rifle/core/parser.c

#include <stdint.h>
#include <stddef.h>
#include <stdbool.h>
#include "rifle/parser.h"
#include "rifle/token.h"
#include "rifle/trace.h"
#include "rifle/lexer.h"
#include "rifle/state.h"
#include "rifle/symbol.h"
#include "rifle/types.h"
#include "rifle/ast.h"
#include "rifle/scope.h"
#include "rifle/codegen.h"

/* Symbolic token */
#define symtok(tok) \
    "[" tok "]"

/* Quoted token */
#define qtok(tok)   \
    "'" tok "'"

/* Convert token to string */
#define tokstr1(tt)  \
    toktab[(tt)]

/* Convert token to string */
#define tokstr(tok)  \
    tokstr1((tok)->type)

/* Unexpected end of file */
#define ueof(state)                 \
    trace_error(                    \
        (state),                    \
        "unexpected end of file\n"  \
    );

#define utok1(state, tok)       \
    trace_error(                \
        (state),                \
        "unexpected token %s\n", \
        tokstr(tok)             \
    )

/* Unexpected token */
#define utok(state, expt, got)          \
    trace_error(                        \
        (state),                        \
        "expected %s, got %s instead\n", \
        (expt),                         \
        (got)                           \
    )

/*
 * Table used to convert token constants into human
 * readable strings
 */
static const char *toktab[] = {
    [TT_NONE]   = symtok("none"),
    [TT_IDENT]  = symtok("ident"),
    [TT_NUMBER] = symtok("number"),
    [TT_PLUS]   = qtok("+"),
    [TT_MINUS]  = qtok("-"),
    [TT_STAR]   = qtok("*"),
    [TT_SLASH]  = qtok("/"),
    [TT_COLON]  = qtok(":"),
    [TT_LPAREN] = qtok("("),
    [TT_RPAREN] = qtok(")"),
    [TT_LBRACE] = qtok("{"),
    [TT_RBRACE] = qtok("}"),
    [TT_SEMI]   = qtok(";"),
    [TT_COMMA]  = qtok(","),
    [TT_F]      = qtok(".f"),
    [TT_STRUCT] = qtok(".struct"),
    [TT_EXTERN] = qtok(".extern"),
    [TT_PUB]    = qtok(".pub"),
    [TT_RETURN] = qtok("return"),
    [TT_BREAK]  = qtok("break"),
    [TT_LOOP]   = qtok("loop"),
    [TT_VOID]   = qtok("void"),
    [TT_U8]     = qtok("u8"),
    [TT_U16]    = qtok("u16"),
    [TT_U32]    = qtok("u32"),
    [TT_U64]    = qtok("u64"),
    [TT_DEFINE] = qtok("#define"),
    [TT_IFDEF]  = qtok("#ifdef"),
    [TT_IFNDEF] = qtok("#ifndef"),
    [TT_ENDIF]  = qtok("#endif")
};

/*
 * Parse-side token scan function
 *
 * @state: Compiler state
 * @tok:   Last token
 *
 * Returns zero on success
 */
static int
parse_scan(struct rifle_state *state, struct token *tok)
{
    struct token *popped;

    if (state == NULL || tok == NULL) {
        return -1;
    }

    switch (state->pass_num) {
    case 0:
        if (lexer_scan(state, tok) < 0) {
            return -1;
        }

        break;
    case 1:
        if ((popped = tokbuf_pop(&state->tokbuf)) == NULL) {
            return -1;
        }

        /*
         * The parser stage is what handles the incrementation of the line
         * number counter. The preprocessor does not filter out newlines for
         * this reason and therefore we are supposed to do it ourselves.
         */
        while (popped->type == TT_NEWLINE) {
            ++state->line_num;
            if ((popped = tokbuf_pop(&state->tokbuf)) == NULL)
                return -1;
        }

