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

stage1 人工重构ast.c

Browse Source
This commit is contained in:
zzy
2026-08-12 10:29:38 +08:00
parent 314100afbc
commit b0e6b406ac
22 changed files with 1619 additions and 2895 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3481
stage1/spl_ast.c
View File

File diff suppressed because it is too large Load Diff

437
stage1/spl_ast.h
View File

@@ -2,31 +2,133 @@
#define __SPL_AST_H__
#include "../stage0/include/utils.h"
#include "spl_dbg.h"
#include "spl_lexer.h"
#include "spl_tok.h"
typedef enum {
SPL_AST_CONTAINER_ITEM,
SPL_AST_FN_DECL,
SPL_AST_FN_DEFINE,
SPL_AST_TYPE_DECL,
SPL_AST_VAR_DECL,
SPL_AST_CONST_DECL,
SPL_AST_MEMBER_DECL,
SPL_AST__COMPTIME_STMT, /*不实现*/
SPL_AST__DIRECTIVE_BLOCK, /*不实现*/
/* clang-format off */
#define SPL_AST_KIND_TABLE \
X(SPL_AST_NONE, V0, none) \
X(SPL_AST_CONTAINER_ITEM, V0, container_item) \
X(SPL_AST_FN_DECL, V0, fn_decl) \
X(SPL_AST_FN_DEFINE, V0, fn_define) \
X(SPL_AST_TYPE_DECL, V0, type_decl) \
X(SPL_AST_VAR_DECL, V0, var_decl) \
X(SPL_AST_CONST_DECL, V0, const_decl) \
X(SPL_AST_MEMBER_DECL, V0, member_decl) \
X(SPL_AST__COMPTIME_STMT, V0, comptime_stmt) \
X(SPL_AST__DIRECTIVE_BLOCK, V0, directive_block) \
X(SPL_AST_PARAM_DECL, V0, param_decl) \
X(SPL_AST_ATTR_ITEM, V0, attr_item) \
X(SPL_AST_ARGG_INIT_ITEM, V0, argg_init_item) \
X(SPL_AST_IF_STATEMENT, V0, if_statement) \
X(SPL_AST_IFVAR_STATEMENT, V0, ifvar_statement) \
X(SPL_AST_WHILE_STATEMENT, V0, while_statement) \
X(SPL_AST_LOOP_STATEMENT, V0, loop_statement) \
X(SPL_AST_FOR_STATEMENT, V0, for_statement) \
X(SPL_AST_MATCH_STATEMENT, V0, match_statement) \
X(SPL_AST_RET_STATEMENT, V0, ret_statement) \
X(SPL_AST_BREAK_STATEMENT, V0, break_statement) \
X(SPL_AST_CONTINUE_STATEMENT, V0, continue_statement) \
X(SPL_AST_DEFER_STATEMENT, V0, defer_statement) \
X(SPL_AST_TRY_STATEMENT, V0, try_statement) \
X(SPL_AST_CATCH_STATEMENT, V0, catch_statement) \
X(SPL_AST_ERRDEFER_STATEMEMT, V0, errdefer_statement) \
X(SPL_AST_EXPR_STATEMENT, V0, expr_statement) \
X(SPL_AST_PACKED_EXPR, V0, packed_expr) \
X(SPL_AST_ASSIGN_EXPR, V0, assign_expr) \
X(SPL_AST_ASSIGN_ADD_EXPR, V0, assign_add_expr) \
X(SPL_AST_ASSIGN_SUB_EXPR, V0, assign_sub_expr) \
X(SPL_AST_ASSIGN_MUL_EXPR, V0, assign_mul_expr) \
X(SPL_AST_ASSIGN_DIV_EXPR, V0, assign_div_expr) \
X(SPL_AST_ASSIGN_MOD_EXPR, V0, assign_mod_expr) \
X(SPL_AST_ASSIGN_AND_EXPR, V0, assign_and_expr) \
X(SPL_AST_ASSIGN_OR_EXPR, V0, assign_or_expr) \
X(SPL_AST_ASSIGN_XOR_EXPR, V0, assign_xor_expr) \
X(SPL_AST_ASSIGN_LSHIFT_EXPR, V0, assign_lshift_expr) \
X(SPL_AST_ASSIGN_USHIFT_EXPR, V0, assign_ushift_expr) \
X(SPL_AST_BOOL_OR_EXPR, V0, bool_or_expr) \
X(SPL_AST_BOOL_AND_EXPR, V0, bool_and_expr) \
X(SPL_AST_BIT_OR_EXPR, V0, bit_or_expr) \
X(SPL_AST_BIT_XOR_EXPR, V0, bit_xor_expr) \
X(SPL_AST_BIT_AND_EXPR, V0, bit_and_expr) \
X(SPL_AST_CMP_EQ_EXPR, V0, cmp_eq_expr) \
X(SPL_AST_CMP_NE_EXPR, V0, cmp_ne_expr) \
X(SPL_AST_CMP_LE_EXPR, V0, cmp_le_expr) \
X(SPL_AST_CMP_GE_EXPR, V0, cmp_ge_expr) \
X(SPL_AST_CMP_LT_EXPR, V0, cmp_lt_expr) \
X(SPL_AST_CMP_GT_EXPR, V0, cmp_gt_expr) \
X(SPL_AST_RANGE_EXPR, V0, range_expr) \
X(SPL_AST_LSHIFT_EXPR, V0, lshift_expr) \
X(SPL_AST_RSHIFT_EXPR, V0, rshift_expr) \
X(SPL_AST_ADD_EXPR, V0, add_expr) \
X(SPL_AST_SUB_EXPR, V0, sub_expr) \
X(SPL_AST_MUL_EXPR, V0, mul_expr) \
X(SPL_AST_DIV_EXPR, V0, div_expr) \
X(SPL_AST_MOD_EXPR, V0, mod_expr) \
