zzy/spl
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

stage1 实现06测试

Browse Source
This commit is contained in:
zzy
2026-07-05 22:47:00 +08:00
parent e2e0ebc21f
commit 67c8a137dd
6 changed files with 365 additions and 132 deletions
Download Patch File Download Diff File Expand all files Collapse all files

363
stage1/spl_stmt.c
View File

@@ -1,33 +1,39 @@
/* spl_stmt.c — Statement parser + codegen */
#include "spl_comp.h"
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
static spl_tok_t *peek(spl_comp_t *ctx) {
return &vec_at(ctx->toks, ctx->tok_idx);
}
static spl_tok_t *peek(spl_comp_t *ctx) { return &vec_at(ctx->toks, ctx->tok_idx); }
static spl_tok_t *advance(spl_comp_t *ctx) {
spl_tok_t *t = &vec_at(ctx->toks, ctx->tok_idx);
if (t->type != TOK_EOF) ctx->tok_idx++;
if (t->type != TOK_EOF)
ctx->tok_idx++;
return t;
}
static int expect(spl_comp_t *ctx, spl_tok_type_t type) {
if (peek(ctx)->type == type) { advance(ctx); return 1; }
spl_comp_error(ctx, "expected '%s', got '%s'", spl_tok_type_name(type), spl_tok_type_name(peek(ctx)->type));
if (peek(ctx)->type == type) {
advance(ctx);
return 1;
}
spl_comp_error(ctx, "expected '%s', got '%s'", spl_tok_type_name(type),
spl_tok_type_name(peek(ctx)->type));
return 0;
}
static int match(spl_comp_t *ctx, spl_tok_type_t type) {
if (peek(ctx)->type == type) { advance(ctx); return 1; }
if (peek(ctx)->type == type) {
advance(ctx);
return 1;
}
return 0;
}
static void skip_nl(spl_comp_t *ctx) {
while (peek(ctx)->type == TOK_ENDLINE) advance(ctx);
while (peek(ctx)->type == TOK_ENDLINE)
advance(ctx);
}
/* ============================================================
@@ -48,7 +54,8 @@ static void parse_ret_stmt(spl_comp_t *ctx) {
spl_type_t rt = ctx->current_ret_type ? ctx->current_ret_type->basic_type : SPL_VOID;
spl_emit(ctx, SPL_RET, rt, 0);
}
if (peek(ctx)->type == TOK_SEMICOLON) advance(ctx);
if (peek(ctx)->type == TOK_SEMICOLON)
advance(ctx);
}
/* ============================================================
@@ -62,7 +69,8 @@ static void parse_var_decl(spl_comp_t *ctx, int is_const) {
spl_tok_t *name_tok = advance(ctx);
char vname[256];
usize nlen = name_tok->len < 255 ? name_tok->len : 255;
memcpy(vname, name_tok->lexeme, nlen); vname[nlen] = '\0';
memcpy(vname, name_tok->lexeme, nlen);
vname[nlen] = '\0';
spl_type_info_t *var_type = NULL;
int has_init = 0;
@@ -84,12 +92,15 @@ static void parse_var_decl(spl_comp_t *ctx, int is_const) {
if (peek(ctx)->type == TOK_COLON_ASSIGN || peek(ctx)->type == TOK_ASSIGN) {
has_init = 1;
if (peek(ctx)->type == TOK_COLON_ASSIGN) advance(ctx);
else advance(ctx); /* = */
if (peek(ctx)->type == TOK_COLON_ASSIGN)
advance(ctx);
else
advance(ctx); /* = */
}
/* Allocate stack slot */