        *tok = *popped;
        break;
    }

    return 0;
}

/*
 * Assert that the next token is of a specific type
 *
 * @state: Compiler state
 * @what:  Token to assert that is next
 * @tok:   Last token
 *
 * Returns zero on success
 */
static int
parse_expect(struct rifle_state *state, tt_t what, struct token *tok)
{
    if (state == NULL || tok == NULL) {
        return -1;
    }

    if (parse_scan(state, tok) < 0) {
        ueof(state);
        return -1;
    }

    if (tok->type != what) {
        utok(state, tokstr1(what), tokstr(tok));
        return -1;
    }

    return 0;
}

static int
parse_define(struct rifle_state *state, struct token *tok)
{
    int error;

    if (state == NULL || tok == NULL) {
        return -1;
    }

    if (tok->type != TT_DEFINE) {
        return -1;
    }

    /* EXPECT <IDENT> */
    if (parse_expect(state, TT_IDENT, tok) < 0) {
        return -1;
    }

    error = symbol_new(
        &state->symtab,
        tok->s,
        SYMBOL_MACRO,
        NULL
    );

    if (error < 0) {
        trace_error(state, "failed to create symbol '%s'\n", tok->s);
        return -1;
    }

    return 0;
}

/*
 * Skip to an '#endif' token
 *
 * @state: Compiler state
 * @tok:   Last token
 *
 * Returns zero on success
 */
static int
parse_skip_to_endif(struct rifle_state *state, struct token *tok)
{
    if (state == NULL || tok == NULL) {
        return -1;
    }

    while (parse_scan(state, tok) == 0) {
        if (tok->type == TT_ENDIF) {
            break;
        }
    }

    if (tok->type == TT_ENDIF) {
        --state->ifx_depth;
    }

    return 0;
}

/*
 * Parse an '#ifdef' preprocessing directive
 *
 * @state: Compiler state
 * @tok:   Last token
 *
 * Returns zero on success
 */
static int
parse_ifdef(struct rifle_state *state, struct token *tok)
{
    struct symbol *symbol;

    if (state == NULL || tok == NULL) {
        return -1;
    }

    if (tok->type != TT_IFDEF) {
        return -1;
    }

    /* EXPECT <IDENT> */
    if (parse_expect(state, TT_IDENT, tok) < 0) {
        return -1;
    }

    symbol = symbol_from_name(
        &state->symtab,
        tok->s
    );

    if (symbol == NULL) {
        parse_skip_to_endif(state, tok);
    }

    return 0;
}

/*
 * Handle an '#ifndef' directive
 *
 * @state: Compiler state
 * @tok:   Last token
 *
 * Returns zero on success
 */
static int
parse_ifndef(struct rifle_state *state, struct token *tok)
{
    struct symbol *symbol;

    if (state == NULL || tok == NULL) {
        return -1;
    }

    if (tok->type != TT_IFNDEF) {
        return -1;
    }

    if (parse_expect(state, TT_IDENT, tok) < 0) {
        return -1;
    }

    symbol = symbol_from_name(
        &state->symtab,
        tok->s
    );

    if (symbol != NULL) {
        parse_skip_to_endif(state, tok);
    }

    return 0;
}

/*
 * Handle the preprocessing stage
 *
 * @state: Compiler state
 */
static int
parse_preprocess(struct rifle_state *state)
{
    struct token tok;

    while (parse_scan(state, &tok) == 0) {
        switch (tok.type) {
        case TT_DEFINE:
            if (parse_define(state, &tok) < 0) {
                return -1;
            }

            break;
        case TT_IFDEF:
            ++state->ifx_depth;
            if (parse_ifdef(state, &tok) < 0) {
                return -1;
            }

            break;
        case TT_IFNDEF:
            ++state->ifx_depth;
            if (parse_ifndef(state, &tok) < 0) {
                return -1;
            }

            break;
        case TT_ENDIF:
            if (state->ifx_depth == 0) {
                trace_error(state, "extraneous '#endif' directive\n");
                return -1;
            }

            --state->ifx_depth;
            break;
        default:
            tokbuf_push(&state->tokbuf, &tok);
            break;
        }
    }

    return 0;
}

/*
 * Parse a type definition
 *
 * @state: Compiler state
 * @tok:   Last token
 * @res:   Data type result
 *
 * Returns zero on success
 */
static int
parse_type(struct rifle_state *state, struct token *tok, struct data_type *res)
{
    if (state == NULL || tok == NULL){
        return -1;
    }

    res->ptr_depth = 0;
    res->type = token_to_data_type(tok->type);

    if (res->type == DATA_TYPE_BAD) {
        utok(state, "TYPE", tokstr(tok));
        return -1;
    }

    return 0;
}

/*
 * Parse a lbrace
 *
 * @state:  Compiler state
 * @scope:  Scope token type
 * @tok:    Last token
 *
 * Returns zero on success
 */
static int
parse_lbrace(struct rifle_state *state, tt_t scope, struct token *tok)
{
    if (state == NULL || tok == NULL) {
        return -1;
    }

