536 lines
19 KiB
C
536 lines
19 KiB
C
/* spl_parser.c — Top-level parser: function declarations, type declarations, etc. */
|
|
|
|
#include "spl_comp.h"
|
|
#include "spl_lex_util.h"
|
|
#include <string.h>
|
|
|
|
spl_tok_t *peek(spl_comp_t *ctx) { return &vec_at(ctx->toks, ctx->tok_idx); }
|
|
|
|
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++;
|
|
return t;
|
|
}
|
|
|
|
/* ============================================================
|
|
* Parse function definition
|
|
* fn name(params) ret-type { body }
|
|
* or fn name(params) ret-type; (forward decl, not used for stage1)
|
|
* ============================================================ */
|
|
|
|
/* Shared helper: register a function, declare params, parse body, end function.
|
|
* Used by both top-level fn decl and methods inside type bodies. */
|
|
/* Shared helper: parse function/method parameter list: (name: type, name: type, ...)
|
|
* Returns number of params parsed. Stores names in pnames and types in ptypes
|
|
* (both must be MAX_PARAMS-sized arrays). */
|
|
enum { MAX_PARAMS = 64 };
|
|
|
|
static int parse_params_decl(spl_comp_t *ctx, char pnames[][256], spl_type_info_t *ptypes[]) {
|
|
int nparams = 0;
|
|
skip_nl(ctx);
|
|
if (peek(ctx)->type != TOK_R_PAREN) {
|
|
while (1) {
|
|
spl_tok_t *pname = advance(ctx);
|
|
spl_tok_copy_name(pname, pnames[nparams], 256);
|
|
skip_nl(ctx);
|
|
if (peek(ctx)->type == TOK_COLON) {
|
|
advance(ctx); /* : */
|
|
skip_nl(ctx);
|
|
ptypes[nparams] = spl_parse_type(ctx);
|
|
} else {
|
|
ptypes[nparams] = spl_type_basic(SPL_I32);
|
|
}
|
|
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);
|
|
}
|
|
break;
|
|
}
|
|
}
|
|
expect(ctx, TOK_R_PAREN);
|
|
return nparams;
|
|
}
|
|
|
|
static int parse_fn_body(spl_comp_t *ctx, const char *fn_name, spl_type_info_t *ret_type,
|
|
int nparams, char pnames[][256], spl_type_info_t *ptypes[], int is_pub) {
|
|
int fi = spl_declare_func(ctx, fn_name, ret_type, nparams, 0, is_pub);
|
|
ctx->current_func_idx = fi;
|
|
ctx->current_ret_type = ret_type;
|
|
ctx->current_local_bytes = 0;
|
|
ctx->peak_local_bytes = 0;
|
|
|
|
spl_push_scope(ctx);
|
|
|
|
for (int i = 0; i < nparams; i++) {
|
|
spl_declare_var(ctx, pnames[i], ptypes[i], 0);
|
|
}
|
|
|
|
skip_nl(ctx);
|
|
if (peek(ctx)->type == TOK_L_BRACE) {
|
|
advance(ctx); /* { */
|
|
skip_nl(ctx);
|
|
|
|
spl_val_t alloc_addr = vec_size(ctx->prog.insns);
|
|
spl_emit(ctx, SPL_ALLOC, SPL_VOID, 0);
|
|
|
|
while (peek(ctx)->type != TOK_R_BRACE && peek(ctx)->type != TOK_EOF) {
|
|
spl_parse_stmt(ctx);
|
|
skip_nl(ctx);
|
|
}
|
|
|
|
/* Compute physical param slot count (1 slot = 8 bytes).
