Question

hi im confusing about how to get a char* when i read a specific token…
I look in various sites and they provide suggestions but not complete, i mean, for example yylval and yytext declaration is missing or how to transform the types, etc

What is need in .l file?
What is need in .y file?

What I have

in the .l file:

{WORD}      { yylval = strdup(yytext);return T_ValidWord;}

in the .y file:

%union{
  char *str;
}

%token<str> T_ValidWord

%%

element:
T_OpenTag T_ValidWord ele1 {printf("%s", $2);}
; 

The error:

xml.lex: In function ‘yylex’:
xml.lex:34: error: incompatible types when assigning to type ‘YYSTYPE’ from type ‘char *’

Other thing more that confused me:
In some places i see

yylval->something = yytext
yylval.something = yytext
yylval = yytext

In the manual of bison tell that yylval is a macro, I understand that a macro is text replaced for other text, but in this situation i really don’t get it.

Answer 1

yylval is a union type YYSTYPE. Change your assignment in the scanner to yylval.str = strdup(yytext).

yylval is a union which you can either declare or the bison will declare automatically. Bison’s default yylval is essentially useless. You declare yylval using %union { ... } in your parser because you may need to return other values from the scanner to the parser. For example, when you match a number in your scanner, you’ll want to return a token like T_NUM. But you most probably also want the value of that number, which is where yylval comes handy. Using yylval, if you have an integer field, you could simply do yylval.num = atoi(yytext) inside your scanner, and then use that num field in the parser.

yytext is an array of characters, which acts as a buffer for the input currently being parsed. You cannot declare yytext, and neither should you.

When you use bison to compile the parser into a .tab.c file, use the flags bison -d -t for debugging symbols and the header file. The header file will be called *.tab.h. Include that in your scanner so you only need to declare token names once and have proper use of yylval.