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                 static FactoryCache cache = new FactoryCache (15);
 181                 public static int CacheSize {
 182                         get { return cache.Capacity; }
 183                         set {
 184                                 if (value < 0)
 185                                         throw new ArgumentOutOfRangeException ("CacheSize");
 186
 187                                 cache.Capacity = value;
 188                         }
 189                 }
 190
 191                 // private
 192
 193
 194                 // constructors
 195
 196                 // This constructor is used by compiled regular expressions that are
 197                 // classes derived from Regex class. No initialization required.
 198                 protected Regex ()
 199                 {
 200                 }
 201
 202                 public Regex (string pattern) : this (pattern, RegexOptions.None)
 203                 {
 204                 }
 205
 206                 public Regex (string pattern, RegexOptions options)
 207                 {
 208                         if (pattern == null)
 209                                 throw new ArgumentNullException ("pattern");
 210                         validate_options (options);
 211                         this.pattern = pattern;
 212                         this.roptions = options;
 213                         Init ();
 214                 }
 215
 216                 static void validate_options (RegexOptions options)
 217                 {
 218                         const RegexOptions allopts =
 219                                 RegexOptions.None |
 220                                 RegexOptions.IgnoreCase |
 221                                 RegexOptions.Multiline |
 222                                 RegexOptions.ExplicitCapture |
 223 #if MOBILE || !NET_2_1
 224                                 RegexOptions.Compiled |
 225 #endif
 226                                 RegexOptions.Singleline |
 227                                 RegexOptions.IgnorePatternWhitespace |
 228                                 RegexOptions.RightToLeft |
 229                                 RegexOptions.ECMAScript |
 230                                 RegexOptions.CultureInvariant;
 231
 232                         const RegexOptions ecmaopts =
 233                                 RegexOptions.IgnoreCase |
 234                                 RegexOptions.Multiline |
 235 #if MOBILE || !NET_2_1
 236                                 RegexOptions.Compiled |
 237 #endif
 238                                 RegexOptions.ECMAScript;
 239
 240                         if ((options & ~allopts) != 0)
 241                                 throw new ArgumentOutOfRangeException ("options");
 242                         if ((options & RegexOptions.ECMAScript) != 0 && (options & ~ecmaopts) != 0)
 243                                 throw new ArgumentOutOfRangeException ("options");
 244                 }
 245
 246 #if !TARGET_JVM
 247                 private void Init ()
 248                 {
 249                         this.machineFactory = cache.Lookup (this.pattern, this.roptions);
 250
 251                         if (this.machineFactory == null) {
 252                                 InitNewRegex();
 253                         } else {
 254                                 this.group_count = this.machineFactory.GroupCount;
 255                                 this.gap = this.machineFactory.Gap;
 256                                 this.mapping = this.machineFactory.Mapping;
 257                                 this.group_names = this.machineFactory.NamesMapping;
 258                         }
 259                 }
 260 #endif
 261
 262                 private void InitNewRegex ()
 263                 {
 264                         this.machineFactory = CreateMachineFactory (this.pattern, this.roptions);
 265                         cache.Add (this.pattern, this.roptions, this.machineFactory);
 266                         this.group_count = machineFactory.GroupCount;
 267                         this.gap = this.machineFactory.Gap;
 268                         this.mapping = machineFactory.Mapping;
 269                         this.group_names = this.machineFactory.NamesMapping;
 270                 }
 271
 272 #if !NET_2_1
 273                 // The new rx engine seems to be working now, but
 274                 // potential problems are being tracked down here:
 275                 // https://bugzilla.novell.com/show_bug.cgi?id=470827
 276                 static readonly bool old_rx =
 277                         Environment.GetEnvironmentVariable ("MONO_NEW_RX") == null;
 278 #endif
 279
 280                 private static IMachineFactory CreateMachineFactory (string pattern, RegexOptions options)
 281                 {
 282                         Parser psr = new Parser ();
 283                         RegularExpression re = psr.ParseRegularExpression (pattern, options);
 284
 285 #if NET_2_1
 286                         ICompiler cmp = new PatternCompiler ();
 287 #else
 288                         ICompiler cmp;
 289                         if (!old_rx) {
 290                                 if ((options & RegexOptions.Compiled) != 0)
