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                         // FIXME: looks like a hack: if '-' is here '->' won't be tokenized
 175                         // maybe we should treat item reference as a token
 176                         if (Char.IsDigit (ch) || ch == '-') {
 177                                 StringBuilder sb = new StringBuilder ();
 178
 179                                 sb.Append (ch);
 180                                 ReadChar ();
 181
 182                                 while ((i = PeekChar ()) != -1) {
 183                                         ch = (char) i;
 184
 185                                         if (Char.IsDigit (ch) || ch == '.')
 186                                                 sb.Append ((char) ReadChar ());
 187                                         else
 188                                                 break;
 189                                 }
 190
 191                                 token = new Token (sb.ToString (), TokenType.Number);
 192                                 return;
 193                         }
 194
 195                         if (ch == '\'') {
 196                                 StringBuilder sb = new StringBuilder ();
 197                                 string temp;
 198
 199                                 sb.Append (ch);
 200                                 ReadChar ();
 201
 202                                 while ((i = PeekChar ()) != -1) {
 203                                         ch = (char) i;
 204
 205                                         sb.Append ((char) ReadChar ());
 206
 207                                         if (ch == '\'')
 208                                                 break;
 209                                 }
 210
 211                                 temp = sb.ToString ();
 212
 213                                 // FIXME: test extreme cases
 214                                 // it fails on '$(something) == ''
 215                                 token = new Token (temp.Substring (1, temp.Length - 2), TokenType.String);
 216
 217                                 return;
 218                         }
 219
 220                         if (ch == '_' || Char.IsLetter (ch)) {
 221                                 StringBuilder sb = new StringBuilder ();
 222
 223                                 sb.Append ((char) ch);
 224                                 ReadChar ();
 225
 226                                 while ((i = PeekChar ()) != -1) {
 227                                         if ((char) i == '_' || Char.IsLetterOrDigit ((char) i))
 228                                                 sb.Append ((char) ReadChar ());
 229                                         else
 230                                                 break;
 231                                 }
 232
 233                                 string temp = sb.ToString ();
 234
 235                                 if (keywords.ContainsKey (temp))
 236                                         token = new Token (temp, keywords [temp]);
 237                                 else
 238                                         token = new Token (temp, TokenType.String);
 239
 240                                 return;
 241                         }
 242
 243                         ReadChar ();
 244
 245                         if (ch == '!' && PeekChar () == (int) '=') {
 246                                 token = new Token ("!=", TokenType.NotEqual);
 247                                 ReadChar ();
 248                                 return;
 249                         }
 250
 251                         if (ch == '<' && PeekChar () == (int) '=') {
 252                                 token = new Token ("<=", TokenType.LessOrEqual);
 253                                 ReadChar ();
 254                                 return;
 255                         }
 256
 257                         if (ch == '>' && PeekChar () == (int) '=') {
 258                                 token = new Token (">=", TokenType.GreaterOrEqual);
 259                                 ReadChar ();
 260                                 return;
 261                         }
 262
 263                         if (ch == '=' && PeekChar () == (int) '=') {
 264                                 token = new Token ("==", TokenType.Equal);
 265                                 ReadChar ();
 266                                 return;
 267                         }
 268
 269                         if (ch == '-' && PeekChar () == (int) '>') {
 270                                 token = new Token ("->", TokenType.Transform);
 271                                 ReadChar ();
 272                                 return;
 273                         }
 274
 275                         if (ch >= 32 && ch < 128) {
 276                                 if (charIndexToTokenType [ch] != TokenType.Invalid) {
 277                                         token = new Token (new String (ch, 1), charIndexToTokenType [ch]);
 278                                         return;
 279                                 } else
 280                                         throw new ExpressionParseException (String.Format ("Invalid punctuation: {0}", ch));
 281                         }
 282
 283                         throw new ExpressionParseException (String.Format ("Invalid token: {0}", ch));
 284                 }
 285
 286                 public int TokenPosition {
 287                         get { return tokenPosition; }
 288                 }
 289
 290                 public Token Token {
 291                         get { return token; }
 292                 }
 293
 294                 public bool IgnoreWhiteSpace {
 295                         get { return ignoreWhiteSpace; }
 296                         set { ignoreWhiteSpace = value; }
 297                 }
 298
 299                 struct CharToTokenType {
 300                         public char ch;
 301                         public TokenType tokenType;
 302
 303                         public CharToTokenType (char ch, TokenType tokenType)
 304                         {
 305                                 this.ch = ch;
 306                                 this.tokenType = tokenType;
 307                         }
 308                 }
 309
 310                 static CharToTokenType[] charToTokenType = {
 311                         new CharToTokenType ('<', TokenType.Less),
 312                         new CharToTokenType ('>', TokenType.Greater),
 313                         new CharToTokenType ('=', TokenType.Equal),
 314                         new CharToTokenType ('(', TokenType.LeftParen),
 315                         new CharToTokenType (')', TokenType.RightParen),
 316                         new CharToTokenType ('.', TokenType.Dot),
 317                         new CharToTokenType (',', TokenType.Comma),
 318                         new CharToTokenType ('!', TokenType.Not),
 319                         new CharToTokenType ('@', TokenType.Item),
 320                         new CharToTokenType ('$', TokenType.Property),
 321                         new CharToTokenType ('%', TokenType.Metadata),
 322                         new CharToTokenType ('\'', TokenType.Apostrophe),
 323                 };
 324         }
 325 }
 326
 327 #endif