if (!var_type) var_type = spl_type_basic(SPL_I32); /* default type */
if (!var_type)
var_type = spl_type_basic(SPL_I32); /* default type */
int slot = spl_declare_var(ctx, vname, var_type, is_const);
skip_nl(ctx);
@@ -126,7 +137,8 @@ static void parse_var_decl(spl_comp_t *ctx, int is_const) {
}
}
if (peek(ctx)->type == TOK_SEMICOLON) advance(ctx);
if (peek(ctx)->type == TOK_SEMICOLON)
advance(ctx);
}
/* ============================================================
@@ -145,7 +157,8 @@ void spl_parse_block(spl_comp_t *ctx) {
}
spl_pop_scope(ctx);
if (peek(ctx)->type == TOK_R_BRACE) advance(ctx);
if (peek(ctx)->type == TOK_R_BRACE)
advance(ctx);
} else {
/* Single statement */
spl_parse_stmt(ctx);
@@ -257,68 +270,218 @@ static void parse_for_stmt(spl_comp_t *ctx) {
skip_nl(ctx);
/* Parse the iteration expression(s) */
spl_expr_result_t start = spl_parse_expr(ctx, PREC_MIN);
(void)start;
spl_expr_result_t start_expr = spl_parse_expr(ctx, PREC_MIN);
if (peek(ctx)->type == TOK_RANGE) {
/* for begin..end as i { body } */
advance(ctx); /* .. */
spl_expr_result_t end = spl_parse_expr(ctx, PREC_MIN);
(void)end;
/* ===== Numeric range: for begin..end as i { body } ===== */
advance(ctx); /* .. */
spl_parse_expr(ctx, PREC_MIN); /* end expression */
/* Stack: [begin, end] */
skip_nl(ctx);
if (peek(ctx)->type == KW_AS) {
advance(ctx); /* as */
spl_tok_t *ivar = advance(ctx);
char iname[256];
usize inl = ivar->len < 255 ? ivar->len : 255;
memcpy(iname, ivar->lexeme, inl); iname[inl] = '\0';
/* Declare loop variable */
spl_type_info_t *itype = spl_type_basic(SPL_I32);
int islot = spl_declare_var(ctx, iname, itype, 0);
/* The loop variable was already set up by the range operator */
/* Store initial value */
/* For now: we need to emit proper loop code. This is a simplification. */
int saved_loop = ctx->in_loop;
usize saved_continue = ctx->continue_target;
usize saved_bp_count = ctx->break_patch_count;
ctx->in_loop = 1;
ctx->continue_target = 0; /* will set after store */
/* Loop: store begin to i, check i < end, body, i++ */
/* Stack has: begin, end (from parsing the range expression) */
/* Actually we need to emit proper code here. For bootstrap,
let me just handle the simple numeric for loop. */
/* For now, emit a basic loop body */
if (peek(ctx)->type != KW_AS) {
skip_nl(ctx);
spl_parse_block(ctx);
return;
}
advance(ctx); /* as */
spl_tok_t *ivar = advance(ctx);
char iname[256];
usize inl = ivar->len < 255 ? ivar->len : 255;
memcpy(iname, ivar->lexeme, inl);
iname[inl] = '\0';
spl_push_scope(ctx);
int islot = spl_declare_var(ctx, iname, spl_type_basic(SPL_I32), 0);
/* Stack: [begin, end]; swap so TOS = begin */
spl_emit(ctx, SPL_SWAP, SPL_VOID, 0);
/* Store begin to i */
spl_emit(ctx, SPL_LADDR, SPL_PTR, islot);
