong>case</strong> 14:
<a name="261" href="#261">261</a>             date = ArchiveUtils.parse14DigitDate(d);
<a name="262" href="#262">262</a>             <strong>break</strong>;
<a name="263" href="#263">263</a> 
<a name="264" href="#264">264</a>         <strong>case</strong> 17:
<a name="265" href="#265">265</a>             date = ArchiveUtils.parse17DigitDate(d);
<a name="266" href="#266">266</a>             <strong>break</strong>;
<a name="267" href="#267">267</a> 
<a name="268" href="#268">268</a>         <strong>case</strong> 12:
<a name="269" href="#269">269</a>             date = ArchiveUtils.parse12DigitDate(d);
<a name="270" href="#270">270</a>             <strong>break</strong>;
<a name="271" href="#271">271</a>            
<a name="272" href="#272">272</a>         <strong>case</strong> 0:
<a name="273" href="#273">273</a>         <strong>case</strong> 1:
<a name="274" href="#274">274</a>         <strong>case</strong> 2:
<a name="275" href="#275">275</a>         <strong>case</strong> 3:
<a name="276" href="#276">276</a>             <strong>throw</strong> <strong>new</strong> ParseException(<span class="string">"Date string must at least contain a"</span> +
<a name="277" href="#277">277</a>                 <span class="string">"year: "</span> + d, d.length());
<a name="278" href="#278">278</a>             
<a name="279" href="#279">279</a>         <strong>default</strong>:
<a name="280" href="#280">280</a>             <strong>if</strong> (!(d.startsWith(<span class="string">"19"</span>) || d.startsWith(<span class="string">"20"</span>))) {
<a name="281" href="#281">281</a>                 <strong>throw</strong> <strong>new</strong> ParseException(<span class="string">"Unrecognized century: "</span> + d, 0);
<a name="282" href="#282">282</a>             }
<a name="283" href="#283">283</a>             <strong>if</strong> (d.length() &lt; 8 &amp;&amp; (d.length() % 2) != 0) {
<a name="284" href="#284">284</a>                 <strong>throw</strong> <strong>new</strong> ParseException(<span class="string">"Incomplete month/date: "</span> + d,
<a name="285" href="#285">285</a>                     d.length());
<a name="286" href="#286">286</a>             }
<a name="287" href="#287">287</a>             StringBuilder sb = <strong>new</strong> StringBuilder(d);
<a name="288" href="#288">288</a>             <strong>if</strong> (sb.length() &lt; 8) {
<a name="289" href="#289">289</a>                 <strong>for</strong> (<strong>int</strong> i = sb.length(); sb.length() &lt; 8; i += 2) {
<a name="290" href="#290">290</a>                     sb.append(<span class="string">"01"</span>);
<a name="291" href="#291">291</a>                 }
<a name="292" href="#292">292</a>             }
<a name="293" href="#293">293</a>             <strong>if</strong> (sb.length() &lt; 12) {
<a name="294" href="#294">294</a>                 <strong>for</strong> (<strong>int</strong> i = sb.length(); sb.length() &lt; 12; i++) {
<a name="295" href="#295">295</a>                     sb.append(<span class="string">"0"</span>);