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 using System;
  10 using System.Collections;
  11 using System.IO;
  12 using System.Text;
  13
  14 namespace System.Web.Compilation
  15 {
  16         delegate void ParseErrorHandler (ILocation location, string message);
  17         delegate void TextParsedHandler (ILocation location, string text);
  18         delegate void TagParsedHandler (ILocation location, TagType tagtype, string id, TagAttributes attributes);
  19
  20         class AspParser : ILocation
  21         {
  22                 AspTokenizer tokenizer;
  23                 int beginLine, endLine;
  24                 int beginColumn, endColumn;
  25                 int beginPosition, endPosition;
  26                 string filename;
  27                 string fileText;
  28                 string verbatimID;
  29
  30                 public AspParser (string filename, TextReader input)
  31                 {
  32                         this.filename = filename;
  33                         fileText = input.ReadToEnd ();
  34                         StringReader reader = new StringReader (fileText);
  35                         tokenizer = new AspTokenizer (reader);
  36                 }
  37
  38                 public int BeginLine {
  39                         get { return beginLine; }
  40                 }
  41
  42                 public int BeginColumn {
  43                         get { return beginColumn; }
  44                 }
  45
  46                 public int EndLine {
  47                         get { return endLine; }
  48                 }
  49
  50                 public int EndColumn {
  51                         get { return endColumn; }
  52                 }
  53
  54                 public string PlainText {
  55                         get {
  56                                 if (beginPosition >= endPosition)
  57                                         return null;
  58
  59                                 return fileText.Substring (beginPosition, endPosition - beginPosition);
  60                         }
  61                 }
  62
  63                 public string Filename {
  64                         get { return filename; }
  65                 }
  66
  67                 public string VerbatimID {
  68                         set {
  69                                 tokenizer.Verbatim = true;
  70                                 verbatimID = value.ToUpper ();
  71                         }
  72                 }
  73
  74                 bool Eat (int expected_token)
  75                 {
  76                         if (tokenizer.get_token () != expected_token) {
  77                                 tokenizer.put_back ();
  78                                 return false;
  79                         }
  80
  81                         endLine = tokenizer.EndLine;
  82                         endColumn = tokenizer.EndColumn;
  83                         return true;
  84                 }
  85
  86                 void BeginElement ()
  87                 {
  88                         beginLine = tokenizer.BeginLine;
  89                         beginColumn = tokenizer.BeginColumn;
  90                         beginPosition = tokenizer.Position - 1;
  91                 }
  92
  93                 void EndElement ()
  94                 {
  95                         endLine = tokenizer.EndLine;
  96                         endColumn = tokenizer.EndColumn;
  97                         endPosition = tokenizer.Position;
  98                 }
  99
 100                 public void Parse ()
 101                 {
 102                         int token;
 103                         string id;
 104                         TagAttributes attributes;
 105                         TagType tagtype;
 106                         StringBuilder text =  new StringBuilder ();
 107
 108                         while ((token = tokenizer.get_token ()) != Token.EOF) {
 109                                 BeginElement ();
 110
 111                                 if (tokenizer.Verbatim){
 112                                         string end_verbatim = "</" + verbatimID + ">";
 113                                         string verbatim_text = GetVerbatim (token, end_verbatim);
 114
 115                                         if (verbatim_text == null)
 116                                                 OnError ("Unexpected EOF processing " + verbatimID);
 117
 118                                         tokenizer.Verbatim = false;
 119
 120                                         EndElement ();
 121                                         endPosition -= end_verbatim.Length;
 122                                         OnTextParsed (verbatim_text);
 123                                         beginPosition = endPosition;
 124                                         endPosition += end_verbatim.Length;
 125                                         OnTagParsed (TagType.Close, verbatimID, null);
 126                                         continue;
 127                                 }
 128
 129                                 if (token == '<') {
 130                                         GetTag (out tagtype, out id, out attributes);
 131                                         EndElement ();
 132                                         if (tagtype == TagType.ServerComment)
 133                                                 continue;
 134
 135                                         if (tagtype == TagType.Text)
 136                                                 OnTextParsed (id);
 137                                         else