    /* EXPECT '{' */
    if (parse_expect(state, TT_LBRACE, tok) < 0) {
        return -1;
    }

    if (scope_push(state, scope) < 0) {
        return -1;
    }

    return 0;
}

/*
 * Parse a function
 *
 * @state: Compiler state
 * @tok:   Last token
 * @res:   AST node
 *
 * Returns zero on success
 */
static int
parse_func(struct rifle_state *state, struct token *tok, struct ast_node **res)
{
    struct token *prevtok;
    struct ast_node *root;
    struct symbol *symbol;
    bool is_pub = false;
    int error;

    if (state == NULL || tok == NULL) {
        return -1;
    }

    if (res == NULL) {
        return -1;
    }

    if (tok->type != TT_F) {
        return -1;
    }

    if (state->cur_func != NULL) {
        trace_error(state, "nested functions not supported\n");
        return -1;
    }

    /* Is this marked as public? */
    if ((prevtok = tokbuf_lookbehind(&state->tokbuf, 1)) != NULL) {
        if (prevtok->type == TT_PUB)
            is_pub = true;
    }

   /* EXPECT <IDENT> */
    if (parse_expect(state, TT_IDENT, tok) < 0) {
        return -1;
    }

    error = symbol_new(
        &state->symtab,
        tok->s,
        SYMBOL_FUNC,
        &symbol
    );

    if (error < 0) {
        trace_error(state, "failed to create symbol '%s'\n", tok->s);
        return error;
    }

    if (ast_alloc_node(state, AST_FUNC, &root) < 0) {
        trace_error(state, "failed to allocate AST_FUNC\n");
        return -1;
    }

    symbol->pub = is_pub;
    root->symbol = symbol;

    /* EXPECT '(' */
    if (parse_expect(state, TT_LPAREN, tok) < 0) {
        return -1;
    }

    /* EXPECT ')' : TODO: ARGUMENTS */
    if (parse_expect(state, TT_RPAREN, tok) < 0) {
        return -1;
    }

    /* EXPECT ':' */
    if (parse_expect(state, TT_COLON, tok) < 0) {
        return -1;
    }

    if (parse_scan(state, tok) < 0) {
        ueof(state);
        return -1;
    }

    /* Parse the return type */
    if (parse_type(state, tok, &symbol->dtype) < 0) {
        return -1;
    }

    /* Expect '{' */
    if (parse_lbrace(state, TT_F, tok) < 0) {
        return -1;
    }

    state->cur_func = symbol;
    *res = root;
    return 0;
}

/*
 * Parse a '{' token
 *
 * @state: Compiler state
 * @tok:   Last token
 * @res:   AST result
 */
static int
parse_rbrace(struct rifle_state *state, struct token *tok, struct ast_node **res)
{
    struct ast_node *root = NULL;
    tt_t scope;

    if (state == NULL || tok == NULL) {
        return -1;
    }

    if (res == NULL) {
        return -1;
    }

    if (tok->type != TT_RBRACE) {
        return -1;
    }

    scope = scope_pop(state);
    switch (scope) {
    case TT_F:
        state->cur_func = NULL;

        /* Don't double return */
        if (state->have_ret) {
            state->have_ret = 0;
            break;
        }

        if (ast_alloc_node(state, AST_FUNC, &root) < 0) {
            trace_error(state, "failed to allocate AST_FUNC\n");
            return -1;
        }

        root->epilogue = 1;
        break;
    case TT_LOOP:
        if (ast_alloc_node(state, AST_LOOP, &root) < 0) {
            trace_error(state, "failed to allocate AST_LOOP\n");
            return -1;
        }

        root->epilogue = 1;
        break;
    default:
        break;
    }

    *res = root;
    return 0;
}

/*
 * Parse a value
 *
 * @state: Compiler state
 * @tok:   Last token
 * @res:   AST result
 *
 * Returns zero on success
 */
static int
parse_value(struct rifle_state *state, struct token *tok, struct ast_node **res)
{
    struct ast_node *root;

