mcs/class/corlib/System.Globalization/StringInfo.cs

   1 //
   2 // System.Globalization.StringInfo.cs
   3 //
   4 // Author:
   5 //      Dick Porter (dick@ximian.com)
   6 //
   7 // (C) 2002 Ximian, Inc.
   8 // (C) 2004 Novell, Inc.
   9 //
  10
  11 using System.Collections;
  12
  13 namespace System.Globalization {
  14
  15         [Serializable]
  16         public class StringInfo {
  17                 public StringInfo()
  18                 {
  19                 }
  20
  21                 public static string GetNextTextElement(string str)
  22                 {
  23                         if(str == null || str.Length == 0) {
  24                                 throw new ArgumentNullException("string is null");
  25                         }
  26                         return(GetNextTextElement (str, 0));
  27                 }
  28
  29                 public static string GetNextTextElement(string str, int index)
  30                 {
  31                         if(str == null) {
  32                                 throw new ArgumentNullException("string is null");
  33                         }
  34
  35                         if(index < 0 || index >= str.Length) {
  36                                 throw new ArgumentOutOfRangeException ("Index is not valid");
  37                         }
  38
  39                         /* Find the next base character, surrogate
  40                          * pair or combining character sequence
  41                          */
  42
  43                         char ch = str[index];
  44                         UnicodeCategory cat = char.GetUnicodeCategory (ch);
  45
  46                         if (cat == UnicodeCategory.Surrogate) {
  47                                 /* Check that it's a high surrogate
  48                                  * followed by a low surrogate
  49                                  */
  50                                 if (ch >= 0xD800 && ch <= 0xDBFF) {
  51                                         if ((index + 1) < str.Length &&
  52                                             str[index + 1] >= 0xDC00 &&
  53                                             str[index + 1] <= 0xDFFF) {
  54                                                 /* A valid surrogate pair */
  55                                                 return(str.Substring (index, 2));
  56                                         } else {
  57                                                 /* High surrogate on its own */
  58                                                 return(new String (ch, 1));
  59                                         }
  60                                 } else {
  61                                         /* Low surrogate on its own */
  62                                         return(new String (ch, 1));
  63                                 }
  64                         } else {
  65                                 /* Look for a base character, which
  66                                  * may or may not be followed by a
  67                                  * series of combining characters
  68                                  */
  69
  70                                 if (cat == UnicodeCategory.NonSpacingMark ||
  71                                     cat == UnicodeCategory.SpacingCombiningMark ||
  72                                     cat == UnicodeCategory.EnclosingMark) {
  73                                         /* Not a base character */
  74                                         return(new String (ch, 1));
  75                                 }
  76
  77                                 int count = 1;
  78
  79                                 while (index + count < str.Length) {
  80                                         cat = char.GetUnicodeCategory (str[index + count]);
  81                                         if (cat != UnicodeCategory.NonSpacingMark &&
  82                                             cat != UnicodeCategory.SpacingCombiningMark &&
  83                                             cat != UnicodeCategory.EnclosingMark) {
  84                                                 /* Finished the sequence */
  85                                                 break;
  86                                         }
  87                                         count++;
  88                                 }
  89
  90                                 return(str.Substring (index, count));
  91                         }
  92                 }
  93
  94                 public static TextElementEnumerator GetTextElementEnumerator(string str)
  95                 {
  96                         if(str == null || str.Length == 0) {
  97                                 throw new ArgumentNullException("string is null");
  98                         }
  99                         return(new TextElementEnumerator (str, 0));
 100                 }
 101
 102                 public static TextElementEnumerator GetTextElementEnumerator(string str, int index)
 103                 {
 104                         if(str == null) {
 105                                 throw new ArgumentNullException("string is null");
 106                         }
 107
 108                         if(index < 0 || index >= str.Length) {
 109                                 throw new ArgumentOutOfRangeException ("Index is not valid");
 110                         }
 111
 112                         return(new TextElementEnumerator (str, index));
 113                 }
 114
 115                 public static int[] ParseCombiningCharacters(string str)
 116                 {
 117                         if(str == null) {
 118                                 throw new ArgumentNullException("string is null");
 119                         }
 120
 121                         ArrayList indices = new ArrayList (str.Length);
 122                         TextElementEnumerator tee = GetTextElementEnumerator (str);
 123
 124                         tee.Reset ();
 125                         while(tee.MoveNext ()) {
 126                                 indices.Add (tee.ElementIndex);
 127                         }
 128
 129                         return((int[])indices.ToArray (typeof (int)));
 130                 }
 131         }
 132 }