 291                                         cmp = new CILCompiler ();
 292                                 else
 293                                         cmp = new RxCompiler ();
 294                         } else {
 295                                 cmp = new PatternCompiler ();
 296                         }
 297 #endif
 298
 299                         re.Compile (cmp, (options & RegexOptions.RightToLeft) != 0);
 300
 301                         IMachineFactory machineFactory = cmp.GetMachineFactory ();
 302                         Hashtable mapping = new Hashtable ();
 303                         machineFactory.Gap = psr.GetMapping (mapping);
 304                         machineFactory.Mapping = mapping;
 305                         machineFactory.NamesMapping = GetGroupNamesArray (machineFactory.GroupCount, machineFactory.Mapping);
 306
 307                         return machineFactory;
 308                 }
 309
 310                 protected Regex (SerializationInfo info, StreamingContext context) :
 311                         this (info.GetString ("pattern"),
 312                               (RegexOptions) info.GetValue ("options", typeof (RegexOptions)))
 313                 {
 314                 }
 315
 316                 // public instance properties
 317
 318                 public RegexOptions Options {
 319                         get { return roptions; }
 320                 }
 321
 322                 public bool RightToLeft {
 323                         get { return (roptions & RegexOptions.RightToLeft) != 0; }
 324                 }
 325
 326                 // public instance methods
 327
 328                 public string [] GetGroupNames ()
 329                 {
 330                         string [] names = new string [1 + group_count];
 331                         Array.Copy (group_names, names, 1 + group_count);
 332                         return names;
 333                 }
 334
 335                 public int [] GetGroupNumbers ()
 336                 {
 337                         int [] numbers = new int [1 + group_count];
 338                         Array.Copy (GroupNumbers, numbers, 1 + group_count);
 339                         return numbers;
 340                 }
 341
 342                 public string GroupNameFromNumber (int i)
 343                 {
 344                         i = GetGroupIndex (i);
 345                         if (i < 0)
 346                                 return "";
 347
 348                         return group_names [i];
 349                 }
 350
 351                 public int GroupNumberFromName (string name)
 352                 {
 353                         if (!mapping.Contains (name))
 354                                 return -1;
 355                         int i = (int) mapping [name];
 356                         if (i >= gap)
 357                                 i = Int32.Parse (name);
 358                         return i;
 359                 }
 360
 361                 internal int GetGroupIndex (int number)
 362                 {
 363                         if (number < gap)
 364                                 return number;
 365                         if (gap > group_count)
 366                                 return -1;
 367                         return Array.BinarySearch (GroupNumbers, gap, group_count - gap + 1, number);
 368                 }
 369
 370                 int default_startat (string input)
 371                 {
 372                         return (RightToLeft && input != null) ? input.Length : 0;
 373                 }
 374
 375                 // match methods
 376
 377                 public bool IsMatch (string input)
 378                 {
 379                         return IsMatch (input, default_startat (input));
 380                 }
 381
 382                 public bool IsMatch (string input, int startat)
 383                 {
 384                         return Match (input, startat).Success;
 385                 }
 386
 387                 public Match Match (string input)
 388                 {
 389                         return Match (input, default_startat (input));
 390                 }
 391
 392                 public Match Match (string input, int startat)
 393                 {
 394                         if (input == null)
 395                                 throw new ArgumentNullException ("input");
 396                         if (startat < 0 || startat > input.Length)
 397                                 throw new ArgumentOutOfRangeException ("startat");
 398                         return CreateMachine ().Scan (this, input, startat, input.Length);
 399                 }
 400
 401                 public Match Match (string input, int beginning, int length)
 402                 {
 403                         if (input == null)
 404                                 throw new ArgumentNullException ("input");
 405                         if (beginning < 0 || beginning > input.Length)
 406                                 throw new ArgumentOutOfRangeException ("beginning");
 407                         if (length < 0 || length > input.Length - beginning)
 408                                 throw new ArgumentOutOfRangeException ("length");