    if (state == NULL || tok == NULL) {
        return -1;
    }

    if (res == NULL) {
        return -1;
    }

    switch (tok->type) {
    case TT_NUMBER:
        if (ast_alloc_node(state, AST_NUMBER, &root) < 0) {
            trace_error(state, "failed to allocate AST_NUMBER\n");
            return -1;
        }

        root->v = tok->v;
        *res = root;
        return 0;
    default:
        utok(state, symtok("number"), tokstr(tok));
        break;
    }

    return -1;
}

/*
 * Parse a binary expression
 *
 * @state: Compiler state
 * @tok:   Last token
 * @res:   AST result
 *
 * Returns zero on success
 */
static int
parse_binexpr(struct rifle_state *state, struct token *tok, struct ast_node **res)
{
    struct ast_node *root;

    if (state == NULL || tok == NULL) {
        return -1;
    }

    if (res == NULL) {
        return -1;
    }

    if (parse_value(state, tok, &root) < 0) {
        return -1;
    }

    /* TODO: Allow actual binary expressions */
    *res = root;
    return 0;
}

/*
 * Parse a 'return' token
 *
 * @state: Compiler state
 * @tok:   Last token
 * @res:   AST result
 *
 * Returns zero on success
 */
static int
parse_return(struct rifle_state *state, struct token *tok, struct ast_node **res)
{
    struct ast_node *root, *rhs;

    if (state == NULL || tok == NULL) {
        return -1;
    }

    if (res == NULL) {
        return -1;
    }

    if (state->cur_func == NULL) {
        trace_error(state, "'return' not in function\n");
        return -1;
    }

    if (parse_scan(state, tok) < 0) {
        ueof(state);
        return -1;
    }

    if (parse_binexpr(state, tok, &rhs) < 0) {
        return -1;
    }

    if (ast_alloc_node(state, AST_RETURN, &root) < 0) {
        trace_error(state, "failed to allocate AST_RETURN\n");
        return -1;
    }

    if (parse_expect(state, TT_SEMI, tok) < 0) {
        return -1;
    }

    state->have_ret = 1;
    root->right = rhs;
    *res = root;
    return 0;
}

/*
 * Parse a loop statement
 *
 * @state: Compiler state
 * @tok:   Last token
 * @res:   AST result
 *
 * Returns zero on success
 */
static int
parse_loop(struct rifle_state *state, struct token *tok, struct ast_node **res)
{
    struct ast_node *root;

    if (state == NULL || tok == NULL) {
        return -1;
    }

    if (res == NULL) {
        return -1;
    }

    if (tok->type != TT_LOOP) {
        return -1;
    }

    if (parse_lbrace(state, TT_LOOP, tok) < 0) {
        return -1;
    }

    if (ast_alloc_node(state, AST_LOOP, &root) < 0) {
        trace_error(state, "failed to allocate AST_LOOP\n");
        return -1;
    }

    *res = root;
    return 0;
}

static int
parse_begin(struct rifle_state *state)
{
    struct token tok;
    struct ast_node *root = NULL;

    while (parse_scan(state, &tok) == 0) {
        switch (tok.type) {
        case TT_F:
            if (parse_func(state, &tok, &root) < 0) {
                return -1;
            }

            break;
        case TT_RBRACE:
            if (parse_rbrace(state, &tok, &root) < 0) {
                return -1;
            }

            break;
        case TT_RETURN:
            if (parse_return(state, &tok, &root) < 0) {
                return -1;
            }

            break;
        case TT_LOOP:
            if (parse_loop(state, &tok, &root) < 0) {
                return -1;
            }

            break;
        case TT_PUB:
            break;
        default:
            utok1(state, &tok);
            return -1;
        }

        if (root != NULL) {
            if (cg_resolve_node(state, root) < 0)
                return -1;
        }
    }

    return 0;
}

int
parser_parse(struct rifle_state *state)
{
    if (state == NULL) {
        return 0;
    }

    switch (state->pass_num) {
    case 0:
        /* Pre-processor */
        if (parse_preprocess(state) < 0) {
            return -1;
        }

        if (state->ifx_depth > 0) {
            trace_error(state, "unterminated '#if' directive\n");
            return -1;
        }
        break;
    case 1:
        /* Parse loop */
        if (parse_begin(state) < 0) {
            return -1;
        }

        break;
    }

    ++state->pass_num;
    return 0;
}