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.Collections;
  32 using System.Globalization;
  33 using System.IO;
  34 using System.Text;
  35 using System.Web.Util;
  36
  37 namespace System.Web.Compilation
  38 {
  39         delegate void ParseErrorHandler (ILocation location, string message);
  40         delegate void TextParsedHandler (ILocation location, string text);
  41         delegate void TagParsedHandler (ILocation location, TagType tagtype, string id, TagAttributes attributes);
  42
  43         class AspParser : ILocation
  44         {
  45                 AspTokenizer tokenizer;
  46                 int beginLine, endLine;
  47                 int beginColumn, endColumn;
  48                 int beginPosition, endPosition;
  49                 string filename;
  50                 string fileText;
  51                 string verbatimID;
  52
  53                 public AspParser (string filename, TextReader input)
  54                 {
  55                         this.filename = filename;
  56                         fileText = input.ReadToEnd ();
  57                         StringReader reader = new StringReader (fileText);
  58                         tokenizer = new AspTokenizer (reader);
  59                 }
  60
  61                 public int BeginLine {
  62                         get { return beginLine; }
  63                 }
  64
  65                 public int BeginColumn {
  66                         get { return beginColumn; }
  67                 }
  68
  69                 public int EndLine {
  70                         get { return endLine; }
  71                 }
  72
  73                 public int EndColumn {
  74                         get { return endColumn; }
  75                 }
  76
  77                 public string PlainText {
  78                         get {
  79                                 if (beginPosition >= endPosition)
  80                                         return null;
  81
  82                                 return fileText.Substring (beginPosition, endPosition - beginPosition);
  83                         }
  84                 }
  85
  86                 public string Filename {
  87                         get { return filename; }
  88                 }
  89
  90                 public string VerbatimID {
  91                         set {
  92                                 tokenizer.Verbatim = true;
  93                                 verbatimID = value;
  94                         }
  95                 }
  96
  97                 bool Eat (int expected_token)
  98                 {
  99                         if (tokenizer.get_token () != expected_token) {
 100                                 tokenizer.put_back ();
 101                                 return false;
 102                         }
 103
 104                         endLine = tokenizer.EndLine;
 105                         endColumn = tokenizer.EndColumn;
 106                         return true;
 107                 }
 108
 109                 void BeginElement ()
 110                 {
 111                         beginLine = tokenizer.BeginLine;
 112                         beginColumn = tokenizer.BeginColumn;
 113                         beginPosition = tokenizer.Position - 1;
 114                 }
 115
 116                 void EndElement ()
 117                 {
 118                         endLine = tokenizer.EndLine;
 119                         endColumn = tokenizer.EndColumn;
 120                         endPosition = tokenizer.Position;
 121                 }
 122
 123                 public void Parse ()
 124                 {
 125                         int token;
 126                         string id;
 127                         TagAttributes attributes;
 128                         TagType tagtype = TagType.Text;
 129                         StringBuilder text =  new StringBuilder ();
 130
 131                         while ((token = tokenizer.get_token ()) != Token.EOF) {
 132                                 BeginElement ();
 133
 134                                 if (tokenizer.Verbatim){
 135                                         string end_verbatim = "</" + verbatimID + ">";
 136                                         string verbatim_text = GetVerbatim (token, end_verbatim);
 137
 138                                         if (verbatim_text == null)
 139                                                 OnError ("Unexpected EOF processing " + verbatimID);
 140
 141                                         tokenizer.Verbatim = false;
 142
 143                                         EndElement ();
 144                                         endPosition -= end_verbatim.Length;
 145                                         OnTextParsed (verbatim_text);
 146                                         beginPosition = endPosition;
 147                                         endPosition += end_verbatim.Length;
 148                                         OnTagParsed (TagType.Close, verbatimID, null);
 149                                         continue;
 150                                 }
 151
 152                                 if (token == '<') {
 153                                         GetTag (out tagtype, out id, out attributes);
 154                                         EndElement ();
 155                                         if (tagtype == TagType.ServerComment)
 156                                                 continue;
 157
 158                                         if (tagtype == TagType.Text)
 159                                                 OnTextParsed (id);
 160                                         else
 161                                                 OnTagParsed (tagtype, id, attributes);
 162
 163                                         continue;
 164                                 }