 138                                                 OnTagParsed (tagtype, id, attributes);
 139
 140                                         continue;
 141                                 }
 142
 143                                 text.Length = 0;
 144                                 do {
 145                                         text.Append (tokenizer.Value);
 146                                         token = tokenizer.get_token ();
 147                                 } while (token != '<' && token != Token.EOF);
 148
 149                                 tokenizer.put_back ();
 150                                 EndElement ();
 151                                 OnTextParsed (text.ToString ());
 152                         }
 153                 }
 154
 155                 bool GetInclude (string str, out string pathType, out string filename)
 156                 {
 157                         pathType = null;
 158                         filename = null;
 159                         str = str.Substring (2).Trim ();
 160                         int len = str.Length;
 161                         int lastQuote = str.LastIndexOf ('"');
 162                         if (len < 10 || lastQuote != len - 1 || !str.StartsWith ("#include "))
 163                                 return false;
 164
 165                         str = str.Substring (9).Trim ();
 166                         bool isfile = (str.StartsWith ("file"));
 167                         if (!isfile && !str.StartsWith ("virtual"))
 168                                 return false;
 169
 170                         pathType = (isfile) ? "file" : "virtual";
 171                         if (str.Length < pathType.Length + 3)
 172                                 return false;
 173
 174                         str = str.Substring (pathType.Length).Trim ();
 175                         if (str.Length < 3 || str [0] != '=')
 176                                 return false;
 177
 178                         int index = 1;
 179                         for (; index < str.Length; index++) {
 180                                 if (Char.IsWhiteSpace (str [index]))
 181                                         index++;
 182                                 else if (str [index] == '"')
 183                                         break;
 184                         }
 185
 186                         if (index == str.Length || index == lastQuote)
 187                                 return false;
 188
 189                         str = str.Substring (index);
 190                         if (str.Length == 2) { // only quotes
 191                                 OnError ("Empty file name.");
 192                                 return false;
 193                         }
 194
 195                         filename = str.Trim ().Substring (index, str.Length - 2);
 196                         if (filename.LastIndexOf  ('"') != -1)
 197                                 return false; // file=""" -> no error
 198
 199                         return true;
 200                 }
 201
 202                 void GetTag (out TagType tagtype, out string id, out TagAttributes attributes)
 203                 {
 204                         int token = tokenizer.get_token ();
 205
 206                         tagtype = TagType.ServerComment;
 207                         id = null;
 208                         attributes = null;
 209                         switch (token){
 210                         case '%':
 211                                 GetServerTag (out tagtype, out id, out attributes);
 212                                 break;
 213                         case '/':
 214                                 if (!Eat (Token.IDENTIFIER))
 215                                         OnError ("expecting TAGNAME");
 216
 217                                 id = tokenizer.Value;
 218                                 if (!Eat ('>'))
 219                                         OnError ("expecting '>'. Got '" + id + "'");
 220
 221                                 tagtype = TagType.Close;
 222                                 break;
 223                         case '!':
 224                                 bool double_dash = Eat (Token.DOUBLEDASH);
 225                                 if (double_dash)
 226                                         tokenizer.put_back ();
 227
 228                                 tokenizer.Verbatim = true;
 229                                 string end = double_dash ? "-->" : ">";
 230                                 string comment = GetVerbatim (tokenizer.get_token (), end);
 231                                 tokenizer.Verbatim = false;
 232                                 if (comment == null)
 233                                         OnError ("Unfinished HTML comment/DTD");
 234
 235                                 string pathType, filename;
 236                                 if (double_dash && GetInclude (comment, out pathType, out filename)) {
 237                                         tagtype = TagType.Include;
 238                                         attributes = new TagAttributes ();
 239                                         attributes.Add (pathType, filename);
 240                                 } else {
 241                                         tagtype = TagType.Text;
 242                                         id = "<!" + comment + end;
 243                                 }
 244                                 break;
 245                         case Token.IDENTIFIER:
 246                                 if (this.filename == "@@inner_string@@") {
 247                                         // Actually not tag but "xxx < yyy" stuff in inner_string!