X(SPL_AST_MINUS_EXPR, V0, minus_expr) \
X(SPL_AST_NOT_EXPR, V0, not_expr) \
X(SPL_AST_BIT_NOT_EXPR, V0, bit_not_expr) \
X(SPL_AST_ADDRESS_EXPR, V0, address_expr) \
X(SPL_AST_CALL_EXPR, V0, call_expr) \
X(SPL_AST_FIELD_EXPR, V0, field_expr) \
X(SPL_AST_DEREF_EXPR, V0, deref_expr) \
X(SPL_AST_INDEX_EXPR, V0, index_expr) \
X(SPL_AST_SLICE_EXPR, V0, slice_expr) \
X(SPL_AST_AS_EXPR, V0, as_expr) \
X(SPL_AST_EXPR_INTEGER_LIT, V0, integer_lit) \
X(SPL_AST_EXPR_FLOAT_LIT, V0, float_lit) \
X(SPL_AST_EXPR_CHAR_LIT, V0, char_lit) \
X(SPL_AST_EXPR_STRING_LIT, V0, string_lit) \
X(SPL_AST_EXPR_TRUE, V0, true_lit) \
X(SPL_AST_EXPR_FALSE, V0, false_lit) \
X(SPL_AST_EXPR_NULL, V0, null_lit) \
X(SPL_AST_EXPR_UNDEFINED, V0, undefined_lit) \
X(SPL_AST_EXPR_IDENT, V0, ident_expr) \
X(SPL_AST_ARGGREGATE_INIT, V0, aggregate_init) \
X(SPL_AST_EXPR_EXPR, V0, expr_expr) \
X(SPL_AST_ARRAY_LIT, V0, array_lit) \
X(SPL_AST_BUILTIN_EXPR, V0, builtin_expr) \
X(SPL_AST_BLOCK_EXPR, V0, block_expr) \
X(SPL_AST_BASE_TYPE_FN, V0, base_type_fn) \
X(SPL_AST_BASE_TYPE_PATH, V0, base_type_path) \
X(SPL_AST_TYPE_POINTER, V0, type_pointer) \
X(SPL_AST_TYPE_ARRAY, V0, type_array) \
X(SPL_AST_TYPE_SLICE, V0, type_slice) \
X(SPL_AST_TYPE_STRUCT, V0, type_struct) \
X(SPL_AST_TYPE_UNION, V0, type_union) \
X(SPL_AST_TYPE_ENUM, V0, type_enum) \
X(SPL_AST_TYPE_VOID, V0, type_void) \
X(SPL_AST_TYPE_BOOL, V0, type_bool) \
X(SPL_AST_TYPE_OPAQUE, V0, type_opaque) \
X(SPL_AST_TYPE_I8, V0, type_i8) \
X(SPL_AST_TYPE_U8, V0, type_u8) \
X(SPL_AST_TYPE_I16, V0, type_i16) \
X(SPL_AST_TYPE_U16, V0, type_u16) \
X(SPL_AST_TYPE_I32, V0, type_i32) \
X(SPL_AST_TYPE_U32, V0, type_u32) \
X(SPL_AST_TYPE_I64, V0, type_i64) \
X(SPL_AST_TYPE_U64, V0, type_u64) \
X(SPL_AST_TYPE_ISIZE, V0, type_isize) \
X(SPL_AST_TYPE_USIZE, V0, type_usize) \
X(SPL_AST_TYPE__F32, V0, type_f32) \
X(SPL_AST_TYPE__F64, V0, type_f64) \
X(SPL_AST_TYPE_ANY, V0, type_any) \
X(SPL_AST_TYPE_IDENT, V0, type_ident) \
SPL_AST_BLOCK_ITEM,
SPL_AST_EXPR,
SPL_AST_TYPE_EXPR,
SPL_AST_ATTR_LIST,
/* clang-format on*/
typedef enum {
#ifdef X
#undef X
#endif
#define X(name, ...) name,
SPL_AST_KIND_TABLE
#undef X
SPL_AST_COUNT,
} spl_ast_node_kind_t;
typedef struct {
const char *fname;
int line;
int col;
} spl_ast_loc_t;
const char *spl_ast_kind_name(spl_ast_node_kind_t kind);
struct spl_ast_node;
typedef struct spl_ast_node spl_ast_node_t;
@@ -34,13 +136,12 @@ typedef usize spl_ast_node_ref_t;
typedef VEC(spl_ast_node_ref_t) spl_ast_node_ref_vec_t;
struct spl_ast_node {
spl_ast_node_kind_t kind;
spl_ast_loc_t loc;
spl_dbg_node_t dbg;
usize resolved_def_id;
union {
struct {
spl_ast_node_ref_vec_t attr_list; /* attr_item */
spl_ast_node_ref_t self; /* self */
spl_ast_node_ref_vec_t members; /* container_decl 列表 */
} container_item;
@@ -54,7 +155,7 @@ struct spl_ast_node {
const char *name;
spl_ast_node_ref_vec_t param_list; /* param_decl */
spl_ast_node_ref_t type_expr;
spl_ast_node_ref_vec_t block; /* block_item */
spl_ast_node_ref_vec_t block; /* 语句/尾表达式 */
} fn_decl;
struct {
spl_ast_node_ref_vec_t attr_list; /* attr_item */
@@ -77,149 +178,60 @@ struct spl_ast_node {
const char *name;
spl_ast_node_ref_t type_expr;
spl_ast_node_ref_t expr;
} var_decl;
struct {
spl_ast_node_ref_vec_t attr_list; /* attr_item */
spl_ast_node_ref_t type_expr;
const char *name;
spl_ast_node_ref_t expr;
} const_decl;
} var_const_decl;
/* 语句数据：kind 决定解释哪个成员 */
struct {
enum {