 165
 166                                 if (tokenizer.Value.Trim () == "" && tagtype == TagType.Directive) {
 167                                         continue;
 168                                 }
 169
 170                                 text.Length = 0;
 171                                 do {
 172                                         text.Append (tokenizer.Value);
 173                                         token = tokenizer.get_token ();
 174                                 } while (token != '<' && token != Token.EOF);
 175
 176                                 tokenizer.put_back ();
 177                                 EndElement ();
 178                                 OnTextParsed (text.ToString ());
 179                         }
 180                 }
 181
 182                 bool GetInclude (string str, out string pathType, out string filename)
 183                 {
 184                         pathType = null;
 185                         filename = null;
 186                         str = str.Substring (2).Trim ();
 187                         int len = str.Length;
 188                         int lastQuote = str.LastIndexOf ('"');
 189                         if (len < 10 || lastQuote != len - 1)
 190                                 return false;
 191
 192                         if (!StrUtils.StartsWith (str, "#include ", true))
 193                                 return false;
 194
 195                         str = str.Substring (9).Trim ();
 196                         bool isfile = (StrUtils.StartsWith (str ,"file", true));
 197                         if (!isfile && !StrUtils.StartsWith (str, "virtual", true))
 198                                 return false;
 199
 200                         pathType = (isfile) ? "file" : "virtual";
 201                         if (str.Length < pathType.Length + 3)
 202                                 return false;
 203
 204                         str = str.Substring (pathType.Length).Trim ();
 205                         if (str.Length < 3 || str [0] != '=')
 206                                 return false;
 207
 208                         int index = 1;
 209                         for (; index < str.Length; index++) {
 210                                 if (Char.IsWhiteSpace (str [index]))
 211                                         continue;
 212                                 else if (str [index] == '"')
 213                                         break;
 214                         }
 215
 216                         if (index == str.Length || index == lastQuote)
 217                                 return false;
 218
 219                         str = str.Substring (index);
 220                         if (str.Length == 2) { // only quotes
 221                                 OnError ("Empty file name.");
 222                                 return false;
 223                         }
 224
 225                         filename = str.Trim ().Substring (index, str.Length - 2);
 226                         if (filename.LastIndexOf  ('"') != -1)
 227                                 return false; // file=""" -> no error
 228
 229                         return true;
 230                 }
 231
 232                 void GetTag (out TagType tagtype, out string id, out TagAttributes attributes)
 233                 {
 234                         int token = tokenizer.get_token ();
 235
 236                         tagtype = TagType.ServerComment;
 237                         id = null;
 238                         attributes = null;
 239                         switch (token){
 240                         case '%':
 241                                 GetServerTag (out tagtype, out id, out attributes);
 242                                 break;
 243                         case '/':
 244                                 if (!Eat (Token.IDENTIFIER))
 245                                         OnError ("expecting TAGNAME");
 246
 247                                 id = tokenizer.Value;
 248                                 if (!Eat ('>'))
 249                                         OnError ("expecting '>'. Got '" + id + "'");
 250
 251                                 tagtype = TagType.Close;
 252                                 break;
 253                         case '!':
 254                                 bool double_dash = Eat (Token.DOUBLEDASH);
 255                                 if (double_dash)
 256                                         tokenizer.put_back ();
 257
 258                                 tokenizer.Verbatim = true;
 259                                 string end = double_dash ? "-->" : ">";
 260                                 string comment = GetVerbatim (tokenizer.get_token (), end);
 261                                 tokenizer.Verbatim = false;
 262                                 if (comment == null)
 263                                         OnError ("Unfinished HTML comment/DTD");
 264
 265                                 string pathType, filename;
 266                                 if (double_dash && GetInclude (comment, out pathType, out filename)) {
 267                                         tagtype = TagType.Include;
 268                                         attributes = new TagAttributes ();
 269                                         attributes.Add (pathType, filename);
 270                                 } else {
 271                                         tagtype = TagType.Text;
 272                                         id = "<!" + comment + end;
 273                                 }
 274                                 break;
 275                         case Token.IDENTIFIER:
 276                                 if (this.filename == "@@inner_string@@") {
 277                                         // Actually not tag but "xxx < yyy" stuff in inner_string!
