path: root/bootstrap/parsestmt.c

/*
 * Statement parsing routines for the bootstrap compiler.
 *
 * Copyright © 2025 Samuel Lidén Borell <samuel@kodafritt.se>
 *
 * SPDX-License-Identifier: EUPL-1.2+ OR LGPL-2.1-or-later
 */
#include <assert.h>
#include <string.h>
#include "compiler.h"
#include "token.h"

static int current_stmt_id = 0;
static struct Stmt *current_loop = NULL;
static struct Var **nextptr_vardecl = NULL;

static enum Token parse_stmt_block(struct Stmt **stmt_out);
static void parser_enter_scope(void);
static void parser_leave_scope(void);
static struct Var *parse_local_var(enum VarType vartype);

struct Scope {
    struct Scope *outer;
    struct TreeNode *vars;
};
static struct Scope *current_scope = NULL;

void parse_func_body(void)
{
    struct Stmt **stmt_ptr = &current_func->code;
    struct Section **section_ptr = &current_func->section_first;
    current_stmt_id = 0;
    assert(current_func->vardecls == NULL);
    parser_enter_scope(); /* params scope */
    current_scope->vars = current_funcparams;
    nextptr_vardecl = &current_func->vardecls;
    for (;;) {
        enum Token t = parse_stmt_block(stmt_ptr);
        if (t == T_KW_end) {
            assert(current_loop == NULL);
            parser_leave_scope();
            assert(current_scope == NULL);
            return;
        } else if (t == T_KW_section) {
            struct LexemeInfo li;
            struct Section *section;
            expect(&li, T_LowerIdent, "Expected lowercase section name");
            section = (struct Section *)tree_insert_str(
                    &current_func->section_by_name,
                    li.string, li.len, NULL, sizeof(struct Section));
            if (section->ident.node.is_defined) {
                error("Duplicate section name");
            }
            srcloc_init(&section->ident.srcloc);
            section->ident.node.is_defined = 1;
            *section_ptr = section;
            section_ptr = &section->next;
            section->next = NULL;
            section->code = NULL;
            stmt_ptr = &section->code;
            expect_next_line();
        } else {
            error(tokenizer_line_is_indented() ?
                  "Too many `end`s" :
                  "Expected `end` or `section xxx` after code block");
        }
    }
}

static struct Stmt *new_stmt(void)
{
    struct Stmt *s = malloc(sizeof(struct Stmt));
    NO_NULL(s);
    s->next = NULL;
    s->id = current_stmt_id++;
    return s;
}

static enum Token parse_stmt_block(struct Stmt **stmt_out)
{
    *stmt_out = NULL;
    parser_enter_scope();
    for (;;) {
        struct Stmt *s;
        struct LexemeInfo li;
        enum Token t, endt;