SPL_AST_IF_STATEMENT,
SPL_AST_IFVAR_STATEMENT,
SPL_AST_WHILE_STATEMENT,
SPL_AST_LOOP_STATEMENT,
SPL_AST_FOR_STATEMENT,
SPL_AST_MATCH_STATEMENT,
SPL_AST_RET_STATEMENT,
SPL_AST_BREAK_STATEMENT,
SPL_AST_CONTINUE_STATEMENT,
SPL_AST_DEFER_STATEMENT,
SPL_AST_VARDECL,
SPL_AST_CONSTDECL,
SPL_AST_TYPEDECL,
SPL_AST_EXPR_STATEMENT,
} kind;
union {
struct {
spl_ast_node_ref_t expr;
spl_ast_node_ref_vec_t if_block; /* block_item */
spl_ast_node_ref_vec_t else_block; /* block_item */
} if_statement;
struct {
spl_ast_node_ref_t packed_expr;
spl_ast_node_ref_vec_t if_block; /* block_item */
spl_ast_node_ref_vec_t else_block; /* block_item */
} ifvar_statement;
struct {
spl_ast_node_ref_t expr;
spl_ast_node_ref_vec_t while_block; /* block_item */
} while_statement;
struct {
spl_ast_node_ref_vec_t loop_block; /* block_item */
} loop_statement;
struct {
spl_ast_node_ref_vec_t expr_vec;
VEC(char *) ident_vec;
spl_ast_node_ref_vec_t block; /* block_item */
} for_statement;
struct {
spl_ast_node_ref_t expr;
spl_ast_node_ref_vec_t paced_exprs; /* packed_expr */
spl_ast_node_ref_vec_t match_block; /* block_item */
} match_statement;
struct {
spl_ast_node_ref_t expr;
} ret_statement;
struct {
} break_statement;
struct {
} continue_statement;
struct {
spl_ast_node_ref_vec_t block_or_statement; /* block_item/statement */
} defer_statement;
spl_ast_node_ref_t var_decl;
spl_ast_node_ref_t const_decl;
spl_ast_node_ref_t type_decl;
spl_ast_node_ref_t expr_statement;
};
} block_item;
spl_ast_node_ref_t expr;
spl_ast_node_ref_vec_t if_block; /* 语句 */
spl_ast_node_ref_vec_t else_block; /* 语句 */
} if_statement;
struct {
spl_ast_node_ref_t packed_expr;
spl_ast_node_ref_vec_t if_block; /* 语句 */
spl_ast_node_ref_vec_t else_block; /* 语句 */
} ifvar_statement;
struct {
spl_ast_node_ref_t expr;
spl_ast_node_ref_vec_t while_block; /* 语句 */
} while_statement;
struct {
spl_ast_node_ref_vec_t loop_block; /* 语句 */
} loop_statement;
struct {
spl_ast_node_ref_vec_t expr_vec;
VEC(char *) ident_vec;
spl_ast_node_ref_vec_t block; /* 语句 */
} for_statement;
struct {
spl_ast_node_ref_t expr;
spl_ast_node_ref_vec_t paced_exprs; /* packed_expr */
spl_ast_node_ref_vec_t match_block; /* 语句 */
} match_statement;
struct {
spl_ast_node_ref_t expr;
} ret_statement;
struct {
} break_statement;
struct {
} continue_statement;
struct {
spl_ast_node_ref_vec_t block_or_statement; /* 语句 */
} defer_statement;
struct {
const char *ident;
const char *bind_ident;
spl_ast_node_ref_t expr;
} packed_expr;
struct {
enum {
SPL_AST_ASSIGN_EXPR,
SPL_AST_ASSIGN_ADD_EXPR,
SPL_AST_ASSIGN_sUB_EXPR,
SPL_AST_ASSIGN_MUL_EXPR,
SPL_AST_ASSIGN_DIV_EXPR,
SPL_AST_ASSIGN_MOD_EXPR,
SPL_AST_ASSIGN_AND_EXPR,
SPL_AST_ASSIGN_OR_EXPR,
SPL_AST_ASSIGN_XOR_EXPR,
SPL_AST_ASSIGN_LSHIFT_EXPR,
SPL_AST_ASSIGN_USHIFT_EXPR,
SPL_AST_BOOLOR_EXPR,
SPL_AST_BOOLAND_EXPR,
SPL_AST_BITOR_EXPR,
SPL_AST_BITXOR_EXPR,
SPL_AST_BITAND_EXPR,
SPL_AST_CMPEQ_EXPR,
SPL_AST_CMPNE_EXPR,
SPL_AST_CMP_LE_EXPR,
SPL_AST_CMP_GE_EXPR,
SPL_AST_CMP_LT_EXPR,
SPL_AST_CMP_GT_EXPR,
SPL_AST_RANGE_EXPR,
SPL_AST_LSHIFT_EXPR,
SPL_AST_RSHIFT_EXPR,
SPL_AST_ADD_EXPR,
SPL_AST_SUB_EXPR,
SPL_AST_MUL_EXPR,
SPL_AST_DIV_EXPR,
SPL_AST_MOD_EXPR,
SPL_AST_PREFIX_EXPR, /* left ref node */
SPL_AST_POSTFIX_EXPR, /* left ref node */
SPL_AST_PRIMARY_EXPR, /* left ref node */
} op;
struct {
spl_ast_node_ref_t left;
spl_ast_node_ref_t right;
} op_expr;
} expr;
spl_ast_node_ref_t left;
spl_ast_node_ref_t right;
} op_expr;
