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 using System.Runtime.InteropServices;
  36
  37 namespace System.Globalization {
  38
  39         [Serializable]
  40 #if NET_2_0
  41         [ComVisible(true)]
  42 #endif
  43         public class StringInfo {
  44                 public StringInfo()
  45                 {
  46                 }
  47
  48 #if NET_2_0
  49                 string s;
  50                 int length;
  51
  52                 public StringInfo (string value)
  53                 {
  54                         // Argument check in property
  55                         String = value;
  56                 }
  57
  58                 [ComVisible (false)]
  59                 public override bool Equals (object value)
  60                 {
  61                         StringInfo other = value as StringInfo;
  62                         return other != null && s == other.s;
  63                 }
  64
  65                 [ComVisible (false)]
  66                 public override int GetHashCode ()
  67                 {
  68                         return s.GetHashCode ();
  69                 }
  70
  71                 public int LengthInTextElements {
  72                         get {
  73                                 if (length < 0) {
  74                                         length = 0;
  75                                         for (int idx = 0; idx < s.Length; length++)
  76                                                 idx += GetNextTextElementLength (s, idx);
  77                                 }
  78                                 return length;
  79                         }
  80                 }
  81
  82                 public string String {
  83                         get { return s; }
  84                         set {
  85                                 if (value == null)
  86                                         throw new ArgumentNullException ("value");
  87                                 length = -1;
  88                                 s = value;
  89                         }
  90                 }
  91
  92                 public string SubstringByTextElements (int startingTextElement)
  93                 {
  94                         if (startingTextElement < 0 || s.Length == 0)
  95                                 throw new ArgumentOutOfRangeException ("startingTextElement");
  96                         int idx = 0;
  97                         for (int i = 0; i < startingTextElement; i++) {
  98                                 if (idx >= s.Length)
  99                                         throw new ArgumentOutOfRangeException ("startingTextElement");
 100                                 idx += GetNextTextElementLength (s, idx);
 101                         }
 102                         return s.Substring (idx);
 103                 }
 104
 105                 public string SubstringByTextElements (int startingTextElement, int lengthInTextElements)
 106                 {
 107                         if (startingTextElement < 0 || s.Length == 0)
 108                                 throw new ArgumentOutOfRangeException ("startingTextElement");
 109                         if (lengthInTextElements < 0)
 110                                 throw new ArgumentOutOfRangeException ("lengthInTextElements");
 111                         int idx = 0;
 112                         for (int i = 0; i < startingTextElement; i++) {
 113                                 if (idx >= s.Length)
 114                                         throw new ArgumentOutOfRangeException ("startingTextElement");
 115                                 idx += GetNextTextElementLength (s, idx);
 116                         }
 117                         int start = idx;
 118                         for (int i = 0; i < lengthInTextElements; i++) {
 119                                 if (idx >= s.Length)
 120                                         throw new ArgumentOutOfRangeException ("lengthInTextElements");
 121                                 idx += GetNextTextElementLength (s, idx);
 122                         }
 123                         return s.Substring (start, idx - start);
 124                 }
 125 #endif
 126
 127                 public static string GetNextTextElement(string str)
 128                 {
 129                         if(str == null || str.Length == 0) {
 130                                 throw new ArgumentNullException("string is null");
 131                         }
 132                         return(GetNextTextElement (str, 0));
 133                 }
 134
 135                 public static string GetNextTextElement(string str, int index)
 136                 {
 137                         int len = GetNextTextElementLength (str, index);
 138                         return len != 1 ? str.Substring (index, len) : new string (str [index], 1);
 139                 }
 140
 141                 static int GetNextTextElementLength(string str, int index)
 142                 {
 143                         if(str == null) {
 144                                 throw new ArgumentNullException("string is null");
 145                         }
 146
 147 #if NET_2_0
 148                         if(index >= str.Length)
 149                                 return 0;
 150                         if(index < 0)
 151 #else
 152                         if(index < 0 || index >= str.Length)
 153 #endif
 154                                 throw new ArgumentOutOfRangeException ("Index is not valid");
 155
 156                         /* Find the next base character, surrogate
 157                          * pair or combining character sequence
 158                          */
 159
 160                         char ch = str[index];
 161                         UnicodeCategory cat = char.GetUnicodeCategory (ch);
 162
 163                         if (cat == UnicodeCategory.Surrogate) {
 164                                 /* Check that it's a high surrogate
 165                                  * followed by a low surrogate
 166                                  */
 167                                 if (ch >= 0xD800 && ch <= 0xDBFF) {
 168                                         if ((index + 1) < str.Length &&
 169                                             str[index + 1] >= 0xDC00 &&
 170                                             str[index + 1] <= 0xDFFF) {
 171                                                 /* A valid surrogate pair */
 172                                                 return 2;
 173                                         } else {
 174                                                 /* High surrogate on its own */
 175                                                 return 1;
 176                                         }
 177                                 } else {
 178                                         /* Low surrogate on its own */
 179                                         return 1;
 180                                 }
 181                         } else {
 182                                 /* Look for a base character, which
 183                                  * may or may not be followed by a
 184                                  * series of combining characters
 185                                  */
 186
 187                                 if (cat == UnicodeCategory.NonSpacingMark ||
 188                                     cat == UnicodeCategory.SpacingCombiningMark ||
 189                                     cat == UnicodeCategory.EnclosingMark) {
 190                                         /* Not a base character */
 191                                         return 1;
 192                                 }
 193
 194                                 int count = 1;
 195
 196                                 while (index + count < str.Length) {
 197                                         cat = char.GetUnicodeCategory (str[index + count]);
 198                                         if (cat != UnicodeCategory.NonSpacingMark &&
 199                                             cat != UnicodeCategory.SpacingCombiningMark &&
 200                                             cat != UnicodeCategory.EnclosingMark) {
 201                                                 /* Finished the sequence */
 202                                                 break;
 203                                         }
 204                                         count++;
 205                                 }
 206
 207                                 return count;
 208                         }
 209                 }
 210
 211                 public static TextElementEnumerator GetTextElementEnumerator(string str)
 212                 {
 213                         if(str == null || str.Length == 0) {
 214                                 throw new ArgumentNullException("string is null");
 215                         }
 216                         return(new TextElementEnumerator (str, 0));
 217                 }
 218
 219                 public static TextElementEnumerator GetTextElementEnumerator(string str, int index)
 220                 {
 221                         if(str == null) {
 222                                 throw new ArgumentNullException("string is null");
 223                         }
 224
 225                         if(index < 0 || index >= str.Length) {
 226                                 throw new ArgumentOutOfRangeException ("Index is not valid");
 227                         }
 228
 229                         return(new TextElementEnumerator (str, index));
 230                 }
 231
 232                 public static int[] ParseCombiningCharacters(string str)
 233                 {
 234                         if(str == null) {
 235                                 throw new ArgumentNullException("string is null");
 236                         }
 237
 238                         ArrayList indices = new ArrayList (str.Length);
 239                         TextElementEnumerator tee = GetTextElementEnumerator (str);
 240
 241                         tee.Reset ();
 242                         while(tee.MoveNext ()) {
 243                                 indices.Add (tee.ElementIndex);
 244                         }
 245
 246                         return((int[])indices.ToArray (typeof (int)));
 247                 }
 248         }
 249 }