mcs/class/Microsoft.Build.Engine/Microsoft.Build.BuildEngine/ConditionTokenizer.cs

   1 //
   2 // ConditionTokenizer.cs
   3 //
   4 // Author:
   5 //   Marek Sieradzki (marek.sieradzki@gmail.com)
   6 //   Jaroslaw Kowalski <jaak@jkowalski.net>
   7 //
   8 // (C) 2006 Marek Sieradzki
   9 // (C) 2004-2006 Jaroslaw Kowalski
  10 //
  11 // Permission is hereby granted, free of charge, to any person obtaining
  12 // a copy of this software and associated documentation files (the
  13 // "Software"), to deal in the Software without restriction, including
  14 // without limitation the rights to use, copy, modify, merge, publish,
  15 // distribute, sublicense, and/or sell copies of the Software, and to
  16 // permit persons to whom the Software is furnished to do so, subject to
  17 // the following conditions:
  18 //
  19 // The above copyright notice and this permission notice shall be
  20 // included in all copies or substantial portions of the Software.
  21 //
  22 // THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
  23 // EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
  24 // MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
  25 // NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
  26 // LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
  27 // OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
  28 // WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  29
  30 #if NET_2_0
  31
  32 using System;
  33 using System.Collections;
  34 using System.Collections.Generic;
  35 using System.Collections.Specialized;
  36 using System.Text;
  37
  38 namespace Microsoft.Build.BuildEngine {
  39
  40         internal sealed class ConditionTokenizer {
  41
  42                 string  inputString = null;
  43                 int     position = 0;
  44                 int     tokenPosition = 0;
  45
  46                 Token   token;
  47
  48                 bool    ignoreWhiteSpace = true;
  49
  50                 static TokenType[] charIndexToTokenType = new TokenType[128];
  51                 static Dictionary <string, TokenType> keywords = new Dictionary <string, TokenType> (StringComparer.InvariantCultureIgnoreCase);
  52
  53                 static ConditionTokenizer ()
  54                 {
  55                         for (int i = 0; i < 128; i++)
  56                                 charIndexToTokenType [i] = TokenType.Invalid;
  57
  58                         foreach (CharToTokenType cht in charToTokenType)
  59                                 charIndexToTokenType [(int) cht.ch] = cht.tokenType;
  60
  61                         keywords.Add ("and", TokenType.And);
  62                         keywords.Add ("or", TokenType.Or);
  63                 }
  64
  65                 public ConditionTokenizer ()
  66                 {
  67                         this.ignoreWhiteSpace = true;
  68                 }
  69
  70                 public void Tokenize (string s)
  71                 {
  72                         if (s == null)
  73                                 throw new ArgumentNullException ("s");
  74
  75                         this.inputString = s;
  76                         this.position = 0;
  77                         this.token = new Token (null, TokenType.BOF);
  78
  79                         GetNextToken ();
  80                 }
  81
  82                 private void SkipWhiteSpace ()
  83                 {
  84                         int ch;
  85
  86                         while ((ch = PeekChar ()) != -1) {
  87                                 if (!Char.IsWhiteSpace ((char)ch))
  88                                         break;
  89                                 ReadChar ();
  90                         }
  91                 }
  92
  93                 private int PeekChar ()
  94                 {
  95                         if (position < inputString.Length)
  96                                 return (int) inputString [position];
  97                         else
  98                                 return -1;
  99                 }
 100
 101                 private int ReadChar ()
 102                 {
 103                         if (position < inputString.Length)
 104                                 return (int) inputString [position++];
 105                         else
 106                                 return -1;
 107                 }
 108
 109                 public void Expect (TokenType type)
 110                 {
 111                         if (token.Type != type)
 112                                 throw new ExpressionParseException ("Expected token type of type: " + type + ", got " + token.Type +
 113                                         " (" + token.Value + ") .");
 114
 115                         GetNextToken ();
 116                 }
 117
 118                 public bool IsEOF ()
 119                 {
 120                         return token.Type == TokenType.EOF;
 121                 }
 122
 123                 public bool IsNumber ()
 124                 {
 125                         return token.Type == TokenType.Number;
 126                 }
 127
 128                 public bool IsToken (TokenType type)
 129                 {
 130                         return token.Type == type;
 131                 }
 132
 133                 public bool IsPunctation ()
 134                 {
 135                         return (token.Type >= TokenType.FirstPunct && token.Type < TokenType.LastPunct);
 136                 }
 137
 138                 // FIXME: add 'and' and 'or' tokens
 139                 public void GetNextToken ()
 140                 {
 141                         if (token.Type == TokenType.EOF)
 142                                 throw new ExpressionParseException ("Cannot read past the end of stream.");
 143
 144                         if (IgnoreWhiteSpace)
 145                                 SkipWhiteSpace ();
 146
 147                         tokenPosition = position;
 148
 149                         int i = PeekChar ();
 150
 151                         if (i == -1) {
 152                                 token = new Token (null, TokenType.EOF);
 153                                 return;
 154                         }
 155
 156                         char ch = (char) i;
 157
 158                         if (IgnoreWhiteSpace == false && Char.IsWhiteSpace (ch)) {
 159                                 StringBuilder sb = new StringBuilder ();
 160                                 int ch2;
 161
 162                                 while ((ch2 = PeekChar ()) != -1)  {
 163                                         if (!Char.IsWhiteSpace ((char) ch2))
 164                                                 break;
 165
 166                                         sb.Append ((char)ch2);
 167                                         ReadChar();
 168                                 }
 169
 170                                 token = new Token (sb.ToString (), TokenType.WhiteSpace);
 171                                 return;
 172                         }
 173
 174                         if (Char.IsDigit (ch)) {
 175                                 StringBuilder sb = new StringBuilder ();
 176
 177                                 sb.Append (ch);
 178                                 ReadChar ();
 179
 180                                 while ((i = PeekChar ()) != -1) {
 181                                         ch = (char) i;
 182
 183                                         if (Char.IsDigit (ch) || ch == '.')
