view lwcc/cc-parse.c @ 583:000381ee2d5c default tip

Guard against single operand multiplication when detecting like terms This *shouldn't* happen, but it apparently does in some pathological cases so guard against a single operand multiplication to prevent a crash.
author William Astle <lost@l-w.ca>
date Mon, 04 Nov 2024 23:48:23 -0700
parents 7e8298f7bc0a
children
line wrap: on
line source

/*
lwcc/cc-parse.c

Copyright © 2019 William Astle

This file is part of LWTOOLS.

LWTOOLS is free software: you can redistribute it and/or modify it under the
terms of the GNU General Public License as published by the Free Software
Foundation, either version 3 of the License, or (at your option) any later
version.

This program is distributed in the hope that it will be useful, but WITHOUT
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
more details.

You should have received a copy of the GNU General Public License along with
this program. If not, see <http://www.gnu.org/licenses/>.
*/

#include <string.h>

#include <lw_alloc.h>
#include <lw_string.h>

#include "cpp.h"
#include "tree.h"

#define TOK_KW_IF       -1
#define TOK_KW_ELSE     -2
#define TOK_KW_WHILE    -3
#define TOK_KW_DO       -4
#define TOK_KW_FOR      -5
#define TOK_KW_VOID     -6
#define TOK_KW_INT      -7
#define TOK_KW_CHAR     -8
#define TOK_KW_SHORT    -9
#define TOK_KW_LONG     -10
#define TOK_KW_UNSIGNED -11
#define TOK_KW_SIGNED   -12
#define TOK_KW_FLOAT    -13
#define TOK_KW_DOUBLE   -14
#define TOK_KW_STRUCT   -15
#define TOK_KW_UNION    -16
#define TOK_KW_TYPEDEF  -17
#define TOK_KW_STATIC   -18
#define TOK_KW_SWITCH   -19
#define TOK_KW_CASE     -20
#define TOK_KW_DEFAULT  -21
#define TOK_KW_BREAK    -22
#define TOK_KW_CONTINUE -23
#define TOK_KW_CONST    -24
#define TOK_KW_AUTO     -25
#define TOK_KW_ENUM     -26
#define TOK_KW_REGISTER -27
#define TOK_KW_SIZEOF   -28
#define TOK_KW_VOLATILE -29
#define TOK_KW_RETURN   -30
#define TOK_KW_EXTERN   -31
#define TOK_KW_GOTO     -32
#define TOK_TYPENAME    -100
#define TOK_CONST_INT   -150

static struct { int tok; char *word; } keyword_list[] = {
    { TOK_KW_IF, "if" },
    { TOK_KW_ELSE, "else" },
    { TOK_KW_WHILE, "while" },
    { TOK_KW_DO, "do" },
    { TOK_KW_FOR, "for" },
    { TOK_KW_VOID, "void" },
    { TOK_KW_INT, "int" },
    { TOK_KW_CHAR, "char" },
    { TOK_KW_SHORT, "short" },
    { TOK_KW_LONG, "long" },
    { TOK_KW_UNSIGNED, "unsigned" },
    { TOK_KW_SIGNED, "signed" },
    { TOK_KW_FLOAT, "float" },
    { TOK_KW_DOUBLE, "double" },
    { TOK_KW_STRUCT, "struct" },
    { TOK_KW_UNION, "union" },
    { TOK_KW_TYPEDEF, "typedef" },
    { TOK_KW_STATIC, "static" },
    { TOK_KW_SWITCH, "switch" },
    { TOK_KW_CASE, "case" },
    { TOK_KW_DEFAULT, "default" },
    { TOK_KW_BREAK, "break" },
    { TOK_KW_CONTINUE, "continue" },
    { TOK_KW_CONST, "const" },
    { TOK_KW_AUTO, "auto" },
    { TOK_KW_ENUM, "enum" },
    { TOK_KW_REGISTER, "register" },
    { TOK_KW_SIZEOF, "sizeof" },
    { TOK_KW_VOLATILE, "volatile" },
    { TOK_KW_RETURN, "return" },
    { TOK_KW_EXTERN, "extern" },
    { TOK_KW_GOTO, "goto" },
    { TOK_NONE, "" }
};


struct parser_state
{
    struct preproc_info *pp;                // preprocessor data
    struct token *curtok;                   // the current token
};


struct token *parse_next(struct parser_state *ps)
{
    struct token *tok;
    int i;
    
