/AvalonEdit/ICSharpCode.AvalonEdit/Xml/TagReader.cs
C# | 740 lines | 646 code | 41 blank | 53 comment | 187 complexity | 77c8afecec91eee81431195dac97a856 MD5 | raw file
1// Copyright (c) AlphaSierraPapa for the SharpDevelop Team (for details please see \doc\copyright.txt) 2// This code is distributed under the GNU LGPL (for details please see \doc\license.txt) 3 4using System; 5using System.Collections.Generic; 6using System.Globalization; 7using System.Linq; 8using System.Text; 9 10namespace ICSharpCode.AvalonEdit.Xml 11{ 12 class TagReader: TokenReader 13 { 14 AXmlParser parser; 15 TrackedSegmentCollection trackedSegments; 16 string input; 17 18 public TagReader(AXmlParser parser, string input): base(input) 19 { 20 this.parser = parser; 21 this.trackedSegments = parser.TrackedSegments; 22 this.input = input; 23 } 24 25 bool TryReadFromCacheOrNew<T>(out T res) where T: AXmlObject, new() 26 { 27 return TryReadFromCacheOrNew(out res, t => true); 28 } 29 30 bool TryReadFromCacheOrNew<T>(out T res, Predicate<T> condition) where T: AXmlObject, new() 31 { 32 T cached = trackedSegments.GetCachedObject<T>(this.CurrentLocation, 0, condition); 33 if (cached != null) { 34 Skip(cached.Length); 35 AXmlParser.Assert(cached.Length > 0, "cached elements must not have zero length"); 36 res = cached; 37 return true; 38 } else { 39 res = new T(); 40 return false; 41 } 42 } 43 44 void OnParsed(AXmlObject obj) 45 { 46 AXmlParser.Log("Parsed {0}", obj); 47 trackedSegments.AddParsedObject(obj, this.MaxTouchedLocation > this.CurrentLocation ? (int?)this.MaxTouchedLocation : null); 48 } 49 50 /// <summary> 51 /// Read all tags in the document in a flat sequence. 52 /// It also includes the text between tags and possibly some properly nested Elements from cache. 53 /// </summary> 54 public List<AXmlObject> ReadAllTags() 55 { 56 List<AXmlObject> stream = new List<AXmlObject>(); 57 58 while(true) { 59 if (IsEndOfFile()) { 60 break; 61 } else if (TryPeek('<')) { 62 AXmlElement elem; 63 if (TryReadFromCacheOrNew(out elem, e => e.IsProperlyNested)) { 64 stream.Add(elem); 65 } else { 66 stream.Add(ReadTag()); 67 } 68 } else { 69 stream.AddRange(ReadText(TextType.CharacterData)); 70 } 71 } 72 73 return stream; 74 } 75 76 /// <summary> 77 /// Context: "<" 78 /// </summary> 79 AXmlTag ReadTag() 80 { 81 AssertHasMoreData(); 82 83 AXmlTag tag; 84 if (TryReadFromCacheOrNew(out tag)) return tag; 85 86 tag.StartOffset = this.CurrentLocation; 87 88 // Read the opening bracket 89 // It identifies the type of tag and parsing behavior for the rest of it 90 tag.OpeningBracket = ReadOpeningBracket(); 91 92 if (tag.IsUnknownBang && !TryPeekWhiteSpace()) 93 OnSyntaxError(tag, tag.StartOffset, this.CurrentLocation, "Unknown tag"); 94 95 if (tag.IsStartOrEmptyTag || tag.IsEndTag || tag.IsProcessingInstruction) { 96 // Read the name 97 string name; 98 if (TryReadName(out name)) { 99 if (!IsValidName(name)) { 100 OnSyntaxError(tag, this.CurrentLocation - name.Length, this.CurrentLocation, "The name '{0}' is invalid", name); 101 } 102 } else { 103 OnSyntaxError(tag, "Element name expected"); 104 } 105 tag.Name = name; 106 } else { 107 tag.Name = string.Empty; 108 } 109 110 bool isXmlDeclr = tag.StartOffset == 0 && tag.Name == "xml"; 111 112 if (tag.IsStartOrEmptyTag || tag.IsEndTag || isXmlDeclr) { 113 // Read attributes for the tag 