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 MOBILE || !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 MOBILE || !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 seems to be working now, but
 278                 // potential problems are being tracked down here:
 279                 // https://bugzilla.novell.com/show_bug.cgi?id=470827
 280                 static readonly bool old_rx =
 281                         Environment.GetEnvironmentVariable ("MONO_NEW_RX") == null;
 282 #endif
 283
 284                 private static IMachineFactory CreateMachineFactory (string pattern, RegexOptions options)
 285                 {
 286                         Parser psr = new Parser ();
 287                         RegularExpression re = psr.ParseRegularExpression (pattern, options);
 288
 289 #if NET_2_1
 290                         ICompiler cmp = new PatternCompiler ();
 291 #else
 292                         ICompiler cmp;
 293                         if (!old_rx) {
 294                                 if ((options & RegexOptions.Compiled) != 0)
 295                                         cmp = new CILCompiler ();
 296                                 else
 297                                         cmp = new RxCompiler ();
 298                         } else {
 299                                 cmp = new PatternCompiler ();
 300                         }
 301 #endif
 302
 303                         re.Compile (cmp, (options & RegexOptions.RightToLeft) != 0);
 304
 305                         IMachineFactory machineFactory = cmp.GetMachineFactory ();
 306                         Hashtable mapping = new Hashtable ();
 307                         machineFactory.Gap = psr.GetMapping (mapping);
 308                         machineFactory.Mapping = mapping;
 309                         machineFactory.NamesMapping = GetGroupNamesArray (machineFactory.GroupCount, machineFactory.Mapping);
 310
 311                         return machineFactory;
 312                 }
 313
 314 #if NET_2_0
 315                 protected
 316 #else
 317                 private
 318 #endif
 319                 Regex (SerializationInfo info, StreamingContext context) :
 320                         this (info.GetString ("pattern"),
 321                               (RegexOptions) info.GetValue ("options", typeof (RegexOptions)))
 322                 {
 323                 }
 324
 325 #if ONLY_1_1 && !TARGET_JVM
 326                 // fixes public API signature
 327                 ~Regex ()
 328                 {
 329                 }
 330 #endif
 331                 // public instance properties
 332
 333                 public RegexOptions Options {
 334                         get { return roptions; }
 335                 }
 336
 337                 public bool RightToLeft {
 338                         get { return (roptions & RegexOptions.RightToLeft) != 0; }
 339                 }
 340
 341                 // public instance methods
 342
 343                 public string [] GetGroupNames ()
 344                 {
 345                         string [] names = new string [1 + group_count];
 346                         Array.Copy (group_names, names, 1 + group_count);
 347                         return names;
 348                 }
 349
 350                 public int [] GetGroupNumbers ()
 351                 {
 352                         int [] numbers = new int [1 + group_count];
 353                         Array.Copy (GroupNumbers, numbers, 1 + group_count);
 354                         return numbers;
 355                 }
 356
 357                 public string GroupNameFromNumber (int i)
 358                 {
 359                         i = GetGroupIndex (i);
 360                         if (i < 0)
 361                                 return "";
 362
 363                         return group_names [i];
 364                 }
 365
 366                 public int GroupNumberFromName (string name)
 367                 {
 368                         if (!mapping.Contains (name))
 369                                 return -1;
 370                         int i = (int) mapping [name];
 371                         if (i >= gap)
 372                                 i = Int32.Parse (name);
 373                         return i;
 374                 }
 375
 376                 internal int GetGroupIndex (int number)
 377                 {
 378                         if (number < gap)
 379                                 return number;
 380                         if (gap > group_count)
 381                                 return -1;
 382                         return Array.BinarySearch (GroupNumbers, gap, group_count - gap + 1, number);
 383                 }
 384
 385                 int default_startat (string input)
 386                 {
 387                         return (RightToLeft && input != null) ? input.Length : 0;
 388                 }
 389
 390                 // match methods
 391
 392                 public bool IsMatch (string input)
 393                 {
 394                         return IsMatch (input, default_startat (input));
 395                 }
 396
 397                 public bool IsMatch (string input, int startat)
 398                 {
 399                         return Match (input, startat).Success;
 400                 }
 401
 402                 public Match Match (string input)
 403                 {
 404                         return Match (input, default_startat (input));
 405                 }
 406
 407                 public Match Match (string input, int startat)
 408                 {
 409                         if (input == null)
 410                                 throw new ArgumentNullException ("input");
 411                         if (startat < 0 || startat > input.Length)
 412                                 throw new ArgumentOutOfRangeException ("startat");
 413                         return CreateMachine ().Scan (this, input, startat, input.Length);
