Coverage Report

Coverage Report - yarfraw.utils.CommonUtils

Classes in this File

83%

90%

 package yarfraw.utils;
 
 import static yarfraw.utils.CommonConstants.MIN_PER_DAY;
 import static yarfraw.utils.CommonConstants.MIN_PER_MONTH;
 import static yarfraw.utils.CommonConstants.MIN_PER_WEEK;
 import static yarfraw.utils.CommonConstants.MIN_PER_YEAR;
 
 import java.math.BigInteger;
 import java.text.ParseException;
 import java.text.SimpleDateFormat;
 import java.util.Date;
 
 import org.apache.commons.lang.StringUtils;
 import org.apache.commons.logging.Log;
 import org.apache.commons.logging.LogFactory;
 
 import yarfraw.core.datamodel.FeedFormat;
 import yarfraw.core.datamodel.YarfrawException;
 import yarfraw.generated.rss10.elements.UpdatePeriodEnum;
 
 /**
  * Utilities methods.
  * 
  * @author jliang
  *
  */
 public class CommonUtils{
   private static final Log LOG = LogFactory.getLog(CommonUtils.class);
   
   public static final String RSS20_JAXB_CONTEXT = "yarfraw.generated.rss20.elements";
   public static final String RSS10_JAXB_CONTEXT = "yarfraw.generated.rss10.elements";
   public static final String ATOM10_JAXB_CONTEXT = "yarfraw.generated.atom10.elements";
   public static final String ATOM03_JAXB_CONTEXT = "yarfraw.generated.atom03.elements";
   
   /////////////////////DATE PARSING///////////////////////////////////
   public static final String RFC822DATE_PATTERN = "EEE, dd MMM yyyy HH:mm:ss zzz";
   public static final String ISO8601DATE_PATTERN = "yyyy-MM-dd'T'HH:mm:ssZ";
   public static final String ISO_8601_LVL1_PATTERN = "yyyy";
   public static final String ISO_8601_LVL2_PATTERN = "yyyy-MM";
   public static final String ISO_8601_LVL3_PATTERN = "yyyy-MM-dd";
   public static final String ISO_8601_LVL4_PATTERN = "yyyy-MM-dd'T'HH:mmZ";
   public static final String ISO_8601_LVL5_PATTERN = "yyyy-MM-dd'T'HH:mm:ssZ";
   public static final String ISO_8601_LVL6_PATTERN = "yyyy-MM-dd'T'HH:mm:ss.sZ";
   
   public static final String NON_STANDARD_PATTERN1 = "EEE, dd MMM yyyy HH:mm:ss z";
   public static final String NON_STANDARD_PATTERN2 = "EEE, dd MMM yyyy HH:mm zzzz";
   public static final String NON_STANDARD_PATTERN3 = "EEE, dd MMM yy HH:mm:ss z";
   public static final String NON_STANDARD_PATTERN4 = "yyyy-MM-dd'T'HH:mm:ss.SSSzzzz";
   public static final String NON_STANDARD_PATTERN5 = "yyyy-MM-dd'T'HH:mm:sszzzz";
   public static final String NON_STANDARD_PATTERN6 = "yyyy-MM-dd'T'HH:mm:ss";
   public static final String NON_STANDARD_PATTERN7 = "yyyy-MM-dd'T'HH:mm:ss.sZ";
   
   
   //6 level of ISO 8601 Date
   private static final SimpleDateFormat ISO_8601_LVL1 = new SimpleDateFormat(ISO_8601_LVL1_PATTERN);
   private static final SimpleDateFormat ISO_8601_LVL2 = new SimpleDateFormat(ISO_8601_LVL2_PATTERN);
   private static final SimpleDateFormat ISO_8601_LVL3 = new SimpleDateFormat(ISO_8601_LVL3_PATTERN);
   private static final SimpleDateFormat ISO_8601_LVL4 = new SimpleDateFormat(ISO_8601_LVL4_PATTERN);
   private static final SimpleDateFormat ISO_8601_LVL5 = new SimpleDateFormat(ISO_8601_LVL5_PATTERN);
   private static final SimpleDateFormat ISO_8601_LVL6 = new SimpleDateFormat(ISO_8601_LVL6_PATTERN);
   