    for (;;)
    {
        tok = preproc_next(ps -> pp);
        if (tok -> ttype == TOK_WSPACE)
            continue;
        if (tok -> ttype == TOK_EOL)
            continue;
        if (tok -> ttype == TOK_CHAR)
        {
            // random character
            fprintf(stderr, "Random character %02x\n", tok -> strval[0]);
            if (tok -> strval[0] < 32 || tok -> strval[0] > 126)
                    continue;
        }
        break;
    }
    if (tok -> ttype == TOK_IDENT)
    {
        // convert identifier tokens to their respective meanings
        for (i = 0; keyword_list[i].tok != TOK_NONE; i++)
        {
            if (strcmp(keyword_list[i].word, tok -> strval) == 0)
            {
                tok -> ttype = keyword_list[i].tok;
                goto out;
            }
        }
        // check for registered types here
    }
    else if (tok -> ttype == TOK_NUMBER)
    {
        // look for anything that isn't 0-9
        for (i = 0; tok -> strval[i]; i++)
        {
            if (tok -> strval[i] < '0' || tok -> strval[i] > '9')
                break;
        }
        if (tok -> strval[i] == 0)
            tok -> ttype = TOK_CONST_INT;
    }
out:
    fprintf(stderr, "Lexed: ");
    token_print(tok, stderr);
    fprintf(stderr, " (%d)\n", tok -> ttype);
    if (ps -> curtok)
        token_free(ps -> curtok);
    ps -> curtok = tok;
    return tok;
}

void parse_generr(struct parser_state *ps, char *tag)
{
    fprintf(stderr, "(%s) Unexpected token (%d): ", tag, ps -> curtok -> ttype);
    token_print(ps -> curtok, stderr);
    fprintf(stderr, "\n");

}

node_t *parse_expr_real(struct parser_state *ps, int prec);

// parse an elementary type (int, etc.)
node_t *parse_elem_type(struct parser_state *ps)
{
    int sgn = -1;
    int nt = -1;
    int nn = 1;

    if (ps -> curtok -> ttype == TOK_KW_SIGNED)
    {
        sgn = 1;
        parse_next(ps);
    }
    else if (ps -> curtok -> ttype == TOK_KW_UNSIGNED)
    {
        sgn = 0;
        parse_next(ps);
    }
    
    switch (ps -> curtok -> ttype)
    {
    // NOTE: char is unsigned by default
    case TOK_KW_CHAR:
        if (sgn == -1 || sgn == 0)
            nt = NODE_TYPE_UCHAR;
        else
            nt = NODE_TYPE_CHAR;
        break;
    
    case TOK_KW_SHORT:
        nt = sgn ? NODE_TYPE_SHORT : NODE_TYPE_USHORT;
        break;
    
    case TOK_KW_INT:
        nt = sgn ? NODE_TYPE_INT : NODE_TYPE_UINT;
        break;
    
    case TOK_KW_LONG:
        parse_next(ps);
        if (ps -> curtok -> ttype == TOK_KW_LONG)
        {
            nt = sgn ? NODE_TYPE_LONGLONG : NODE_TYPE_ULONGLONG;
            break;
        }
        nn = 0;
        nt = sgn ? NODE_TYPE_LONG : NODE_TYPE_ULONG;
        break;
    
    }
    if (nt == -1)
    {
        if (sgn == -1)
        {
            return NULL;
        }
        else
        {
            nt = sgn ? NODE_TYPE_INT : NODE_TYPE_UINT;
        }
    }
    else if (nn)
    {
        parse_next(ps);
    }
    return node_create(nt);
}

// if ident is non-zero, accept an identifier as part of the type; otherwise
// do not accept an identifier; currently a stub
node_t *parse_type(struct parser_state *ps, int ident)
{
    node_t *rv;

    // see if we have an elementary type
    rv = parse_elem_type(ps);

    // look for "struct", etc.

    // look for pointer indicator(s)

    // look for identifier if wanted/allowed

    // look for array indicator or function parameter list
    return rv;
}

node_t *parse_term_real(struct parser_state *ps)
{
    node_t *rv, *rv2;

    switch (ps -> curtok -> ttype)
    {
    case TOK_CONST_INT:
        rv = node_create(NODE_CONST_INT, ps -> curtok -> strval);
        parse_next(ps);
        return rv;
     
    // opening paren: either grouping or type cast
    case TOK_OPAREN:
        parse_next(ps);
        // parse a type without an identifier
        rv2 = parse_type(ps, 0);
        if (rv2)
        {
            if (ps -> curtok -> ttype != TOK_CPAREN)
            {   
                node_destroy(rv2);
                parse_generr(ps, "missing ) on type cast");
                return NULL;
            }
            parse_next(ps);
            // detect C99 compound literal here
            rv = parse_expr_real(ps, 175);
            if (!rv)
            {
                node_destroy(rv);
                return NULL;
            }
            return node_create(NODE_TYPECAST, rv2, rv);
        }
        // grouping
        rv = parse_expr_real(ps, 0);
        if (ps -> curtok -> ttype != TOK_CPAREN)
        {
            node_destroy(rv);
            parse_generr(ps, "missing ) on expression grouping");
            return NULL;
        }
        parse_next(ps);
        return rv;
    }