|
|
* Multi-slot struct params occupy ceil(size/8) slots; scalar params occupy 1 each. */
|
|
int total_phys_slots = 0;
|
|
for (int i = 0; i < nparams; i++) {
|
|
usize psz = spl_type_size(ptypes[i]);
|
|
total_phys_slots += (int)((psz + sizeof(spl_val_t) - 1) / sizeof(spl_val_t));
|
|
}
|
|
/* Ensure ALLOC provides enough space for multi-slot return value
|
|
* pre-placed at fp[0..nslots-1] (above saved_sp). */
|
|
if (ret_type && !spl_type_is_scalar(ret_type) &&
|
|
spl_type_size(ret_type) > sizeof(spl_val_t)) {
|
|
usize min_bytes = spl_type_slot_count(ret_type) * sizeof(spl_val_t);
|
|
if ((usize)ctx->peak_local_bytes < min_bytes)
|
|
ctx->peak_local_bytes = (int)min_bytes;
|
|
}
|
|
spl_patch(ctx, alloc_addr,
|
|
ctx->peak_local_bytes / (int)sizeof(spl_val_t) - total_phys_slots);
|
|
expect(ctx, TOK_R_BRACE);
|
|
}
|
|
|
|
spl_emit_defer_epilogue(ctx, ctx->scope_depth);
|
|
spl_pop_scope(ctx);
|
|
|
|
/* Store param_types for call-site type checking and multi-slot expansion */
|
|
{
|
|
spl_func_info_t *f = &vec_at(ctx->funcs, fi);
|
|
f->param_types = calloc(nparams, sizeof(spl_type_info_t *));
|
|
f->param_names = calloc(nparams, sizeof(char *));
|
|
for (int i = 0; i < nparams; i++) {
|
|
f->param_types[i] = ptypes[i];
|
|
f->param_names[i] = strdup(pnames[i]);
|
|
}
|
|
}
|
|
|
|
/* Epilogue return: for multi-slot returns, return address of fp[1]
|
|
* (fp[0] is a gap slot that absorbs the RET PUSH, data at fp[1..nslots]). */
|
|
if (ctx->current_ret_type && !spl_type_is_scalar(ctx->current_ret_type) &&
|
|
spl_type_size(ctx->current_ret_type) > sizeof(spl_val_t)) {
|
|
spl_emit(ctx, SPL_LADDR, SPL_PTR, (int)sizeof(spl_val_t));
|
|
spl_emit(ctx, SPL_RET, SPL_PTR, 0);
|
|
} else {
|
|
spl_emit(ctx, SPL_RET, SPL_VOID, 0);
|
|
}
|
|
spl_prog_end_func(&ctx->prog, fi);
|
|
ctx->current_func_idx = -1;
|
|
ctx->current_ret_type = NULL;
|
|
return fi;
|
|
}
|
|
|
|
static void parse_fn_decl(spl_comp_t *ctx, int is_extern, int is_pub) {
|
|
advance(ctx); /* fn */
|
|
skip_nl(ctx);
|
|
|
|
spl_tok_t *fname_tok = advance(ctx);
|
|
char fn_name[256];
|
|
spl_tok_copy_name(fname_tok, fn_name, sizeof(fn_name));
|
|
skip_nl(ctx);
|
|
expect(ctx, TOK_L_PAREN);
|
|
|
|
/* Parse parameters — collect names and types */
|
|
char pnames[MAX_PARAMS][256];
|
|
spl_type_info_t *ptypes[MAX_PARAMS];
|
|
int nparams = parse_params_decl(ctx, pnames, ptypes);
|
|
skip_nl(ctx);
|
|
|
|
/* Return type (default: void) */
|
|
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);
|
|
skip_nl(ctx);
|
|
}
|
|
|
|
/* Extern function: register as native, no body */
|
|
if (is_extern) {
|
|
spl_declare_func(ctx, fn_name, ret_type, nparams, 1, is_pub);
|
|
spl_ensure_native(ctx, fn_name);
|
|
if (peek(ctx)->type == TOK_SEMICOLON)
|
|
advance(ctx);
|
|
return;
|
|
}
|
|
|
|
/* Check for forward declaration (just semicolon, skip) */
|
|
if (peek(ctx)->type == TOK_SEMICOLON) {
|
|
advance(ctx);
|
|
return;
|
|
}
|
|
|
|
skip_nl(ctx);
|
|
|
|
/* Use shared helper for function body parsing */
|
|
parse_fn_body(ctx, fn_name, ret_type, nparams, pnames, ptypes, is_pub);
|
|
}
|
|
|
|
/* ============================================================
|
|
* Shared helper: parse a method declaration inside a type body.
|
|
* Used by both struct_body and enum_body parsers.
