mcs/class/System/System.Text.RegularExpressions/Regex.cs

   1 //
   2 // assembly:    System
   3 // namespace:   System.Text.RegularExpressions
   4 // file:        regex.cs
   5 //
   6 // author:      Dan Lewis (dlewis@gmx.co.uk)
   7 //              (c) 2002
   8
   9 //
  10 // Permission is hereby granted, free of charge, to any person obtaining
  11 // a copy of this software and associated documentation files (the
  12 // "Software"), to deal in the Software without restriction, including
  13 // without limitation the rights to use, copy, modify, merge, publish,
  14 // distribute, sublicense, and/or sell copies of the Software, and to
  15 // permit persons to whom the Software is furnished to do so, subject to
  16 // the following conditions:
  17 //
  18 // The above copyright notice and this permission notice shall be
  19 // included in all copies or substantial portions of the Software.
  20 //
  21 // THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
  22 // EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
  23 // MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
  24 // NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
  25 // LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
  26 // OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
  27 // WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  28 //
  29
  30 using System;
  31 using System.Text;
  32 using System.Collections;
  33 using System.Reflection;
  34 using System.Reflection.Emit;
  35 using System.Runtime.Serialization;
  36
  37 using RegularExpression = System.Text.RegularExpressions.Syntax.RegularExpression;
  38 using Parser = System.Text.RegularExpressions.Syntax.Parser;
  39
  40 using System.Diagnostics;
  41
  42
  43 namespace System.Text.RegularExpressions {
  44
  45         [Serializable]
  46         public partial class Regex : ISerializable {
  47
  48 #if !TARGET_JVM
  49                 [MonoTODO]
  50                 public static void CompileToAssembly (RegexCompilationInfo [] regexes, AssemblyName aname)
  51                 {
  52                         Regex.CompileToAssembly(regexes, aname, new CustomAttributeBuilder [] {}, null);
  53                 }
  54
  55                 [MonoTODO]
  56                 public static void CompileToAssembly (RegexCompilationInfo [] regexes, AssemblyName aname,
  57                                                       CustomAttributeBuilder [] attribs)
  58                 {
  59                         Regex.CompileToAssembly(regexes, aname, attribs, null);
  60                 }
  61
  62                 [MonoTODO]
  63                 public static void CompileToAssembly (RegexCompilationInfo [] regexes, AssemblyName aname,
  64                                                       CustomAttributeBuilder [] attribs, string resourceFile)
  65                 {
  66                         throw new NotImplementedException ();
  67                         // TODO : Make use of attribs and resourceFile parameters
  68                         /*
  69                         AssemblyBuilder asmBuilder = AppDomain.CurrentDomain.DefineDynamicAssembly (aname, AssemblyBuilderAccess.RunAndSave);
  70                         ModuleBuilder modBuilder = asmBuilder.DefineDynamicModule("InnerRegexModule",aname.Name);
  71                         Parser psr = new Parser ();
  72
  73                         System.Console.WriteLine("CompileToAssembly");
  74
  75                         for(int i=0; i < regexes.Length; i++)
  76                                 {
  77                                         System.Console.WriteLine("Compiling expression :" + regexes[i].Pattern);
  78                                         RegularExpression re = psr.ParseRegularExpression (regexes[i].Pattern, regexes[i].Options);
  79
  80                                         // compile
  81
  82                                         CILCompiler cmp = new CILCompiler (modBuilder, i);
  83                                         bool reverse = (regexes[i].Options & RegexOptions.RightToLeft) !=0;
  84                                         re.Compile (cmp, reverse);
  85                                         cmp.Close();
  86
  87                                 }
  88
  89
  90                         // Define a runtime class with specified name and attributes.
