mcs/class/System.Net.Http/System.Net.Http.Headers/Lexer.cs

   1 //
   2 // Lexer.cs
   3 //
   4 // Authors:
   5 //      Marek Safar  <marek.safar@gmail.com>
   6 //
   7 // Copyright (C) 2011 Xamarin Inc (http://www.xamarin.com)
   8 //
   9 // Permission is hereby granted, free of charge, to any person obtaining
  10 // a copy of this software and associated documentation files (the
  11 // "Software"), to deal in the Software without restriction, including
  12 // without limitation the rights to use, copy, modify, merge, publish,
  13 // distribute, sublicense, and/or sell copies of the Software, and to
  14 // permit persons to whom the Software is furnished to do so, subject to
  15 // the following conditions:
  16 //
  17 // The above copyright notice and this permission notice shall be
  18 // included in all copies or substantial portions of the Software.
  19 //
  20 // THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
  21 // EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
  22 // MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
  23 // NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
  24 // LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
  25 // OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
  26 // WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  27 //
  28
  29 using System.Globalization;
  30
  31 namespace System.Net.Http.Headers
  32 {
  33         struct Token
  34         {
  35                 public enum Type
  36                 {
  37                         Error,
  38                         End,
  39                         Token,
  40                         QuotedString,
  41                         SeparatorEqual,
  42                         SeparatorSemicolon,
  43                         SeparatorSlash,
  44                         SeparatorDash,
  45                         SeparatorComma,
  46                         OpenParens,
  47                 }
  48
  49                 readonly Type type;
  50
  51                 public Token (Type type, int startPosition, int endPosition)
  52                         : this ()
  53                 {
  54                         this.type = type;
  55                         StartPosition = startPosition;
  56                         EndPosition = endPosition;
  57                 }
  58
  59                 public int StartPosition { get; private set; }
  60                 public int EndPosition { get; private set; }
  61
  62                 public Type Kind {
  63                         get {
  64                                 return type;
  65                         }
  66                 }
  67
  68                 public static implicit operator Token.Type (Token token)
  69                 {
  70                         return token.type;
  71                 }
  72
  73                 public override string ToString ()
  74                 {
  75                         return type.ToString ();
  76                 }
  77         }
  78
  79         class Lexer
  80         {
  81                 // any CHAR except CTLs or separators
  82                 static readonly bool[] token_chars = {
  83                         /*0*/   false, false, false, false, false, false, false, false, false, false,
  84                         /*10*/  false, false, false, false, false, false, false, false, false, false,
  85                         /*20*/  false, false, false, false, false, false, false, false, false, false,
  86                         /*30*/  false, false, false, true, false, true, true, true, true, true,
  87                         /*40*/  false, false, true, true, false, true, true, false, true, true,
  88                         /*50*/  true, true, true, true, true, true, true, true, false, false,
  89                         /*60*/  false, false, false, false, false, true, true, true, true, true,
  90                         /*70*/  true, true, true, true, true, true, true, true, true, true,
  91                         /*80*/  true, true, true, true, true, true, true, true, true, true,
  92                         /*90*/  true, false, false, false, true, true, true, true, true, true,
  93                         /*100*/ true, true, true, true, true, true, true, true, true, true,
  94                         /*110*/ true, true, true, true, true, true, true, true, true, true,
  95                         /*120*/ true, true, true, false, true, false
  96                         };
  97
  98                 static readonly int last_token_char = token_chars.Length;
  99                 static readonly string[] dt_formats = new[] {
 100                                 "r",
 101                                 "dddd, dd'-'MMM'-'yy HH:mm:ss 'GMT'",
 102                                 "ddd MMM d HH:mm:ss yyyy",
 103                                 "d MMM yy H:m:s",
 104                                 "ddd, d MMM yyyy H:m:s zzz"
 105                 };
 106
 107                 readonly string s;
 108                 int pos;
 109
 110                 public Lexer (string stream)
 111                 {
 112                         this.s = stream;
 113                 }
 114
 115                 public string GetStringValue (Token token)
 116                 {
 117                         return s.Substring (token.StartPosition, token.EndPosition - token.StartPosition);
 118                 }
 119
 120                 public string GetStringValue (Token start, Token end)
 121                 {
 122                         return s.Substring (start.StartPosition, end.EndPosition - start.StartPosition);
 123                 }
 124
 125                 public string GetQuotedStringValue (Token start)
 126                 {
 127                         return s.Substring (start.StartPosition + 1, start.EndPosition - start.StartPosition - 2);
 128                 }
 129
 130                 public string GetRemainingStringValue (int position)
 131                 {
 132                         return position > s.Length ? null : s.Substring (position);
 133                 }
 134
 135                 public bool IsStarStringValue (Token token)
 136                 {
 137                         return (token.EndPosition - token.StartPosition) == 1 && s[token.StartPosition] == '*';
 138                 }
 139
 140                 public bool TryGetNumericValue (Token token, out int value)
 141                 {
 142                         return int.TryParse (GetStringValue (token), NumberStyles.None, CultureInfo.InvariantCulture, out value);
 143                 }
 144
 145                 public TimeSpan? TryGetTimeSpanValue (Token token)
 146                 {
 147                         int seconds;
 148                         if (TryGetNumericValue (token, out seconds)) {
 149                                 return TimeSpan.FromSeconds (seconds);
 150                         }
 151
 152                         return null;
 153                 }
 154
 155                 public bool TryGetDateValue (Token token, out DateTimeOffset value)
 156                 {
 157                         string text = token == Token.Type.QuotedString ?
