scanner: scanner.l hush-3.1/grammar/ml/html

html_scanner

  %class html_scanner
  
  include <ml/html/scanner.h>
  
  static int delim(char c) {
          return (c=='\0'||c==' '||c=='\n'||c=='\t');
          }
  
  

  
  
  escape ("\\\"")
  ident [A-Za-z0-9]+
  stringelem ({escape}|[^\"\n])
  string (\"{stringelem}+\")
  
  path  [A-Za-z0-9\~\#\:\-\.\/\?]+
  
  value  ({path}|{string})
  attrib {ident}"="{value}
  
  attribs ([ \t]|{attrib})*
  
  simple   ("<"{ident}">")
  begintag ("<"{ident}{attribs}">")
  endtag   "</"{ident}">")
  beginpr  "<?"{ident}{attribs}">"
  endpr    "<?/"{ident}">"
  doctype  "<!doctype"[^\>]">"
  element  "<!element"[^\>]">"
  attlist  "<!attlist"(^\>]">"
  
  begincmt "<!--"
  endcmt   "-->"
  
  any ({beginpr}|{endpr}|{doctype}|{element}|{attlist})
  
  

  
  {begintag}  { 
          char tag[BUFSIZ];
          char atts[2*BUFSIZ];
          char* p = yytext+1;
          yytext[yyleng-1]='\0';
          while (p && !delim(*p)) p++;
          if (*p != '\0') {
  		*p = '\0';
                  strcpy(tag,yytext+1);
                  strcpy(atts,p+1);
          } else {
                  strcpy(tag,yytext+1);
                  strcpy(atts,"");
          }
                  
          cout << "TAG: " << tag << endl;
          cout << "ATTS: " << atts << endl;
          //VW->open_element(tag, atts );
          //cout << "SC:begin simple " << yytext;
          }
  "</"{ident}">"    {
          char* tag = yytext+2; yytext[yyleng-1]='\0';
          cout << "/TAG: " << tag << endl;
          //VW->close_element(tag);
          }
  {begincmt}  { 
          in_cmt++; //cout << "begin cmt " <, endl;
          }
  {endcmt}  { 
          in_cmt--; cout << "begin cmt " << endl; 
          }
  any     { 
          //cout << "Any " << yytext << endl;
          }
  {value}|\n|. {
          cout << "DATA: " << yytext << endl;
          //VW->new_data(yytext);
          }
  

Hush Online Technology hush@cs.vu.nl 11/03/98