114 while(true) { 115 // Chech for all forbiden 'name' charcters first - see ReadName 116 if (IsEndOfFile()) break; 117 if (TryPeekWhiteSpace()) { 118 tag.AddChildren(ReadText(TextType.WhiteSpace)); 119 continue; // End of file might be next 120 } 121 if (TryPeek('<')) break; 122 string endBr; 123 int endBrStart = this.CurrentLocation; // Just peek 124 if (TryReadClosingBracket(out endBr)) { // End tag 125 GoBack(endBrStart); 126 break; 127 } 128 129 // We have "=\'\"" or name - read attribute 130 AXmlAttribute attr = ReadAttribulte(); 131 tag.AddChild(attr); 132 if (tag.IsEndTag) 133 OnSyntaxError(tag, attr.StartOffset, attr.EndOffset, "Attribute not allowed in end tag."); 134 } 135 } else if (tag.IsDocumentType) { 136 tag.AddChildren(ReadContentOfDTD()); 137 } else { 138 int start = this.CurrentLocation; 139 IEnumerable<AXmlObject> text; 140 if (tag.IsComment) { 141 text = ReadText(TextType.Comment); 142 } else if (tag.IsCData) { 143 text = ReadText(TextType.CData); 144 } else if (tag.IsProcessingInstruction) { 145 text = ReadText(TextType.ProcessingInstruction); 146 } else if (tag.IsUnknownBang) { 147 text = ReadText(TextType.UnknownBang); 148 } else { 149 throw new InternalException(string.Format(CultureInfo.InvariantCulture, "Unknown opening bracket '{0}'", tag.OpeningBracket)); 150 } 151 // Enumerate 152 text = text.ToList(); 153 // Backtrack at complete start 154 if (IsEndOfFile() || (tag.IsUnknownBang && TryPeek('<'))) { 155 GoBack(start); 156 } else { 157 tag.AddChildren(text); 158 } 159 } 160 161 // Read closing bracket 162 string bracket; 163 TryReadClosingBracket(out bracket); 164 tag.ClosingBracket = bracket; 165 166 // Error check 167 int brStart = this.CurrentLocation - (tag.ClosingBracket ?? string.Empty).Length; 168 int brEnd = this.CurrentLocation; 169 if (tag.Name == null) { 170 // One error was reported already 171 } else if (tag.IsStartOrEmptyTag) { 172 if (tag.ClosingBracket != ">" && tag.ClosingBracket != "/>") OnSyntaxError(tag, brStart, brEnd, "'>' or '/>' expected"); 173 } else if (tag.IsEndTag) { 174 if (tag.ClosingBracket != ">") OnSyntaxError(tag, brStart, brEnd, "'>' expected"); 175 } else if (tag.IsComment) { 176 if (tag.ClosingBracket != "-->") OnSyntaxError(tag, brStart, brEnd, "'-->' expected"); 177 } else if (tag.IsCData) { 178 if (tag.ClosingBracket != "]]>") OnSyntaxError(tag, brStart, brEnd, "']]>' expected"); 179 } else if (tag.IsProcessingInstruction) { 180 if (tag.ClosingBracket != "?>") OnSyntaxError(tag, brStart, brEnd, "'?>' expected"); 181 } else if (tag.IsUnknownBang) { 182 if (tag.ClosingBracket != ">") OnSyntaxError(tag, brStart, brEnd, "'>' expected"); 183 } else if (tag.IsDocumentType) { 184 if (tag.ClosingBracket != ">") OnSyntaxError(tag, brStart, brEnd, "'>' expected"); 185 } else { 186 throw new InternalException(string.Format(CultureInfo.InvariantCulture, "Unknown opening bracket '{0}'", tag.OpeningBracket)); 187 } 188 189 // Attribute name may not apper multiple times 190 var duplicates = tag.Children.OfType<AXmlAttribute>().GroupBy(attr => attr.Name).SelectMany(g => g.Skip(1)); 191 foreach(AXmlAttribute attr in duplicates) { 192 OnSyntaxError(tag, attr.StartOffset, attr.EndOffset, "Attribute with name '{0}' already exists", attr.Name); 193 } 194 195 tag.EndOffset = this.CurrentLocation; 196 197 OnParsed(tag); 198 return tag; 199 } 200 201 /// <summary> 202 /// Reads any of the know opening brackets. (only full bracket) 203 /// Context: "<" 204 /// </summary> 205 string ReadOpeningBracket() 206 { 207 // We are using a lot of string literals so that the memory instances are shared 208 //int start = this.CurrentLocation; 209 if (TryRead('<')) { 210 if (TryRead('/')) { 211 return "</"; 212 } else if (TryRead('?')) { 213 return "<?"; 214 } else if (TryRead('!')) { 215 if (TryRead("--")) { 216 return "<!