spl_emit(ctx, SPL_SWAP, SPL_VOID, 0);
spl_emit(ctx, SPL_STORE, SPL_I32, 0);
/* Stack: [end] */
spl_val_t loop_start = vec_size(ctx->prog.insns);
/* Condition: i < end */
spl_emit(ctx, SPL_LADDR, SPL_PTR, islot);
spl_emit(ctx, SPL_LOAD, SPL_I32, 0);
spl_emit(ctx, SPL_PICK, SPL_VOID, 1);
spl_emit(ctx, SPL_SLT, SPL_I32, 0);
spl_val_t bz_addr = spl_emit_bz(ctx);
int saved_loop = ctx->in_loop;
usize saved_continue = ctx->continue_target;
usize saved_bp_count = ctx->break_patch_count;
ctx->in_loop = 1;
ctx->continue_target = loop_start;
ctx->in_loop = saved_loop;
ctx->continue_target = saved_continue;
ctx->break_patch_count = saved_bp_count;
}
} else if (peek(ctx)->type == TOK_COMMA) {
/* for slice, 0.. as val, idx { body } — skip for now */
advance(ctx); /* , */
spl_parse_expr(ctx, PREC_MIN); /* skip the range */
if (peek(ctx)->type == KW_AS) {
advance(ctx);
advance(ctx); /* val */
if (peek(ctx)->type == TOK_COMMA) {
advance(ctx);
advance(ctx); /* idx */
}
}
skip_nl(ctx);
spl_parse_block(ctx);
} else {
/* for ident in ... — skip */
skip_nl(ctx);
spl_parse_block(ctx);
spl_parse_block(ctx); /* body */
/* Increment: i = i + 1 */
spl_emit(ctx, SPL_LADDR, SPL_PTR, islot);
spl_emit(ctx, SPL_LADDR, SPL_PTR, islot);
spl_emit(ctx, SPL_LOAD, SPL_I32, 0);
spl_emit(ctx, SPL_PUSH, SPL_I32, 1);
spl_emit(ctx, SPL_ADD, SPL_I32, 0);
spl_emit(ctx, SPL_STORE, SPL_I32, 0);
/* JMP back to condition */
spl_val_t jmp_here = vec_size(ctx->prog.insns);
spl_emit(ctx, SPL_JMP, SPL_VOID, (spl_val_t)((isize)loop_start - (isize)jmp_here - 1));
/* Exit: patch bz, drop end */
spl_patch_to_here(ctx, bz_addr);
spl_emit(ctx, SPL_DROP, SPL_VOID, 0);
/* Patch breaks */
for (usize i = saved_bp_count; i < ctx->break_patch_count; i++)
spl_patch_to_here(ctx, ctx->break_patches[i]);
ctx->break_patch_count = saved_bp_count;
ctx->in_loop = saved_loop;
ctx->continue_target = saved_continue;
spl_pop_scope(ctx);
return;
}
/* ===== Slice iteration: for slice [as val |, range as val, idx] ===== */
int has_idx = 0;
if (peek(ctx)->type == TOK_COMMA) {
advance(ctx); /* , */
spl_parse_expr(ctx, PREC_MIN); /* parse start of range (e.g. 0) */
if (peek(ctx)->type == TOK_RANGE) {
advance(ctx); /* .. */
/* consume optional end expression */
if (peek(ctx)->type != KW_AS && peek(ctx)->type != TOK_COMMA &&
peek(ctx)->type != TOK_L_BRACE && peek(ctx)->type != TOK_ENDLINE &&
peek(ctx)->type != TOK_EOF)
spl_parse_expr(ctx, PREC_MIN);
}
has_idx = 1;
/* Range pushed a value; we manage our own idx, drop it */
spl_emit(ctx, SPL_DROP, SPL_VOID, 0);
}
if (peek(ctx)->type != KW_AS) {
skip_nl(ctx);
spl_parse_block(ctx);
return;
}
advance(ctx); /* as */
/* Parse val variable name */
spl_tok_t *vtok = advance(ctx);
char vname[256];