  91                         TypeBuilder builder = modBuilder.DefineType("ITest");
  92                         builder.CreateType();
  93                         asmBuilder.Save(aname.Name);
  94                         */
  95                 }
  96 #endif
  97
  98                 public static string Escape (string str)
  99                 {
 100                         if (str == null)
 101                                 throw new ArgumentNullException ("str");
 102                         return Parser.Escape (str);
 103                 }
 104
 105                 public static string Unescape (string str)
 106                 {
 107                         if (str == null)
 108                                 throw new ArgumentNullException ("str");
 109                         return Parser.Unescape (str);
 110                 }
 111
 112                 public static bool IsMatch (string input, string pattern)
 113                 {
 114                         return IsMatch (input, pattern, RegexOptions.None);
 115                 }
 116
 117                 public static bool IsMatch (string input, string pattern, RegexOptions options)
 118                 {
 119                         Regex re = new Regex (pattern, options);
 120                         return re.IsMatch (input);
 121                 }
 122
 123                 public static Match Match (string input, string pattern)
 124                 {
 125                         return Regex.Match (input, pattern, RegexOptions.None);
 126                 }
 127
 128                 public static Match Match (string input, string pattern, RegexOptions options)
 129                 {
 130                         Regex re = new Regex (pattern, options);
 131                         return re.Match (input);
 132                 }
 133
 134                 public static MatchCollection Matches (string input, string pattern)
 135                 {
 136                         return Matches (input, pattern, RegexOptions.None);
 137                 }
 138
 139                 public static MatchCollection Matches (string input, string pattern, RegexOptions options)
 140                 {
 141                         Regex re = new Regex (pattern, options);
 142                         return re.Matches (input);
 143                 }
 144
 145                 public static string Replace (string input, string pattern, MatchEvaluator evaluator)
 146                 {
 147                         return Regex.Replace (input, pattern, evaluator, RegexOptions.None);
 148                 }
 149
 150                 public static string Replace (string input, string pattern, MatchEvaluator evaluator,
 151                                               RegexOptions options)
 152                 {
 153                         Regex re = new Regex (pattern, options);
 154                         return re.Replace (input, evaluator);
 155                 }
 156
 157                 public static string Replace (string input, string pattern, string replacement)
 158                 {
 159                         return Regex.Replace (input, pattern, replacement, RegexOptions.None);
 160                 }
 161
 162                 public static string Replace (string input, string pattern, string replacement,
 163                                               RegexOptions options)
 164                 {
 165                         Regex re = new Regex (pattern, options);
 166                         return re.Replace (input, replacement);
 167                 }
 168
 169                 public static string [] Split (string input, string pattern)
 170                 {
 171                         return Regex.Split (input, pattern, RegexOptions.None);
 172                 }
 173
 174                 public static string [] Split (string input, string pattern, RegexOptions options)
 175                 {
 176                         Regex re = new Regex (pattern, options);
 177                         return re.Split (input);
 178                 }
 179
 180 #if NET_2_0
 181                 static FactoryCache cache = new FactoryCache (15);
 182                 public static int CacheSize {
 183                         get { return cache.Capacity; }
 184                         set {
 185                                 if (value < 0)
 186                                         throw new ArgumentOutOfRangeException ("CacheSize");
 187
 188                                 cache.Capacity = value;
 189                         }
 190                 }
 191 #else
 192                 static FactoryCache cache = new FactoryCache (200);
 193 #endif
 194
 195                 // private
 196
 197
 198                 // constructors
 199
 200                 // This constructor is used by compiled regular expressions that are
 201                 // classes derived from Regex class. No initialization required.
 202                 protected Regex ()
 203                 {
 204                 }
 205
 206                 public Regex (string pattern) : this (pattern, RegexOptions.None)
 207                 {
 208                 }
 209
 210                 public Regex (string pattern, RegexOptions options)
 211                 {
 212                         if (pattern == null)
 213                                 throw new ArgumentNullException ("pattern");
 214                         validate_options (options);
 215                         this.pattern = pattern;
 216                         this.roptions = options;
 217                         Init ();
 218                 }
 219
 220                 static void validate_options (RegexOptions options)
 221                 {
 222                         const RegexOptions allopts =
 223                                 RegexOptions.None |
 224                                 RegexOptions.IgnoreCase |
 225                                 RegexOptions.Multiline |
 226                                 RegexOptions.ExplicitCapture |
 227 #if !NET_2_1
 228                                 RegexOptions.Compiled |
 229 #endif
 230                                 RegexOptions.Singleline |
 231                                 RegexOptions.IgnorePatternWhitespace |
 232                                 RegexOptions.RightToLeft |
 233                                 RegexOptions.ECMAScript |
 234                                 RegexOptions.CultureInvariant;
 235
 236                         const RegexOptions ecmaopts =
 237                                 RegexOptions.IgnoreCase |
 238                                 RegexOptions.Multiline |
 239 #if !NET_2_1
 240                                 RegexOptions.Compiled |
 241 #endif
