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                 public static readonly Token Empty = new Token (Type.Token, 0, 0);
  50
  51                 readonly Type type;
  52
  53                 public Token (Type type, int startPosition, int endPosition)
  54                         : this ()
  55                 {
  56                         this.type = type;
  57                         StartPosition = startPosition;
  58                         EndPosition = endPosition;
  59                 }
  60
  61                 public int StartPosition { get; private set; }
  62                 public int EndPosition { get; private set; }
  63
  64                 public Type Kind {
  65                         get {
  66                                 return type;
  67                         }
  68                 }
  69
  70                 public static implicit operator Token.Type (Token token)
  71                 {
  72                         return token.type;
  73                 }
  74
  75                 public override string ToString ()
  76                 {
  77                         return type.ToString ();
  78                 }
  79         }
  80
  81         class Lexer
  82         {
  83                 // any CHAR except CTLs or separators
  84                 static readonly bool[] token_chars = {
  85                         /*0*/   false, false, false, false, false, false, false, false, false, false,
  86                         /*10*/  false, false, false, false, false, false, false, false, false, false,
  87                         /*20*/  false, false, false, false, false, false, false, false, false, false,
  88                         /*30*/  false, false, false, true, false, true, true, true, true, true,
  89                         /*40*/  false, false, true, true, false, true, true, false, true, true,
  90                         /*50*/  true, true, true, true, true, true, true, true, false, false,
  91                         /*60*/  false, false, false, false, false, true, true, true, true, true,
  92                         /*70*/  true, true, true, true, true, true, true, true, true, true,
  93                         /*80*/  true, true, true, true, true, true, true, true, true, true,
  94                         /*90*/  true, false, false, false, true, true, true, true, true, true,
  95                         /*100*/ true, true, true, true, true, true, true, true, true, true,
  96                         /*110*/ true, true, true, true, true, true, true, true, true, true,
  97                         /*120*/ true, true, true, false, true, false, true
  98                         };
  99
 100                 static readonly int last_token_char = token_chars.Length;
 101                 static readonly string[] dt_formats = new[] {
 102                                 "r",
 103                                 "dddd, dd'-'MMM'-'yy HH:mm:ss 'GMT'",
 104                                 "ddd MMM d HH:mm:ss yyyy",
 105                                 "d MMM yy H:m:s",
 106                                 "ddd, d MMM yyyy H:m:s zzz"
 107                 };
 108
 109                 readonly string s;
 110                 int pos;
 111
 112                 public Lexer (string stream)
 113                 {
 114                         this.s = stream;
 115                 }
 116
 117                 public int Position {
 118                         get {
 119                                 return pos;
 120                         }
 121                         set {
 122                                 pos = value;
 123                         }
 124                 }
 125
 126                 public string GetStringValue (Token token)
 127                 {
 128                         return s.Substring (token.StartPosition, token.EndPosition - token.StartPosition);
 129                 }
 130
 131                 public string GetStringValue (Token start, Token end)
 132                 {
 133                         return s.Substring (start.StartPosition, end.EndPosition - start.StartPosition);
 134                 }
 135
 136                 public string GetQuotedStringValue (Token start)
 137                 {
 138                         return s.Substring (start.StartPosition + 1, start.EndPosition - start.StartPosition - 2);
 139                 }
 140
 141                 public string GetRemainingStringValue (int position)
 142                 {
 143                         return position > s.Length ? null : s.Substring (position);
 144                 }
 145
 146                 public bool IsStarStringValue (Token token)
 147                 {
 148                         return (token.EndPosition - token.StartPosition) == 1 && s[token.StartPosition] == '*';
 149                 }
 150
 151                 public bool TryGetNumericValue (Token token, out int value)
 152                 {
 153                         return int.TryParse (GetStringValue (token), NumberStyles.None, CultureInfo.InvariantCulture, out value);
 154                 }
 155
 156                 public TimeSpan? TryGetTimeSpanValue (Token token)
 157                 {
 158                         int seconds;
 159                         if (TryGetNumericValue (token, out seconds)) {
 160                                 return TimeSpan.FromSeconds (seconds);
 161                         }
 162
 163                         return null;
 164                 }
 165
 166                 public bool TryGetDateValue (Token token, out DateTimeOffset value)
 167                 {
 168                         string text = token == Token.Type.QuotedString ?