|
|
* fn name(params) ret-type { body }
|
|
* ============================================================ */
|
|
|
|
static void parse_method_decl(spl_comp_t *ctx, spl_type_info_t *container) {
|
|
advance(ctx); /* fn */
|
|
skip_nl(ctx);
|
|
spl_tok_t *mname_tok = advance(ctx);
|
|
char mname[256];
|
|
spl_tok_copy_name(mname_tok, mname, sizeof(mname));
|
|
/* Build qualified name: TypeName.method_name */
|
|
char qualified[512];
|
|
snprintf(qualified, sizeof(qualified), "%s.%s", container->name ? container->name : "anon",
|
|
mname);
|
|
|
|
skip_nl(ctx);
|
|
expect(ctx, TOK_L_PAREN);
|
|
|
|
/* Parse parameters using shared helper */
|
|
char pnames[MAX_PARAMS][256];
|
|
spl_type_info_t *ptypes[MAX_PARAMS];
|
|
int nparams = parse_params_decl(ctx, pnames, ptypes);
|
|
skip_nl(ctx);
|
|
|
|
/* Return type (default: void) */
|
|
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);
|
|
skip_nl(ctx);
|
|
}
|
|
|
|
int fi = parse_fn_body(ctx, qualified, ret_type, nparams, pnames, ptypes, 0);
|
|
|
|
spl_type_add_method(container, mname, fi);
|
|
}
|
|
|
|
/* ============================================================
|
|
* Parse type container body (shared for struct, union, enum)
|
|
*
|
|
* For struct/union: fields are added for identifiers
|
|
* For enum: variants are added for identifiers
|
|
*
|
|
* Body supports:
|
|
* var name: type; — field declarations (struct/union only)
|
|
* name: type, — field/variant declarations
|
|
* name: Type, — enum variant with data
|
|
* name, — simple enum variant
|
|
* type Name = ...; — nested type declarations
|
|
* fn name(...) type { } — methods
|
|
* ============================================================ */
|
|
|
|
/* Skip past a nested type declaration in pass 2 without re-parsing */
|
|
static void skip_type_decl(spl_comp_t *ctx) {
|
|
advance(ctx); /* type */
|
|
advance(ctx); /* name */
|
|
skip_nl(ctx);
|
|
expect(ctx, TOK_ASSIGN);
|
|
skip_nl(ctx);
|
|
|
|
spl_tok_type_t tt = peek(ctx)->type;
|
|
if (tt == KW_STRUCT || tt == KW_UNION || tt == KW_ENUM) {
|
|
advance(ctx);
|
|
skip_nl(ctx);
|
|
}
|
|
|
|
if (peek(ctx)->type == TOK_L_BRACE) {
|
|
int bd = 1;
|
|
advance(ctx);
|
|
while (bd > 0 && ctx->tok_idx < vec_size(ctx->toks)) {
|
|
if (peek(ctx)->type == TOK_L_BRACE)
|
|
bd++;
|
|
else if (peek(ctx)->type == TOK_R_BRACE)
|
|
bd--;
|
|
advance(ctx);
|
|
}
|
|
} else {
|
|
while (peek(ctx)->type != TOK_SEMICOLON && peek(ctx)->type != TOK_EOF)
|
|
advance(ctx);
|
|
}
|
|
|
|
skip_nl(ctx);
|
|
if (peek(ctx)->type == TOK_SEMICOLON)
|
|
advance(ctx);
|
|
}
|
|
|
|
static void parse_type_body(spl_comp_t *ctx, spl_type_info_t *container, int is_enum) {
|
|
if (peek(ctx)->type != TOK_L_BRACE)
|
|
return;
|
|
advance(ctx); /* { */
|
|
|
|
/* Push type onto namespace chain for short-name resolution inside body */
|
|
spl_ns_push(ctx, container);
|
|
|
|
/* === Pass 1: Parse all nested type declarations first ===
|
|
* This allows fields/variants to reference types defined later. */
|
|
{
|
|
usize saved = ctx->tok_idx;
|
|
int depth = 1;
|
|
while (depth > 0 && ctx->tok_idx < vec_size(ctx->toks)) {
|
|
spl_tok_type_t tt = peek(ctx)->type;
|
|
if (tt == TOK_L_BRACE) {
|
|
depth++;
|
|
advance(ctx);
|
|
} else if (tt == TOK_R_BRACE) {
|
|
depth--;
|
|
if (depth == 0)
|
|
break;
|
|
advance(ctx);
|
|
} else if (tt == KW_TYPE && depth == 1) {
|
|
parse_type_decl(ctx);
|
|
} else {
|
|
advance(ctx);
|
|
}
|
|
}
|
