EscapeUtil.java 4.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152
  1. package com.shs.official.common.utils.html;
  2. import com.shs.official.common.utils.StringUtils;
  3. /**
  4. * 转义和反转义工具类
  5. *
  6. * @author shs
  7. */
  8. public class EscapeUtil
  9. {
  10. public static final String RE_HTML_MARK = "(<[^<]*?>)|(<[\\s]*?/[^<]*?>)|(<[^<]*?/[\\s]*?>)";
  11. private static final char[][] TEXT = new char[64][];
  12. static
  13. {
  14. for (int i = 0; i < 64; i++)
  15. {
  16. TEXT[i] = new char[] { (char) i };
  17. }
  18. // special HTML characters
  19. TEXT['\''] = "&#039;".toCharArray(); // 单引号
  20. TEXT['"'] = "&#34;".toCharArray(); // 单引号
  21. TEXT['&'] = "&#38;".toCharArray(); // &符
  22. TEXT['<'] = "&#60;".toCharArray(); // 小于号
  23. TEXT['>'] = "&#62;".toCharArray(); // 大于号
  24. }
  25. /**
  26. * 转义文本中的HTML字符为安全的字符
  27. *
  28. * @param text 被转义的文本
  29. * @return 转义后的文本
  30. */
  31. public static String escape(String text)
  32. {
  33. return encode(text);
  34. }
  35. /**
  36. * 还原被转义的HTML特殊字符
  37. *
  38. * @param content 包含转义符的HTML内容
  39. * @return 转换后的字符串
  40. */
  41. public static String unescape(String content)
  42. {
  43. return decode(content);
  44. }
  45. /**
  46. * 清除所有HTML标签,但是不删除标签内的内容
  47. *
  48. * @param content 文本
  49. * @return 清除标签后的文本
  50. */
  51. public static String clean(String content)
  52. {
  53. return new HTMLFilter().filter(content);
  54. }
  55. /**
  56. * Escape编码
  57. *
  58. * @param text 被编码的文本
  59. * @return 编码后的字符
  60. */
  61. private static String encode(String text)
  62. {
  63. int len;
  64. if ((text == null) || ((len = text.length()) == 0))
  65. {
  66. return StringUtils.EMPTY;
  67. }
  68. StringBuilder buffer = new StringBuilder(len + (len >> 2));
  69. char c;
  70. for (int i = 0; i < len; i++)
  71. {
  72. c = text.charAt(i);
  73. if (c < 64)
  74. {
  75. buffer.append(TEXT[c]);
  76. }
  77. else
  78. {
  79. buffer.append(c);
  80. }
  81. }
  82. return buffer.toString();
  83. }
  84. /**
  85. * Escape解码
  86. *
  87. * @param content 被转义的内容
  88. * @return 解码后的字符串
  89. */
  90. public static String decode(String content)
  91. {
  92. if (StringUtils.isEmpty(content))
  93. {
  94. return content;
  95. }
  96. StringBuilder tmp = new StringBuilder(content.length());
  97. int lastPos = 0, pos = 0;
  98. char ch;
  99. while (lastPos < content.length())
  100. {
  101. pos = content.indexOf("%", lastPos);
  102. if (pos == lastPos)
  103. {
  104. if (content.charAt(pos + 1) == 'u')
  105. {
  106. ch = (char) Integer.parseInt(content.substring(pos + 2, pos + 6), 16);
  107. tmp.append(ch);
  108. lastPos = pos + 6;
  109. }
  110. else
  111. {
  112. ch = (char) Integer.parseInt(content.substring(pos + 1, pos + 3), 16);
  113. tmp.append(ch);
  114. lastPos = pos + 3;
  115. }
  116. }
  117. else
  118. {
  119. if (pos == -1)
  120. {
  121. tmp.append(content.substring(lastPos));
  122. lastPos = content.length();
  123. }
  124. else
  125. {
  126. tmp.append(content.substring(lastPos, pos));
  127. lastPos = pos;
  128. }
  129. }
  130. }
  131. return tmp.toString();
  132. }
  133. public static void main(String[] args)
  134. {
  135. String html = "alert('11111');";
  136. System.out.println(EscapeUtil.clean(html));
  137. System.out.println(EscapeUtil.escape(html));
  138. System.out.println(EscapeUtil.unescape(html));
  139. }
  140. }