 184                                                 sb.Append ((char) ReadChar ());
 185                                         else
 186                                                 break;
 187                                 }
 188
 189                                 token = new Token (sb.ToString (), TokenType.Number);
 190                                 return;
 191                         }
 192
 193                         if (ch == '\'') {
 194                                 StringBuilder sb = new StringBuilder ();
 195                                 string temp;
 196
 197                                 sb.Append (ch);
 198                                 ReadChar ();
 199
 200                                 while ((i = PeekChar ()) != -1) {
 201                                         ch = (char) i;
 202
 203                                         sb.Append ((char) ReadChar ());
 204
 205                                         if (ch == '\'')
 206                                                 break;
 207                                 }
 208
 209                                 temp = sb.ToString ();
 210
 211                                 // FIXME: test extreme cases
 212                                 // it fails on '$(something) == ''
 213                                 token = new Token (temp.Substring (1, temp.Length - 2), TokenType.String);
 214
 215                                 return;
 216                         }
 217
 218                         if (ch == '_' || Char.IsLetter (ch)) {
 219                                 StringBuilder sb = new StringBuilder ();
 220
 221                                 sb.Append ((char) ch);
 222                                 ReadChar ();
 223
 224                                 while ((i = PeekChar ()) != -1) {
 225                                         if ((char) i == '_' || Char.IsLetterOrDigit ((char) i))
 226                                                 sb.Append ((char) ReadChar ());
 227                                         else
 228                                                 break;
 229                                 }
 230
 231                                 string temp = sb.ToString ();
 232
 233                                 if (keywords.ContainsKey (temp))
 234                                         token = new Token (temp, keywords [temp]);
 235                                 else
 236                                         token = new Token (temp, TokenType.String);
 237
 238                                 return;
 239                         }
 240
 241                         ReadChar ();
 242
 243                         if (ch == '!' && PeekChar () == (int) '=') {
 244                                 token = new Token ("!=", TokenType.NotEqual);
 245                                 ReadChar ();
 246                                 return;
 247                         }
 248
 249                         if (ch == '<' && PeekChar () == (int) '=') {
 250                                 token = new Token ("<=", TokenType.LessOrEqual);
 251                                 ReadChar ();
 252                                 return;
 253                         }
 254
 255                         if (ch == '>' && PeekChar () == (int) '=') {
 256                                 token = new Token (">=", TokenType.GreaterOrEqual);
 257                                 ReadChar ();
 258                                 return;
 259                         }
 260
 261                         if (ch == '=' && PeekChar () == (int) '=') {
 262                                 token = new Token ("==", TokenType.Equal);
 263                                 ReadChar ();
 264                                 return;
 265                         }
 266
 267                         if (ch == '-' && PeekChar () == (int) '>') {
 268                                 token = new Token ("->", TokenType.Transform);
 269                                 ReadChar ();
 270                                 return;
 271                         }
 272
 273                         if (ch >= 32 && ch < 128) {
 274                                 if (charIndexToTokenType [ch] != TokenType.Invalid) {
 275                                         token = new Token (new String (ch, 1), charIndexToTokenType [ch]);
 276                                         return;
 277                                 } else
 278                                         throw new ExpressionParseException (String.Format ("Invalid punctuation: {0}", ch));
 279                         }
 280
 281                         throw new ExpressionParseException (String.Format ("Invalid token: {0}", ch));
 282                 }
 283
 284                 public int TokenPosition {
 285                         get { return tokenPosition; }
 286                 }
 287
 288                 public Token Token {
 289                         get { return token; }
 290                 }
 291
 292                 public bool IgnoreWhiteSpace {
 293                         get { return ignoreWhiteSpace; }
 294                         set { ignoreWhiteSpace = value; }
 295                 }
 296
 297                 struct CharToTokenType {
 298                         public char ch;
 299                         public TokenType tokenType;
 300
 301                         public CharToTokenType (char ch, TokenType tokenType)
 302                         {
 303                                 this.ch = ch;
 304                                 this.tokenType = tokenType;
 305                         }
 306                 }
 307
 308                 static CharToTokenType[] charToTokenType = {
 309                         new CharToTokenType ('<', TokenType.Less),
 310                         new CharToTokenType ('>', TokenType.Greater),
 311                         new CharToTokenType ('=', TokenType.Equal),
 312                         new CharToTokenType ('(', TokenType.LeftParen),
 313                         new CharToTokenType (')', TokenType.RightParen),
 314                         new CharToTokenType ('.', TokenType.Dot),
 315                         new CharToTokenType (',', TokenType.Comma),
 316                         new CharToTokenType ('!', TokenType.Not),
 317                         new CharToTokenType ('@', TokenType.Item),
 318                         new CharToTokenType ('$', TokenType.Property),
 319                         new CharToTokenType ('%', TokenType.Metadata),
 320                         new CharToTokenType ('\'', TokenType.Apostrophe),
 321                 };
 322         }
 323 }
 324
 325 #endif