--"; 217 } else if (TryRead("[CDATA[")) { 218 return "<![CDATA["; 219 } else { 220 foreach(string dtdName in AXmlTag.DtdNames) { 221 // the dtdName includes "<!" 222 if (TryRead(dtdName.Remove(0, 2))) return dtdName; 223 } 224 return "<!"; 225 } 226 } else { 227 return "<"; 228 } 229 } else { 230 throw new InternalException("'<' expected"); 231 } 232 } 233 234 /// <summary> 235 /// Reads any of the know closing brackets. (only full bracket) 236 /// Context: any 237 /// </summary> 238 bool TryReadClosingBracket(out string bracket) 239 { 240 // We are using a lot of string literals so that the memory instances are shared 241 if (TryRead('>')) { 242 bracket = ">"; 243 } else if (TryRead("/>")) { 244 bracket = "/>"; 245 } else if (TryRead("?>")) { 246 bracket = "?>"; 247 } else if (TryRead("-->")) { 248 bracket = "-->"; 249 } else if (TryRead("]]>")) { 250 bracket = "]]>"; 251 } else { 252 bracket = string.Empty; 253 return false; 254 } 255 return true; 256 } 257 258 IEnumerable<AXmlObject> ReadContentOfDTD() 259 { 260 int start = this.CurrentLocation; 261 while(true) { 262 if (IsEndOfFile()) break; // End of file 263 TryMoveToNonWhiteSpace(); // Skip whitespace 264 if (TryRead('\'')) TryMoveTo('\''); // Skip single quoted string TODO: Bug 265 if (TryRead('\"')) TryMoveTo('\"'); // Skip single quoted string 266 if (TryRead('[')) { // Start of nested infoset 267 // Reading infoset 268 while(true) { 269 if (IsEndOfFile()) break; 270 TryMoveToAnyOf('<', ']'); 271 if (TryPeek('<')) { 272 if (start != this.CurrentLocation) { // Two following tags 273 yield return MakeText(start, this.CurrentLocation); 274 } 275 yield return ReadTag(); 276 start = this.CurrentLocation; 277 } 278 if (TryPeek(']')) break; 279 } 280 } 281 TryRead(']'); // End of nested infoset 282 if (TryPeek('>')) break; // Proper closing 283 if (TryPeek('<')) break; // Malformed XML 284 TryMoveNext(); // Skip anything else 285 } 286 if (start != this.CurrentLocation) { 287 yield return MakeText(start, this.CurrentLocation); 288 } 289 } 290 291 /// <summary> 292 /// Context: name or "=\'\"" 293 /// </summary> 294 AXmlAttribute ReadAttribulte() 295 { 296 AssertHasMoreData(); 297 298 AXmlAttribute attr; 299 if (TryReadFromCacheOrNew(out attr)) return attr; 300 301 attr.StartOffset = this.CurrentLocation; 302 303 // Read name 304 string name; 305 if (TryReadName(out name)) { 306 if (!IsValidName(name)) { 307 OnSyntaxError(attr, this.CurrentLocation - name.Length, this.CurrentLocation, "The name '{0}' is invalid", name); 308 } 309 } else { 310 OnSyntaxError(attr, "Attribute name expected"); 311 } 312 attr.Name = name; 313 314 // Read equals sign and surrounding whitespace 315 int checkpoint = this.CurrentLocation; 316 TryMoveToNonWhiteSpace(); 317 if (TryRead('=')) { 318 int chk2 = this.CurrentLocation; 319 TryMoveToNonWhiteSpace(); 320 if (!TryPeek('"') && !TryPeek('\'')) { 