 278                                         tagtype = TagType.Text;
 279                                         tokenizer.InTag = false;
 280                                         id = "<" + tokenizer.Odds + tokenizer.Value;
 281                                 } else {
 282                                         id = tokenizer.Value;
 283                                         try {
 284                                                 attributes = GetAttributes ();
 285                                         } catch (Exception e) {
 286                                                 OnError (e.Message);
 287                                                 break;
 288                                         }
 289
 290                                         tagtype = TagType.Tag;
 291                                         if (Eat ('/') && Eat ('>')) {
 292                                                 tagtype = TagType.SelfClosing;
 293                                         } else if (!Eat ('>')) {
 294                                                 if (attributes.IsRunAtServer ()) {
 295                                                         OnError ("The server tag is not well formed.");
 296                                                         break;
 297                                                 }
 298                                                 tokenizer.Verbatim = true;
 299                                                 attributes.Add ("", GetVerbatim (tokenizer.get_token (), ">") + ">");
 300                                                 tokenizer.Verbatim = false;
 301                                         }
 302                                 }
 303
 304                                 break;
 305                         default:
 306                                 tagtype = TagType.Text;
 307                                 tokenizer.InTag = false;
 308                                 id = "<" + tokenizer.Value;
 309                                 break;
 310                         }
 311                 }
 312
 313                 TagAttributes GetAttributes ()
 314                 {
 315                         int token;
 316                         TagAttributes attributes;
 317                         string id;
 318                         bool wellFormedForServer = true;
 319
 320                         attributes = new TagAttributes ();
 321                         while ((token = tokenizer.get_token ()) != Token.EOF){
 322                                 if (token == '<' && Eat ('%')) {
 323                                         tokenizer.Verbatim = true;
 324                                         attributes.Add ("", "<%" +
 325                                                         GetVerbatim (tokenizer.get_token (), "%>") + "%>");
 326                                         tokenizer.Verbatim = false;
 327                                         tokenizer.InTag = true;
 328                                         continue;
 329                                 }
 330
 331                                 if (token != Token.IDENTIFIER)
 332                                         break;
 333
 334                                 id = tokenizer.Value;
 335                                 if (Eat ('=')){
 336                                         if (Eat (Token.ATTVALUE)){
 337                                                 attributes.Add (id, tokenizer.Value);
 338                                                 wellFormedForServer &= tokenizer.AlternatingQuotes;
 339                                         } else if (Eat ('<') && Eat ('%')) {
 340                                                 tokenizer.Verbatim = true;
 341                                                 attributes.Add (id, "<%" +
 342                                                                 GetVerbatim (tokenizer.get_token (), "%>") + "%>");
 343                                                 tokenizer.Verbatim = false;
 344                                                 tokenizer.InTag = true;
 345                                         } else {
 346                                                 OnError ("expected ATTVALUE");
 347                                                 return null;
 348                                         }
 349                                 } else {
 350                                         attributes.Add (id, null);
 351                                 }
 352                         }
 353
 354                         tokenizer.put_back ();
 355
 356                         if (attributes.IsRunAtServer () && !wellFormedForServer) {
 357                                 OnError ("The server tag is not well formed.");
 358                                 return null;
 359                         }
 360
 361                         return attributes;
 362                 }
 363
 364                 string GetVerbatim (int token, string end)
 365                 {
 366                         StringBuilder vb_text = new StringBuilder ();
 367                         int i = 0;
 368
 369                         if (tokenizer.Value.Length > 1){
 370                                 // May be we have a put_back token that is not a single character