struct {
enum {
SPL_AST_MINUS_EXPR,
SPL_AST_BANG_EXPR,
SPL_AST_TILDE_EXPR,
SPL_AST_AMPERSAND_EXPR,
SPL_AST_ASTERISK_EXPR,
} kind;
spl_ast_node_ref_t postfix_expr;
} prefix_expr;
struct {
enum {
SPL_AST_CALL_EXPR,
SPL_AST_FIELD_EXPR,
SPL_AST_DEREF_EXPR,
SPL_AST_INDEX_EXPR,
SPL_AST_SLICE_EXPR,
SPL_AST_AS_EXPR,
} kind;
spl_ast_node_ref_t primary_expr;
union {
spl_ast_node_ref_vec_t call_expr;
@@ -232,66 +244,53 @@ struct spl_ast_node {
spl_ast_node_ref_t type_expr;
};
} postfix_expr;
struct {
enum {
SPL_AST_INTEGER,
SPL_AST_FLOAT,
SPL_AST_CHAR_LIT,
SPL_AST_STRING_LIT,
SPL_AST_TRUE,
SPL_AST_FALSE,
SPL_AST_NULL,
SPL_AST_IDENT,
SPL_AST_ARGGREGATE_INIT,
SPL_AST_EXPR_EXPR,
SPL_AST_ARRAY_LIT,
SPL_AST_BUILTIN_EXPR,
SPL_AST_BLOCK_EXPR,
} kind;
union {
isize integer_expr;
double float_expr;
char char_lit_expr;
const char *string_lit_expr; /* parsed c string */
isize integer_expr;
double float_expr;
char char_lit_expr;
const char *string_lit_expr; /* parsed c string */
const char *ident;
struct {
const char *name;
spl_ast_node_ref_vec_t expr; /* aggregate_init_item */
} aggregate_init;
spl_ast_node_ref_t expr;
struct {
isize integer;
spl_ast_node_ref_t type_expr;
spl_ast_node_ref_vec_t expr_list;
} array_lit_expr;
struct {
const char *ident;
struct {
const char *name;
spl_ast_node_ref_vec_t expr; /* aggregate_init_item */
} aggregate_init;
spl_ast_node_ref_t expr;
struct {
isize integer;
spl_ast_node_ref_t type_expr;
spl_ast_node_ref_vec_t expr_list;
} array_lit_expr;
struct {
const char *ident;
spl_ast_node_ref_vec_t expr_list;
} builtin_expr;
spl_ast_node_ref_vec_t block_expr; /* block_item */
};
spl_ast_node_ref_vec_t expr_list;
} builtin_expr;
spl_ast_node_ref_vec_t block_expr; /* 语句 */
} primary_expr;
struct {
const char *ident;
spl_ast_node_ref_t expr;
} aggregate_init_item;
struct {
spl_ast_node_ref_vec_t type_prefixs; /* prefix_type */
spl_ast_node_ref_vec_t attr_list; /* attr_item */
enum {
SPL_AST_BASE_TYPE_FN,
SPL_AST_BASE_TYPE_PATH,
SPL_AST_TYPE_STRUCT,
SPL_AST_TYPE_UNION,
SPL_AST_TYPE_ENUM,
} kind;
const char *spl_base_type;
spl_ast_node_ref_vec_t attr_list; /* attr_item */
union {
spl_ast_node_ref_vec_t type_path; /* type_atom */
struct {
VEC(const char *) ident_vec;
} type_path;
struct {
spl_ast_node_ref_t element;
spl_ast_node_ref_t size;
} array_type;
struct {
spl_ast_node_ref_t element;
} slice_type;
struct {
spl_ast_node_ref_t pointee;
} pointer_type;
struct {
spl_ast_node_ref_vec_t param_list; /* param_decl */
spl_ast_node_ref_t type_expr;
@@ -299,42 +298,16 @@ struct spl_ast_node {
spl_ast_node_ref_vec_t aggregate_list; /* container_decl */
};
} type_expr;
struct {
/* ASTERISK = 1, [] = 2, else = 0*/
int pointer;
/* if array_size != 0 then pointer == 2 */
int array_size;
} prefix_type;
struct {
enum {
SPL_AST_TYPE_VOID,
SPL_AST_TYPE_BOOL,
SPL_AST_TYPE_I8,
SPL_AST_TYPE_U8,
SPL_AST_TYPE_I16,
SPL_AST_TYPE_U16,
SPL_AST_TYPE_I32,
SPL_AST_TYPE_U32,
SPL_AST_TYPE_I64,
SPL_AST_TYPE_U64,
SPL_AST_TYPE_ISIZE,
SPL_AST_TYPE_USIZE,
SPL_AST_TYPE__F32, /*不实现*/
SPL_AST_TYPE__F64, /*不实现*/
SPL_AST_TYPE_PTR,
SPL_AST_TYPE_ANY,
SPL_AST_TYPE_IDENT,
} kind;
const char *ident;
} type_atom;
};
};
typedef VEC(spl_ast_node_t) spl_ast_node_vec_t;
typedef VEC(char*)spl_ast_cstr_ref_vec_t;
typedef struct {
int parsed;
spl_tok_vec_t input;
spl_ast_node_vec_t buckets;
spl_ast_node_vec_t node_buckets;
spl_ast_cstr_ref_vec_t str_buckets;
spl_ast_node_ref_t root;
} spl_ast_t;

