archiveutils.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="163" href="#163">163</a> <em>     * Utility function for creating log timestamps, in</em>
<a name="164" href="#164">164</a> <em>     * W3C/ISO8601 format, assuming UTC. </em>
<a name="165" href="#165">165</a> <em>     * </em>
<a name="166" href="#166">166</a> <em>     * Format is yyyy-MM-dd'T'HH:mm:ss'Z'</em>
<a name="167" href="#167">167</a> <em>     * @param date Date to format.</em>
<a name="168" href="#168">168</a> <em>     * </em>
<a name="169" href="#169">169</a> <em>     * @return the date stamp</em>
<a name="170" href="#170">170</a> <em>     */</em>
<a name="171" href="#171">171</a>     <strong>public</strong> <strong>static</strong> String getLog14Date(<strong>long</strong> date){
<a name="172" href="#172">172</a>         <strong>return</strong> TIMESTAMP14ISO8601Z.format(<strong>new</strong> Date(date));
<a name="173" href="#173">173</a>     }
<a name="174" href="#174">174</a>     
<a name="175" href="#175">175</a>     <em>/**<em>*</em></em>
<a name="176" href="#176">176</a> <em>     * Utility function for creating arc-style date stamps</em>
<a name="177" href="#177">177</a> <em>     * in the format yyyMMddHHmmssSSS.</em>
<a name="178" href="#178">178</a> <em>     * Date stamps are in the UTC time zone</em>
<a name="179" href="#179">179</a> <em>     *</em>
<a name="180" href="#180">180</a> <em>     * @param date milliseconds since epoc</em>
<a name="181" href="#181">181</a> <em>     * @return the date stamp</em>
<a name="182" href="#182">182</a> <em>     */</em>
<a name="183" href="#183">183</a>     <strong>public</strong> <strong>static</strong> String get17DigitDate(<strong>long</strong> date){
<a name="184" href="#184">184</a>         <strong>return</strong> TIMESTAMP17.format(<strong>new</strong> Date(date));
<a name="185" href="#185">185</a>     }
<a name="186" href="#186">186</a>     
<a name="187" href="#187">187</a>     <strong>public</strong> <strong>static</strong> String get17DigitDate(Date date){
<a name="188" href="#188">188</a>         <strong>return</strong> TIMESTAMP17.format(date);
<a name="189" href="#189">189</a>     }
<a name="190" href="#190">190</a> 
<a name="191" href="#191">191</a>     <em>/**<em>*</em></em>
<a name="192" href="#192">192</a> <em>     * Utility function for creating arc-style date stamps</em>
<a name="193" href="#193">193</a> <em>     * in the format yyyMMddHHmmss.</em>
<a name="194" href="#194">194</a> <em>     * Date stamps are in the UTC time zone</em>
<a name="195" href="#195">195</a> <em>     *</em>
<a name="196" href="#196">196</a> <em>     * @param date milliseconds since epoc</em>
<a name="197" href="#197">197</a> <em>     * @return the date stamp</em>
<a name="198" href="#198">198</a> <em>     */</em>
<a name="199" href="#199">199</a>     <strong>public</strong> <strong>static</strong> String get14DigitDate(<strong>long</strong> date){
<a name="200" href="#200">200</a>         <strong>return</strong> TIMESTAMP14.format(<strong>new</strong> Date(date));
<a name="201" href="#201">201</a>     }
<a name="202" href="#202">202</a> 
<a name="203" href="#203">203</a>     <strong>public</strong> <strong>static</strong> String get14DigitDate(Date d) {
<a name="204" href="#204">204</a>         <strong>return</strong> TIMESTAMP14.format(d);
<a name="205" href="#205">205</a>     }
<a name="206" href="#206">206</a> 
<a name="207" href="#207">207</a>     <em>/**<em>*</em></em>
<a name="208" href="#208">208</a> <em>     * Utility function for creating arc-style date stamps</em>
<a name="209" href="#209">209</a> <em>     * in the format yyyMMddHHmm.</em>