 371                                 vb_text.Append (tokenizer.Value);
 372                                 token = tokenizer.get_token ();
 373                         }
 374
 375                         end = end.ToLower (CultureInfo.InvariantCulture);
 376                         while (token != Token.EOF){
 377                                 if (Char.ToLower ((char) token, CultureInfo.InvariantCulture) == end [i]){
 378                                         if (++i >= end.Length)
 379                                                 break;
 380                                         token = tokenizer.get_token ();
 381                                         continue;
 382                                 } else if (i > 0) {
 383                                         for (int j = 0; j < i; j++)
 384                                                 vb_text.Append (end [j]);
 385                                         i = 0;
 386                                 }
 387
 388                                 vb_text.Append ((char) token);
 389                                 token = tokenizer.get_token ();
 390                         }
 391
 392                         if (token == Token.EOF)
 393                                 OnError ("Expecting " + end + " and got EOF.");
 394
 395                         return RemoveComments (vb_text.ToString ());
 396                 }
 397
 398                 string RemoveComments (string text)
 399                 {
 400                         int end;
 401                         int start = text.IndexOf ("<%--");
 402
 403                         while (start != -1) {
 404                                 end = text.IndexOf ("--%>");
 405                                 if (end == -1 || end <= start + 1)
 406                                         break;
 407
 408                                 text = text.Remove (start, end - start + 4);
 409                                 start = text.IndexOf ("<%--");
 410                         }
 411
 412                         return text;
 413                 }
 414
 415                 void GetServerTag (out TagType tagtype, out string id, out TagAttributes attributes)
 416                 {
 417                         string inside_tags;
 418                         bool old = tokenizer.ExpectAttrValue;
 419
 420                         tokenizer.ExpectAttrValue = false;
 421                         if (Eat ('@')){
 422                                 tokenizer.ExpectAttrValue = old;
 423                                 tagtype = TagType.Directive;
 424                                 id = "";
 425                                 if (Eat (Token.DIRECTIVE))
 426                                         id = tokenizer.Value;
 427
 428                                 attributes = GetAttributes ();
 429                                 if (!Eat ('%') || !Eat ('>'))
 430                                         OnError ("expecting '%>'");
 431
 432                                 return;
 433                         }
 434
 435                         if (Eat (Token.DOUBLEDASH)) {
 436                                 tokenizer.ExpectAttrValue = old;
 437                                 tokenizer.Verbatim = true;
 438                                 inside_tags = GetVerbatim (tokenizer.get_token (), "--%>");
 439                                 tokenizer.Verbatim = false;
 440                                 id = null;
 441                                 attributes = null;
 442                                 tagtype = TagType.ServerComment;
 443                                 return;
 444                         }
 445
 446                         tokenizer.ExpectAttrValue = old;
 447                         bool varname;
 448                         bool databinding;
 449                         varname = Eat ('=');
 450                         databinding = !varname && Eat ('#');
 451
 452                         tokenizer.Verbatim = true;
 453                         inside_tags = GetVerbatim (tokenizer.get_token (), "%>");
 454                         tokenizer.Verbatim = false;
 455                         id = inside_tags;
 456                         attributes = null;
 457                         tagtype = (databinding ? TagType.DataBinding :
 458                                   (varname ? TagType.CodeRenderExpression : TagType.CodeRender));
 459                 }
 460
 461                 public event ParseErrorHandler Error;
 462                 public event TagParsedHandler TagParsed;
 463                 public event TextParsedHandler TextParsed;
 464
 465                 void OnError (string msg)
 466                 {
 467                         if (Error != null)
 468                                 Error (this, msg);
 469                 }
 470
 471                 void OnTagParsed (TagType tagtype, string id, TagAttributes attributes)
 472                 {
 473                         if (TagParsed != null)
 474                                 TagParsed (this, tagtype, id, attributes);
 475                 }
 476
 477                 void OnTextParsed (string text)
 478                 {
 479                         if (TextParsed != null)
 480                                 TextParsed (this, text);
 481                 }
 482         }
 483
 484 }
 485