 242                                 RegexOptions.ECMAScript;
 243
 244                         if ((options & ~allopts) != 0)
 245                                 throw new ArgumentOutOfRangeException ("options");
 246                         if ((options & RegexOptions.ECMAScript) != 0 && (options & ~ecmaopts) != 0)
 247                                 throw new ArgumentOutOfRangeException ("options");
 248                 }
 249
 250 #if !TARGET_JVM
 251                 private void Init ()
 252                 {
 253                         this.machineFactory = cache.Lookup (this.pattern, this.roptions);
 254
 255                         if (this.machineFactory == null) {
 256                                 InitNewRegex();
 257                         } else {
 258                                 this.group_count = this.machineFactory.GroupCount;
 259                                 this.gap = this.machineFactory.Gap;
 260                                 this.mapping = this.machineFactory.Mapping;
 261                                 this.group_names = this.machineFactory.NamesMapping;
 262                         }
 263                 }
 264 #endif
 265
 266                 private void InitNewRegex ()
 267                 {
 268                         this.machineFactory = CreateMachineFactory (this.pattern, this.roptions);
 269                         cache.Add (this.pattern, this.roptions, this.machineFactory);
 270                         this.group_count = machineFactory.GroupCount;
 271                         this.gap = this.machineFactory.Gap;
 272                         this.mapping = machineFactory.Mapping;
 273                         this.group_names = this.machineFactory.NamesMapping;
 274                 }
 275
 276 #if !NET_2_1
 277                 // The new rx engine has blocking bugs like
 278                 // https://bugzilla.novell.com/show_bug.cgi?id=470827
 279                 static readonly bool old_rx =
 280                         Environment.GetEnvironmentVariable ("MONO_NEW_RX") == null;
 281 #endif
 282
 283                 private static IMachineFactory CreateMachineFactory (string pattern, RegexOptions options)
 284                 {
 285                         Parser psr = new Parser ();
 286                         RegularExpression re = psr.ParseRegularExpression (pattern, options);
 287
 288 #if NET_2_1
 289                         ICompiler cmp = new PatternCompiler ();
 290 #else
 291                         ICompiler cmp;
 292                         if (!old_rx) {
 293                                 if ((options & RegexOptions.Compiled) != 0)
 294                                         cmp = new CILCompiler ();
 295                                 else
 296                                         cmp = new RxCompiler ();
 297                         } else {
 298                                 cmp = new PatternCompiler ();
 299                         }
 300 #endif
 301
 302                         re.Compile (cmp, (options & RegexOptions.RightToLeft) != 0);
 303
 304                         IMachineFactory machineFactory = cmp.GetMachineFactory ();
 305                         Hashtable mapping = new Hashtable ();
 306                         machineFactory.Gap = psr.GetMapping (mapping);
 307                         machineFactory.Mapping = mapping;
 308                         machineFactory.NamesMapping = GetGroupNamesArray (machineFactory.GroupCount, machineFactory.Mapping);
 309
 310                         return machineFactory;
 311                 }
 312
 313 #if NET_2_0
 314                 protected
 315 #else
 316                 private
 317 #endif
 318                 Regex (SerializationInfo info, StreamingContext context) :
 319                         this (info.GetString ("pattern"),
 320                               (RegexOptions) info.GetValue ("options", typeof (RegexOptions)))
 321                 {
 322                 }
 323
 324 #if ONLY_1_1 && !TARGET_JVM
 325                 // fixes public API signature
 326                 ~Regex ()
 327                 {
 328                 }
 329 #endif
 330                 // public instance properties
 331
 332                 public RegexOptions Options {
 333                         get { return roptions; }
 334                 }
 335
 336                 public bool RightToLeft {
 337                         get { return (roptions & RegexOptions.RightToLeft) != 0; }
 338                 }
 339
 340                 // public instance methods
 341
 342                 public string [] GetGroupNames ()
 343                 {
 344                         string [] names = new string [1 + group_count];
 345                         Array.Copy (group_names, names, 1 + group_count);
 346                         return names;
 347                 }
 348
 349                 public int [] GetGroupNumbers ()
 350                 {
 351                         int [] numbers = new int [1 + group_count];
 352                         Array.Copy (GroupNumbers, numbers, 1 + group_count);
 353                         return numbers;
 354                 }
 355
 356                 public string GroupNameFromNumber (int i)
 357                 {
 358                         i = GetGroupIndex (i);
 359                         if (i < 0)
 360                                 return "";
 361
 362                         return group_names [i];
 363                 }
 364
 365                 public int GroupNumberFromName (string name)
 366                 {
 367                         if (!mapping.Contains (name))
 368                                 return -1;
 369                         int i = (int) mapping [name];
 370                         if (i >= gap)
 371                                 i = Int32.Parse (name);
 372                         return i;
 373                 }
 374
 375                 internal int GetGroupIndex (int number)
 376                 {
 377                         if (number < gap)
 378                                 return number;
 379                         if (gap > group_count)
 380                                 return -1;
 381                         return Array.BinarySearch (GroupNumbers, gap, group_count - gap + 1, number);
 382                 }
 383
 384                 int default_startat (string input)
 385                 {
 386                         return (RightToLeft && input != null) ? input.Length : 0;
 387                 }
 388
 389                 // match methods
 390
 391                 public bool IsMatch (string input)
 392                 {
 393                         return IsMatch (input, default_startat (input));
 394                 }
 395
 396                 public bool IsMatch (string input, int startat)
