mcs/class/System.Web/System.Web.Compilation/AspParser.cs

   1 //
   2 // System.Web.Compilation.AspParser
   3 //
   4 // Authors:
   5 //      Gonzalo Paniagua Javier (gonzalo@ximian.com)
   6 //
   7 // (C) 2002,2003 Ximian, Inc (http://www.ximian.com)
   8 //
   9
  10 //
  11 // Permission is hereby granted, free of charge, to any person obtaining
  12 // a copy of this software and associated documentation files (the
  13 // "Software"), to deal in the Software without restriction, including
  14 // without limitation the rights to use, copy, modify, merge, publish,
  15 // distribute, sublicense, and/or sell copies of the Software, and to
  16 // permit persons to whom the Software is furnished to do so, subject to
  17 // the following conditions:
  18 //
  19 // The above copyright notice and this permission notice shall be
  20 // included in all copies or substantial portions of the Software.
  21 //
  22 // THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
  23 // EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
  24 // MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
  25 // NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
  26 // LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
  27 // OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
  28 // WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  29 //
  30 using System;
  31 using System.ComponentModel;
  32 using System.Collections;
  33 using System.Globalization;
  34 using System.IO;
  35 using System.Text;
  36 using System.Web.Util;
  37 using System.Security.Cryptography;
  38
  39 namespace System.Web.Compilation
  40 {
  41         delegate void ParseErrorHandler (ILocation location, string message);
  42         delegate void TextParsedHandler (ILocation location, string text);
  43         delegate void TagParsedHandler (ILocation location, TagType tagtype, string id, TagAttributes attributes);
  44
  45         class AspParser : ILocation
  46         {
  47                 static readonly object errorEvent = new object ();
  48                 static readonly object tagParsedEvent = new object ();
  49                 static readonly object textParsedEvent = new object ();
  50
  51 #if NET_2_0
  52                 byte[] md5checksum;
  53 #endif
  54                 AspTokenizer tokenizer;
  55                 int beginLine, endLine;
  56                 int beginColumn, endColumn;
  57                 int beginPosition, endPosition;
  58                 string filename;
  59                 string fileText;
  60                 string verbatimID;
  61
  62                 EventHandlerList events = new EventHandlerList ();
  63
  64                 public event ParseErrorHandler Error {
  65                         add { events.AddHandler (errorEvent, value); }
  66                         remove { events.RemoveHandler (errorEvent, value); }
  67                 }
  68
  69                 public event TagParsedHandler TagParsed {
  70                         add { events.AddHandler (tagParsedEvent, value); }
  71                         remove { events.RemoveHandler (tagParsedEvent, value); }
  72                 }
  73
  74                 public event TextParsedHandler TextParsed {
  75                         add { events.AddHandler (textParsedEvent, value); }
  76                         remove { events.RemoveHandler (textParsedEvent, value); }
  77                 }
  78
  79                 public AspParser (string filename, TextReader input)
  80                 {
  81                         this.filename = filename;
  82                         fileText = input.ReadToEnd ();
  83 #if NET_2_0
  84                         MD5 md5 = MD5.Create ();
  85                         md5checksum = md5.ComputeHash (Encoding.UTF8.GetBytes (fileText));
  86 #endif
  87                         StringReader reader = new StringReader (fileText);
  88                         tokenizer = new AspTokenizer (reader);
  89                 }
  90
  91 #if NET_2_0
  92                 public byte[] MD5Checksum {
  93                         get { return md5checksum; }
  94                 }
  95 #endif
  96
  97                 public int BeginLine {
  98                         get { return beginLine; }
  99                 }
 100
 101                 public int BeginColumn {
 102                         get { return beginColumn; }
 103                 }
 104
 105                 public int EndLine {
 106                         get { return endLine; }
 107                 }
 108
 109                 public int EndColumn {
 110                         get { return endColumn; }
 111                 }
 112
 113                 public string FileText {
 114                         get { return fileText; }
 115                 }
 116
 117                 public string PlainText {
 118                         get {
 119                                 if (beginPosition >= endPosition)
 120                                         return null;
 121
 122                                 return fileText.Substring (beginPosition, endPosition - beginPosition);
 123                         }
 124                 }
 125
 126                 public string Filename {
 127                         get { return filename; }
 128                 }
 129
 130                 public string VerbatimID {
 131                         set {
 132                                 tokenizer.Verbatim = true;
 133                                 verbatimID = value;
 134                         }
 135                 }
 136
 137                 bool Eat (int expected_token)
 138                 {
 139                         if (tokenizer.get_token () != expected_token) {
 140                                 tokenizer.put_back ();
 141                                 return false;
 142                         }
 143
 144                         endLine = tokenizer.EndLine;
 145                         endColumn = tokenizer.EndColumn;
 146                         return true;
 147                 }
 148
 149                 void BeginElement ()
 150                 {
 151                         beginLine = tokenizer.BeginLine;
 152                         beginColumn = tokenizer.BeginColumn;
 153                         beginPosition = tokenizer.Position - 1;
 154                 }
 155
 156                 void EndElement ()
 157                 {