usize vnl = vtok->len < 255 ? vtok->len : 255;
memcpy(vname, vtok->lexeme, vnl);
vname[vnl] = '\0';
/* Parse optional idx variable name */
char iname[256] = {0};
if (peek(ctx)->type == TOK_COMMA) {
advance(ctx);
spl_tok_t *itok = advance(ctx);
usize inl = itok->len < 255 ? itok->len : 255;
memcpy(iname, itok->lexeme, inl);
iname[inl] = '\0';
}
/* Stack: [slice_addr] or whatever the slice expression left */
spl_push_scope(ctx);
int idx_slot = -1;
if (iname[0])
idx_slot = spl_declare_var(ctx, iname, spl_type_basic(SPL_I32), 0);
spl_type_info_t *elem_type = NULL;
if (start_expr.type) {
if (start_expr.type->kind == TYPE_SLICE)
elem_type = start_expr.type->elem;
else if (start_expr.type->kind == TYPE_PTR && start_expr.type->elem)
elem_type = start_expr.type->elem;
}
if (!elem_type)
elem_type = spl_type_basic(SPL_I32);
int val_slot = spl_declare_var(ctx, vname, elem_type, 0);
/* Extract ptr and len from slice, push index=0: stack [ptr, len, idx] */
spl_emit(ctx, SPL_DUP, SPL_VOID, 0);
spl_emit(ctx, SPL_LOAD, SPL_PTR, 0);
spl_emit(ctx, SPL_SWAP, SPL_VOID, 0);
spl_emit(ctx, SPL_PUSH, SPL_U64, sizeof(spl_val_t));
spl_emit(ctx, SPL_ADD, SPL_U64, 0);
spl_emit(ctx, SPL_LOAD, SPL_I32, 0);
spl_emit(ctx, SPL_PUSH, SPL_I32, 0);
/* Stack: [ptr, len, idx=0] */
spl_val_t loop_start = vec_size(ctx->prog.insns);
/* Condition: idx < len */
spl_emit(ctx, SPL_PICK, SPL_VOID, 1); /* copy len */
spl_emit(ctx, SPL_PICK, SPL_VOID, 1); /* copy idx */
spl_emit(ctx, SPL_SWAP, SPL_VOID, 0); /* [idx, len] → [len, idx] → SWAP → [idx, len] */
spl_emit(ctx, SPL_SLT, SPL_I32, 0);
spl_val_t bz_addr = spl_emit_bz(ctx);
/* Store current idx to idx variable */
if (idx_slot >= 0) {
spl_emit(ctx, SPL_PICK, SPL_VOID, 0); /* copy idx (TOS) */
spl_emit(ctx, SPL_LADDR, SPL_PTR, idx_slot);
spl_emit(ctx, SPL_SWAP, SPL_VOID, 0);
spl_emit(ctx, SPL_STORE, SPL_I32, 0);
}
/* Load slice[idx] and store to val */
spl_emit(ctx, SPL_PICK, SPL_VOID, 2); /* copy ptr: [ptr, len, idx, ptr] */
spl_emit(ctx, SPL_PICK, SPL_VOID, 1); /* copy idx: [ptr, len, idx, ptr, idx] */
usize elem_byte_size = (elem_type && elem_type->byte_size) ? elem_type->byte_size : 4;
spl_emit(ctx, SPL_PUSH, SPL_U64, sizeof(spl_val_t));
spl_emit(ctx, SPL_MUL, SPL_U64, 0);
spl_emit(ctx, SPL_ADD, SPL_U64, 0);
spl_emit(ctx, SPL_LOAD, SPL_I32, 0);
spl_emit(ctx, SPL_LADDR, SPL_PTR, val_slot);
spl_emit(ctx, SPL_SWAP, SPL_VOID, 0);
spl_emit(ctx, SPL_STORE, SPL_I32, 0);
/* Loop context */
int saved_loop = ctx->in_loop;
usize saved_continue = ctx->continue_target;
usize saved_bp_count = ctx->break_patch_count;
ctx->in_loop = 1;
ctx->continue_target = loop_start;
skip_nl(ctx);
spl_parse_block(ctx); /* body */
/* Increment idx */
spl_emit(ctx, SPL_PICK, SPL_VOID, 0); /* copy idx */
