%option noyywrap yylineno

%{
/* -*- mode: c -*- */
#include <stdio.h>
#include <string.h>
#include "macro.h"
#include "AST.h"

#include "y.tab.h"

extern YYSTYPE yylval;
extern YYLTYPE yylloc;
extern int debug;

int charnum = 0;
int linenum = 0;

#define YY_USER_ACTION do { yylloc.first_line = yylloc.last_line = yylineno; yylloc.first_column = charnum; } while (0);
%}


ID       [_a-z]+
NUM      [0-9]+
INVALID  {ID}{NUM}|{NUM}{ID}
COMMENTL \[
COMMENTR \]

OPERATOR  "+"|"-"|"*"|"/"|"%"
COND      "="|"!="|"<"|">"|"<="|">="
CONDEXP   IF|THEN|ELSE|ENDIF
LOOP      WLOOP|FLOOP
WLOOP     WHILE|DO|ENDWHILE
FLOOP     FOR|DOWN|FROM|TO|ENDFOR
IO        GET|PUT
ASSIGN    ":="
DECLARE   DECLARE
IN        IN
END       END
SEMICOLON ";"
PAR       "("|")"
KEYWORD   {OPERATOR}|{COND}|{CONDEXP}|{LOOP}|{IO}|{ASSIGN}|{DECLARE}|{IN}|{END}|{SEMICOLON}|{PAR}|{INVALID}

%x COMMENT
%x ERROR

%%

{KEYWORD} {
        charnum += (strlen(yytext));
	debug("KEYWORD %s", yytext);
        //yylval.string = (char *)strdup(yytext);
        //return yytext;
        REJECT;
}

"+" { return PLUS; }
"-" { return MINUS; }
"*" { return TIMES; }
"/" { return DIV; }
"%" { return MOD; }

"=" { return EQ; }
"!=" { return DIFF; }
"<" { return LE; }
">" { return GE; }
"<=" { return LEQ; }
">=" { return GEQ; }

IF { return IF; }
THEN { return THEN; }
ELSE { return ELSE; }
ENDIF { return ENDIF; }

WHILE { return WHILE; }
DO { return DO; }
ENDWHILE { return ENDWHILE; }

FOR { return FOR; }
DOWN { return DOWN; }
FROM { return FROM; }
TO { return TO; }
DO { return DO; }
ENDFOR { return ENDFOR; }

GET { return GET; }
PUT { return PUT; }

:= { return ASSIGN; }

DECLARE { return DECLARE; }
IN { return IN; }
END { return END; }

\( { return PARL; }
\) { return PARR; }

{SEMICOLON} { return SEMICOLON; }

{INVALID} {
        error("%d:%d: unrecognized token \"%s\"", linenum, charnum, yytext);
        exit(1);
}

{ID} {
        debug("ID %s", yytext);
        charnum += (strlen(yytext));

        yylval.ast_node = new_stringnode('I', strdup(yytext)); // IDENTIFIER
        return PIDENTIFIER;
}

{NUM} {
        debug("INTEGER %s", yytext);
        charnum += (strlen(yytext));
        //yylval = atoi(yytext);

        yylval.ast_node = new_stringnode('N', strdup(yytext)); // NUM
        return NUM;
}


{COMMENTL} {
        BEGIN(COMMENT);
        charnum += (strlen(yytext));
}

<COMMENT>\n {
        /* eat multiline comments */
        charnum = 0;
        ++linenum;
}

<COMMENT>{COMMENTR} {
        BEGIN(INITIAL);
        charnum += (strlen(yytext));
}

<COMMENT>. {
        /* eat multiline comments */
        charnum += (strlen(yytext));
}

<COMMENT>{COMMENTL} {
        error("%d:%d: illegal nested comment",
              linenum, charnum);
        charnum += (strlen(yytext));
        exit(1);
}

<COMMENT><<EOF>> {
        error("%d:%d: unexpected EOF (in comment)",
              linenum, charnum);
        BEGIN(INITIAL);
        exit(1);
}


[ \t] {
        /* eat whitespaces */
        charnum += (strlen(yytext));
}

\n {
        /* eat whitespaces */
        charnum = 0;
        ++linenum;
}


. {
        BEGIN(ERROR);
        yymore();
}

<ERROR>[ \t] {
        error("%d:%d: unrecognized token \"%s\"",
              linenum, charnum, yytext);
        BEGIN(INITIAL);
        charnum += (strlen(yytext));
        exit(1);
}

<ERROR>\n {
        error("%d:%d: unrecognized token \"%s\"",
              linenum, charnum, yytext);
        BEGIN(INITIAL);
        charnum = 0;
        ++linenum;
        exit(1);
}

<ERROR><<EOF>> {
        error("%d:%d: unrecognized token \"%s\"",
              linenum, charnum);
        exit(1);
}

<ERROR>. {
        yymore();
}
%%