|
ctx->tok_idx = saved;
|
|
}
|
|
|
|
/* === Pre-register method names for forward references === */
|
|
if (container->name) {
|
|
usize saved = ctx->tok_idx;
|
|
int depth = 1;
|
|
while (depth > 0 && ctx->tok_idx < vec_size(ctx->toks)) {
|
|
spl_tok_type_t tt = peek(ctx)->type;
|
|
if (tt == TOK_L_BRACE) {
|
|
depth++;
|
|
advance(ctx);
|
|
} else if (tt == TOK_R_BRACE) {
|
|
depth--;
|
|
if (depth == 0)
|
|
break;
|
|
advance(ctx);
|
|
} else if (tt == KW_FN && depth == 1) {
|
|
advance(ctx); /* fn */
|
|
skip_nl(ctx);
|
|
spl_tok_t *name_tok = advance(ctx);
|
|
char mname[256];
|
|
spl_tok_copy_name(name_tok, mname, sizeof(mname));
|
|
char qualified[512];
|
|
snprintf(qualified, sizeof(qualified), "%s.%s", container->name, mname);
|
|
int fi = spl_declare_func(ctx, qualified, NULL, 0, 0, 0);
|
|
/* Register in type->methods immediately so ns_chain lookup works */
|
|
spl_type_add_method(container, mname, fi);
|
|
} else {
|
|
advance(ctx);
|
|
}
|
|
}
|
|
ctx->tok_idx = saved;
|
|
}
|
|
|
|
/* === Pass 2: Parse fields/variants and methods === */
|
|
{
|
|
int depth = 1;
|
|
while (depth > 0 && ctx->tok_idx < vec_size(ctx->toks)) {
|
|
skip_nl(ctx);
|
|
spl_tok_type_t tt = peek(ctx)->type;
|
|
if (tt == TOK_L_BRACE) {
|
|
depth++;
|
|
advance(ctx);
|
|
} else if (tt == TOK_R_BRACE) {
|
|
depth--;
|
|
if (depth == 0) {
|
|
advance(ctx);
|
|
break;
|
|
}
|
|
advance(ctx);
|
|
} else if (tt == KW_TYPE && depth == 1) {
|
|
skip_type_decl(ctx);
|
|
} else if (tt == KW_VAR && depth == 1 && !is_enum) {
|
|
/* var name: type; (struct/union only) */
|
|
advance(ctx); /* var */
|
|
skip_nl(ctx);
|
|
spl_tok_t *ftok = advance(ctx);
|
|
skip_nl(ctx);
|
|
if (peek(ctx)->type == TOK_COLON) {
|
|
advance(ctx); /* : */
|
|
skip_nl(ctx);
|
|
spl_type_info_t *ftype = spl_parse_type(ctx);
|
|
char fname[256];
|
|
spl_tok_copy_name(ftok, fname, sizeof(fname));
|
|
spl_type_add_field(container, fname, ftype);
|
|
}
|
|
skip_nl(ctx);
|
|
if (peek(ctx)->type == TOK_SEMICOLON || peek(ctx)->type == TOK_COMMA)
|
|
advance(ctx);
|
|
} else if (tt == TOK_IDENT && depth == 1) {
|
|
if (is_enum) {
|
|
/* Variant: Name or Name: Type */
|
|
spl_tok_t *vtok = advance(ctx);
|
|
skip_nl(ctx);
|
|
if (peek(ctx)->type == TOK_COLON) {
|
|
advance(ctx); /* : */
|
|
skip_nl(ctx);
|
|
spl_type_info_t *dtype = spl_parse_type(ctx);
|
|
char vname[256];
|
|
spl_tok_copy_name(vtok, vname, sizeof(vname));
|
|
spl_type_add_variant(container, vname, dtype);
|
|
} else {
|
|
char vname[256];
|
|
spl_tok_copy_name(vtok, vname, sizeof(vname));
|
|
spl_type_add_variant(container, vname, NULL);
|
|
}
|
|
} else {
|
|
/* Old-style field: name: type, */
|
|
spl_tok_t *ftok = advance(ctx);
|
|
skip_nl(ctx);
|
|
if (peek(ctx)->type == TOK_COLON) {
|
|
advance(ctx); /* : */
|
|
skip_nl(ctx);
|
|
spl_type_info_t *ftype = spl_parse_type(ctx);
|
|
char fname[256];
|
|
spl_tok_copy_name(ftok, fname, sizeof(fname));
|
|
spl_type_add_field(container, fname, ftype);
|
|
}
|
|
}
|
|
skip_nl(ctx);
|
|
if (peek(ctx)->type == TOK_SEMICOLON || peek(ctx)->type == TOK_COMMA)
|
|
advance(ctx);
|
|
} else if (tt == KW_FN && depth == 1) {
|
|
/* Compute layout before compiling methods so
|
|
* field offsets are correct during codegen */
|
|
spl_type_compute_layout(container);
|
|
parse_method_decl(ctx, container);
|
|
} else {
|
|
advance(ctx);
|
|
}
|
|
}
|
|
}
|
|
|
|
spl_ns_pop(ctx);
|
|
spl_type_compute_layout(container);
|
