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 //
  12 // Copyright (C) 2004 Novell, Inc (http://www.novell.com)
  13 //
  14 // Permission is hereby granted, free of charge, to any person obtaining
  15 // a copy of this software and associated documentation files (the
  16 // "Software"), to deal in the Software without restriction, including
  17 // without limitation the rights to use, copy, modify, merge, publish,
  18 // distribute, sublicense, and/or sell copies of the Software, and to
  19 // permit persons to whom the Software is furnished to do so, subject to
  20 // the following conditions:
  21 //
  22 // The above copyright notice and this permission notice shall be
  23 // included in all copies or substantial portions of the Software.
  24 //
  25 // THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
  26 // EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
  27 // MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
  28 // NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
  29 // LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
  30 // OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
  31 // WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  32 //
  33
  34 using System.Collections;
  35
  36 namespace System.Globalization {
  37
  38         [Serializable]
  39 #if NET_2_0
  40         [System.Runtime.InteropServices.ComVisible(true)]
  41 #endif
  42         public class StringInfo {
  43                 public StringInfo()
  44                 {
  45                 }
  46
  47                 public static string GetNextTextElement(string str)
  48                 {
  49                         if(str == null || str.Length == 0) {
  50                                 throw new ArgumentNullException("string is null");
  51                         }
  52                         return(GetNextTextElement (str, 0));
  53                 }
  54
  55                 public static string GetNextTextElement(string str, int index)
  56                 {
  57                         if(str == null) {
  58                                 throw new ArgumentNullException("string is null");
  59                         }
  60
  61                         if(index < 0 || index >= str.Length) {
  62                                 throw new ArgumentOutOfRangeException ("Index is not valid");
  63                         }
  64
  65                         /* Find the next base character, surrogate
  66                          * pair or combining character sequence
  67                          */
  68
  69                         char ch = str[index];
  70                         UnicodeCategory cat = char.GetUnicodeCategory (ch);
  71
  72                         if (cat == UnicodeCategory.Surrogate) {
  73                                 /* Check that it's a high surrogate
  74                                  * followed by a low surrogate
  75                                  */
  76                                 if (ch >= 0xD800 && ch <= 0xDBFF) {
  77                                         if ((index + 1) < str.Length &&
  78                                             str[index + 1] >= 0xDC00 &&
  79                                             str[index + 1] <= 0xDFFF) {
  80                                                 /* A valid surrogate pair */
  81                                                 return(str.Substring (index, 2));
  82                                         } else {
  83                                                 /* High surrogate on its own */
  84                                                 return(new String (ch, 1));
  85                                         }
  86                                 } else {
  87                                         /* Low surrogate on its own */
  88                                         return(new String (ch, 1));
  89                                 }
  90                         } else {
  91                                 /* Look for a base character, which
  92                                  * may or may not be followed by a
  93                                  * series of combining characters
  94                                  */
  95
  96                                 if (cat == UnicodeCategory.NonSpacingMark ||
  97                                     cat == UnicodeCategory.SpacingCombiningMark ||
  98                                     cat == UnicodeCategory.EnclosingMark) {
  99                                         /* Not a base character */
 100                                         return(new String (ch, 1));
 101                                 }
 102
 103                                 int count = 1;
 104
 105                                 while (index + count < str.Length) {
 106                                         cat = char.GetUnicodeCategory (str[index + count]);
 107                                         if (cat != UnicodeCategory.NonSpacingMark &&
 108                                             cat != UnicodeCategory.SpacingCombiningMark &&
 109                                             cat != UnicodeCategory.EnclosingMark) {
 110                                                 /* Finished the sequence */
 111                                                 break;
 112                                         }
 113                                         count++;
 114                                 }
 115
 116                                 return(str.Substring (index, count));
 117                         }
 118                 }
 119
 120                 public static TextElementEnumerator GetTextElementEnumerator(string str)
 121                 {
 122                         if(str == null || str.Length == 0) {
 123                                 throw new ArgumentNullException("string is null");
 124                         }
 125                         return(new TextElementEnumerator (str, 0));
 126                 }
 127
 128                 public static TextElementEnumerator GetTextElementEnumerator(string str, int index)
 129                 {
 130                         if(str == null) {
 131                                 throw new ArgumentNullException("string is null");
 132                         }
 133
 134                         if(index < 0 || index >= str.Length) {
 135                                 throw new ArgumentOutOfRangeException ("Index is not valid");
 136                         }
 137
 138                         return(new TextElementEnumerator (str, index));
 139                 }
 140
 141                 public static int[] ParseCombiningCharacters(string str)
 142                 {
 143                         if(str == null) {
 144                                 throw new ArgumentNullException("string is null");
 145                         }
 146
 147                         ArrayList indices = new ArrayList (str.Length);
 148                         TextElementEnumerator tee = GetTextElementEnumerator (str);
 149
 150                         tee.Reset ();
 151                         while(tee.MoveNext ()) {
 152                                 indices.Add (tee.ElementIndex);
 153                         }
 154
 155                         return((int[])indices.ToArray (typeof (int)));
 156                 }
 157         }
 158 }