0
votes

I'm trying to write a programming language using flex/bison in C. However, it reports 117 shift/reduce conflicts in my grammar. Of these, 44 are from the expression grammar. I don't understand why, since I have specified the precedence.

I have already tried adding precedence rules in my bison grammar file:

%token INTEGER DOUBLE IDENTIFIER STRING RUNE TUPLEID
%token BEGIN END WHILE IF CLASS METHOD

%nonassoc BLOCKSTMT
%nonassoc ELSE
%nonassoc IFX
%right '=' INC DEC ASSIGN INCN DECN MULN DIVN MODN
%nonassoc ASSIGNBLOCK
%right STOP
%left GE LE EQ NE
%left '+' '-'
%left '*' '/' '%'
%nonassoc UMINUS

This did not fix the problem. I have looked here on Stack Overflow for answers, but I find none that address the type of grammar I have written (which is pretty much just expanded on from a yacc tutorial).

expr:
    INTEGER {
        node_t *newInt = initNode(line, LEAFINT);
        newInt->leaf.Integer = $1;
        $$ = newInt;
    }
    | DOUBLE {
        node_t *newDouble = initNode(line, LEAFDOUBLE);
        newDouble->leaf.Double = $1;
        $$ = newDouble;
    }
    | id { $$ = $1; }
    | STRING {
        node_t *newStr = initNode(line, LEAFSTRING);
        newStr->leaf.String = $1;
        $$ = newStr;
    }
    | RUNE {
        node_t *newRune = initNode(line, LEAFRUNE);
        if (strcmp($1, "\\n") == 0) {
            newRune->leaf.Rune = '\n';
        } else if (strcmp($1, "\\t") == 0) {
            newRune->leaf.Rune = '\t';
        } else if (strcmp($1, "\\\\") == 0) {
            newRune->leaf.Rune = '\\';
        } else if (strcmp($1, "\\'") == 0) {
            newRune->leaf.Rune = '\'';
        } else if (strlen($1) < 2) {
            newRune->leaf.Rune = *$1;
        } else {
            yyerror("Error: Invalid rune literal.", line);
        }
        $$ = newRune;
    }
    | list { $$ = $1; }
    | '-' expr %prec UMINUS {
        node_t *negative = initNode(line, OPERATOR);
        negative->operation = "-";
        addOperand(negative, $2);
        $$ = negative;
    }
    | expr '=' expr { $$ = binOpr(line, "=", $1, $3); }
    | expr '+' expr { $$ = binOpr(line, "+", $1, $3); }
    | expr '-' expr { $$ = binOpr(line, "-", $1, $3); }
    | expr '*' expr { $$ = binOpr(line, "*", $1, $3); }
    | expr '/' expr { $$ = binOpr(line, "/", $1, $3); }
    | expr '%' expr { $$ = binOpr(line, "%", $1, $3); }
    | expr EQ expr { $$ = binOpr(line, "==", $1, $3); }
    | expr NE expr { $$ = binOpr(line, "!=", $1, $3); }
    | expr '<' expr { $$ = binOpr(line, "<", $1, $3); }
    | expr '>' expr { $$ = binOpr(line, ">", $1, $3); }
    | expr LE expr { $$ = binOpr(line, "<=", $1, $3); }
    | expr GE expr { $$ = binOpr(line, ">=", $1, $3); }
    | expr INC {
        node_t *increment = initNode(line, OPERATOR);
        increment->operation = "++";
        addOperand(increment, $1);
        $$ = increment;
    }
    | INC expr {
        node_t *increment = initNode(line, OPERATOR);
        increment->operation = "++";
        addOperand(increment, $2);
        $$ = increment;
    }
    | expr DEC {
        node_t *decrement = initNode(line, OPERATOR);
        decrement->operation = "--";
        addOperand(decrement, $1);
        $$ = decrement;
    }
    | DEC expr {
        node_t *decrement = initNode(line, OPERATOR);
        decrement->operation = "--";
        addOperand(decrement, $2);
        $$ = decrement;
    }
    | expr INCN expr { $$ = binOpr(line, "+=", $1, $3); }
    | expr DECN expr { $$ = binOpr(line, "-=", $1, $3); }
    | expr MULN expr { $$ = binOpr(line, "*=", $1, $3); }
    | expr DIVN expr { $$ = binOpr(line, "/=", $1, $3); }
    | expr MODN expr { $$ = binOpr(line, "%=", $1, $3); }
    | '(' expr ')' { $$ = $2 }
    | expr STOP IDENTIFIER {
        node_t *newID = initNode(line, LEAFID);
        newID->leaf.Identifier = $3;
        $$ = binOpr(line, ".", $1, newID);
    }
    | expr '[' expr ']' %prec STOP { $$ = binOpr(line, "index", $1, $3); }
    | expr '(' expr ')' {
        node_t *newList = initNode(line, LIST);
        addOperand(newList, $3);
        node_t *newCall = initNode(line, CALL);
        addOperand(newCall, $1);
        addOperand(newCall, newList);
        $$ = newCall;
    }
    | expr parenlist {
        node_t *newCall = initNode(line, CALL);
        addOperand(newCall, $1);
        addOperand(newCall, $2);
        $$ = newCall;
    }