|
}
|
|
|
|
/* ============================================================
|
|
* Parse type declaration
|
|
* type Name = struct { ... };
|
|
* type Name = union { ... };
|
|
* type Name = enum { ... };
|
|
* type Name = ExistingType;
|
|
* ============================================================ */
|
|
|
|
void parse_type_decl(spl_comp_t *ctx) {
|
|
advance(ctx); /* type */
|
|
spl_tok_t *name_tok = advance(ctx);
|
|
char tname[256];
|
|
spl_tok_copy_name(name_tok, tname, sizeof(tname));
|
|
|
|
skip_nl(ctx);
|
|
expect(ctx, TOK_ASSIGN);
|
|
skip_nl(ctx);
|
|
|
|
/* Determine parent type if inside a type body (for nested type registration) */
|
|
spl_type_info_t *parent =
|
|
vec_size(ctx->ns_chain) > 0 ? vec_at(ctx->ns_chain, vec_size(ctx->ns_chain) - 1) : NULL;
|
|
|
|
if (peek(ctx)->type == KW_STRUCT) {
|
|
advance(ctx);
|
|
spl_type_info_t *st = spl_type_struct(tname);
|
|
/* Register type early to allow self-referential fields */
|
|
map_put(ctx->type_defs, strdup(tname), st);
|
|
if (parent)
|
|
map_put(parent->children, strdup(tname), st);
|
|
skip_nl(ctx);
|
|
parse_type_body(ctx, st, 0);
|
|
} else if (peek(ctx)->type == KW_UNION) {
|
|
advance(ctx);
|
|
spl_type_info_t *ut = spl_type_union(tname);
|
|
map_put(ctx->type_defs, strdup(tname), ut);
|
|
if (parent)
|
|
map_put(parent->children, strdup(tname), ut);
|
|
skip_nl(ctx);
|
|
parse_type_body(ctx, ut, 0);
|
|
} else if (peek(ctx)->type == KW_ENUM) {
|
|
advance(ctx);
|
|
spl_type_info_t *et = spl_type_enum(tname);
|
|
/* Register type early to allow self-referential variants */
|
|
map_put(ctx->type_defs, strdup(tname), et);
|
|
if (parent)
|
|
map_put(parent->children, strdup(tname), et);
|
|
skip_nl(ctx);
|
|
parse_type_body(ctx, et, 1);
|
|
} else if (peek(ctx)->type == TOK_IDENT ||
|
|
(peek(ctx)->type >= KW_AS && peek(ctx)->type <= KW_ANY)) {
|
|
spl_type_info_t *base = spl_parse_type(ctx);
|
|
if (base) {
|
|
spl_type_info_t *alias = spl_type_clone(base);
|
|
alias->name = strdup(tname);
|
|
alias->kind = TYPE_NAME;
|
|
map_put(ctx->type_defs, strdup(tname), alias);
|
|
}
|
|
}
|
|
|
|
skip_nl(ctx);
|
|
if (peek(ctx)->type == TOK_SEMICOLON)
|
|
advance(ctx);
|
|
}
|
|
|
|
/* ============================================================
|
|
* Parse top-level program
|
|
* ============================================================ */
|
|
|
|
void spl_parse_prog(spl_comp_t *ctx) {
|
|
while (peek(ctx)->type != TOK_EOF) {
|
|
skip_nl(ctx);
|
|
if (peek(ctx)->type == TOK_EOF)
|
|
break;
|
|
|
|
switch (peek(ctx)->type) {
|
|
case KW_FN:
|
|
parse_fn_decl(ctx, 0, 0);
|
|
break;
|
|
case KW_TYPE:
|
|
parse_type_decl(ctx);
|
|
break;
|
|
case KW_PUB: {
|
|
advance(ctx); /* skip pub */
|
|
skip_nl(ctx);
|
|
if (peek(ctx)->type == KW_FN)
|
|
parse_fn_decl(ctx, 0, 1);
|
|
else if (peek(ctx)->type == KW_TYPE)
|
|
parse_type_decl(ctx);
|
|
break;
|
|
}
|
|
case TOK_SHARP:
|
|
/* #[extern("vm")] fn ... */
|
|
advance(ctx); /* # */
|
|
if (peek(ctx)->type == TOK_L_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);
|
|
if (peek(ctx)->type == KW_FN)
|
|
parse_fn_decl(ctx, 1, 0);
|
|
break;
|
|
default: {
|
|
int prev = ctx->tok_idx; /* Safety: prevent infinite loop */
|
|
/* Try to parse as a statement */
|
|
spl_parse_stmt(ctx);
|
|
/* Safety: prevent infinite loop on unrecognized tokens */
|
|
if (ctx->tok_idx == prev)
|
|
advance(ctx);
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
}
|