   private static final SimpleDateFormat RFC822DATE_FORMAT = new SimpleDateFormat(RFC822DATE_PATTERN);
 
   private static final SimpleDateFormat[] RFC_FORMAT = new SimpleDateFormat[]{
     RFC822DATE_FORMAT, 
     new SimpleDateFormat(NON_STANDARD_PATTERN1),
     new SimpleDateFormat(NON_STANDARD_PATTERN2),
     new SimpleDateFormat(NON_STANDARD_PATTERN3)
   };
   
   private static final SimpleDateFormat[] NON_STANDARD_ISO_FORMAT = new SimpleDateFormat[]{
     new SimpleDateFormat(NON_STANDARD_PATTERN4),
     new SimpleDateFormat(NON_STANDARD_PATTERN5),
     new SimpleDateFormat(NON_STANDARD_PATTERN6),
     new SimpleDateFormat(NON_STANDARD_PATTERN7)
   };
 
   private CommonUtils(){}
 
   /**
    * Determines whether the input <code>dateString</code> is valid based on the input <code>FeedFormat</code>.
    * @param dateString
    * @param format
    * @return
    */
   public static synchronized boolean isDateFormatValid(String dateString, FeedFormat format){
     if(format == FeedFormat.ATOM10 || format == FeedFormat.RSS10 || format == FeedFormat.ATOM03){
       try {
         return tryParseISODate(dateString) != null ;
       } catch (Exception e) {
         //non strict ISO format
         return tryParseNonStandardIsoDates(dateString) != null;
       }
     }else if(format == FeedFormat.RSS20){
       try{
         return tryParseRfcDates(dateString) != null;
       } catch (Exception e) {
         return false;
       }
     }else{
       throw new IllegalArgumentException("Unsupported format: "+ format);
     }
   }
   
   /**
    * Remove last occurrence of the character c in s
    */
   private static String removeLast(String s, char c){
     int idx = s.lastIndexOf(c);
     if(idx < 0){
       return s;//nothing to remove
     }
     return s.substring(0, idx)+ (idx==s.length()-1? StringUtils.EMPTY: s.substring(idx+1));
   }
 
 
   /**
    * Parse a date string using both ISO and RFC formats and some non standard formats.
    * 
    * @param dateString
    * @return {@link Date} representation of the input string. null if
    * unable to parse input string.
    */
   public synchronized static Date tryParseDate(String dateString){
     Date ret = null;
     try {
       ret = tryParseISODate(dateString);
       return ret;
     } catch (Exception e) {
       ret = tryParseNonStandardIsoDates(dateString);
       if(ret != null){
         return ret;
       }else{
         ret = tryParseRfcDates(dateString);
       }
     }
     if(ret == null){
       LOG.warn("Unparsable dateString "+dateString+", returning null");
     }
     return ret;
   }
 
   /**
    * Try to parse using rfc format
    * @param dateString
    * @return
    */
   private static Date tryParseRfcDates(String dateString){
     Date ret = null;
     for(SimpleDateFormat format : RFC_FORMAT){
       try {
         ret = format.parse(dateString);
         return ret;
       } catch (Exception ee) {
         //keep trying
       }
     }
     return ret;
   }
   
   /**
    * Try to parse using non standard ISO formats
    * @param dateString
    * @return
    */
   private static Date tryParseNonStandardIsoDates(String dateString){
     Date ret = null;
     for(SimpleDateFormat format : NON_STANDARD_ISO_FORMAT){
       try {
         ret = format.parse(dateString);
         return ret;
       } catch (Exception ee) {
         //keep trying
       }
     }
     return ret;
   }
   