 409                         return CreateMachine ().Scan (this, input, beginning, beginning + length);
 410                 }
 411
 412                 public MatchCollection Matches (string input)
 413                 {
 414                         return Matches (input, default_startat (input));
 415                 }
 416
 417                 public MatchCollection Matches (string input, int startat)
 418                 {
 419                         Match m = Match (input, startat);
 420                         return new MatchCollection (m);
 421                 }
 422
 423                 // replace methods
 424
 425                 public string Replace (string input, MatchEvaluator evaluator)
 426                 {
 427                         return Replace (input, evaluator, Int32.MaxValue, default_startat (input));
 428                 }
 429
 430                 public string Replace (string input, MatchEvaluator evaluator, int count)
 431                 {
 432                         return Replace (input, evaluator, count, default_startat (input));
 433                 }
 434
 435                 class Adapter {
 436                         MatchEvaluator ev;
 437                         public Adapter (MatchEvaluator ev) { this.ev = ev; }
 438                         public void Evaluate (Match m, StringBuilder sb) { sb.Append (ev (m)); }
 439                 }
 440
 441                 public string Replace (string input, MatchEvaluator evaluator, int count, int startat)
 442                 {
 443                         if (input == null)
 444                                 throw new ArgumentNullException ("input");
 445                         if (evaluator == null)
 446                                 throw new ArgumentNullException ("evaluator");
 447                         if (count < -1)
 448                                 throw new ArgumentOutOfRangeException ("count");
 449                         if (startat < 0 || startat > input.Length)
 450                                 throw new ArgumentOutOfRangeException ("startat");
 451
 452                         BaseMachine m = (BaseMachine)CreateMachine ();
 453
 454                         if (RightToLeft)
 455                                 return m.RTLReplace (this, input, evaluator, count, startat);
 456
 457                         // NOTE: If this is a cause of a lot of allocations, we can convert it to
 458                         //       use a ThreadStatic allocation mitigator
 459                         Adapter a = new Adapter (evaluator);
 460
 461                         return m.LTRReplace (this, input, new BaseMachine.MatchAppendEvaluator (a.Evaluate),
 462                                                                  count, startat);
 463                 }
 464
 465                 public string Replace (string input, string replacement)
 466                 {
 467                         return Replace (input, replacement, Int32.MaxValue, default_startat (input));
 468                 }
 469
 470                 public string Replace (string input, string replacement, int count)
 471                 {
 472                         return Replace (input, replacement, count, default_startat (input));
 473                 }
 474
 475                 public string Replace (string input, string replacement, int count, int startat)
 476                 {
 477                         if (input == null)
 478                                 throw new ArgumentNullException ("input");
 479                         if (replacement == null)
 480                                 throw new ArgumentNullException ("replacement");
 481                         if (count < -1)
 482                                 throw new ArgumentOutOfRangeException ("count");
 483                         if (startat < 0 || startat > input.Length)
 484                                 throw new ArgumentOutOfRangeException ("startat");
 485
 486                         return CreateMachine ().Replace (this, input, replacement, count, startat);
 487                 }
 488
 489                 // split methods
 490
 491                 public string [] Split (string input)
 492                 {
 493                         return Split (input, Int32.MaxValue, default_startat (input));
 494                 }
 495
 496                 public string [] Split (string input, int count)
 497                 {
 498                         return Split (input, count, default_startat (input));
 499                 }
 500
 501                 public string [] Split (string input, int count, int startat)
 502                 {
 503                         if (input == null)
 504                                 throw new ArgumentNullException ("input");
 505                         if (count < 0)
 506                                 throw new ArgumentOutOfRangeException ("count");
 507                         if (startat < 0 || startat > input.Length)
 508                                 throw new ArgumentOutOfRangeException ("startat");
 509
 510                         return CreateMachine ().Split (this, input, count, startat);
 511                 }
 512
 513                 // This method is called at the end of the constructor of compiled
 514                 // regular expression classes to do internal initialization.
