RegexUtil


1   /*
2    * Created on Nov 8, 2003
3    *
4    */
5   package org.roller.util;
6   import org.apache.commons.codec.binary.Hex;
7   
8   import java.io.UnsupportedEncodingException  ;
9   import java.util.ArrayList  ;
10  import java.util.regex.Matcher  ;
11  import java.util.regex.Pattern  ;
12  
13  
14  /**
15   * @author lance
16   */
17  public class RegexUtil
18  {
19      public static final Pattern   mailtoPattern = Pattern.compile("mailto:([a-zA-Z0-9\\.]+@[a-zA-Z0-9\\.]+\\.[a-zA-Z0-9]+)");
20      public static final Pattern   emailPattern = Pattern.compile("\\b[a-zA-Z0-9\\.]+(@)([a-zA-Z0-9\\.]+)(\\.)([a-zA-Z0-9]+)\\b");
21      
22      public static String   encodeEmail(String   str)
23      {
24          // obfuscate mailto's: turns them into hex encoded,
25          // so that browsers can still understand the mailto link
26          Matcher   mailtoMatch = mailtoPattern.matcher(str);
27          while (mailtoMatch.find())
28          {
29              String   email = mailtoMatch.group(1);
30              //System.out.println("email=" + email);
31              String   hexed = encode(email);
32              str = str.replaceFirst("mailto:"+email, "mailto:"+hexed);
33          }
34          
35          return obfuscateEmail(str);
36      }
37  
38      /**
39       * obfuscate plaintext emails: makes them
40       * "human-readable" - still too easy for
41       * machines to parse however.
42       * 
43       * @param str
44       * @return
45       */
46      public static String   obfuscateEmail(String   str)
47      {
48          Matcher   emailMatch = emailPattern.matcher(str);
49          while (emailMatch.find())
50          {
51              String   at = emailMatch.group(1);
52              //System.out.println("at=" + at);
53              str = str.replaceFirst(at, "-AT-");
54              
55              String   dot = emailMatch.group(2) + emailMatch.group(3) + emailMatch.group(4);
56              String   newDot = emailMatch.group(2) + "-DOT-" + emailMatch.group(4);
57              //System.out.println("dot=" + dot);
58              str = str.replaceFirst(dot, newDot);
59          }
60          return str;
61      }
62      
63      /**
64       * Return the specified match "groups" from the pattern.
65       * For each group matched a String will be entered in the ArrayList.
66       * 
67       * @param pattern The Pattern to use.
68       * @param match The String to match against.
69       * @param group The group number to return in case of a match.
70       * @return
71       */
72      public static ArrayList   getMatches(Pattern   pattern, String   match, int group)
73      {
74          ArrayList   matches = new ArrayList  ();
75          Matcher   matcher = pattern.matcher(match);
76          while (matcher.find()) 
77          {
78              matches.add( matcher.group(group) );
79          }
80          return matches;
81      }
82  
83      /**
84       * Thanks to the folks at Blojsom (http://sf.net/projects/blojsom)
85       * for showing me what I was doing wrong with the Hex class.
86       * 
87       * @param email
88       * @return
89       */
90      public static String   encode(String   email)
91      {
92          StringBuffer   result = new StringBuffer  ();
93          try {
94              char[] hexString = Hex.encodeHex(email.getBytes("UTF8"));
95              for (int i = 0; i < hexString.length; i++) {
96                  if (i % 2 == 0) {
97                      result.append("%");
98                  }
99                  result.append(hexString[i]);
100             }
101         } catch (UnsupportedEncodingException   e) {
102             return email;
103         }
104 
105         return result.toString();
106     }
107 }
108
A to Z: JavaDoc & Examples Daily Java News & Articles Open Source Projects Open Source Codes Free Computer Books Remove Frame
Popular Tags