321 // Do not read whitespace if quote does not follow 322 GoBack(chk2); 323 } 324 attr.EqualsSign = GetText(checkpoint, this.CurrentLocation); 325 } else { 326 GoBack(checkpoint); 327 OnSyntaxError(attr, "'=' expected"); 328 attr.EqualsSign = string.Empty; 329 } 330 331 // Read attribute value 332 int start = this.CurrentLocation; 333 char quoteChar = TryPeek('"') ? '"' : '\''; 334 bool startsWithQuote; 335 if (TryRead(quoteChar)) { 336 startsWithQuote = true; 337 int valueStart = this.CurrentLocation; 338 TryMoveToAnyOf(quoteChar, '<'); 339 if (TryRead(quoteChar)) { 340 if (!TryPeekAnyOf(' ', '\t', '\n', '\r', '/', '>', '?')) { 341 if (TryPeekPrevious('=', 2) || (TryPeekPrevious('=', 3) && TryPeekPrevious(' ', 2))) { 342 // This actually most likely means that we are in the next attribute value 343 GoBack(valueStart); 344 ReadAttributeValue(quoteChar); 345 if (TryRead(quoteChar)) { 346 OnSyntaxError(attr, "White space or end of tag expected"); 347 } else { 348 OnSyntaxError(attr, "Quote {0} expected (or add whitespace after the following one)", quoteChar); 349 } 350 } else { 351 OnSyntaxError(attr, "White space or end of tag expected"); 352 } 353 } 354 } else { 355 // '<' or end of file 356 GoBack(valueStart); 357 ReadAttributeValue(quoteChar); 358 OnSyntaxError(attr, "Quote {0} expected", quoteChar); 359 } 360 } else { 361 startsWithQuote = false; 362 int valueStart = this.CurrentLocation; 363 ReadAttributeValue(null); 364 TryRead('\"'); 365 TryRead('\''); 366 if (valueStart == this.CurrentLocation) { 367 OnSyntaxError(attr, "Attribute value expected"); 368 } else { 369 OnSyntaxError(attr, valueStart, this.CurrentLocation, "Attribute value must be quoted"); 370 } 371 } 372 attr.QuotedValue = GetText(start, this.CurrentLocation); 373 attr.Value = Unquote(attr.QuotedValue); 374 attr.Value = Dereference(attr, attr.Value, startsWithQuote ? start + 1 : start); 375 376 attr.EndOffset = this.CurrentLocation; 377 378 OnParsed(attr); 379 return attr; 380 } 381 382 /// <summary> 383 /// Read everything up to quote (excluding), opening/closing tag or attribute signature 384 /// </summary> 385 void ReadAttributeValue(char? quote) 386 { 387 while(true) { 388 if (IsEndOfFile()) return; 389 // What is next? 390 int start = this.CurrentLocation; 391 TryMoveToNonWhiteSpace(); // Read white space (if any) 392 if (quote.HasValue) { 393 if (TryPeek(quote.Value)) return; 394 } else { 395 if (TryPeek('"') || TryPeek('\'')) return; 396 } 397 // Opening/closing tag 398 string endBr; 399 if (TryPeek('<') || TryReadClosingBracket(out endBr)) { 400 GoBack(start); 401 return; 402 } 403 // Try reading attribute signature 404 string name; 405 if (TryReadName(out name)) { 406 int nameEnd = this.CurrentLocation; 407 if (TryMoveToNonWhiteSpace() && TryRead("=") && 408 TryMoveToNonWhiteSpace() && TryPeekAnyOf('"', '\'')) 409 { 410 // Start of attribute. Great 411 GoBack(start); 412 return; // Done 413 } else { 414 // Just some gargabe - make it part of the value 415 GoBack(nameEnd); 416 continue; // Read more 417 } 418 } 419 TryMoveNext(); // Accept everyting else 420 } 421 } 422 423 AXmlText MakeText(int start, int end) 424 { 425 AXmlParser.DebugAssert(end > start, "Empty text"); 426 427 AXmlText text = new AXmlText() { 428 StartOffset = start, 429 EndOffset = end, 430 EscapedValue = GetText(start, end), 431 Type = TextType.Other 432 }; 433 434 OnParsed(text); 435 return text; 436 } 437 438 const int maxEntityLength = 16; // The longest build-in one is 10 ("") 439 const int maxTextFragmentSize = 64; 440 const int lookAheadLength = (3 * maxTextFragmentSize) / 2; // More so that we do not get small "what was inserted" fragments 441 442 /// <summary> 443 /// Reads text and optionaly separates it into fragments. 