 397                 {
 398                         return Match (input, startat).Success;
 399                 }
 400
 401                 public Match Match (string input)
 402                 {
 403                         return Match (input, default_startat (input));
 404                 }
 405
 406                 public Match Match (string input, int startat)
 407                 {
 408                         if (input == null)
 409                                 throw new ArgumentNullException ("input");
 410                         if (startat < 0 || startat > input.Length)
 411                                 throw new ArgumentOutOfRangeException ("startat");
 412                         return CreateMachine ().Scan (this, input, startat, input.Length);
 413                 }
 414
 415                 public Match Match (string input, int startat, int length)
 416                 {
 417                         if (input == null)
 418                                 throw new ArgumentNullException ("input");
 419                         if (startat < 0 || startat > input.Length)
 420                                 throw new ArgumentOutOfRangeException ("startat");
 421                         if (length < 0 || length > input.Length - startat)
 422                                 throw new ArgumentOutOfRangeException ("length");
 423                         return CreateMachine ().Scan (this, input, startat, startat + length);
 424                 }
 425
 426                 public MatchCollection Matches (string input)
 427                 {
 428                         return Matches (input, default_startat (input));
 429                 }
 430
 431                 public MatchCollection Matches (string input, int startat)
 432                 {
 433                         Match m = Match (input, startat);
 434                         return new MatchCollection (m);
 435                 }
 436
 437                 // replace methods
 438
 439                 public string Replace (string input, MatchEvaluator evaluator)
 440                 {
 441                         return Replace (input, evaluator, Int32.MaxValue, default_startat (input));
 442                 }
 443
 444                 public string Replace (string input, MatchEvaluator evaluator, int count)
 445                 {
 446                         return Replace (input, evaluator, count, default_startat (input));
 447                 }
 448
 449                 class Adapter {
 450                         MatchEvaluator ev;
 451                         public Adapter (MatchEvaluator ev) { this.ev = ev; }
 452                         public void Evaluate (Match m, StringBuilder sb) { sb.Append (ev (m)); }
 453                 }
 454
 455                 public string Replace (string input, MatchEvaluator evaluator, int count, int startat)
 456                 {
 457                         if (input == null)
 458                                 throw new ArgumentNullException ("input");
 459                         if (evaluator == null)
 460                                 throw new ArgumentNullException ("evaluator");
 461                         if (count < -1)
 462                                 throw new ArgumentOutOfRangeException ("count");
 463                         if (startat < 0 || startat > input.Length)
 464                                 throw new ArgumentOutOfRangeException ("startat");
 465
 466                         BaseMachine m = (BaseMachine)CreateMachine ();
 467
 468                         if (RightToLeft)
 469                                 return m.RTLReplace (this, input, evaluator, count, startat);
 470
 471                         // NOTE: If this is a cause of a lot of allocations, we can convert it to
 472                         //       use a ThreadStatic allocation mitigator
 473                         Adapter a = new Adapter (evaluator);
 474
 475                         return m.LTRReplace (this, input, new BaseMachine.MatchAppendEvaluator (a.Evaluate),
 476                                                                  count, startat);
 477                 }
 478
 479                 public string Replace (string input, string replacement)
 480                 {
 481                         return Replace (input, replacement, Int32.MaxValue, default_startat (input));
 482                 }
 483
 484                 public string Replace (string input, string replacement, int count)
 485                 {
 486                         return Replace (input, replacement, count, default_startat (input));
 487                 }
 488
 489                 public string Replace (string input, string replacement, int count, int startat)
 490                 {
 491                         if (input == null)
 492                                 throw new ArgumentNullException ("input");
 493                         if (replacement == null)
 494                                 throw new ArgumentNullException ("replacement");
 495                         if (count < -1)
 496                                 throw new ArgumentOutOfRangeException ("count");
 497                         if (startat < 0 || startat > input.Length)
 498                                 throw new ArgumentOutOfRangeException ("startat");
 499
 500                         return CreateMachine ().Replace (this, input, replacement, count, startat);
 501                 }
 502
 503                 // split methods
 504
 505                 public string [] Split (string input)
 506                 {
 507                         return Split (input, Int32.MaxValue, default_startat (input));
 508                 }
 509
 510                 public string [] Split (string input, int count)
 511                 {
 512                         return Split (input, count, default_startat (input));
 513                 }
 514
 515                 public string [] Split (string input, int count, int startat)
 516                 {
 517                         if (input == null)
 518                                 throw new ArgumentNullException ("input");
 519                         if (count < 0)
 520                                 throw new ArgumentOutOfRangeException ("count");
 521                         if (startat < 0 || startat > input.Length)
 522                                 throw new ArgumentOutOfRangeException ("startat");
 523
 524                         return CreateMachine ().Split (this, input, count, startat);
 525                 }
 526
 527                 // This method is called at the end of the constructor of compiled
 528                 // regular expression classes to do internal initialization.
