AspTokenizer.cs 4.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225
  1. //
  2. // System.Web.Compilation.AspTokenizer
  3. //
  4. // Authors:
  5. // Gonzalo Paniagua Javier ([email protected])
  6. //
  7. // (C) 2002 Ximian, Inc (http://www.ximian.com)
  8. //
  9. using System;
  10. using System.Collections;
  11. using System.IO;
  12. using System.Text;
  13. namespace System.Web.Compilation {
  14. class Token
  15. {
  16. public const int EOF = 0;
  17. public const int IDENTIFIER = 1000;
  18. public const int DIRECTIVE = 1001;
  19. public const int ATTVALUE = 1002;
  20. public const int TEXT = 1003;
  21. public const int DOUBLEDASH = 1004;
  22. public const int CLOSING = 1005;
  23. }
  24. class AspTokenizer {
  25. private StreamReader sr;
  26. private int current_token;
  27. private StringBuilder sb;
  28. private int col, line;
  29. private bool inTag;
  30. private bool hasPutBack;
  31. private bool verbatim;
  32. private string filename;
  33. public AspTokenizer (string filename, Stream stream)
  34. {
  35. if (filename == null || stream == null)
  36. throw new ArgumentNullException ();
  37. this.sr = new StreamReader (stream);
  38. this.filename = filename;
  39. sb = new StringBuilder ();
  40. col = line = 1;
  41. hasPutBack = inTag = false;
  42. }
  43. public bool Verbatim
  44. {
  45. get { return verbatim; }
  46. set { verbatim = value; }
  47. }
  48. public void put_back ()
  49. {
  50. if (hasPutBack)
  51. throw new ApplicationException ("put_back called twice!");
  52. hasPutBack = true;
  53. }
  54. public int get_token ()
  55. {
  56. if (hasPutBack){
  57. hasPutBack = false;
  58. return current_token;
  59. }
  60. current_token = NextToken ();
  61. return current_token;
  62. }
  63. bool is_identifier_start_character (char c)
  64. {
  65. return (Char.IsLetter (c) || c == '_' );
  66. }
  67. bool is_identifier_part_character (char c)
  68. {
  69. return (Char.IsLetterOrDigit (c) || c == '_' || c == '-');
  70. }
  71. private int read_char ()
  72. {
  73. int c = sr.Read ();
  74. if (c == '\r' && sr.Peek () == '\n')
  75. c = sr.Read ();
  76. if (c == '\n'){
  77. col = 0;
  78. line++;
  79. }
  80. else if (c != -1)
  81. col++;
  82. return c;
  83. }
  84. private int NextToken ()
  85. {
  86. int c;
  87. sb.Length = 0;
  88. while ((c = read_char ()) != -1){
  89. if (verbatim){
  90. inTag = false;
  91. sb.Append ((char) c);
  92. return c;
  93. }
  94. if (inTag && (c == '"' || c == '\'')){
  95. int previous;
  96. int start = c;
  97. previous = 0;
  98. while ((c = sr.Peek ()) != -1) {
  99. if (c == start && previous != '\\'){
  100. read_char ();
  101. break;
  102. }
  103. sb.Append ((char) read_char ());
  104. previous = c;
  105. }
  106. return Token.ATTVALUE;
  107. }
  108. if (c == '<'){
  109. inTag = true;
  110. sb.Append ((char) c);
  111. return c;
  112. }
  113. if (c == '>'){
  114. inTag = false;
  115. sb.Append ((char) c);
  116. return c;
  117. }
  118. if (current_token == '<' && "%/!".IndexOf ((char) c) != -1){
  119. sb.Append ((char) c);
  120. return c;
  121. }
  122. if (inTag && current_token == '%' && "@#=".IndexOf ((char) c) != -1){
  123. sb.Append ((char) c);
  124. return c;
  125. }
  126. if (inTag && c == '-' && sr.Peek () == '-'){
  127. sb.Append ("--");
  128. read_char ();
  129. return Token.DOUBLEDASH;
  130. }
  131. if (!inTag){
  132. sb.Append ((char) c);
  133. while ((c = sr.Peek ()) != -1 && c != '<')
  134. sb.Append ((char) read_char ());
  135. return (c != -1 || sb.Length > 0) ? Token.TEXT : Token.EOF;
  136. }
  137. if (inTag && current_token == '=' && !Char.IsWhiteSpace ((char) c)){
  138. sb.Append ((char) c);
  139. while ((c = sr.Peek ()) != -1) {
  140. if (Char.IsWhiteSpace ((char) c) || c == '/' || c == '>')
  141. break;
  142. sb.Append ((char) read_char ());
  143. }
  144. return Token.ATTVALUE;
  145. }
  146. if (inTag && is_identifier_start_character ((char) c)){
  147. sb.Append ((char) c);
  148. while ((c = sr.Peek ()) != -1) {
  149. if (!is_identifier_part_character ((char) c) && c != ':')
  150. break;
  151. sb.Append ((char) read_char ());
  152. }
  153. if (current_token == '@' && Directive.IsDirectiveID (sb.ToString ()))
  154. return Token.DIRECTIVE;
  155. return Token.IDENTIFIER;
  156. }
  157. if (!Char.IsWhiteSpace ((char) c))
  158. return c;
  159. }
  160. return Token.EOF;
  161. }
  162. public string value {
  163. get { return sb.ToString (); }
  164. }
  165. public int Line {
  166. get {
  167. return line;
  168. }
  169. }
  170. public int Column {
  171. get {
  172. return col;
  173. }
  174. }
  175. public string Location {
  176. get {
  177. string msg = filename;
  178. msg += " (" + line + ", " + col + "): " + sb.ToString ();
  179. return msg;
  180. }
  181. }
  182. }
  183. }