444 /// It can also return empty set for no appropriate text input. 445 /// Make sure you enumerate it only once 446 /// </summary> 447 IEnumerable<AXmlObject> ReadText(TextType type) 448 { 449 bool lookahead = false; 450 while(true) { 451 AXmlText text; 452 if (TryReadFromCacheOrNew(out text, t => t.Type == type)) { 453 // Cached text found 454 yield return text; 455 continue; // Read next fragment; the method can handle "no text left" 456 } 457 text.Type = type; 458 459 // Limit the reading to just a few characters 460 // (the first character not to be read) 461 int fragmentEnd = Math.Min(this.CurrentLocation + maxTextFragmentSize, this.InputLength); 462 463 // Look if some futher text has been already processed and align so that 464 // we hit that chache point. It is expensive so it is off for the first run 465 if (lookahead) { 466 // Note: Must fit entity 467 AXmlObject nextFragment = trackedSegments.GetCachedObject<AXmlText>(this.CurrentLocation + maxEntityLength, lookAheadLength - maxEntityLength, t => t.Type == type); 468 if (nextFragment != null) { 469 fragmentEnd = Math.Min(nextFragment.StartOffset, this.InputLength); 470 AXmlParser.Log("Parsing only text ({0}-{1}) because later text was already processed", this.CurrentLocation, fragmentEnd); 471 } 472 } 473 lookahead = true; 474 475 text.StartOffset = this.CurrentLocation; 476 int start = this.CurrentLocation; 477 478 // Whitespace would be skipped anyway by any operation 479 TryMoveToNonWhiteSpace(fragmentEnd); 480 int wsEnd = this.CurrentLocation; 481 482 // Try move to the terminator given by the context 483 if (type == TextType.WhiteSpace) { 484 TryMoveToNonWhiteSpace(fragmentEnd); 485 } else if (type == TextType.CharacterData) { 486 while(true) { 487 if (!TryMoveToAnyOf(new char[] {'<', ']'}, fragmentEnd)) break; // End of fragment 488 if (TryPeek('<')) break; 489 if (TryPeek(']')) { 490 if (TryPeek("]]>")) { 491 OnSyntaxError(text, this.CurrentLocation, this.CurrentLocation + 3, "']]>' is not allowed in text"); 492 } 493 TryMoveNext(); 494 continue; 495 } 496 throw new Exception("Infinite loop"); 497 } 498 } else if (type == TextType.Comment) { 499 // Do not report too many errors 500 bool errorReported = false; 501 while(true) { 502 if (!TryMoveTo('-', fragmentEnd)) break; // End of fragment 503 if (TryPeek("-->")) break; 504 if (TryPeek("--") && !errorReported) { 505 OnSyntaxError(text, this.CurrentLocation, this.CurrentLocation + 2, "'--' is not allowed in comment"); 506 errorReported = true; 507 } 508 TryMoveNext(); 509 } 510 } else if (type == TextType.CData) { 511 while(true) { 512 // We can not use use TryMoveTo("]]>", fragmentEnd) because it may incorectly accept "]" at the end of fragment 513 if (!TryMoveTo(']', fragmentEnd)) break; // End of fragment 514 if (TryPeek("]]>")) break; 515 TryMoveNext(); 516 } 517 } else if (type == TextType.ProcessingInstruction) { 518 while(true) { 519 if (!TryMoveTo('?', fragmentEnd)) break; // End of fragment 520 if (TryPeek("?