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+
 */
#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);


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);
    nextptr_vardecl = &current_func->vardecls;
    for (;;) {
        enum Token t = parse_stmt_block(stmt_ptr);
        if (t == T_KW_end) {
            assert(current_loop == 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");
            }
            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)
{
    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 */
            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;
            /* TODO loop var should go out of scope after the loop */
            var = parse_var(&current_func->vars, VAR_DECL_ONLY,
                            nextptr_vardecl);
            var->is_funcparam = 0;
            nextptr_vardecl = &var->next;
            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_);
            }
            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 */
            struct Var *var;
            s->kind = S_VARDECL;
            /* TODO "unassign" vars when they go out of scope
                    (maybe allow re-assignment in another scope, but only
                     with exactly the same type) */
            unread_token();
            var = parse_var(&current_func->vars, VAR_ALLOW_INITVAL,
                            nextptr_vardecl);
            NO_NULL(var);
            var->is_funcparam = 0;
            nextptr_vardecl = &var->next;
            s->u.var = var;
            break; }
        case T_LowerIdent:
            /* Function call or assignment expression */
            unread_token();
            s->kind = S_EXPR;
            s->u.expr = parse_expr();
            break;
        default:
            error("Unexpected token at start of statement");
        }
        expect_next_line();
        stmt_out = &s->next;
    }
}