   /**
    * Format a {@link Date} object to string based on the input {@link FeedFormat}.
    * For Atom 1.0 and RSS 1.0 it will be formatted as ISO8601 Level 5 string. <br/>
    * http://www.w3.org/TR/NOTE-datetime
    * <br/>
    * For RSS 2.0, it will be formatted as RFC 822 date string.
    * <br/>
    * http://www.faqs.org/rfcs/rfc822.html
    * @param date any date 
    * @param format any {@link FeedFormat}
    * @return null if either input is null. Formatted date string otherwise.
    */
   public static synchronized String formatDate(Date date, FeedFormat format){
     if(date == null || format == null){
       return null;
     }
     if(format == FeedFormat.ATOM10 || format == FeedFormat.RSS20 || format == FeedFormat.ATOM03){
       return getDateAsISO8601String(date);
     }else if(format == FeedFormat.RSS10 ){
       return RFC822DATE_FORMAT.format(date);
     }else{
       throw new IllegalArgumentException("Unsupported format: "+ format);
     }
   }
   
   /**
    * Simple date format does not support ISO0861 date, so i have to do some hacking
    * <br/>
    * return null if date is null
    */
   private static String getDateAsISO8601String(Date date)
   { 
     if(date == null){
       return null;
     }
     String result = ISO_8601_LVL5.format(date);
     //convert YYYYMMDDTHH:mm:ss+HH00 into YYYYMMDDTHH:mm:ss+HH:00
     //- note the added colon for the Timezone
     result = result.substring(0, result.length()-2)
       + ":" + result.substring(result.length()-2);
     return result;
   }
   
   /**
    * Try to parse a date string using different formatting string.
    * <br/>
    * return null if dateString is null
    * @throws YarfrawException 
    * @throws ParseException 
    */
   private synchronized static Date tryParseISODate(String dateString) throws YarfrawException, ParseException{
     
     if(dateString == null){
       return null;
     }else if(dateString.length() == 4){
       return ISO_8601_LVL1.parse(dateString);
     }else if(dateString.length() == 7){
       return ISO_8601_LVL2.parse(dateString);
     }else if(dateString.length() == 10){
       return ISO_8601_LVL3.parse(dateString);
     }else if(dateString.length() == 22){
       return ISO_8601_LVL4.parse(removeLast(dateString, ':'));
     }else if(dateString.length() == 25){
       return ISO_8601_LVL5.parse(removeLast(dateString, ':'));
     }else if(dateString.length() == 28){
       return ISO_8601_LVL6.parse(removeLast(dateString, ':'));
     }else{
       throw new YarfrawException("Invalid ISO 8601 Date format: "+dateString);
     }
   }
   
 /////////////////////DATE PARSING///////////////////////////////////
   
   
   /**
    * calculate the ttl value from updatePeriod and updateFrequency
    * @return null if anything unexpcted occurs
    */
   public static Integer calculateTtl(UpdatePeriodEnum updatePeriod, BigInteger updateFrequency){
     if(updatePeriod == null && updateFrequency == null){
       return null;
     }
     int freq = updateFrequency == null ? 1: updateFrequency.intValue();
     if(updatePeriod == UpdatePeriodEnum.HOURLY){
       return Math.max(1, 60/freq);
     }else if(updatePeriod == UpdatePeriodEnum.DAILY){
       return Math.max(1, MIN_PER_DAY/freq);
     }else if(updatePeriod == UpdatePeriodEnum.MONTHLY){
       return Math.max(1, MIN_PER_MONTH/freq);
     }else if(updatePeriod == UpdatePeriodEnum.WEEKLY){
       return Math.max(1, MIN_PER_WEEK/freq);
     }else if(updatePeriod == UpdatePeriodEnum.YEARLY){
       return Math.max(1, MIN_PER_YEAR/freq);
     }else{
       return null;
     }
   }
 }