spl_emit(ctx, SPL_PUSH, SPL_I32, 1);
spl_emit(ctx, SPL_ADD, SPL_I32, 0);
spl_emit(ctx, SPL_SWAP, SPL_VOID, 0);
spl_emit(ctx, SPL_DROP, SPL_VOID, 0); /* replace old idx with new */
/* JMP back to condition */
spl_val_t jmp_here = vec_size(ctx->prog.insns);
spl_emit(ctx, SPL_JMP, SPL_VOID, (spl_val_t)((isize)loop_start - (isize)jmp_here - 1));
/* Exit: patch bz, drop idx, len, ptr */
spl_patch_to_here(ctx, bz_addr);
spl_emit(ctx, SPL_DROP, SPL_VOID, 0); /* drop idx */
spl_emit(ctx, SPL_DROP, SPL_VOID, 0); /* drop len */
spl_emit(ctx, SPL_DROP, SPL_VOID, 0); /* drop ptr */
/* Patch breaks */
for (usize i = saved_bp_count; i < ctx->break_patch_count; i++)
spl_patch_to_here(ctx, ctx->break_patches[i]);
ctx->break_patch_count = saved_bp_count;
ctx->in_loop = saved_loop;
ctx->continue_target = saved_continue;
spl_pop_scope(ctx);
}
/* ============================================================
@@ -338,7 +501,8 @@ static void parse_break_stmt(spl_comp_t *ctx) {
ctx->break_patch_cap = new_cap;
}
ctx->break_patches[ctx->break_patch_count++] = addr;
if (peek(ctx)->type == TOK_SEMICOLON) advance(ctx);
if (peek(ctx)->type == TOK_SEMICOLON)
advance(ctx);
}
static void parse_continue_stmt(spl_comp_t *ctx) {
@@ -349,7 +513,8 @@ static void parse_continue_stmt(spl_comp_t *ctx) {
/* Emit JMP with relative offset to continue_target */
spl_val_t here = vec_size(ctx->prog.insns);
spl_emit(ctx, SPL_JMP, SPL_VOID, (spl_val_t)((isize)ctx->continue_target - (isize)here - 1));
if (peek(ctx)->type == TOK_SEMICOLON) advance(ctx);
if (peek(ctx)->type == TOK_SEMICOLON)
advance(ctx);
}
/* ============================================================
@@ -379,8 +544,10 @@ static void parse_extern_decl(spl_comp_t *ctx) {
advance(ctx); /* # */
expect(ctx, TOK_L_BRACKET);
/* Skip extern("vm") or just look for fn */
while (peek(ctx)->type != TOK_R_BRACKET && peek(ctx)->type != TOK_EOF) advance(ctx);
if (peek(ctx)->type == TOK_R_BRACKET) advance(ctx);
while (peek(ctx)->type != TOK_R_BRACKET && peek(ctx)->type != TOK_EOF)
advance(ctx);
if (peek(ctx)->type == TOK_R_BRACKET)
advance(ctx);
skip_nl(ctx);
@@ -390,7 +557,8 @@ static void parse_extern_decl(spl_comp_t *ctx) {
spl_tok_t *fname_tok = advance(ctx);
char fn_name[256];
usize fnl = fname_tok->len < 255 ? fname_tok->len : 255;
memcpy(fn_name, fname_tok->lexeme, fnl); fn_name[fnl] = '\0';
memcpy(fn_name, fname_tok->lexeme, fnl);
fn_name[fnl] = '\0';
skip_nl(ctx);
expect(ctx, TOK_L_PAREN);
@@ -409,8 +577,15 @@ static void parse_extern_decl(spl_comp_t *ctx) {
}
nparams++;
skip_nl(ctx);
if (peek(ctx)->type == TOK_COMMA) { advance(ctx); skip_nl(ctx); continue; }
if (peek(ctx)->type == TOK_ELLIPSIS) { advance(ctx); skip_nl(ctx); } /* variadic */