<a name="210" href="#210">210</a> <em>     * Date stamps are in the UTC time zone</em>
<a name="211" href="#211">211</a> <em>     *</em>
<a name="212" href="#212">212</a> <em>     * @param date milliseconds since epoc</em>
<a name="213" href="#213">213</a> <em>     * @return the date stamp</em>
<a name="214" href="#214">214</a> <em>     */</em>
<a name="215" href="#215">215</a>     <strong>public</strong> <strong>static</strong> String get12DigitDate(<strong>long</strong> date){
<a name="216" href="#216">216</a>         <strong>return</strong> TIMESTAMP12.format(<strong>new</strong> Date(date));
<a name="217" href="#217">217</a>     }
<a name="218" href="#218">218</a>     
<a name="219" href="#219">219</a>     <strong>public</strong> <strong>static</strong> String get12DigitDate(Date d) {
<a name="220" href="#220">220</a>         <strong>return</strong> TIMESTAMP12.format(d);
<a name="221" href="#221">221</a>     }
<a name="222" href="#222">222</a> 
<a name="223" href="#223">223</a>     <em>/**<em>*</em></em>
<a name="224" href="#224">224</a> <em>     * Utility function for parsing arc-style date stamps</em>
<a name="225" href="#225">225</a> <em>     * in the format yyyMMddHHmmssSSS.</em>
<a name="226" href="#226">226</a> <em>     * Date stamps are in the UTC time zone.  The whole string will not be</em>
<a name="227" href="#227">227</a> <em>     * parsed, only the first 17 digits.</em>
<a name="228" href="#228">228</a> <em>     *</em>
<a name="229" href="#229">229</a> <em>     * @param date an arc-style formatted date stamp</em>
<a name="230" href="#230">230</a> <em>     * @return the Date corresponding to the date stamp string</em>
<a name="231" href="#231">231</a> <em>     * @throws ParseException if the inputstring was malformed</em>
<a name="232" href="#232">232</a> <em>     */</em>
<a name="233" href="#233">233</a>     <strong>public</strong> <strong>static</strong> Date parse17DigitDate(String date) throws ParseException{
<a name="234" href="#234">234</a>         <strong>return</strong> TIMESTAMP17.parse(date);
<a name="235" href="#235">235</a>     }
<a name="236" href="#236">236</a> 
<a name="237" href="#237">237</a>     <em>/**<em>*</em></em>
<a name="238" href="#238">238</a> <em>     * Utility function for parsing arc-style date stamps</em>
<a name="239" href="#239">239</a> <em>     * in the format yyyMMddHHmmss.</em>
<a name="240" href="#240">240</a> <em>     * Date stamps are in the UTC time zone.  The whole string will not be</em>
<a name="241" href="#241">241</a> <em>     * parsed, only the first 14 digits.</em>
<a name="242" href="#242">242</a> <em>     *</em>
<a name="243" href="#243">243</a> <em>     * @param date an arc-style formatted date stamp</em>
<a name="244" href="#244">244</a> <em>     * @return the Date corresponding to the date stamp string</em>
<a name="245" href="#245">245</a> <em>     * @throws ParseException if the inputstring was malformed</em>
<a name="246" href="#246">246</a> <em>     */</em>
<a name="247" href="#247">247</a>     <strong>public</strong> <strong>static</strong> Date parse14DigitDate(String date) throws ParseException{
<a name="248" href="#248">248</a>         <strong>return</strong> TIMESTAMP14.parse(date);
<a name="249" href="#249">249</a>     }
<a name="250" href="#250">250</a> 
<a name="251" href="#251">251</a>     <em>/**<em>*</em></em>
<a name="252" href="#252">252</a> <em>     * Utility function for parsing arc-style date stamps</em>
<a name="253" href="#253">253</a> <em>     * in the format yyyMMddHHmm.</em>
<a name="254" href="#254">254</a> <em>     * Date stamps are in the UTC time zone.  The whole string will not be</em>