 515                 protected void InitializeReferences ()
 516                 {
 517                         if (refsInitialized)
 518                                 throw new NotSupportedException ("This operation is only allowed once per object.");
 519
 520                         refsInitialized = true;
 521
 522                         // Compile pattern that results in performance loss as existing
 523                         // CIL code is ignored but provides support for regular
 524                         // expressions compiled to assemblies.
 525                         Init ();
 526                 }
 527 #if !NET_2_1
 528                 protected bool UseOptionC ()
 529                 {
 530                         return ((roptions & RegexOptions.Compiled) != 0);
 531                 }
 532 #endif
 533                 protected bool UseOptionR ()
 534                 {
 535                         return ((roptions & RegexOptions.RightToLeft) != 0);
 536                 }
 537
 538                 // object methods
 539
 540                 public override string ToString ()
 541                 {
 542                         return pattern;
 543                 }
 544
 545                 // ISerializable interface
 546                 void ISerializable.GetObjectData (SerializationInfo info, StreamingContext context)
 547                 {
 548                         info.AddValue ("pattern", this.ToString (), typeof (string));
 549                         info.AddValue ("options", this.Options, typeof (RegexOptions));
 550                 }
 551
 552                 // internal
 553
 554                 internal int GroupCount {
 555                         get { return group_count; }
 556                 }
 557
 558                 internal int Gap {
 559                         get { return gap; }
 560                 }
 561
 562                 // private
 563
 564                 private IMachine CreateMachine ()
 565                 {
 566                         return machineFactory.NewInstance ();
 567                 }
 568
 569                 private static string [] GetGroupNamesArray (int groupCount, IDictionary mapping)
 570                 {
 571                         string [] group_names = new string [groupCount + 1];
 572                         IDictionaryEnumerator de = mapping.GetEnumerator ();
 573                         while (de.MoveNext ())
 574                                 group_names [(int) de.Value] = (string) de.Key;
 575                         return group_names;
 576                 }
 577
 578                 private int [] GroupNumbers {
 579                         get {
 580                                 if (group_numbers == null) {
 581                                         group_numbers = new int [1 + group_count];
 582                                         for (int i = 0; i < gap; ++i)
 583                                                 group_numbers [i] = i;
 584                                         for (int i = gap; i <= group_count; ++i)
 585                                                 group_numbers [i] = Int32.Parse (group_names [i]);
 586                                         return group_numbers;
 587                                 }
 588                                 return group_numbers;
 589                         }
 590                 }
 591
 592                 private IMachineFactory machineFactory;
 593                 private IDictionary mapping;
 594                 private int group_count;
 595                 private int gap;
 596                 private bool refsInitialized;
 597                 private string [] group_names;
 598                 private int [] group_numbers;
 599
 600                 // protected members
 601
 602                 protected internal string pattern;
 603                 protected internal RegexOptions roptions;
 604
 605                 // MS undocumented members
 606 #if NET_2_1
 607                 [MonoTODO]
 608                 internal System.Collections.Generic.Dictionary<string, int> capnames;
 609                 [MonoTODO]
 610                 internal System.Collections.Generic.Dictionary<int, int> caps;
 611 #else
 612                 [MonoTODO]
 613                 protected internal System.Collections.Hashtable capnames;
 614                 [MonoTODO]
 615                 protected internal System.Collections.Hashtable caps;
 616
 617                 [MonoTODO]
 618                 protected internal RegexRunnerFactory factory;
 619 #endif
 620                 [MonoTODO]
 621                 protected internal int capsize;
 622                 [MonoTODO]
 623                 protected internal string [] capslist;
 624         }
 625 }