 158                         endLine = tokenizer.EndLine;
 159                         endColumn = tokenizer.EndColumn;
 160                         endPosition = tokenizer.Position;
 161                 }
 162
 163                 public void Parse ()
 164                 {
 165                         int token;
 166                         string id;
 167                         TagAttributes attributes;
 168                         TagType tagtype = TagType.Text;
 169                         StringBuilder text =  new StringBuilder ();
 170
 171                         while ((token = tokenizer.get_token ()) != Token.EOF) {
 172                                 BeginElement ();
 173
 174                                 if (tokenizer.Verbatim){
 175                                         string end_verbatim = "</" + verbatimID + ">";
 176                                         string verbatim_text = GetVerbatim (token, end_verbatim);
 177
 178                                         if (verbatim_text == null)
 179                                                 OnError ("Unexpected EOF processing " + verbatimID);
 180
 181                                         tokenizer.Verbatim = false;
 182
 183                                         EndElement ();
 184                                         endPosition -= end_verbatim.Length;
 185                                         OnTextParsed (verbatim_text);
 186                                         beginPosition = endPosition;
 187                                         endPosition += end_verbatim.Length;
 188                                         OnTagParsed (TagType.Close, verbatimID, null);
 189                                         continue;
 190                                 }
 191
 192                                 if (token == '<') {
 193                                         GetTag (out tagtype, out id, out attributes);
 194                                         EndElement ();
 195                                         if (tagtype == TagType.ServerComment)
 196                                                 continue;
 197
 198                                         if (tagtype == TagType.Text)
 199                                                 OnTextParsed (id);
 200                                         else
 201                                                 OnTagParsed (tagtype, id, attributes);
 202
 203                                         continue;
 204                                 }
 205
 206                                 if (tokenizer.Value.Trim () == "" && tagtype == TagType.Directive) {
 207                                         continue;
 208                                 }
 209
 210                                 text.Length = 0;
 211                                 do {
 212                                         text.Append (tokenizer.Value);
 213                                         token = tokenizer.get_token ();
 214                                 } while (token != '<' && token != Token.EOF);
 215
 216                                 tokenizer.put_back ();
 217                                 EndElement ();
 218                                 OnTextParsed (text.ToString ());
 219                         }
 220                 }
 221
 222                 bool GetInclude (string str, out string pathType, out string filename)
 223                 {
 224                         pathType = null;
 225                         filename = null;
 226                         str = str.Substring (2).Trim ();
 227                         int len = str.Length;
 228                         int lastQuote = str.LastIndexOf ('"');
 229                         if (len < 10 || lastQuote != len - 1)
 230                                 return false;
 231
 232                         if (!StrUtils.StartsWith (str, "#include ", true))
 233                                 return false;
 234
 235                         str = str.Substring (9).Trim ();
 236                         bool isfile = (StrUtils.StartsWith (str ,"file", true));
 237                         if (!isfile && !StrUtils.StartsWith (str, "virtual", true))
 238                                 return false;
 239
 240                         pathType = (isfile) ? "file" : "virtual";
 241                         if (str.Length < pathType.Length + 3)
 242                                 return false;
 243
 244                         str = str.Substring (pathType.Length).Trim ();
 245                         if (str.Length < 3 || str [0] != '=')
 246                                 return false;
 247
 248                         int index = 1;
 249                         for (; index < str.Length; index++) {
 250                                 if (Char.IsWhiteSpace (str [index]))
 251                                         continue;
 252                                 else if (str [index] == '"')
 253                                         break;
 254                         }
 255
 256                         if (index == str.Length || index == lastQuote)
 257                                 return false;
 258
 259                         str = str.Substring (index);
 260                         if (str.Length == 2) { // only quotes
 261                                 OnError ("Empty file name.");
 262                                 return false;
 263                         }
 264
 265                         filename = str.Trim ().Substring (index, str.Length - 2);
 266                         if (filename.LastIndexOf  ('"') != -1)
 267                                 return false; // file=""" -> no error
 268
 269                         return true;
 270                 }
 271
 272                 void GetTag (out TagType tagtype, out string id, out TagAttributes attributes)
 273                 {
 274                         int token = tokenizer.get_token ();
 275
 276                         tagtype = TagType.ServerComment;
 277                         id = null;
 278                         attributes = null;
 279                         switch (token){
 280                         case '%':
 281                                 GetServerTag (out tagtype, out id, out attributes);
 282                                 break;
 283                         case '/':
 284                                 if (!Eat (Token.IDENTIFIER))