>")) break; 521 TryMoveNext(); 522 } 523 } else if (type == TextType.UnknownBang) { 524 TryMoveToAnyOf(new char[] {'<', '>'}, fragmentEnd); 525 } else { 526 throw new Exception("Uknown type " + type); 527 } 528 529 text.ContainsOnlyWhitespace = (wsEnd == this.CurrentLocation); 530 531 // Terminal found or real end was reached; 532 bool finished = this.CurrentLocation < fragmentEnd || IsEndOfFile(); 533 534 if (!finished) { 535 // We have to continue reading more text fragments 536 537 // If there is entity reference, make sure the next segment starts with it to prevent framentation 538 int entitySearchStart = Math.Max(start + 1 /* data for us */, this.CurrentLocation - maxEntityLength); 539 int entitySearchLength = this.CurrentLocation - entitySearchStart; 540 if (entitySearchLength > 0) { 541 // Note that LastIndexOf works backward 542 int entityIndex = input.LastIndexOf('&', this.CurrentLocation - 1, entitySearchLength); 543 if (entityIndex != -1) { 544 GoBack(entityIndex); 545 } 546 } 547 } 548 549 text.EscapedValue = GetText(start, this.CurrentLocation); 550 if (type == TextType.CharacterData) { 551 // Normalize end of line first 552 text.Value = Dereference(text, NormalizeEndOfLine(text.EscapedValue), start); 553 } else { 554 text.Value = text.EscapedValue; 555 } 556 text.EndOffset = this.CurrentLocation; 557 558 if (text.EscapedValue.Length > 0) { 559 OnParsed(text); 560 yield return text; 561 } 562 563 if (finished) { 564 yield break; 565 } 566 } 567 } 568 569 #region Helper methods 570 571 void OnSyntaxError(AXmlObject obj, string message, params object[] args) 572 { 573 OnSyntaxError(obj, this.CurrentLocation, this.CurrentLocation + 1, message, args); 574 } 575 576 public static void OnSyntaxError(AXmlObject obj, int start, int end, string message, params object[] args) 577 { 578 if (end <= start) end = start + 1; 579 string formattedMessage = string.Format(CultureInfo.InvariantCulture, message, args); 580 AXmlParser.Log("Syntax error ({0}-{1}): {2}", start, end, formattedMessage); 581 obj.AddSyntaxError(new SyntaxError() { 582 Object = obj, 583 StartOffset = start, 584 EndOffset = end, 585 Message = formattedMessage, 586 }); 587 } 588 589 static bool IsValidName(string name) 590 { 591 try { 592 System.Xml.XmlConvert.VerifyName(name); 593 return true; 594 } catch (System.Xml.XmlException) { 595 return false; 596 } 597 } 598 599 /// <summary> Remove quoting from the given string </summary> 600 static string Unquote(string quoted) 601 { 602 if (string.IsNullOrEmpty(quoted)) return string.Empty; 603 char first = quoted[0]; 604 if (quoted.Length == 1) return (first == '"' || first == '\'') ? string.Empty : quoted; 605 char last = quoted[quoted.Length - 1]; 606 if (first == '"' || first == '\'') { 607 if (first == last) { 608 // Remove both quotes 609 return quoted.Substring(1, quoted.Length - 2); 610 } else { 611 // Remove first quote 612 return quoted.Remove(0, 1); 613 } 614 } else { 615 if (last == '"' || last == '\'') { 616 // Remove last quote 617 return quoted.Substring(0, quoted.Length - 1); 618 } else { 619 // Keep whole string 620 return quoted; 621 } 622 } 623 } 624 625 static string NormalizeEndOfLine(string text) 626 { 627 return text.Replace("\r\n", "\n").Replace("\r", "\n"); 628 } 629 630 string