1		package yarfraw.utils;
2
3		import static yarfraw.utils.CommonConstants.MIN_PER_DAY;
4		import static yarfraw.utils.CommonConstants.MIN_PER_MONTH;
5		import static yarfraw.utils.CommonConstants.MIN_PER_WEEK;
6		import static yarfraw.utils.CommonConstants.MIN_PER_YEAR;
7
8		import java.math.BigInteger;
9		import java.text.ParseException;
10		import java.text.SimpleDateFormat;
11		import java.util.Date;
12
13		import org.apache.commons.lang.StringUtils;
14		import org.apache.commons.logging.Log;
15		import org.apache.commons.logging.LogFactory;
16
17		import yarfraw.core.datamodel.FeedFormat;
18		import yarfraw.core.datamodel.YarfrawException;
19		import yarfraw.generated.rss10.elements.UpdatePeriodEnum;
20
21		/**
22		* Utilities methods.
23		*
24		* @author jliang
25		*
26		*/
27	33	public class CommonUtils{
28	33	private static final Log LOG = LogFactory.getLog(CommonUtils.class);
29
30		public static final String RSS20_JAXB_CONTEXT = "yarfraw.generated.rss20.elements";
31		public static final String RSS10_JAXB_CONTEXT = "yarfraw.generated.rss10.elements";
32		public static final String ATOM10_JAXB_CONTEXT = "yarfraw.generated.atom10.elements";
33		public static final String ATOM03_JAXB_CONTEXT = "yarfraw.generated.atom03.elements";
34
35		/////////////////////DATE PARSING///////////////////////////////////
36		public static final String RFC822DATE_PATTERN = "EEE, dd MMM yyyy HH:mm:ss zzz";
37		public static final String ISO8601DATE_PATTERN = "yyyy-MM-dd'T'HH:mm:ssZ";
38		public static final String ISO_8601_LVL1_PATTERN = "yyyy";
39		public static final String ISO_8601_LVL2_PATTERN = "yyyy-MM";
40		public static final String ISO_8601_LVL3_PATTERN = "yyyy-MM-dd";
41		public static final String ISO_8601_LVL4_PATTERN = "yyyy-MM-dd'T'HH:mmZ";
42		public static final String ISO_8601_LVL5_PATTERN = "yyyy-MM-dd'T'HH:mm:ssZ";
43		public static final String ISO_8601_LVL6_PATTERN = "yyyy-MM-dd'T'HH:mm:ss.sZ";
44
45		public static final String NON_STANDARD_PATTERN1 = "EEE, dd MMM yyyy HH:mm:ss z";
46		public static final String NON_STANDARD_PATTERN2 = "EEE, dd MMM yyyy HH:mm zzzz";
47		public static final String NON_STANDARD_PATTERN3 = "EEE, dd MMM yy HH:mm:ss z";
48		public static final String NON_STANDARD_PATTERN4 = "yyyy-MM-dd'T'HH:mm:ss.SSSzzzz";
49		public static final String NON_STANDARD_PATTERN5 = "yyyy-MM-dd'T'HH:mm:sszzzz";
50		public static final String NON_STANDARD_PATTERN6 = "yyyy-MM-dd'T'HH:mm:ss";
51		public static final String NON_STANDARD_PATTERN7 = "yyyy-MM-dd'T'HH:mm:ss.sZ";
52
53
54		//6 level of ISO 8601 Date
55	33	private static final SimpleDateFormat ISO_8601_LVL1 = new SimpleDateFormat(ISO_8601_LVL1_PATTERN);
56	33	private static final SimpleDateFormat ISO_8601_LVL2 = new SimpleDateFormat(ISO_8601_LVL2_PATTERN);
57	33	private static final SimpleDateFormat ISO_8601_LVL3 = new SimpleDateFormat(ISO_8601_LVL3_PATTERN);
58	33	private static final SimpleDateFormat ISO_8601_LVL4 = new SimpleDateFormat(ISO_8601_LVL4_PATTERN);
59	33	private static final SimpleDateFormat ISO_8601_LVL5 = new SimpleDateFormat(ISO_8601_LVL5_PATTERN);
60	33	private static final SimpleDateFormat ISO_8601_LVL6 = new SimpleDateFormat(ISO_8601_LVL6_PATTERN);
61
62	33	private static final SimpleDateFormat RFC822DATE_FORMAT = new SimpleDateFormat(RFC822DATE_PATTERN);
63
64	66	private static final SimpleDateFormat[] RFC_FORMAT = new SimpleDateFormat[]{
65	33	RFC822DATE_FORMAT,
66	33	new SimpleDateFormat(NON_STANDARD_PATTERN1),