 285                                         OnError ("expecting TAGNAME");
 286
 287                                 id = tokenizer.Value;
 288                                 if (!Eat ('>'))
 289                                         OnError ("expecting '>'. Got '" + id + "'");
 290
 291                                 tagtype = TagType.Close;
 292                                 break;
 293                         case '!':
 294                                 bool double_dash = Eat (Token.DOUBLEDASH);
 295                                 if (double_dash)
 296                                         tokenizer.put_back ();
 297
 298                                 tokenizer.Verbatim = true;
 299                                 string end = double_dash ? "-->" : ">";
 300                                 string comment = GetVerbatim (tokenizer.get_token (), end);
 301                                 tokenizer.Verbatim = false;
 302                                 if (comment == null)
 303                                         OnError ("Unfinished HTML comment/DTD");
 304
 305                                 string pathType, filename;
 306                                 if (double_dash && GetInclude (comment, out pathType, out filename)) {
 307                                         tagtype = TagType.Include;
 308                                         attributes = new TagAttributes ();
 309                                         attributes.Add (pathType, filename);
 310                                 } else {
 311                                         tagtype = TagType.Text;
 312                                         id = "<!" + comment + end;
 313                                 }
 314                                 break;
 315                         case Token.IDENTIFIER:
 316                                 if (this.filename == "@@inner_string@@") {
 317                                         // Actually not tag but "xxx < yyy" stuff in inner_string!
 318                                         tagtype = TagType.Text;
 319                                         tokenizer.InTag = false;
 320                                         id = "<" + tokenizer.Odds + tokenizer.Value;
 321                                 } else {
 322                                         id = tokenizer.Value;
 323                                         try {
 324                                                 attributes = GetAttributes ();
 325                                         } catch (Exception e) {
 326                                                 OnError (e.Message);
 327                                                 break;
 328                                         }
 329
 330                                         tagtype = TagType.Tag;
 331                                         if (Eat ('/') && Eat ('>')) {
 332                                                 tagtype = TagType.SelfClosing;
 333                                         } else if (!Eat ('>')) {
 334                                                 if (attributes.IsRunAtServer ()) {
 335                                                         OnError ("The server tag is not well formed.");
 336                                                         break;
 337                                                 }
 338                                                 tokenizer.Verbatim = true;
 339                                                 attributes.Add ("", GetVerbatim (tokenizer.get_token (), ">") + ">");
 340                                                 tokenizer.Verbatim = false;
 341                                         }
 342                                 }
 343
 344                                 break;
 345                         default:
 346                                 tagtype = TagType.Text;
 347                                 tokenizer.InTag = false;
 348                                 id = "<" + tokenizer.Value;
 349                                 break;
 350                         }
 351                 }
 352
 353                 TagAttributes GetAttributes ()
 354                 {
 355                         int token;
 356                         TagAttributes attributes;
 357                         string id;
 358                         bool wellFormedForServer = true;
 359
 360                         attributes = new TagAttributes ();
 361                         while ((token = tokenizer.get_token ()) != Token.EOF){
 362                                 if (token == '<' && Eat ('%')) {
 363                                         tokenizer.Verbatim = true;
 364                                         attributes.Add ("", "<%" +
 365                                                         GetVerbatim (tokenizer.get_token (), "%>") + "%>");
 366                                         tokenizer.Verbatim = false;
 367                                         tokenizer.InTag = true;
 368                                         continue;
 369                                 }
 370
 371                                 if (token != Token.IDENTIFIER)
 372                                         break;
 373
 374                                 id = tokenizer.Value;
 375                                 if (Eat ('=')){
 376                                         if (Eat (Token.ATTVALUE)){
 377                                                 attributes.Add (id, tokenizer.Value);
 378                                                 wellFormedForServer &= tokenizer.AlternatingQuotes;
 379                                         } else if (Eat ('<') && Eat ('%')) {
 380                                                 tokenizer.Verbatim = true;
 381                                                 attributes.Add (id, "<%" +
 382                                                                 GetVerbatim (tokenizer.get_token (), "%>") + "%>");
 383                                                 tokenizer.Verbatim = false;
 384                                                 tokenizer.InTag = true;
 385                                         } else {
 386                                                 OnError ("expected ATTVALUE");
 387                                                 return null;
 388                                         }
 389                                 } else {
 390                                         attributes.Add (id, null);