<a name="255" href="#255">255</a> <em>     * parsed, only the first 12 digits.</em>
<a name="256" href="#256">256</a> <em>     *</em>
<a name="257" href="#257">257</a> <em>     * @param date an arc-style formatted date stamp</em>
<a name="258" href="#258">258</a> <em>     * @return the Date corresponding to the date stamp string</em>
<a name="259" href="#259">259</a> <em>     * @throws ParseException if the inputstring was malformed</em>
<a name="260" href="#260">260</a> <em>     */</em>
<a name="261" href="#261">261</a>     <strong>public</strong> <strong>static</strong> Date parse12DigitDate(String date) throws ParseException{
<a name="262" href="#262">262</a>         <strong>return</strong> TIMESTAMP12.parse(date);
<a name="263" href="#263">263</a>     }
<a name="264" href="#264">264</a>     
<a name="265" href="#265">265</a>     <em>/**<em>*</em></em>
<a name="266" href="#266">266</a> <em>     * Convert 17-digit date format timestamps (as found in crawl.log, for</em>
<a name="267" href="#267">267</a> <em>     * example) into a GregorianCalendar object. + * Useful so you can convert</em>
<a name="268" href="#268">268</a> <em>     * into milliseconds-since-epoch. Note: it is possible to compute</em>
<a name="269" href="#269">269</a> <em>     * milliseconds-since-epoch + * using {@link #parse17DigitDate}.UTC(), but</em>
<a name="270" href="#270">270</a> <em>     * that method is deprecated in favor of using Calendar.getTimeInMillis(). + *</em>
<a name="271" href="#271">271</a> <em>     * &lt;p/>I probably should have dug into all the utility methods in</em>
<a name="272" href="#272">272</a> <em>     * DateFormat.java to parse the timestamp, but this was + * easier. If</em>
<a name="273" href="#273">273</a> <em>     * someone wants to fix this to use those methods, please have at it! &lt;p/></em>
<a name="274" href="#274">274</a> <em>     * Mike Schwartz, schwartz at CodeOnTheRoad dot com.</em>
<a name="275" href="#275">275</a> <em>     * </em>
<a name="276" href="#276">276</a> <em>     * @param timestamp17String</em>
<a name="277" href="#277">277</a> <em>     * @return Calendar set to &lt;code>timestamp17String&lt;/code>.</em>
<a name="278" href="#278">278</a> <em>     */</em>
<a name="279" href="#279">279</a>     <strong>public</strong> <strong>static</strong> Calendar timestamp17ToCalendar(String timestamp17String) {
<a name="280" href="#280">280</a>         GregorianCalendar calendar = <strong>new</strong> GregorianCalendar();
<a name="281" href="#281">281</a>         <strong>int</strong> year = Integer.parseInt(timestamp17String.substring(0, 4));
<a name="282" href="#282">282</a>         <strong>int</strong> dayOfMonth = Integer.parseInt(timestamp17String.substring(6, 8));
<a name="283" href="#283">283</a>         <em class="comment">// Month is 0-based</em>
<a name="284" href="#284">284</a>         <strong>int</strong> month = Integer.parseInt(timestamp17String.substring(4, 6)) - 1;
<a name="285" href="#285">285</a>         <strong>int</strong> hourOfDay = Integer.parseInt(timestamp17String.substring(8, 10));
<a name="286" href="#286">286</a>         <strong>int</strong> minute = Integer.parseInt(timestamp17String.substring(10, 12));
<a name="287" href="#287">287</a>         <strong>int</strong> second = Integer.parseInt(timestamp17String.substring(12, 14));
<a name="288" href="#288">288</a>         <strong>int</strong> milliseconds = Integer
<a name="289" href="#289">289</a>                 .parseInt(timestamp17String.substring(14, 17));
<a name="290" href="#290">290</a>         calendar.set(Calendar.YEAR, year);
<a name="291" href="#291">291</a>         calendar.set(Calendar.MONTH, month);