I expected it to compile my compiler normally, but instead, when I do "bison -d -v parse.y" it gives me the warning "parse.y: warning: 117 shift/reduce conflicts [-Wconflicts-sr]." The y.output file says the following:

State 122

   40 expr: expr . '=' expr
   41     | expr . '+' expr
   42     | expr . '-' expr
   43     | expr . '*' expr
   44     | expr . '/' expr
   45     | expr . '%' expr
   46     | expr . EQ expr
   47     | expr . NE expr
   48     | expr . '<' expr
   48     | expr '<' expr .
   49     | expr . '>' expr
   50     | expr . LE expr
   51     | expr . GE expr
   52     | expr . INC
   54     | expr . DEC
   56     | expr . INCN expr
   57     | expr . DECN expr
   58     | expr . MULN expr
   59     | expr . DIVN expr
   60     | expr . MODN expr
   62     | expr . STOP IDENTIFIER
   63     | expr . '[' expr ']'
   64     | expr . '(' expr ')'
   65     | expr . parenlist

    '='   shift, and go to state 60
    INC   shift, and go to state 61
    DEC   shift, and go to state 62
    INCN  shift, and go to state 63
    DECN  shift, and go to state 64
    MULN  shift, and go to state 65
    DIVN  shift, and go to state 66
    MODN  shift, and go to state 67
    STOP  shift, and go to state 68
    GE    shift, and go to state 69
    LE    shift, and go to state 70
    EQ    shift, and go to state 71
    NE    shift, and go to state 72
    '+'   shift, and go to state 73
    '-'   shift, and go to state 74
    '*'   shift, and go to state 75
    '/'   shift, and go to state 76
    '%'   shift, and go to state 77
    '('   shift, and go to state 79
    '<'   shift, and go to state 80
    '>'   shift, and go to state 81
    '['   shift, and go to state 82

    '='       [reduce using rule 48 (expr)]
    INC       [reduce using rule 48 (expr)]
    DEC       [reduce using rule 48 (expr)]
    INCN      [reduce using rule 48 (expr)]
    DECN      [reduce using rule 48 (expr)]
    MULN      [reduce using rule 48 (expr)]
    DIVN      [reduce using rule 48 (expr)]
    MODN      [reduce using rule 48 (expr)]
    STOP      [reduce using rule 48 (expr)]
    GE        [reduce using rule 48 (expr)]
    LE        [reduce using rule 48 (expr)]
    EQ        [reduce using rule 48 (expr)]
    NE        [reduce using rule 48 (expr)]
    '+'       [reduce using rule 48 (expr)]
    '-'       [reduce using rule 48 (expr)]
    '*'       [reduce using rule 48 (expr)]
    '/'       [reduce using rule 48 (expr)]
    '%'       [reduce using rule 48 (expr)]
    '('       [reduce using rule 48 (expr)]
    '<'       [reduce using rule 48 (expr)]
    '>'       [reduce using rule 48 (expr)]
    '['       [reduce using rule 48 (expr)]
    $default  reduce using rule 48 (expr)

    parenlist  go to state 83
    arglist    go to state 84

The next state, State 123, appears to be identical. State 124 is also identical but without the conflicts.

1

1 Answers

0
votes

The token '<' has no precedence, so rule 48 has no precedence, and all expressions of the form "E < E op E" are ambiguous, leading to the shift/reduce conflicts you see in this state. Add a precedence for '<' The other states with conflicts are probably similar for other rules.