 414                 }
 415
 416                 public Match Match (string input, int beginning, int length)
 417                 {
 418                         if (input == null)
 419                                 throw new ArgumentNullException ("input");
 420                         if (beginning < 0 || beginning > input.Length)
 421                                 throw new ArgumentOutOfRangeException ("beginning");
 422                         if (length < 0 || length > input.Length - beginning)
 423                                 throw new ArgumentOutOfRangeException ("length");
 424                         return CreateMachine ().Scan (this, input, beginning, beginning + length);
 425                 }
 426
 427                 public MatchCollection Matches (string input)
 428                 {
 429                         return Matches (input, default_startat (input));
 430                 }
 431
 432                 public MatchCollection Matches (string input, int startat)
 433                 {
 434                         Match m = Match (input, startat);
 435                         return new MatchCollection (m);
 436                 }
 437
 438                 // replace methods
 439
 440                 public string Replace (string input, MatchEvaluator evaluator)
 441                 {
 442                         return Replace (input, evaluator, Int32.MaxValue, default_startat (input));
 443                 }
 444
 445                 public string Replace (string input, MatchEvaluator evaluator, int count)
 446                 {
 447                         return Replace (input, evaluator, count, default_startat (input));
 448                 }
 449
 450                 class Adapter {
 451                         MatchEvaluator ev;
 452                         public Adapter (MatchEvaluator ev) { this.ev = ev; }
 453                         public void Evaluate (Match m, StringBuilder sb) { sb.Append (ev (m)); }
 454                 }
 455
 456                 public string Replace (string input, MatchEvaluator evaluator, int count, int startat)
 457                 {
 458                         if (input == null)
 459                                 throw new ArgumentNullException ("input");
 460                         if (evaluator == null)
 461                                 throw new ArgumentNullException ("evaluator");
 462                         if (count < -1)
 463                                 throw new ArgumentOutOfRangeException ("count");
 464                         if (startat < 0 || startat > input.Length)
 465                                 throw new ArgumentOutOfRangeException ("startat");
 466
 467                         BaseMachine m = (BaseMachine)CreateMachine ();
 468
 469                         if (RightToLeft)
 470                                 return m.RTLReplace (this, input, evaluator, count, startat);
 471
 472                         // NOTE: If this is a cause of a lot of allocations, we can convert it to
 473                         //       use a ThreadStatic allocation mitigator
 474                         Adapter a = new Adapter (evaluator);
 475
 476                         return m.LTRReplace (this, input, new BaseMachine.MatchAppendEvaluator (a.Evaluate),
 477                                                                  count, startat);
 478                 }
 479
 480                 public string Replace (string input, string replacement)
 481                 {
 482                         return Replace (input, replacement, Int32.MaxValue, default_startat (input));
 483                 }
 484
 485                 public string Replace (string input, string replacement, int count)
 486                 {
 487                         return Replace (input, replacement, count, default_startat (input));
 488                 }
 489
 490                 public string Replace (string input, string replacement, int count, int startat)
 491                 {
 492                         if (input == null)
 493                                 throw new ArgumentNullException ("input");
 494                         if (replacement == null)
 495                                 throw new ArgumentNullException ("replacement");
 496                         if (count < -1)
 497                                 throw new ArgumentOutOfRangeException ("count");
 498                         if (startat < 0 || startat > input.Length)
 499                                 throw new ArgumentOutOfRangeException ("startat");
 500
 501                         return CreateMachine ().Replace (this, input, replacement, count, startat);
 502                 }
 503
 504                 // split methods
 505
 506                 public string [] Split (string input)
 507                 {
 508                         return Split (input, Int32.MaxValue, default_startat (input));
 509                 }
 510
 511                 public string [] Split (string input, int count)
 512                 {
 513                         return Split (input, count, default_startat (input));
 514                 }
 515
 516                 public string [] Split (string input, int count, int startat)
 517                 {
 518                         if (input == null)
 519                                 throw new ArgumentNullException ("input");
 520                         if (count < 0)
 521                                 throw new ArgumentOutOfRangeException ("count");
 522                         if (startat < 0 || startat > input.Length)
 523                                 throw new ArgumentOutOfRangeException ("startat");
 524
 525                         return CreateMachine ().Split (this, input, count, startat);
 526                 }
 527
 528                 // This method is called at the end of the constructor of compiled
 529                 // regular expression classes to do internal initialization.