67	33	new SimpleDateFormat(NON_STANDARD_PATTERN2),
68	33	new SimpleDateFormat(NON_STANDARD_PATTERN3)
69		};
70
71	66	private static final SimpleDateFormat[] NON_STANDARD_ISO_FORMAT = new SimpleDateFormat[]{
72	33	new SimpleDateFormat(NON_STANDARD_PATTERN4),
73	33	new SimpleDateFormat(NON_STANDARD_PATTERN5),
74	33	new SimpleDateFormat(NON_STANDARD_PATTERN6),
75	33	new SimpleDateFormat(NON_STANDARD_PATTERN7)
76		};
77
78	0	private CommonUtils(){}
79
80		/**
81		* Determines whether the input <code>dateString</code> is valid based on the input <code>FeedFormat</code>.
82		* @param dateString
83		* @param format
84		* @return
85		*/
86		public static synchronized boolean isDateFormatValid(String dateString, FeedFormat format){
87	798	if(format == FeedFormat.ATOM10 \|\| format == FeedFormat.RSS10 \|\| format == FeedFormat.ATOM03){
88		try {
89	195	return tryParseISODate(dateString) != null ;
90	63	} catch (Exception e) {
91		//non strict ISO format
92	63	return tryParseNonStandardIsoDates(dateString) != null;
93		}
94	603	}else if(format == FeedFormat.RSS20){
95		try{
96	603	return tryParseRfcDates(dateString) != null;
97	0	} catch (Exception e) {
98	0	return false;
99		}
100		}else{
101	0	throw new IllegalArgumentException("Unsupported format: "+ format);
102		}
103		}
104
105		/**
106		* Remove last occurrence of the character c in s
107		*/
108		private static String removeLast(String s, char c){
109	138	int idx = s.lastIndexOf(c);
110	138	if(idx < 0){
111	0	return s;//nothing to remove
112		}
113	138	return s.substring(0, idx)+ (idx==s.length()-1? StringUtils.EMPTY: s.substring(idx+1));
114		}
115
116
117		/**
118		* Parse a date string using both ISO and RFC formats and some non standard formats.
119		*
120		* @param dateString
121		* @return {@link Date} representation of the input string. null if
122		* unable to parse input string.
123		*/
124		public synchronized static Date tryParseDate(String dateString){
125	195	Date ret = null;
126		try {
127	195	ret = tryParseISODate(dateString);
128	75	return ret;
129	120	} catch (Exception e) {
130	120	ret = tryParseNonStandardIsoDates(dateString);
131	120	if(ret != null){
132	3	return ret;
133		}else{
134	117	ret = tryParseRfcDates(dateString);
135		}
136		}
137	117	if(ret == null){
138	3	LOG.warn("Unparsable dateString "+dateString+", returning null");
139		}
140	117	return ret;
141		}
142
143		/**
144		* Try to parse using rfc format
145		* @param dateString
146		* @return
147		*/
148		private static Date tryParseRfcDates(String dateString){
149	720	Date ret = null;
150	732	for(SimpleDateFormat format : RFC_FORMAT){
151		try {
152	729	ret = format.parse(dateString);
153	717	return ret;
154	12	} catch (Exception ee) {
155		//keep trying
156		}
157		}
158	3	return ret;
159		}
160
161		/**
162		* Try to parse using non standard ISO formats
163		* @param dateString
164		* @return
165		*/
166		private static Date tryParseNonStandardIsoDates(String dateString){
167	183	Date ret = null;
168	885	for(SimpleDateFormat format : NON_STANDARD_ISO_FORMAT){
169		try {
170	717	ret = format.parse(dateString);
171	15	return ret;
172	702	} catch (Exception ee) {
173		//keep trying
174		}
175		}
176	168	return ret;
177		}
178
179		/**
180		* Format a {@link Date} object to string based on the input {@link FeedFormat}.
181		* For Atom 1.0 and RSS 1.0 it will be formatted as ISO8601 Level 5 string. <br/>
182		* http://www.w3.org/TR/NOTE-datetime
183		* <br/>