 248                                         tagtype = TagType.Text;
 249                                         tokenizer.InTag = false;
 250                                         id = "<" + tokenizer.Odds + tokenizer.Value;
 251                                 } else {
 252                                 id = tokenizer.Value;
 253                                 try {
 254                                         attributes = GetAttributes ();
 255                                 } catch (Exception e) {
 256                                         OnError (e.Message);
 257                                         break;
 258                                 }
 259
 260                                 tagtype = TagType.Tag;
 261                                 if (Eat ('/') && Eat ('>'))
 262                                         tagtype = TagType.SelfClosing;
 263                                 else if (!Eat ('>'))
 264                                         OnError ("expecting '>'. Got '" + tokenizer.Value + "'");
 265                                 }
 266
 267                                 break;
 268                         default:
 269                                 tagtype = TagType.Text;
 270                                 tokenizer.InTag = false;
 271                                 id = "<" + tokenizer.Value;
 272                                 break;
 273                         }
 274                 }
 275
 276                 TagAttributes GetAttributes ()
 277                 {
 278                         int token;
 279                         TagAttributes attributes;
 280                         string id;
 281
 282                         attributes = new TagAttributes ();
 283                         while ((token = tokenizer.get_token ()) != Token.EOF){
 284                                 if (token != Token.IDENTIFIER)
 285                                         break;
 286                                 id = tokenizer.Value;
 287                                 if (Eat ('=')){
 288                                         if (Eat (Token.ATTVALUE)){
 289                                                 attributes.Add (id, tokenizer.Value);
 290                                         } else if (Eat ('<') && Eat ('%')) {
 291                                                 attributes.Add (id, "<%" +
 292                                                                 GetVerbatim (tokenizer.get_token (), "%>"));
 293                                         } else {
 294                                                 OnError ("expected ATTVALUE");
 295                                                 return null;
 296                                         }
 297
 298                                 } else {
 299                                         attributes.Add (id, null);
 300                                 }
 301                         }
 302
 303                         tokenizer.put_back ();
 304                         return attributes;
 305                 }
 306
 307                 string GetVerbatim (int token, string end)
 308                 {
 309                         StringBuilder vb_text = new StringBuilder ();
 310                         int i = 0;
 311
 312                         if (tokenizer.Value.Length > 1){
 313                                 // May be we have a put_back token that is not a single character
 314                                 vb_text.Append (tokenizer.Value);
 315                                 token = tokenizer.get_token ();
 316                         }
 317
 318                         while (token != Token.EOF){
 319                                 if (Char.ToUpper ((char) token) == end [i]){
 320                                         if (++i >= end.Length)
 321                                                 break;
 322                                         token = tokenizer.get_token ();
 323                                         continue;
 324                                 } else if (i > 0) {
 325                                         for (int j = 0; j < i; j++)
 326                                                 vb_text.Append (end [j]);
 327                                         i = 0;
 328                                 }
 329
 330                                 vb_text.Append ((char) token);
 331                                 token = tokenizer.get_token ();
 332                         }
 333
 334                         return RemoveComments (vb_text.ToString ());
 335                 }
 336
 337                 string RemoveComments (string text)
 338                 {
 339                         int end;
 340                         int start = text.IndexOf ("<%--");
 341
 342                         while (start != -1) {
 343                                 end = text.IndexOf ("--%>");
 344                                 if (end == -1 || end <= start + 1)
 345                                         break;
 346
 347                                 text = text.Remove (start, end - start + 4);
 348                                 start = text.IndexOf ("<%--");
 349                         }
 350
 351                         return text;
 352                 }
 353
 354                 void GetServerTag (out TagType tagtype, out string id, out TagAttributes attributes)
 355                 {
 356                         string inside_tags;
 357
 358                         if (Eat ('@')){
 359                                 tagtype = TagType.Directive;
 360                                 id = "";
 361                                 if (Eat (Token.DIRECTIVE))
 362                                         id = tokenizer.Value;
 363
 364                                 attributes = GetAttributes ();
 365                                 if (!Eat ('%') || !Eat ('>'))
 366                                         OnError ("expecting '%>'");
 367
 368                                 return;
 369                         }
 370
 371                         if (Eat (Token.DOUBLEDASH)) {
 372                                 tokenizer.Verbatim = true;
 373                                 inside_tags = GetVerbatim (tokenizer.get_token (), "--%>");
 374                                 tokenizer.Verbatim = false;
 375                                 id = null;
 376                                 attributes = null;
 377                                 tagtype = TagType.ServerComment;
 378                                 return;
 379                         }
 380
 381                         bool varname;
 382                         bool databinding;
 383                         varname = Eat ('=');
 384                         databinding = !varname && Eat ('#');
 385
 386                         tokenizer.Verbatim = true;
 387                         inside_tags = GetVerbatim (tokenizer.get_token (), "%>");
 388                         tokenizer.Verbatim = false;
 389                         id = inside_tags;
 390                         attributes = null;
 391                         tagtype = (databinding ? TagType.DataBinding :
 392                                   (varname ? TagType.CodeRenderExpression : TagType.CodeRender));
 393                 }
 394
 395                 public event ParseErrorHandler Error;
 396                 public event TagParsedHandler TagParsed;
 397                 public event TextParsedHandler TextParsed;
 398
 399                 void OnError (string msg)
 400                 {
 401                         if (Error != null)
 402                                 Error (this, msg);
 403                 }
 404
 405                 void OnTagParsed (TagType tagtype, string id, TagAttributes attributes)
 406                 {
 407                         if (TagParsed != null)
 408                                 TagParsed (this, tagtype, id, attributes);
 409                 }
 410
 411                 void OnTextParsed (string text)
 412                 {
 413                         if (TextParsed != null)
 414                                 TextParsed (this, text);
 415                 }
 416         }
 417
 418 }
 419