<a name="292" href="#292">292</a>         calendar.set(Calendar.DAY_OF_MONTH, dayOfMonth);
<a name="293" href="#293">293</a>         calendar.set(Calendar.HOUR_OF_DAY, hourOfDay);
<a name="294" href="#294">294</a>         calendar.set(Calendar.MINUTE, minute);
<a name="295" href="#295">295</a>         calendar.set(Calendar.SECOND, second);
<a name="296" href="#296">296</a>         calendar.set(Calendar.MILLISECOND, milliseconds);
<a name="297" href="#297">297</a>         <strong>return</strong> calendar;
<a name="298" href="#298">298</a>     }
<a name="299" href="#299">299</a>     
<a name="300" href="#300">300</a>     <em>/**<em>*</em></em>
<a name="301" href="#301">301</a> <em>     * @param timestamp A 14-digit timestamp or the suffix for a 14-digit</em>
<a name="302" href="#302">302</a> <em>     * timestamp: E.g. '20010909014640' or '20010101' or '1970'.</em>
<a name="303" href="#303">303</a> <em>     * @return Seconds since the epoch as a string zero-pre-padded so always</em>
<a name="304" href="#304">304</a> <em>     * Integer.MAX_VALUE wide (Makes it so sorting of resultant string works</em>
<a name="305" href="#305">305</a> <em>     * properly).</em>
<a name="306" href="#306">306</a> <em>     * @throws ParseException </em>
<a name="307" href="#307">307</a> <em>     */</em>
<a name="308" href="#308">308</a>     <strong>public</strong> <strong>static</strong> String secondsSinceEpoch(String timestamp)
<a name="309" href="#309">309</a>     throws ParseException {
<a name="310" href="#310">310</a>         <strong>return</strong> zeroPadInteger((<strong>int</strong>)
<a name="311" href="#311">311</a>             (getSecondsSinceEpoch(timestamp).getTime()/1000));
<a name="312" href="#312">312</a>     }
<a name="313" href="#313">313</a>     
<a name="314" href="#314">314</a>     <em>/**<em>*</em></em>
<a name="315" href="#315">315</a> <em>     * @param timestamp A 14-digit timestamp or the suffix for a 14-digit</em>
<a name="316" href="#316">316</a> <em>     * timestamp: E.g. '20010909014640' or '20010101' or '1970'.</em>
<a name="317" href="#317">317</a> <em>     * @return A date.</em>
<a name="318" href="#318">318</a> <em>     * @see #secondsSinceEpoch(String)</em>
<a name="319" href="#319">319</a> <em>     * @throws ParseException </em>
<a name="320" href="#320">320</a> <em>     */</em>
<a name="321" href="#321">321</a>     <strong>public</strong> <strong>static</strong> Date getSecondsSinceEpoch(String timestamp)
<a name="322" href="#322">322</a>     throws ParseException {
<a name="323" href="#323">323</a>         <strong>if</strong> (timestamp.length() &lt; 14) {
<a name="324" href="#324">324</a>             <strong>if</strong> (timestamp.length() &lt; 10 &amp;&amp; (timestamp.length() % 2) == 1) {
<a name="325" href="#325">325</a>                 <strong>throw</strong> <strong>new</strong> IllegalArgumentException(<span class="string">"Must have year, "</span> +
<a name="326" href="#326">326</a>                     <span class="string">"month, date, hour or second granularity: "</span> + timestamp);
<a name="327" href="#327">327</a>             }
<a name="328" href="#328">328</a>             <strong>if</strong> (timestamp.length() == 4) {
<a name="329" href="#329">329</a>                 <em class="comment">// Add first month and first date.</em>
<a name="330" href="#330">330</a>                 timestamp = timestamp + <span class="string">"01010000"</span>;
<a name="331" href="#331">331</a>             }
<a name="332" href="#332">332</a>             <strong>if</strong> (timestamp.length() == 6) {
<a name="333" href="#333">333</a>                 <em class="comment">// Add a date of the first.</em>