Dereference(AXmlObject owner, string text, int textLocation) 631 { 632 StringBuilder sb = null; // The dereferenced text so far (all up to 'curr') 633 int curr = 0; 634 while(true) { 635 // Reached end of input 636 if (curr == text.Length) { 637 if (sb != null) { 638 return sb.ToString(); 639 } else { 640 return text; 641 } 642 } 643 644 // Try to find reference 645 int start = text.IndexOf('&', curr); 646 647 // No more references found 648 if (start == -1) { 649 if (sb != null) { 650 sb.Append(text, curr, text.Length - curr); // Add rest 651 return sb.ToString(); 652 } else { 653 return text; 654 } 655 } 656 657 // Append text before the enitiy reference 658 if (sb == null) sb = new StringBuilder(text.Length); 659 sb.Append(text, curr, start - curr); 660 curr = start; 661 662 // Process the entity 663 int errorLoc = textLocation + sb.Length; 664 665 // Find entity name 666 int end = text.IndexOfAny(new char[] {'&', ';'}, start + 1, Math.Min(maxEntityLength, text.Length - (start + 1))); 667 if (end == -1 || text[end] == '&') { 668 // Not found 669 OnSyntaxError(owner, errorLoc, errorLoc + 1, "Entity reference must be terminated with ';'"); 670 // Keep '&' 671 sb.Append('&'); 672 curr++; 673 continue; // Restart and next character location 674 } 675 string name = text.Substring(start + 1, end - (start + 1)); 676 677 // Resolve the name 678 string replacement; 679 if (name.Length == 0) { 680 replacement = null; 681 OnSyntaxError(owner, errorLoc + 1, errorLoc + 1, "Entity name expected"); 682 } else if (name == "amp") { 683 replacement = "&"; 684 } else if (name == "lt") { 685 replacement = "<"; 686 } else if (name == "gt") { 687 replacement = ">"; 688 } else if (name == "apos") { 689 replacement = "'"; 690 } else if (name == "quot") { 691 replacement = "\""; 692 } else if (name.Length > 0 && name[0] == '#') { 693 int num; 694 if (name.Length > 1 && name[1] == 'x') { 695 if (!int.TryParse(name.Substring(2), NumberStyles.AllowHexSpecifier, CultureInfo.InvariantCulture.NumberFormat, out num)) { 696 num = -1; 697 OnSyntaxError(owner, errorLoc + 3, errorLoc + 1 + name.Length, "Hexadecimal code of unicode character expected"); 698 } 699 } else { 700 if (!int.TryParse(name.Substring(1), NumberStyles.None, CultureInfo.InvariantCulture.NumberFormat, out num)) { 701 num = -1; 702 OnSyntaxError(owner, errorLoc + 2, errorLoc + 1 + name.Length, "Numeric code of unicode character expected"); 703 } 704 } 705 if (num != -1) { 706 try { 707 replacement = char.ConvertFromUtf32(num); 708 } catch (ArgumentOutOfRangeException) { 709 replacement = null; 710 OnSyntaxError(owner, errorLoc + 2, errorLoc + 1 + name.Length, "Invalid unicode character U+{0:X} ({0})", num); 711 } 712 } else { 713 replacement = null; 714 } 715 } else if (!IsValidName(name)) { 716 replacement = null; 717 OnSyntaxError(owner, errorLoc + 1, errorLoc + 1, "Invalid entity name"); 718 } else { 719 replacement = null; 720 if (parser.UnknownEntityReferenceIsError) { 721 OnSyntaxError(owner, errorLoc, errorLoc + 1 + name.Length + 1, "Unknown entity reference '{0}'", name); 722 } 723 } 724 725 // Append the replacement to output 726 if (replacement != null) { 727 sb.Append(replacement); 728 } else { 729 sb.Append('&'); 730 sb.Append(name); 731 sb.Append(';'); 732 } 733 curr = end + 1; 734 continue; 735 } 736 } 737 738 #endregion 739 } 740}