17
stage1/spl_dbg.h Normal file
View File

@@ -0,0 +1,17 @@
#ifndef __SPL_DBG_H__
#define __SPL_DBG_H__
#include "spl_tok.h"
typedef struct {
const char *dbg_name;
const char *fpath;
int line;
int col;
} spl_dbg_node_t;
#define SPL_FATAL(tok, fmt, ...) \
LOG_FATAL("fatal at %s:%zd:%zd " fmt, ((tok) && (tok)->fname ? (tok)->fname : "<null>"), \
((tok) ? (tok)->line : 0), ((tok) ? (tok)->col : 0), ##__VA_ARGS__)
#endif /* __SPL_DBG_H__ */

16
stage1/spl_ir.h
View File

@@ -2,8 +2,10 @@
#define __SPL_IR_H__
#include "../stage0/include/utils.h"
#include "spl_dbg.h"
#include "spl_type.h"
/* clang-format off */
#define SPL_IR_FN_TABLE \
X(arith.add, V0, SPL_IR_ARITH_ADD) \
@@ -91,7 +93,7 @@ typedef usize spl_ir_func_ref_t; /* 0 is error */
typedef struct {
spl_ir_kind_t kind;
usize src_ref; /* AST 节点 ref（调试行号用，ast2ir 注解） */
spl_dbg_node_t dbg;
union {
struct {
spl_type_id_t tid;
@@ -238,24 +240,12 @@ typedef struct {
} spl_ir_attr_t;
typedef VEC(spl_ir_attr_t) spl_ir_attr_vec_t;
/* 调试：函数内变量（名 → 内存位置）。alloca 节点或参数槽。 */
typedef struct {
const char *name; /* 指向 AST 字符串，不拥有 */
spl_ir_node_ref_t alloca; /* 非参数：alloca 节点 ref */
spl_type_id_t tid;
int is_param; /* 参数则用 param_idx */
usize param_idx;
usize offset; /* fp 字节偏移（LADDR imm 语义；ir2vm 填） */
} spl_ir_dbg_var_t;
typedef VEC(spl_ir_dbg_var_t) spl_ir_dbg_var_vec_t;
typedef struct {
const char *name;
spl_ir_attr_t attr;
spl_type_id_t fn_tid;
spl_ir_node_vec_t nodes;
spl_ir_node_ref_vec_t labels;
spl_ir_dbg_var_vec_t dbg_vars; /* 调试变量表（fp 偏移在 ir2vm 填） */
} spl_ir_func_t;
typedef VEC(spl_ir_func_t) spl_ir_func_vec_t;

0
stage1/spl_layout.c Normal file
View File

0
stage1/spl_layout.h Normal file
View File

6
stage1/spl_lexer.c
View File

@@ -1,4 +1,4 @@
/* spl_lexer.c — SPL lexical analyzer */
/* spl_lexer.c - SPL lexical analyzer */
#include "spl_lexer.h"
#include "spl_tok.h"
@@ -21,7 +21,7 @@ static spl_tok_type_t keyword_type(const char *ident, usize len) {
return TOK_IDENT;
}
/* Escape sequence decoder — returns the decoded character,
/* Escape sequence decoder - returns the decoded character,
* advances *s past the escape sequence, returns 0 on success,
* non-zero on error. */
int spl_decode_escape(const char **s, char *out) {
@@ -85,7 +85,7 @@ spl_tok_vec_t spl_lex(const char *source, const char *fname) {
const char *start = source + offset;
char c = *start;
/* Skip whitespace (but not newlines — emit TOK_ENDLINE) */
/* Skip whitespace (but not newlines - emit TOK_ENDLINE) */
if (c == ' ' || c == '\t' || c == '\r') {
offset++;
col++;

2
stage1/spl_lexer.h
View File

@@ -1,4 +1,4 @@
/* spl_lexer.h — 独立词法分析器 */
/* spl_lexer.h - 独立词法分析器 */
#ifndef __SPL_LEXER_H__
#define __SPL_LEXER_H__