184		* For RSS 2.0, it will be formatted as RFC 822 date string.
185		* <br/>
186		* http://www.faqs.org/rfcs/rfc822.html
187		* @param date any date
188		* @param format any {@link FeedFormat}
189		* @return null if either input is null. Formatted date string otherwise.
190		*/
191		public static synchronized String formatDate(Date date, FeedFormat format){
192	57	if(date == null \|\| format == null){
193	3	return null;
194		}
195	54	if(format == FeedFormat.ATOM10 \|\| format == FeedFormat.RSS20 \|\| format == FeedFormat.ATOM03){
196	6	return getDateAsISO8601String(date);
197	48	}else if(format == FeedFormat.RSS10 ){
198	48	return RFC822DATE_FORMAT.format(date);
199		}else{
200	0	throw new IllegalArgumentException("Unsupported format: "+ format);
201		}
202		}
203
204		/**
205		* Simple date format does not support ISO0861 date, so i have to do some hacking
206		* <br/>
207		* return null if date is null
208		*/
209		private static String getDateAsISO8601String(Date date)
210		{
211	6	if(date == null){
212	0	return null;
213		}
214	6	String result = ISO_8601_LVL5.format(date);
215		//convert YYYYMMDDTHH:mm:ss+HH00 into YYYYMMDDTHH:mm:ss+HH:00
216		//- note the added colon for the Timezone
217	12	result = result.substring(0, result.length()-2)
218	6	+ ":" + result.substring(result.length()-2);
219	6	return result;
220		}
221
222		/**
223		* Try to parse a date string using different formatting string.
224		* <br/>
225		* return null if dateString is null
226		* @throws YarfrawException
227		* @throws ParseException
228		*/
229		private synchronized static Date tryParseISODate(String dateString) throws YarfrawException, ParseException{
230
231	390	if(dateString == null){
232	66	return null;
233	324	}else if(dateString.length() == 4){
234	0	return ISO_8601_LVL1.parse(dateString);
235	324	}else if(dateString.length() == 7){
236	0	return ISO_8601_LVL2.parse(dateString);
237	324	}else if(dateString.length() == 10){
238	3	return ISO_8601_LVL3.parse(dateString);
239	321	}else if(dateString.length() == 22){
240	9	return ISO_8601_LVL4.parse(removeLast(dateString, ':'));
241	312	}else if(dateString.length() == 25){
242	129	return ISO_8601_LVL5.parse(removeLast(dateString, ':'));
243	183	}else if(dateString.length() == 28){
244	0	return ISO_8601_LVL6.parse(removeLast(dateString, ':'));
245		}else{
246	183	throw new YarfrawException("Invalid ISO 8601 Date format: "+dateString);
247		}
248		}
249
250		/////////////////////DATE PARSING///////////////////////////////////
251
252
253		/**
254		* calculate the ttl value from updatePeriod and updateFrequency
255		* @return null if anything unexpcted occurs
256		*/
257		public static Integer calculateTtl(UpdatePeriodEnum updatePeriod, BigInteger updateFrequency){
258	27	if(updatePeriod == null && updateFrequency == null){
259	15	return null;
260		}
261	12	int freq = updateFrequency == null ? 1: updateFrequency.intValue();
262	12	if(updatePeriod == UpdatePeriodEnum.HOURLY){
263	9	return Math.max(1, 60/freq);
264	3	}else if(updatePeriod == UpdatePeriodEnum.DAILY){
265	3	return Math.max(1, MIN_PER_DAY/freq);
266	0	}else if(updatePeriod == UpdatePeriodEnum.MONTHLY){
267	0	return Math.max(1, MIN_PER_MONTH/freq);
268	0	}else if(updatePeriod == UpdatePeriodEnum.WEEKLY){
269	0	return Math.max(1, MIN_PER_WEEK/freq);
270	0	}else if(updatePeriod == UpdatePeriodEnum.YEARLY){
271	0	return Math.max(1, MIN_PER_YEAR/freq);
272		}else{
273	0	return null;
274		}
275		}
276		}