 529                 protected void InitializeReferences ()
 530                 {
 531                         if (refsInitialized)
 532                                 throw new NotSupportedException ("This operation is only allowed once per object.");
 533
 534                         refsInitialized = true;
 535
 536                         // Compile pattern that results in performance loss as existing
 537                         // CIL code is ignored but provides support for regular
 538                         // expressions compiled to assemblies.
 539                         Init ();
 540                 }
 541 #if !NET_2_1
 542                 protected bool UseOptionC ()
 543                 {
 544                         return ((roptions & RegexOptions.Compiled) != 0);
 545                 }
 546 #endif
 547                 protected bool UseOptionR ()
 548                 {
 549                         return ((roptions & RegexOptions.RightToLeft) != 0);
 550                 }
 551
 552                 // object methods
 553
 554                 public override string ToString ()
 555                 {
 556                         return pattern;
 557                 }
 558
 559                 // ISerializable interface
 560                 void ISerializable.GetObjectData (SerializationInfo info, StreamingContext context)
 561                 {
 562                         info.AddValue ("pattern", this.ToString (), typeof (string));
 563                         info.AddValue ("options", this.Options, typeof (RegexOptions));
 564                 }
 565
 566                 // internal
 567
 568                 internal int GroupCount {
 569                         get { return group_count; }
 570                 }
 571
 572                 internal int Gap {
 573                         get { return gap; }
 574                 }
 575
 576                 // private
 577
 578                 private IMachine CreateMachine ()
 579                 {
 580                         return machineFactory.NewInstance ();
 581                 }
 582
 583                 private static string [] GetGroupNamesArray (int groupCount, IDictionary mapping)
 584                 {
 585                         string [] group_names = new string [groupCount + 1];
 586                         IDictionaryEnumerator de = mapping.GetEnumerator ();
 587                         while (de.MoveNext ())
 588                                 group_names [(int) de.Value] = (string) de.Key;
 589                         return group_names;
 590                 }
 591
 592                 private int [] GroupNumbers {
 593                         get {
 594                                 if (group_numbers == null) {
 595                                         group_numbers = new int [1 + group_count];
 596                                         for (int i = 0; i < gap; ++i)
 597                                                 group_numbers [i] = i;
 598                                         for (int i = gap; i <= group_count; ++i)
 599                                                 group_numbers [i] = Int32.Parse (group_names [i]);
 600                                         return group_numbers;
 601                                 }
 602                                 return group_numbers;
 603                         }
 604                 }
 605
 606                 private IMachineFactory machineFactory;
 607                 private IDictionary mapping;
 608                 private int group_count;
 609                 private int gap;
 610                 private bool refsInitialized;
 611                 private string [] group_names;
 612                 private int [] group_numbers;
 613
 614                 // protected members
 615
 616                 protected internal string pattern;
 617                 protected internal RegexOptions roptions;
 618
 619                 // MS undocumented members
 620 #if NET_2_1
 621                 [MonoTODO]
 622                 internal System.Collections.Generic.Dictionary<string, int> capnames;
 623                 [MonoTODO]
 624                 internal System.Collections.Generic.Dictionary<int, int> caps;
 625 #else
 626                 [MonoTODO]
 627                 protected internal System.Collections.Hashtable capnames;
 628                 [MonoTODO]
 629                 protected internal System.Collections.Hashtable caps;
 630
 631                 [MonoTODO]
 632                 protected internal RegexRunnerFactory factory;
 633 #endif
 634                 [MonoTODO]
 635                 protected internal int capsize;
 636                 [MonoTODO]
 637                 protected internal string [] capslist;
 638         }
 639 }