57
stage1/spl_tok.h
View File

@@ -1,4 +1,4 @@
/* spl_tok.h — Token type definitions (extracted from spl_lexer.h) */
/* spl_tok.h - Token type definitions (extracted from spl_lexer.h) */
#ifndef __SPL_TOK_H__
#define __SPL_TOK_H__
@@ -6,31 +6,36 @@
/* clang-format off */
#define KEYWORD_TABLE \
X(as , KW_AS , SPL_V0) \
X(bool , KW_BOOL , SPL_V0) \
X(break , KW_BREAK , SPL_V0) \
X(comptime , KW_COMPTIME , SPL_V0) \
X(const , KW_CONST , SPL_V0) \
X(continue , KW_CONTINUE , SPL_V0) \
X(defer , KW_DEFER , SPL_V0) \
X(else , KW_ELSE , SPL_V0) \
X(enum , KW_ENUM , SPL_V0) \
X(false , KW_FALSE , SPL_V0) \
X(fn , KW_FN , SPL_V0) \
X(for , KW_FOR , SPL_V0) \
X(if , KW_IF , SPL_V0) \
X(loop , KW_LOOP , SPL_V0) \
X(match , KW_MATCH , SPL_V0) \
X(null , KW_NULL , SPL_V0) \
X(ret , KW_RET , SPL_V0) \
X(struct , KW_STRUCT , SPL_V0) \
X(true , KW_TRUE , SPL_V0) \
X(type , KW_TYPE , SPL_V0) \
X(union , KW_UNION , SPL_V0) \
X(var , KW_VAR , SPL_V0) \
X(void , KW_VOID , SPL_V0) \
X(while , KW_WHILE , SPL_V0) \
X(_ , KW_ANY , SPL_V0) \
X(as , KW_AS , SPL_V0) \
X(bool , KW_BOOL , SPL_V0) \
X(break , KW_BREAK , SPL_V0) \
X(catch , KW_CATCH , SPL_V0) \
X(comptime , KW_COMPTIME , SPL_V0) \
X(const , KW_CONST , SPL_V0) \
X(continue , KW_CONTINUE , SPL_V0) \
X(defer , KW_DEFER , SPL_V0) \
X(else , KW_ELSE , SPL_V0) \
X(enum , KW_ENUM , SPL_V0) \
X(errdefer , KW_ERRDEFER , SPL_V0) \
X(false , KW_FALSE , SPL_V0) \
X(fn , KW_FN , SPL_V0) \
X(for , KW_FOR , SPL_V0) \
X(if , KW_IF , SPL_V0) \
X(loop , KW_LOOP , SPL_V0) \
X(match , KW_MATCH , SPL_V0) \
X(null , KW_NULL , SPL_V0) \
X(ret , KW_RET , SPL_V0) \
X(shape , WK_SHAPE , SPL_V0) \
X(struct , KW_STRUCT , SPL_V0) \
X(true , KW_TRUE , SPL_V0) \
X(try , KW_TRY , SPL_V0) \
X(type , KW_TYPE , SPL_V0) \
X(undefined , KW_UNDEFINDED , SPL_V0) \
X(union , KW_UNION , SPL_V0) \
X(var , KW_VAR , SPL_V0) \
X(void , KW_VOID , SPL_V0) \
X(while , KW_WHILE , SPL_V0) \
X(_ , KW_ANY , SPL_V0) \
// KEYWORD_TABLE
#define TOKEN_TABLE \

1
stage1/spl_type.h
View File

@@ -58,7 +58,6 @@ typedef struct {
spl_def_id_t def_id;
spl_type_id_t type_id;
usize scope_id;
int is_const; /* var_def 是否为 const 声明（只读） */
} spl_var_def_t;
typedef VEC(spl_var_def_t) spl_var_def_vec_t;