 391                                 }
 392                         }
 393
 394                         tokenizer.put_back ();
 395
 396                         if (attributes.IsRunAtServer () && !wellFormedForServer) {
 397                                 OnError ("The server tag is not well formed.");
 398                                 return null;
 399                         }
 400
 401                         return attributes;
 402                 }
 403
 404                 string GetVerbatim (int token, string end)
 405                 {
 406                         StringBuilder vb_text = new StringBuilder ();
 407                         StringBuilder tmp = new StringBuilder ();
 408                         int i = 0;
 409
 410                         if (tokenizer.Value.Length > 1){
 411                                 // May be we have a put_back token that is not a single character
 412                                 vb_text.Append (tokenizer.Value);
 413                                 token = tokenizer.get_token ();
 414                         }
 415
 416                         end = end.ToLower (CultureInfo.InvariantCulture);
 417
 418                         while (token != Token.EOF){
 419                                 if (Char.ToLower ((char) token, CultureInfo.InvariantCulture) == end [i]){
 420                                         if (++i >= end.Length)
 421                                                 break;
 422                                         tmp.Append ((char) token);
 423                                         token = tokenizer.get_token ();
 424                                         continue;
 425                                 } else if (i > 0) {
 426                                         vb_text.Append (tmp.ToString ());
 427                                         tmp.Remove (0, tmp.Length);
 428                                         i = 0;
 429                                 }
 430
 431                                 vb_text.Append ((char) token);
 432                                 token = tokenizer.get_token ();
 433                         }
 434
 435                         if (token == Token.EOF)
 436                                 OnError ("Expecting " + end + " and got EOF.");
 437
 438                         return RemoveComments (vb_text.ToString ());
 439                 }
 440
 441                 string RemoveComments (string text)
 442                 {
 443                         int end;
 444                         int start = text.IndexOf ("<%--");
 445
 446                         while (start != -1) {
 447                                 end = text.IndexOf ("--%>");
 448                                 if (end == -1 || end <= start + 1)
 449                                         break;
 450
 451                                 text = text.Remove (start, end - start + 4);
 452                                 start = text.IndexOf ("<%--");
 453                         }
 454
 455                         return text;
 456                 }
 457
 458                 void GetServerTag (out TagType tagtype, out string id, out TagAttributes attributes)
 459                 {
 460                         string inside_tags;
 461                         bool old = tokenizer.ExpectAttrValue;
 462
 463                         tokenizer.ExpectAttrValue = false;
 464                         if (Eat ('@')){
 465                                 tokenizer.ExpectAttrValue = old;
 466                                 tagtype = TagType.Directive;
 467                                 id = "";
 468                                 if (Eat (Token.DIRECTIVE))
 469                                         id = tokenizer.Value;
 470
 471                                 attributes = GetAttributes ();
 472                                 if (!Eat ('%') || !Eat ('>'))
 473                                         OnError ("expecting '%>'");
 474
 475                                 return;
 476                         }
 477
 478                         if (Eat (Token.DOUBLEDASH)) {
 479                                 tokenizer.ExpectAttrValue = old;
 480                                 tokenizer.Verbatim = true;
 481                                 inside_tags = GetVerbatim (tokenizer.get_token (), "--%>");
 482                                 tokenizer.Verbatim = false;
 483                                 id = null;
 484                                 attributes = null;
 485                                 tagtype = TagType.ServerComment;
 486                                 return;
 487                         }
 488
 489                         tokenizer.ExpectAttrValue = old;
 490                         bool varname;
 491                         bool databinding;
 492                         varname = Eat ('=');
 493                         databinding = !varname && Eat ('#');
 494
 495                         tokenizer.Verbatim = true;
 496                         inside_tags = GetVerbatim (tokenizer.get_token (), "%>");
 497                         tokenizer.Verbatim = false;
 498                         id = inside_tags;
 499                         attributes = null;
 500                         tagtype = (databinding ? TagType.DataBinding :
 501                                   (varname ? TagType.CodeRenderExpression : TagType.CodeRender));
 502                 }
 503
 504                 void OnError (string msg)
 505                 {
 506                         ParseErrorHandler eh = events [errorEvent] as ParseErrorHandler;
 507                         if (eh != null)
 508                                 eh (this, msg);
 509                 }
 510
 511                 void OnTagParsed (TagType tagtype, string id, TagAttributes attributes)
 512                 {
 513                         TagParsedHandler eh = events [tagParsedEvent] as TagParsedHandler;
 514                         if (eh != null)
 515                                 eh (this, tagtype, id, attributes);
 516                 }
 517
 518                 void OnTextParsed (string text)
 519                 {
 520                         TextParsedHandler eh = events [textParsedEvent] as TextParsedHandler;
 521                         if (eh != null)
 522                                 eh (this, text);
 523                 }
 524         }
 525
 526 }
 527