    parse_generr(ps, "term");
    return NULL;
}

node_t *parse_expr_fncall(struct parser_state *ps, node_t *term1)
{
    if (ps -> curtok -> ttype != TOK_CPAREN)
    {
        node_destroy(term1);
        parse_generr(ps, "missing )");
        return NULL;
    }
    parse_next(ps);
    return node_create(NODE_OPER_FNCALL, term1, NULL);
}

node_t *parse_expr_postinc(struct parser_state *ps, node_t *term1)
{
    return node_create(NODE_OPER_POSTINC, term1);
}

node_t *parse_expr_postdec(struct parser_state *ps, node_t *term1)
{
    return node_create(NODE_OPER_POSTDEC, term1);
}

node_t *parse_expr_subscript(struct parser_state *ps, node_t *term1)
{
    node_t *term2;
    term2 = parse_expr_real(ps, 0);
    if (!term2)
    {
        node_destroy(term1);
        return NULL;
    }
    if (ps -> curtok -> ttype != TOK_CSQUARE)
    {
        node_destroy(term2);
        node_destroy(term1);
        parse_generr(ps, "missing ]");
        return NULL;
    }
    parse_next(ps);
    return node_create(NODE_OPER_SUBSCRIPT, term1, term2);
}

node_t *parse_expr_cond(struct parser_state *ps, node_t *term1)
{
    node_t *term2, *term3;
    // conditional operator
    // NOTE: the middle operand is evaluated as though it is its own
    // independent expression because the : must appear. The third
    // operand is evaluated at the ternary operator precedence so that
    // subsequent operand binding behaves correctly (if surprisingly). This
    // would be less confusing if the ternary operator was fully bracketed
    // (that is, had a terminator)
    term2 = parse_expr_real(ps, 0);
    if (!term2)
    {
        node_destroy(term1);
        return NULL;
    }
    if (ps -> curtok -> ttype == TOK_COLON)
    {
        parse_next(ps);
        term3 = parse_expr_real(ps, 25);
        if (!term3)
        {
            node_destroy(term1);
            node_destroy(term2);
            return NULL;
        }
        return node_create(NODE_OPER_COND, term1, term2, term3);
    }
    else
    {
        node_destroy(term1);
        node_destroy(term2);
        parse_generr(ps, "missing :");
        return NULL;
    }
}

node_t *parse_expr_real(struct parser_state *ps, int prec)
{
    static struct { int tok; int nodetype; int prec; int ra; node_t *(*spec)(struct parser_state *, node_t *); } operlist[] = {
//        { TOK_OPAREN, NODE_OPER_FNCALL, 200, 0, parse_expr_fncall },
//        { TOK_OSQUARE, NODE_OPER_SUBSCRIPT, 200, 0, parse_expr_subscript },
//        { TOK_ARROW, NODE_OPER_PTRMEM, 200, 0 },
//        { TOK_DOT, NODE_OPER_OBJMEM, 200, 0 },
//        { TOK_DBLADD, NODE_OPER_POSTINC, 200, 0, parse_expr_postinc },
//        { TOK_DBLSUB, NODE_OPER_POSTDEC, 200, 0, parse_expr_postdec },
        { TOK_STAR, NODE_OPER_TIMES, 150 },
        { TOK_DIV, NODE_OPER_DIVIDE, 150 },
        { TOK_MOD, NODE_OPER_MOD, 150 },
        { TOK_ADD, NODE_OPER_PLUS, 100 },
        { TOK_SUB, NODE_OPER_MINUS, 100 },
        { TOK_LSH, NODE_OPER_LSH, 90 },
        { TOK_RSH, NODE_OPER_RSH, 90 },
        { TOK_LT, NODE_OPER_LT, 80 },
        { TOK_LE, NODE_OPER_LE, 80 },
        { TOK_GT, NODE_OPER_GT, 80 },
        { TOK_GE, NODE_OPER_GE, 80 },
        { TOK_EQ, NODE_OPER_EQ, 70 },
        { TOK_NE, NODE_OPER_NE, 70 },
        { TOK_BWAND, NODE_OPER_BWAND, 60},
        { TOK_XOR, NODE_OPER_BWXOR, 55 },
        { TOK_BWOR, NODE_OPER_BWOR, 50 },
        { TOK_BAND, NODE_OPER_BAND, 40 },
        { TOK_BOR, NODE_OPER_BOR, 35 },
        { TOK_QMARK, NODE_OPER_COND, 25, 1, parse_expr_cond },
//        { TOK_ASS, NODE_OPER_ASS, 20, 1 },
//        { TOK_ADDASS, NODE_OPER_ADDASS, 20, 1 },
//        { TOK_SUBASS, NODE_OPER_SUBASS, 20, 1 },
//        { TOK_MULASS, NODE_OPER_MULASS, 20, 1 },
//        { TOK_DIVASS, NODE_OPER_DIVASS, 20, 1 },
//        { TOK_MODASS, NODE_OPER_MODASS, 20, 1 },
//        { TOK_LSHASS, NODE_OPER_LSHASS, 20, 1 },
//        { TOK_RSHASS, NODE_OPER_RSHASS, 20, 1 },
//        { TOK_BWANDASS, NODE_OPER_BWANDASS, 20, 1},
//        { TOK_BWORASS, NODE_OPER_BWORASS, 20, 1 },
//        { TOK_XORASS, NODE_OPER_BWXORASS, 20, 1 },
        { TOK_COMMA, NODE_OPER_COMMA, 1 },
        { 0, 0, 0 }
    };
    node_t *term1, *term2;
    int i;
    