 158                                 s.Substring (token.StartPosition + 1, token.EndPosition - token.StartPosition - 2) :
 159                                 GetStringValue (token);
 160
 161                         return TryGetDateValue (text, out value);
 162                 }
 163
 164                 public static bool TryGetDateValue (string text, out DateTimeOffset value)
 165                 {
 166                         const DateTimeStyles DefaultStyles = DateTimeStyles.AssumeUniversal | DateTimeStyles.AllowWhiteSpaces;
 167
 168                         return DateTimeOffset.TryParseExact (text, dt_formats, DateTimeFormatInfo.InvariantInfo, DefaultStyles, out value);
 169                 }
 170
 171                 public bool TryGetDoubleValue (Token token, out double value)
 172                 {
 173                         string s = GetStringValue (token);
 174                         return double.TryParse (s, NumberStyles.AllowDecimalPoint, CultureInfo.InvariantCulture, out value);
 175                 }
 176
 177                 public static bool IsValidToken (string input)
 178                 {
 179                         int i = 0;
 180                         //
 181                         // any CHAR except CTLs or separator
 182                         //
 183                         for (; i < input.Length; ++i) {
 184                                 char s = input[i];
 185                                 if (s > last_token_char || !token_chars[s])
 186                                         return false;
 187                         }
 188
 189                         return i > 0;
 190                 }
 191
 192                 public void EatChar ()
 193                 {
 194                         ++pos;
 195                 }
 196
 197                 public int PeekChar ()
 198                 {
 199                         return pos < s.Length ? s[pos] : -1;
 200                 }
 201
 202                 public bool ScanCommentOptional (out string value)
 203                 {
 204                         Token t;
 205                         if (ScanCommentOptional (out value, out t))
 206                                 return true;
 207
 208                         return t == Token.Type.End;
 209                 }
 210
 211                 public bool ScanCommentOptional (out string value, out Token readToken)
 212                 {
 213                         readToken = Scan ();
 214                         if (readToken != Token.Type.OpenParens) {
 215                                 value = null;
 216                                 return false;
 217                         }
 218
 219                         while (pos < s.Length) {
 220                                 var ch = s[pos];
 221                                 if (ch == ')') {
 222                                         ++pos;
 223                                         var start = readToken.StartPosition;
 224                                         value = s.Substring (start, pos - start);
 225                                         return true;
 226                                 }
 227
 228                                 // any OCTET except CTLs, but including LWS
 229                                 if (ch < 32 || ch > 126)
 230                                         break;
 231
 232                                 ++pos;
 233                         }
 234
 235                         value = null;
 236                         return false;
 237                 }
 238
 239                 public Token Scan ()
 240                 {
 241                         int start = pos;
 242                         if (s == null)
 243                                 return new Token (Token.Type.Error, 0, 0);
 244
 245                         Token.Type ttype;
 246                         if (pos >= s.Length) {
 247                                 ttype = Token.Type.End;
 248                         } else {
 249                                 ttype = Token.Type.Error;
 250                         start:
 251                                 char ch = s[pos++];
 252                                 switch (ch) {
 253                                 case ' ':
 254                                 case '\t':
 255                                         if (pos == s.Length) {
 256                                                 ttype = Token.Type.End;
 257                                                 break;
 258                                         }
 259
 260                                         goto start;
 261                                 case '=':
 262                                         ttype = Token.Type.SeparatorEqual;
 263                                         break;
 264                                 case ';':
 265                                         ttype = Token.Type.SeparatorSemicolon;
 266                                         break;
 267                                 case '/':
 268                                         ttype = Token.Type.SeparatorSlash;
 269                                         break;
 270                                 case '-':
 271                                         ttype = Token.Type.SeparatorDash;
 272                                         break;
 273                                 case ',':
 274                                         ttype = Token.Type.SeparatorComma;
 275                                         break;
 276                                 case '"':
 277                                         // Quoted string
 278                                         start = pos - 1;
 279                                         while (pos < s.Length) {
 280                                                 ch = s[pos];
 281                                                 if (ch == '"') {
 282                                                         ++pos;
 283                                                         ttype = Token.Type.QuotedString;
 284                                                         break;
 285                                                 }
 286
 287                                                 // any OCTET except CTLs, but including LWS
 288                                                 if (ch < 32 || ch > 126)
 289                                                         break;
 290
 291                                                 ++pos;
 292                                         }
 293
 294                                         break;
 295                                 case '(':
 296                                         start = pos - 1;
 297                                         ttype = Token.Type.OpenParens;
 298                                         break;
 299                                 default:
 300                                         if (ch <= last_token_char && token_chars[ch]) {
 301                                                 start = pos - 1;
 302
 303                                                 ttype = Token.Type.Token;
 304                                                 while (pos < s.Length) {
 305                                                         ch = s[pos];
 306                                                         if (ch > last_token_char || !token_chars[ch]) {
 307                                                                 break;
 308                                                         }
 309
 310                                                         ++pos;
 311                                                 }
 312                                         }
 313
 314                                         break;
 315                                 }
 316                         }
 317
 318                         return new Token (ttype, start, pos);
 319                 }
 320         }
 321 }