emoji过滤

Posted 2020-10-29

因项目要保存用户微信昵称，其中多含emoji表情，mysql版本低于5.5.3无法保存（不敢升），只有转换为unicode和过滤两种解决方法，搜得过滤的文章一篇，稍作改动，自测通过

原文地址：https://www.oschina.net/question/89964_105220

 1 import org.apache.commons.lang.StringUtils;
 2 
 3 public class EmojiUtil {
 4 
 5     /**
 6      * 检测是否有emoji字符
 7      * @param source
 8      * @return 
 9      */
10     public static boolean containsEmoji(String source) {
11         if (StringUtils.isBlank(source)) {
12             return false;
13         }
14 
15         int len = source.length();
16 
17         for (int i = 0; i < len; i++) {
18             char codePoint = source.charAt(i);
19 
20             if (isEmojiCharacter(codePoint)) {
21                 //do nothing，判断到了这里表明，确认有表情字符
22                 return true;
23             }
24         }
25 
26         return false;
27     }
28 
29     private static boolean isNotEmojiCharacter(char codePoint) {
30         return (codePoint == 0x0) ||
31                 (codePoint == 0x9) ||
32                 (codePoint == 0xA) ||
33                 (codePoint == 0xD) ||
34                 ((codePoint >= 0x20) && (codePoint <= 0xD7FF)) ||
35                 ((codePoint >= 0xE000) && (codePoint <= 0xFFFD)) ||
36                 ((codePoint >= 0x10000) && (codePoint <= 0x10FFFF));
37     }
38     
39     private static boolean isEmojiCharacter(char codePoint) {
40         return !isNotEmojiCharacter(codePoint);
41     }
42 
43     /**
44      * 过滤emoji
45      * @param source
46      * @return 
47      */
48     public static String filterEmoji(String source) {
49         if (!containsEmoji(source)) {
50             return source;
51         }
52         StringBuilder sb = new StringBuilder();
53         int len = source.length();
54         for (int i = 0; i < len; i++) {
55             char c = source.charAt(i);
56             if (isEmojiCharacter(c)) {
57                 continue;
58             }
59             sb.append(c);
60         }
61         return sb.toString();
62     }
63     
64     /***
65      * 提取文本中的emoji表情
66      * @param source
67      * @return
68      */
69     public static String extractEmoji(String source) {
70         if (!containsEmoji(source)) {
71             return null;
72         }
73         StringBuilder sb = new StringBuilder();
74         int len = source.length();
75         for (int i = 0; i < len; i++) {
76             char codePoint = source.charAt(i);
77             if (isEmojiCharacter(codePoint)) {
78                 sb.append(codePoint);
79             } 
80         }
81         return sb.toString();
82     }
83     
84     public static void main(String[] args) {
85         String s = "神奇{6}66Abc!";
86         String emoji = UnicodeUtil.unicode2String("\\u4d\\u6f\\u4d\\u6f\\u9171\\ud83c\\udf51");
87         System.out.println(s + "是否包含emoji: " + containsEmoji(s));
88         System.out.println(emoji + "是否包含emoji: " + containsEmoji(emoji));
89         System.out.println(s + "过滤emoji后: " + filterEmoji(s));
90         System.out.println(emoji + "过滤emoji后: " + filterEmoji(emoji));
91         System.out.println(s + "提取出的emoji: " + extractEmoji(s));
92         System.out.println(emoji + "提取出的emoji: " + extractEmoji(emoji));
93     }
94 }