    term1 = parse_term_real(ps);
    if (!term1)
        return NULL;

nextoper:
    for (i = 0; operlist[i].tok; i++)
        if (operlist[i].tok == ps -> curtok -> ttype)
            break;
    fprintf(stderr, "Matched operator: %d, %d\n", operlist[i].tok, operlist[i].prec);
    // if we hit the end of the expression, return
    if (operlist[i].tok == 0)
        return term1;

    // is previous operation higher precedence? If so, just return the first term
    if (operlist[i].prec < prec)
        return term1;

    // is this operator left associative and previous operation is same precedence?
    // if so, just return the first term
    if (operlist[i].ra == 0 && operlist[i].prec == prec)
        return term1;

    // consume the operator
    parse_next(ps);
    
    // special handling
    if (operlist[i].spec)
    {
        term2 = (operlist[i].spec)(ps, term1);
        if (!term2)
        {
            node_destroy(term1);
            return NULL;
        }
        term1 = term2;
        goto nextoper;
    }
    term2 = parse_expr_real(ps, operlist[i].prec);
    if (!term2)
    {
        parse_generr(ps, "expr");
        node_destroy(term1);
    }
    
    term1 = node_create(operlist[i].nodetype, term1, term2);
    term2 = NULL;
    goto nextoper;
}

node_t *parse_expr(struct parser_state *ps)
{
    return parse_expr_real(ps, 0);
}

node_t *parse_statement(struct parser_state *ps)
{
    node_t *rv;
    node_t *n;

    switch (ps -> curtok -> ttype)
    {
    case TOK_KW_RETURN:
        parse_next(ps);
        n = parse_expr(ps);
        if (!n)
        {
            parse_generr(ps, "statement");
            return NULL;
        }
        rv = node_create(NODE_STMT_RETURN);
        node_addchild(rv, n);
        break;

    default:
        return NULL;
    }
    
    if (ps -> curtok -> ttype != TOK_EOS)
        parse_generr(ps, "statement");
    else
        parse_next(ps);
    return rv;
}

node_t *parse_globaldecl(struct parser_state *ps)
{
    node_t *rv = NULL;
    node_t *stmt;
    char *fnname = NULL;
    if (ps -> curtok -> ttype == TOK_KW_INT)
    {
        // variable name
        parse_next(ps);
        if (ps -> curtok -> ttype != TOK_IDENT)
            goto error;
        fnname = lw_strdup(ps -> curtok -> strval);
        parse_next(ps);
        if (ps -> curtok -> ttype != TOK_OPAREN)
            goto error;
        parse_next(ps);
        if (ps -> curtok -> ttype != TOK_CPAREN)
            goto error;
        parse_next(ps);
        if (ps -> curtok -> ttype != TOK_OBRACE)
            goto error;
        parse_next(ps);
        stmt = parse_statement(ps);
        if (!stmt)
            goto error;
        rv = node_create(NODE_FUNDEF, node_create(NODE_TYPE_INT), node_create(NODE_IDENT, fnname), node_create(NODE_FUNARGS), stmt);
        if (ps -> curtok -> ttype != TOK_CBRACE)
            goto error;
        parse_next(ps);
        lw_free(fnname);        
        return rv;
    }        
        

error:
    if (fnname)
        lw_free(fnname);
    parse_generr(ps, "globaldecl");
    return rv;
}

node_t *parse_program(struct preproc_info *pp)
{
    node_t *rv;
    node_t *node;
    struct parser_state ps;
    
    ps.pp = pp;
    ps.curtok = NULL;

    rv = node_create(NODE_PROGRAM);

    // prime the parser
    parse_next(&ps);
    while (ps.curtok -> ttype != TOK_EOF)
    {
        node = parse_globaldecl(&ps);
        if (!node)
            break;
        node_addchild(rv, node);
    }

    return rv;
}