292
stage1/splc0.c
View File

@@ -1,4 +1,4 @@
/* splc0.c — SPL compiler CLI (stage 1, 引导用)
/* splc0.c - SPL compiler CLI (stage 1, 引导用)
*
* splc0 --dump tokens|ast|all <file> dump 前端产物
* splc0 <in> <out> 编译 (阶段 B 实现)
@@ -73,79 +73,79 @@ static void dump_ast(const char *src, const char *fname) {
spl_ast_drop(&ast);
}
static void dump_sema(const char *src, const char *fname) {
spl_tok_vec_t toks = spl_lex(src, fname);
spl_ast_t ast;
spl_ast_init(&ast, &toks);
spl_ast_prase(&ast);
if (ast.parsed < 0) {
printf("parse failed, skip sema dump\n");
spl_ast_drop(&ast);
return;
}
spl_ast_valid(&ast);
spl_sema_t sema;
spl_sema_init(&sema);
sema.ast = &ast;
spl_sema_run(&sema);
spl_sema_check(&sema);
printf("Sema root_scope=%zu scopes=%zu errors=%d\n", sema.root_scope, sema.scopes.size,
sema.error_count);
for (usize i = 0; i < sema.scopes.size; i++) {
printf("scope[%zu] parent=%zu\n", i, sema.scopes.data[i].parent);
map_for(sema.scopes.data[i].symbols, mi) {
printf(" %s -> def#%zu\n", sema.scopes.data[i].symbols.data[mi].key,
sema.scopes.data[i].symbols.data[mi].val);
}
}
printf("TypeTable:\n");
for (usize i = 0; i < sema.type.type_table.size; i++) {
printf(" id#%zu type=", i);
spl_type_pure_dump(&sema.type, i);
printf("\n");
}
printf("DefTable:\n");
for (usize i = 0; i < sema.type.def_table.size; i++) {
printf(" def#%zu ", i);
spl_type_def_dump(&sema.type, i);
printf("\n");
}
spl_sema_drop(&sema);
spl_ast_drop(&ast);
}
// static void dump_sema(const char *src, const char *fname) {
// spl_tok_vec_t toks = spl_lex(src, fname);
// spl_ast_t ast;
// spl_ast_init(&ast, &toks);
// spl_ast_prase(&ast);
// if (ast.parsed < 0) {
// printf("parse failed, skip sema dump\n");
// spl_ast_drop(&ast);
// return;
// }
// spl_ast_valid(&ast);
// spl_sema_t sema;
// spl_sema_init(&sema);
// sema.ast = &ast;
// spl_sema_run(&sema);
// spl_sema_check(&sema);
// printf("Sema root_scope=%zu scopes=%zu errors=%d\n", sema.root_scope, sema.scopes.size,
// sema.error_count);
// for (usize i = 0; i < sema.scopes.size; i++) {
// printf("scope[%zu] parent=%zu\n", i, sema.scopes.data[i].parent);
// map_for(sema.scopes.data[i].symbols, mi) {
// printf(" %s -> def#%zu\n", sema.scopes.data[i].symbols.data[mi].key,
// sema.scopes.data[i].symbols.data[mi].val);
// }
// }
// printf("TypeTable:\n");
// for (usize i = 0; i < sema.type.type_table.size; i++) {
// printf(" id#%zu type=", i);
// spl_type_pure_dump(&sema.type, i);
// printf("\n");
// }
// printf("DefTable:\n");
// for (usize i = 0; i < sema.type.def_table.size; i++) {
// printf(" def#%zu ", i);
// spl_type_def_dump(&sema.type, i);
// printf("\n");
// }
// spl_sema_drop(&sema);
// spl_ast_drop(&ast);
// }
static void dump_ir(const char *src, const char *fname) {
spl_tok_vec_t toks = spl_lex(src, fname);
spl_ast_t ast;
spl_ast_init(&ast, &toks);
spl_ast_prase(&ast);
if (ast.parsed < 0) {
printf("parse failed, skip IR\n");
spl_ast_drop(&ast);
return;
}
spl_ast_valid(&ast);
spl_sema_t sema;
spl_sema_init(&sema);
sema.ast = &ast;
spl_sema_run(&sema);
spl_sema_check(&sema);
if (sema.error_count) {
printf("sema errors=%d, skip IR\n", sema.error_count);
spl_sema_drop(&sema);
spl_ast_drop(&ast);
return;
}
spl_ast2ir_t a2ir;
spl_ast2ir_init(&a2ir, &sema);
spl_ast2ir_run(&a2ir);
if (a2ir.err_count)
printf("ast2ir errors=%d\n", a2ir.err_count);
spl_ir_dump(&a2ir.ir, &sema.type);
spl_ast2ir_drop(&a2ir);
spl_sema_drop(&sema);
spl_ast_drop(&ast);
}
// static void dump_ir(const char *src, const char *fname) {
// spl_tok_vec_t toks = spl_lex(src, fname);
// spl_ast_t ast;
// spl_ast_init(&ast, &toks);
// spl_ast_prase(&ast);
// if (ast.parsed < 0) {
// printf("parse failed, skip IR\n");
// spl_ast_drop(&ast);
// return;
// }
// spl_ast_valid(&ast);
// spl_sema_t sema;
// spl_sema_init(&sema);
// sema.ast = &ast;
// spl_sema_run(&sema);
// spl_sema_check(&sema);
// if (sema.error_count) {
// printf("sema errors=%d, skip IR\n", sema.error_count);
// spl_sema_drop(&sema);
// spl_ast_drop(&ast);
// return;
// }
// spl_ast2ir_t a2ir;
// spl_ast2ir_init(&a2ir, &sema);
// spl_ast2ir_run(&a2ir);
// if (a2ir.err_count)
// printf("ast2ir errors=%d\n", a2ir.err_count);
// spl_ir_dump(&a2ir.ir, &sema.type);
// spl_ast2ir_drop(&a2ir);
// spl_sema_drop(&sema);
// spl_ast_drop(&ast);
// }
static int cmd_dump(const char *flags, const char *path) {
long len;
@@ -160,25 +160,14 @@ static int cmd_dump(const char *flags, const char *path) {
dump_tokens(src, path);
if (do_ast)
dump_ast(src, path);
if (do_sema)
dump_sema(src, path);