 530                 protected void InitializeReferences ()
 531                 {
 532                         if (refsInitialized)
 533                                 throw new NotSupportedException ("This operation is only allowed once per object.");
 534
 535                         refsInitialized = true;
 536
 537                         // Compile pattern that results in performance loss as existing
 538                         // CIL code is ignored but provides support for regular
 539                         // expressions compiled to assemblies.
 540                         Init ();
 541                 }
 542 #if !NET_2_1
 543                 protected bool UseOptionC ()
 544                 {
 545                         return ((roptions & RegexOptions.Compiled) != 0);
 546                 }
 547 #endif
 548                 protected bool UseOptionR ()
 549                 {
 550                         return ((roptions & RegexOptions.RightToLeft) != 0);
 551                 }
 552
 553                 // object methods
 554
 555                 public override string ToString ()
 556                 {
 557                         return pattern;
 558                 }
 559
 560                 // ISerializable interface
 561                 void ISerializable.GetObjectData (SerializationInfo info, StreamingContext context)
 562                 {
 563                         info.AddValue ("pattern", this.ToString (), typeof (string));
 564                         info.AddValue ("options", this.Options, typeof (RegexOptions));
 565                 }
 566
 567                 // internal
 568
 569                 internal int GroupCount {
 570                         get { return group_count; }
 571                 }
 572
 573                 internal int Gap {
 574                         get { return gap; }
 575                 }
 576
 577                 // private
 578
 579                 private IMachine CreateMachine ()
 580                 {
 581                         return machineFactory.NewInstance ();
 582                 }
 583
 584                 private static string [] GetGroupNamesArray (int groupCount, IDictionary mapping)
 585                 {
 586                         string [] group_names = new string [groupCount + 1];
 587                         IDictionaryEnumerator de = mapping.GetEnumerator ();
 588                         while (de.MoveNext ())
 589                                 group_names [(int) de.Value] = (string) de.Key;
 590                         return group_names;
 591                 }
 592
 593                 private int [] GroupNumbers {
 594                         get {
 595                                 if (group_numbers == null) {
 596                                         group_numbers = new int [1 + group_count];
 597                                         for (int i = 0; i < gap; ++i)
 598                                                 group_numbers [i] = i;
 599                                         for (int i = gap; i <= group_count; ++i)
 600                                                 group_numbers [i] = Int32.Parse (group_names [i]);
 601                                         return group_numbers;
 602                                 }
 603                                 return group_numbers;
 604                         }
 605                 }
 606
 607                 private IMachineFactory machineFactory;
 608                 private IDictionary mapping;
 609                 private int group_count;
 610                 private int gap;
 611                 private bool refsInitialized;
 612                 private string [] group_names;
 613                 private int [] group_numbers;
 614
 615                 // protected members
 616
 617                 protected internal string pattern;
 618                 protected internal RegexOptions roptions;
 619
 620                 // MS undocumented members
 621 #if NET_2_1
 622                 [MonoTODO]
 623                 internal System.Collections.Generic.Dictionary<string, int> capnames;
 624                 [MonoTODO]
 625                 internal System.Collections.Generic.Dictionary<int, int> caps;
 626 #else
 627                 [MonoTODO]
 628                 protected internal System.Collections.Hashtable capnames;
 629                 [MonoTODO]
 630                 protected internal System.Collections.Hashtable caps;
 631
 632                 [MonoTODO]
 633                 protected internal RegexRunnerFactory factory;
 634 #endif
 635                 [MonoTODO]
 636                 protected internal int capsize;
 637                 [MonoTODO]
 638                 protected internal string [] capslist;
 639         }
 640 }