        t = tokenize(&li);
        switch ((int)t) {
        case T_EOL:
            expect_next_line();
            continue;
        case T_KW_case:
        case T_KW_default:
        case T_KW_else:
        case T_KW_elif:
        case T_KW_end:
        case T_KW_loopempty:
        case T_KW_loopend:
        case T_KW_section:
            /* End of statement block */
            parser_leave_scope();
            return t;
        }
        s = new_stmt();
        *stmt_out = s;
        s->line = current_line;
        switch ((int)t) {
        case T_KW_assert:
            s->kind = S_ASSERT;
            s->u.expr = parse_expr();
            break;
        case T_KW_break:
            s->kind = S_BREAK;
            if (!current_loop) {
                error("`break` must be inside a loop");
            }
            s->u.break_.loop = current_loop;
            break;
        case T_KW_continue:
            s->kind = S_CONTINUE;
            if (!current_loop) {
                error("`continue` must be inside a loop");
            }
            s->u.break_.loop = current_loop;
            break;
        case T_KW_for: {
            struct Var *var;
            s->kind = S_FOR;
            var = parse_local_var(VAR_DECL_ONLY);
            s->u.for_.var = var;
            expect(&li, T_KW_in, "Expected `in` keyword in `for` statement");
            s->u.for_.loop.cond = parse_expr();
            goto loopbody; }
        case T_KW_if: {
            struct StmtElif **elif_next;
            s->kind = S_IF;
            /* true-block */
            s->u.if_.cond = parse_expr();
            expect_next_line();
            endt = parse_stmt_block(&s->u.if_.true_);
            /* elif-blocks */
            s->u.if_.elifs = NULL;
            elif_next = &s->u.if_.elifs;
            while (endt == T_KW_elif) {
                struct StmtElif *elif = malloc(sizeof(struct StmtElif));
                NO_NULL(elif);
                elif->cond = parse_expr();
                expect_next_line();
                endt = parse_stmt_block(&elif->body);
                elif->next = NULL;
                *elif_next = elif;
                elif_next = &elif->next;
            }
            /* false-block */
            if (endt == T_KW_else) {
                expect_next_line();
                endt = parse_stmt_block(&s->u.if_.false_);
            } else {
                s->u.if_.false_ = NULL;
            }
            if (endt != T_KW_end) {
                error("Expected `end` after `if` block");
            }
            break; }
        case T_KW_return:
            t = tokenize(&li);
            unread_token();
            if (t == T_EOL) {
                s->kind = S_RETURN_NOVALUE;
                if (current_func->returns != NULL) {
                    error("`return` without value in function with returns");
                }
            } else {
                s->kind = S_RETURN_VALUE;
                if (current_func->returns == NULL) {
                    error("`return` with value in function without returns");
                }
                assert(current_func->num_returns >= 1);
                s->u.expr = parse_expr();
            }
            break;
        case T_KW_switch: {
            struct Case **nextptr;
            s->kind = S_SWITCH;
            s->u.switch_.cond = parse_expr();
            s->u.switch_.default_ = NULL;
            s->u.switch_.cases = NULL;
            nextptr = &s->u.switch_.cases;
            expect_next_line();
            t = tokenize(&li);
            for (;;) {
                if (t == T_KW_case) {
                    struct Case *case_ = malloc(sizeof(struct Case));
                    NO_NULL(case_);
                    case_->value = parse_expr();
                    expect_next_line();
                    t = parse_stmt_block(&case_->block);
                    case_->next = NULL;
                    *nextptr = case_;
                    nextptr = &case_->next;
                } else if (t == T_KW_default) {
                    expect_next_line();
                    t = parse_stmt_block(&s->u.switch_.default_);
                    if (t != T_KW_end) {
                        error("Expected `end` after `default` block");
                    }
                } else if (t == T_KW_end) {
                    break;
                } else if (t == T_EOL) {
                    /* Empty / comment-only lines are allowed */
                    expect_next_line();
                    t = tokenize(&li);
                } else {
                    error("Expected `case`, `default` or `end` "
                          "in `switch` block");
                }
            }
            break; }
        case T_KW_while: {
            struct Stmt *outer_loop;
            s->kind = S_WHILE;
            s->u.loop.cond = parse_expr();
          loopbody:
            s->u.loop.loopempty = NULL;
            s->u.loop.loopend = NULL;
            expect_next_line();
            outer_loop = current_loop;
            current_loop = s;
            endt = parse_stmt_block(&s->u.loop.body);
            current_loop = outer_loop;
            /* loopempty block:
               Reached if loop is never executed */
            if (endt == T_KW_loopempty) {
                expect_next_line();
                endt = parse_stmt_block(&s->u.loop.loopempty);
            }
            /* TODO consider renaming this:
                - finish?
                - complete? */
            /* loopend block:
               Reached if loop is not exited (or if empty) */
            if (endt == T_KW_loopend) {
                expect_next_line();
                endt = parse_stmt_block(&s->u.loop.loopend);
            }
            if (endt != T_KW_end) {
                const char *msg;
                if (endt == T_KW_loopempty) {
                    msg="`loopempty` must come before `loopend`";
                } else if (s->u.loop.loopend) {
                    msg="Expected `end` after loop";
                } else if (s->u.loop.loopempty) {
                    msg="Expected `end` or `finish` after loop";
                } else {
                    msg="Expected `end`, `loopempty` or `loopend` after loop";
                }
                error(msg);
            }
            break; }
        TOKEN_CASES_QUALIFIERS
        case T_KW_bool:
        case T_KW_byte:
        case T_KW_int:
        case T_KW_long:
        case T_UpperIdent: {
            /* Variable definition */
            s->kind = S_VARDECL;
            unread_token();
            s->u.var = parse_local_var(VAR_ALLOW_INITVAL);
            break; }
        case T_LowerIdent: {
            struct Expr *expr;
            /* Function call or assignment expression */
            unread_token();
            expr = parse_expr();
            t = tokenize(&li);
            if (t == T_SYM_SingleEqual) {
                struct AssignDestination *dest;
                s->kind = S_ASSIGN;
                dest = &s->u.assign.first_dest;
                for (;;) {
                    struct AssignDestination *nextdest;
                    dest->expr = expr;
                    expr = parse_expr();
                    t = tokenize(&li);
                    if (t != T_SYM_SingleEqual) break;
                    nextdest = malloc(sizeof(struct AssignDestination));
                    NO_NULL(nextdest);
                    dest->next = nextdest;
                    dest = nextdest;
                }
                dest->next = NULL;
                s->u.assign.sourceexpr = expr;
            } else {
                unread_token();
                s->kind = S_EXPR;
                s->u.expr = expr;
            }
            break; }
        default:
            error("Unexpected token at start of statement");
        }
        expect_next_line();
        stmt_out = &s->next;
    }
}

static void parser_enter_scope(void)
{
    struct Scope *scope = malloc(sizeof(struct Scope));
    NO_NULL(scope);
    scope->outer = current_scope;
    scope->vars = NULL;
    current_scope = scope;
}

static void parser_leave_scope(void)
{
    struct Scope *prev;
    assert(current_scope != NULL);
    prev = current_scope;
    /* TODO "unassign" vars when they go out of scope */
    current_scope = current_scope->outer;
    free(prev);
}

static bool exists_in_outer_scope(const struct Var *var)
{
    const struct Scope *scope;
    for (scope = current_scope->outer; scope != NULL; scope = scope->outer) {
        if (tree_search_node(scope->vars, &var->ident.node)) {
            return true;
        }
    }
    return false;
}

static struct Var *parse_local_var(enum VarType vartype)
{
    struct Var *var;
    /* TODO disallow defining the same name with different
            types in different scopes? */
    var = parse_var(&current_scope->vars, vartype, nextptr_vardecl);
    if (exists_in_outer_scope(var)) {
        error("Name shadows an existing variable name in an outer scope");
    }
    var->is_funcparam = 0;
    nextptr_vardecl = &var->next;
    return var;
}

struct Var *lookup_local_var(const char *name, size_t len)
{
    struct Scope *scope;
    HashCode h;

    assert(name != NULL);
    assert(len != 0);
    h = hash_str(name, len);
    /* Should have a funcparam scope + at least one statement scope */
    assert(current_scope != NULL && current_scope->outer != NULL);
    for (scope = current_scope; scope != NULL; scope = scope->outer) {
        struct TreeNode *node = tree_search(scope->vars, h, len, name);
        if (node) {
            return (struct Var *)node;
        }
    }
    return NULL;
}