 169                                 s.Substring (token.StartPosition + 1, token.EndPosition - token.StartPosition - 2) :
 170                                 GetStringValue (token);
 171
 172                         return TryGetDateValue (text, out value);
 173                 }
 174
 175                 public static bool TryGetDateValue (string text, out DateTimeOffset value)
 176                 {
 177                         const DateTimeStyles DefaultStyles = DateTimeStyles.AssumeUniversal | DateTimeStyles.AllowWhiteSpaces;
 178
 179                         return DateTimeOffset.TryParseExact (text, dt_formats, DateTimeFormatInfo.InvariantInfo, DefaultStyles, out value);
 180                 }
 181
 182                 public bool TryGetDoubleValue (Token token, out double value)
 183                 {
 184                         string s = GetStringValue (token);
 185                         return double.TryParse (s, NumberStyles.AllowDecimalPoint, CultureInfo.InvariantCulture, out value);
 186                 }
 187
 188                 public static bool IsValidToken (string input)
 189                 {
 190                         int i = 0;
 191                         //
 192                         // any CHAR except CTLs or separator
 193                         //
 194                         for (; i < input.Length; ++i) {
 195                                 char s = input[i];
 196                                 if (!IsValidCharacter (s))
 197                                         return false;
 198                         }
 199
 200                         return i > 0;
 201                 }
 202
 203                 public static bool IsValidCharacter (char input)
 204                 {
 205                         return input < last_token_char && token_chars[input];
 206                 }
 207
 208                 public void EatChar ()
 209                 {
 210                         ++pos;
 211                 }
 212
 213                 public int PeekChar ()
 214                 {
 215                         return pos < s.Length ? s[pos] : -1;
 216                 }
 217
 218                 public bool ScanCommentOptional (out string value)
 219                 {
 220                         Token t;
 221                         if (ScanCommentOptional (out value, out t))
 222                                 return true;
 223
 224                         return t == Token.Type.End;
 225                 }
 226
 227                 public bool ScanCommentOptional (out string value, out Token readToken)
 228                 {
 229                         readToken = Scan ();
 230                         if (readToken != Token.Type.OpenParens) {
 231                                 value = null;
 232                                 return false;
 233                         }
 234
 235                         int parens = 1;
 236                         while (pos < s.Length) {
 237                                 var ch = s[pos];
 238                                 if (ch == '(') {
 239                                         ++parens;
 240                                         ++pos;
 241                                         continue;
 242                                 }
 243
 244                                 if (ch == ')') {
 245                                         ++pos;
 246                                         if (--parens > 0)
 247                                                 continue;
 248
 249                                         var start = readToken.StartPosition;
 250                                         value = s.Substring (start, pos - start);
 251                                         return true;
 252                                 }
 253
 254                                 // any OCTET except CTLs, but including LWS
 255                                 if (ch < 32 || ch > 126)
 256                                         break;
 257
 258                                 ++pos;
 259                         }
 260
 261                         value = null;
 262                         return false;
 263                 }
 264
 265                 public Token Scan (bool recognizeDash = false)
 266                 {
 267                         int start = pos;
 268                         if (s == null)
 269                                 return new Token (Token.Type.Error, 0, 0);
 270
 271                         Token.Type ttype;
 272                         if (pos >= s.Length) {
 273                                 ttype = Token.Type.End;
 274                         } else {
 275                                 ttype = Token.Type.Error;
 276                         start:
 277                                 char ch = s[pos++];
 278                                 switch (ch) {
 279                                 case ' ':
 280                                 case '\t':
 281                                         if (pos == s.Length) {
 282                                                 ttype = Token.Type.End;
 283                                                 break;
 284                                         }
 285
 286                                         goto start;
 287                                 case '=':
 288                                         ttype = Token.Type.SeparatorEqual;
 289                                         break;
 290                                 case ';':
 291                                         ttype = Token.Type.SeparatorSemicolon;
 292                                         break;
 293                                 case '/':
 294                                         ttype = Token.Type.SeparatorSlash;
 295                                         break;
 296                                 case '-':
 297                                         if (recognizeDash) {
 298                                                 ttype = Token.Type.SeparatorDash;
 299                                                 break;
 300                                         }
 301
 302                                         goto default;
 303                                 case ',':
 304                                         ttype = Token.Type.SeparatorComma;
 305                                         break;
 306                                 case '"':
 307                                         // Quoted string
 308                                         start = pos - 1;
 309                                         while (pos < s.Length) {
 310                                                 ch = s [pos++];
 311                                                 if (ch == '"') {
 312                                                         ttype = Token.Type.QuotedString;
 313                                                         break;
 314                                                 }
 315                                         }
 316
 317                                         break;
 318                                 case '(':
 319                                         start = pos - 1;
 320                                         ttype = Token.Type.OpenParens;
 321                                         break;
 322                                 default:
 323                                         if (ch < last_token_char && token_chars[ch]) {
 324                                                 start = pos - 1;
 325
 326                                                 ttype = Token.Type.Token;
 327                                                 while (pos < s.Length) {
 328                                                         ch = s[pos];
 329                                                         if (ch >= last_token_char || !token_chars[ch]) {
 330                                                                 break;
 331                                                         }
 332
 333                                                         ++pos;
 334                                                 }
 335                                         }
 336
 337                                         break;
 338                                 }
 339                         }
 340
 341                         return new Token (ttype, start, pos);
 342                 }
 343         }
 344 }