if (do_ir)
dump_ir(src, path);
// if (do_sema)
// dump_sema(src, path);
// if (do_ir)
// dump_ir(src, path);
free(src);
return 0;
}
static spl_ast_node_t *ast_node_at(spl_ast_t *ast, spl_ast_node_ref_t ref) {
if (ref == 0 || ref >= ast->buckets.size)
return NULL;
return &ast->buckets.data[ref];
}
static int ast_line(spl_ast_t *ast, spl_ast_node_ref_t ref) {
spl_ast_node_t *n = ast_node_at(ast, ref);
return n ? n->loc.line : 0;
}
/* -g：在 .sir 尾部追加 debug 段（文本：IR 行 + VAR 行），spl_cli -g 读取 */
static void gen_debug_map(const char *outpath, spl_ast_t *ast, const spl_ir_t *ir,
const spl_ir2vm_t *ir2vm) {
@@ -186,73 +175,74 @@ static void gen_debug_map(const char *outpath, spl_ast_t *ast, const spl_ir_t *i
if (!f)
return;
fprintf(f, "SPLDBG\n");
for (usize i = 0; i < ir2vm->fdbg.size; i++) {
const spl_ir2vm_fdbg_t *fd = &ir2vm->fdbg.data[i];
if (fd->fid >= ir->funcs.size)
continue;
const spl_ir_func_t *fn = &ir->funcs.data[fd->fid];
const char *fname = fn->name ? fn->name : "?";
for (usize ref = 1; ref < fd->node_first_ip.size; ref++) {
usize ip = fd->node_first_ip.data[ref];
if (ip == (usize)-1)
continue;
const spl_ir_node_t *n = &fn->nodes.data[ref];
fprintf(f, "IR %zu %zu %d %s\n", ip, ref, ast_line(ast, n->src_ref),
spl_ir_kind_name(n->kind));
}
for (usize j = 0; j < fn->dbg_vars.size; j++) {
const spl_ir_dbg_var_t *dv = &fn->dbg_vars.data[j];
if (!dv->name)
continue;
fprintf(f, "VAR %s %s %zu %zu %d\n", fname, dv->name, dv->offset, dv->tid,
dv->is_param);
}
}
// for (usize i = 0; i < ir2vm->fdbg.size; i++) {
// const spl_ir2vm_fdbg_t *fd = &ir2vm->fdbg.data[i];
// if (fd->fid >= ir->funcs.size)
// continue;
// const spl_ir_func_t *fn = &ir->funcs.data[fd->fid];
// const char *fname = fn->name ? fn->name : "?";
// for (usize ref = 1; ref < fd->node_first_ip.size; ref++) {
// usize ip = fd->node_first_ip.data[ref];
// if (ip == (usize)-1)
// continue;
// const spl_ir_node_t *n = &fn->nodes.data[ref];
// fprintf(f, "IR %zu %zu %d %s\n", ip, ref, ast_line(ast, n->src_ref),
// spl_ir_kind_name(n->kind));
// }
// for (usize j = 0; j < fn->dbg_vars.size; j++) {
// const spl_ir_dbg_var_t *dv = &fn->dbg_vars.data[j];
// if (!dv->name)
// continue;
// fprintf(f, "VAR %s %s %zu %zu %d\n", fname, dv->name, dv->offset, dv->tid,
// dv->is_param);
// }
// }
fclose(f);
}
static int compile_spl(const char *src, const char *fname, const char *outpath, int gen_debug) {
spl_tok_vec_t toks = spl_lex(src, fname);
spl_ast_t ast;
spl_ast_init(&ast, &toks);
spl_ast_prase(&ast);
if (ast.parsed < 0) {
printf("parse failed, no output\n");
spl_ast_drop(&ast);
return 1;
}
spl_ast_valid(&ast);
spl_sema_t sema;
spl_sema_init(&sema);
sema.ast = &ast;
spl_sema_run(&sema);
spl_sema_check(&sema);
if (sema.error_count) {
printf("sema errors=%d, no output\n", sema.error_count);
spl_sema_drop(&sema);
spl_ast_drop(&ast);
return 1;
}
spl_ast2ir_t a2ir;
spl_ast2ir_init(&a2ir, &sema);
spl_ast2ir_run(&a2ir);
if (a2ir.err_count) {
printf("ast2ir errors=%d, no output\n", a2ir.err_count);
spl_ast2ir_drop(&a2ir);
spl_sema_drop(&sema);
spl_ast_drop(&ast);
return 1;
}
spl_ir2vm_t ir2vm;
spl_ir2vm_init(&ir2vm, &a2ir.ir, &sema.type);
int rc = spl_ir2vm_run(&ir2vm, outpath);
if (gen_debug && rc == 0)
gen_debug_map(outpath, &ast, &a2ir.ir, &ir2vm);
spl_ir2vm_drop(&ir2vm);
spl_ast2ir_drop(&a2ir);
spl_sema_drop(&sema);
spl_ast_drop(&ast);
return rc ? 1 : 0;
// spl_tok_vec_t toks = spl_lex(src, fname);
// spl_ast_t ast;
// spl_ast_init(&ast, &toks);
// spl_ast_prase(&ast);
// if (ast.parsed < 0) {
// printf("parse failed, no output\n");
// spl_ast_drop(&ast);
// return 1;
// }
// spl_ast_valid(&ast);
// spl_sema_t sema;
// spl_sema_init(&sema);
// sema.ast = &ast;
// spl_sema_run(&sema);
// spl_sema_check(&sema);
// if (sema.error_count) {
// printf("sema errors=%d, no output\n", sema.error_count);
// spl_sema_drop(&sema);
// spl_ast_drop(&ast);
// return 1;
// }
// spl_ast2ir_t a2ir;
// spl_ast2ir_init(&a2ir, &sema);
// spl_ast2ir_run(&a2ir);
// if (a2ir.err_count) {
// printf("ast2ir errors=%d, no output\n", a2ir.err_count);
// spl_ast2ir_drop(&a2ir);
// spl_sema_drop(&sema);
// spl_ast_drop(&ast);
// return 1;
// }
// spl_ir2vm_t ir2vm;
// spl_ir2vm_init(&ir2vm, &a2ir.ir, &sema.type);
// int rc = spl_ir2vm_run(&ir2vm, outpath);
// if (gen_debug && rc == 0)
// gen_debug_map(outpath, &ast, &a2ir.ir, &ir2vm);
// spl_ir2vm_drop(&ir2vm);
// spl_ast2ir_drop(&a2ir);
// spl_sema_drop(&sema);
// spl_ast_drop(&ast);
// return rc ? 1 : 0;
return 0;
}
int main(int argc, char **argv) {