if (peek(ctx)->type == TOK_COMMA) {
advance(ctx);
skip_nl(ctx);
continue;
}
if (peek(ctx)->type == TOK_ELLIPSIS) {
advance(ctx);
skip_nl(ctx);
} /* variadic */
break;
}
}
@@ -421,13 +596,15 @@ static void parse_extern_decl(spl_comp_t *ctx) {
spl_type_info_t *ret_type = spl_type_basic(SPL_VOID);
if (peek(ctx)->type != TOK_SEMICOLON && peek(ctx)->type != TOK_L_BRACE) {
ret_type = spl_parse_type(ctx);
if (!ret_type) ret_type = spl_type_basic(SPL_VOID);
if (!ret_type)
ret_type = spl_type_basic(SPL_VOID);
skip_nl(ctx);
}
spl_declare_func(ctx, fn_name, ret_type, nparams, 1, 0);
if (peek(ctx)->type == TOK_SEMICOLON) advance(ctx);
if (peek(ctx)->type == TOK_SEMICOLON)
advance(ctx);
}
}
@@ -437,10 +614,10 @@ static void parse_extern_decl(spl_comp_t *ctx) {
/* Check if token type is an assignment operator */
static int is_assign_op(spl_tok_type_t t) {
return t == TOK_ASSIGN || t == TOK_ASSIGN_ADD || t == TOK_ASSIGN_SUB ||
t == TOK_ASSIGN_MUL || t == TOK_ASSIGN_DIV || t == TOK_ASSIGN_MOD ||
t == TOK_ASSIGN_AND || t == TOK_ASSIGN_OR || t == TOK_ASSIGN_XOR ||
t == TOK_ASSIGN_L_SH || t == TOK_ASSIGN_R_SH;
return t == TOK_ASSIGN || t == TOK_ASSIGN_ADD || t == TOK_ASSIGN_SUB || t == TOK_ASSIGN_MUL ||
t == TOK_ASSIGN_DIV || t == TOK_ASSIGN_MOD || t == TOK_ASSIGN_AND ||
t == TOK_ASSIGN_OR || t == TOK_ASSIGN_XOR || t == TOK_ASSIGN_L_SH ||
t == TOK_ASSIGN_R_SH;
}
static void parse_expr_stmt(spl_comp_t *ctx) {
@@ -462,14 +639,19 @@ static void parse_expr_stmt(spl_comp_t *ctx) {
is_assign = is_assign_op(vec_at(ctx->toks, look).type);
}
if (is_assign) ctx->addr_of_mode = 1;
if (is_assign)
ctx->addr_of_mode = 1;
spl_expr_result_t expr = spl_parse_expr(ctx, PREC_MIN);
if (is_assign) ctx->addr_of_mode = 0;
if (is_assign)
ctx->addr_of_mode = 0;
if (!is_assign && expr.type) spl_emit(ctx, SPL_DROP, SPL_VOID, 0);
if (peek(ctx)->type == TOK_SEMICOLON) advance(ctx);
if (!is_assign && expr.type)
spl_emit(ctx, SPL_DROP, SPL_VOID, 0);
if (peek(ctx)->type == TOK_SEMICOLON)
advance(ctx);
/* Safety: if no token was consumed, advance to prevent infinite loop */
if (ctx->tok_idx == prev) advance(ctx);
if (ctx->tok_idx == prev)
advance(ctx);
}
/* ============================================================
@@ -479,7 +661,8 @@ static void parse_expr_stmt(spl_comp_t *ctx) {
void spl_parse_stmt(spl_comp_t *ctx) {
skip_nl(ctx);
if (peek(ctx)->type == TOK_EOF || peek(ctx)->type == TOK_R_BRACE) return;
if (peek(ctx)->type == TOK_EOF || peek(ctx)->type == TOK_R_BRACE)
return;
switch (peek(ctx)->type) {
case KW_RET:
@@ -521,7 +704,7 @@ void spl_parse_stmt(spl_comp_t *ctx) {
case TOK_LINE_COMMENT:
advance(ctx);
break;
case TOK_SHARP: /* #[extern(...)] */
case TOK_SHARP: /* #[extern(...)] */
parse_extern_decl(ctx);
break;
default: