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

stage1 重构代码 语法上重新实现解包语法

Browse Source
This commit is contained in:
zzy committed 2026-09-17 00:33:03 +08:00
1 parent 2a84b61e3d
commit 54425a50c7
10 files changed
+2660 -2457
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

gramma.pest
+86 -44
View File
@@ -1,4 +1,3 @@
// 顶层容器
Root = { SOI ~ ContainerMembers ~ EOI }
ContainerMembers = { ExternalHeaderList ~ (ContainerDeclaration ~ (COMMA | SEMICOLON)*)* }
@@ -22,7 +21,6 @@ ContainerDeclaration = {
ComptimeDecl = { KEYWORD_comptime ~ (Block | Expr) }
// 函数声明（可带 comptime 前缀，支持修饰符）
FnDecl = {
DeclHeader ~ KEYWORD_comptime? ~ KEYWORD_fn ~ IDENTIFIER ~ LPAREN ~ ParamDeclList ~ RPAREN ~ (COLON ~ TypeExpr)? ~ (SEMICOLON | Block)
}
@@ -30,35 +28,38 @@ FnDecl = {
ParamDeclList = { (ParamDecl ~ COMMA)* ~ (ParamDecl | DOT3)? }
ParamDecl = { HeaderList? ~ BindingIdent ~ COLON ~ TypeExpr }
// 变量声明（支持修饰符、元组解构与类型推导）
VarDecl = { DeclHeader ~ KEYWORD_var ~ VarBody }
ConstDecl = { DeclHeader ~ KEYWORD_const ~ VarBody }
VarBody = { VarPattern ~ (COLON ~ TypeExpr)? ~ (EQUAL ~ Expr)? }
VarBody = { Pattern ~ (COLON ~ TypeExpr)? ~ (EQUAL ~ Expr)? }
// 类型声明（支持修饰符）
TypeDecl = { DeclHeader ~ KEYWORD_type ~ IDENTIFIER ~ EQUAL ~ TypeExpr }
// 容器成员（支持修饰符）
MemberDecl = { DeclHeader ~ IDENTIFIER ~ (COLON ~ TypeExpr)? ~ (EQUAL ~ Expr)? }
// 变量声明左值模式（允许单独标识符）
VarPattern = { VarPatternAtom ~ (COMMA ~ VarPatternAtom)* }
VarPatternAtom = {
BindingIdent
| DOT ~ IDENTIFIER ~ (LBRACKET ~ VarPattern ~ RBRACKET)?
| LPAREN ~ VarPattern ~ RPAREN
}
// 匹配模式（不允许单独标识符）
MatchPattern = { MatchPatternAtom ~ (COMMA ~ MatchPatternAtom)* }
MatchPatternAtom = {
Pattern = { PatternAtom ~ (COMMA ~ PatternAtom)* }
PatternAtom = {
UNDERSCORE
| Expr
| DOT ~ IDENTIFIER ~ (LBRACKET ~ MatchPattern ~ RBRACKET)?
| LPAREN ~ MatchPattern ~ RPAREN
| AnonymousAggregatePattern
| DOT ~ IDENTIFIER ~ (LPAREN ~ Pattern? ~ RPAREN)?
| LPAREN ~ Pattern? ~ RPAREN
| PatternLiteral
| IDENTIFIER
}
AnonymousAggregatePattern = { DOT ~ LBRACE ~ PatternFieldList? ~ RBRACE }
PatternFieldList = { PatternField ~ ((COMMA | SEMICOLON) ~ PatternField)* ~ (COMMA | SEMICOLON)? }
PatternField = { DOT ~ IDENTIFIER ~ (EQUAL ~ PatternAtom)? }
PatternLiteral = {
FLOAT_LITERAL
| INTEGER_LITERAL
| CHAR_LITERAL
| STRING_LITERAL
| KEYWORD_true
| KEYWORD_false
| KEYWORD_null
}
// 块内语句
Block = { LBRACE ~ Statement* ~ Expr? ~ RBRACE }
Statement = {
RetStatement
@@ -70,7 +71,6 @@ Statement = {
| ComptimeStatement
}
// 声明语句（内部声明已包含 HeaderList）
DeclStatement = { (VarDecl | ConstDecl | TypeDecl) ~ (COMMA | SEMICOLON)? }
ExprStatement = { Expr ~ (COMMA | SEMICOLON)? }
ComptimeStatement = { ComptimeDecl ~ (COMMA | SEMICOLON)? }
@@ -80,11 +80,8 @@ BreakStatement = { KEYWORD_break ~ Label? ~ Expr? ~ (COMMA | SEMICOLON)? }
ContinueStatement = { KEYWORD_continue ~ Label? ~ (COMMA | SEMICOLON)? }
DeferStatement = { KEYWORD_defer ~ (Block | Statement) ~ (COMMA | SEMICOLON)? }
// 标签定义（$ 前缀）
Label = { DOLLAR ~ IDENTIFIER }
// ============ 表达式 ============
Expr = { AssignExpr }
AssignExpr = { BoolOrExpr ~ (AssignOp ~ AssignExpr)? }
@@ -125,8 +122,8 @@ PrimaryExpr = {
| AnonymousAggregateLiteral
| ArrayLiteral
| AnonymousArrayLiteral
| ComptimeCallExpr // @name(args)
| VariantConstruct
| ComptimeCallExpr
| IDENTIFIER
| Block
| IfExpr
@@ -140,26 +137,74 @@ PrimaryExpr = {
TupleLiteral = { ParenExpr }
// ============ 聚合字面量 ============
AggregateLiteral = { IDENTIFIER ~ LBRACE ~ StructInitList? ~ RBRACE }
AnonymousAggregateLiteral = { DOT ~ LBRACE ~ StructInitList? ~ RBRACE }
StructInitList = { StructInit ~ ((COMMA | SEMICOLON) ~ StructInit)* ~ (COMMA | SEMICOLON)? }
StructInit = { DOT ~ IDENTIFIER ~ (EQUAL ~ Expr)? }
// ============ 数组/切片字面量 ============
VariantConstruct = { DOT ~ IDENTIFIER ~ (LPAREN ~ ExprList? ~ RPAREN)? }
ArrayLiteral = {
// 切片字面量: []T[1,2,3]
// 数组字面量: [N]T[1,2,3] 或 [_]T[1,2,3]
LBRACKET ~ (INTEGER_LITERAL | UNDERSCORE)? ~ RBRACKET ~ TypeExpr ~ LBRACKET ~ ExprList? ~ RBRACKET
LBRACKET ~ RBRACKET ~ TypeExpr ~ LBRACKET ~ ExprList? ~ RBRACKET
| LBRACKET ~ (INTEGER_LITERAL | UNDERSCORE) ~ RBRACKET ~ TypeExpr ~ LBRACKET ~ ExprList? ~ RBRACKET
}
AnonymousArrayLiteral = {
DOT ~ LBRACKET ~ ExprList? ~ RBRACKET
}
// ============ 类型表达式（独立，限制语法） ============
ExprNoStruct = { AssignExprNoStruct }
AssignExprNoStruct = { BoolOrExprNoStruct ~ (AssignOp ~ AssignExprNoStruct)? }
BoolOrExprNoStruct = { BoolAndExprNoStruct ~ (BOOL_OR ~ BoolAndExprNoStruct)* }
BoolAndExprNoStruct = { BitOrExprNoStruct ~ (BOOL_AND ~ BitOrExprNoStruct)* }
BitOrExprNoStruct = { BitXorExprNoStruct ~ (PIPE ~ BitXorExprNoStruct)* }
BitXorExprNoStruct = { BitAndExprNoStruct ~ (CARET ~ BitAndExprNoStruct)* }
BitAndExprNoStruct = { CmpEqExprNoStruct ~ (AMPERSAND ~ CmpEqExprNoStruct)* }
CmpEqExprNoStruct = { CmpExprNoStruct ~ ((EQ_EQ | BANG_EQUAL) ~ CmpExprNoStruct)* }
CmpExprNoStruct = { ArrowExprNoStruct ~ ((L_ARROW_EQ | L_ARROW | R_ARROW_EQ | R_ARROW) ~ ArrowExprNoStruct)* }
ArrowExprNoStruct = { RangeExprNoStruct ~ ((THIN_R_ARROW | THIN_L_ARROW | FAT_D_ARROW | THIN_D_ARROW) ~ RangeExprNoStruct)* }
RangeExprNoStruct = { ShiftExprNoStruct ~ (DOT2 ~ ShiftExprNoStruct?)? }
ShiftExprNoStruct = { AddExprNoStruct ~ ((L_ARROW2_EQ | L_ARROW2 | R_ARROW2_EQ | R_ARROW2) ~ AddExprNoStruct)* }
AddExprNoStruct = { MulExprNoStruct ~ ((PLUS | MINUS) ~ MulExprNoStruct)* }
MulExprNoStruct = { CastExprNoStruct ~ ((ASTERISK | SLASH | PERCENT) ~ CastExprNoStruct)* }
CastExprNoStruct = { PrefixExprNoStruct ~ (KEYWORD_as ~ TypeExpr)* }
PrefixExprNoStruct = { PrefixOp* ~ PostfixExprNoStruct }
PostfixExprNoStruct = {
PrimaryExprNoStruct ~ (LPAREN ~ ExprList ~ RPAREN | DOT ~ IDENTIFIER | DOT ~ ASTERISK | LBRACKET ~ Expr ~ RBRACKET | LBRACKET ~ Expr ~ DOT2 ~ Expr? ~ RBRACKET)*
}
PrimaryExprNoStruct = {
FLOAT_LITERAL
| INTEGER_LITERAL
| CHAR_LITERAL
| STRING_LITERAL
| KEYWORD_true
| KEYWORD_false
| KEYWORD_null
| TupleLiteral
| AnonymousAggregateLiteral
| ArrayLiteral
| AnonymousArrayLiteral
| VariantConstruct
| ComptimeCallExpr
| IDENTIFIER
| Block
| IfExpr
| IfVarExpr
| WhileExpr
| LoopExpr
| ForExpr
| MatchExpr
| FnTypeExpr
}
VarBodyNoStruct = { Pattern ~ (COLON ~ TypeExpr)? ~ (EQUAL ~ ExprNoStruct)? }
TypeExpr = { HeaderList? ~ TypePrefixExpr }
TypePrefixExpr = { ASTERISK* ~ TypePostfixExpr }
@@ -208,26 +253,23 @@ FnTypeExpr = { KEYWORD_comptime? ~ KEYWORD_fn ~ LPAREN ~ TypeExprList? ~ RPAREN
AggregateTypeExpr = { (KEYWORD_struct | KEYWORD_union | KEYWORD_enum | KEYWORD_shape) ~ LBRACE ~ ContainerMembers ~ RBRACE }
// ============ 控制流 ============
IfExpr = { KEYWORD_if ~ Expr ~ Block ~ (KEYWORD_else ~ (Block | IfExpr | IfVarExpr))? }
IfVarExpr = { KEYWORD_if ~ VarBody ~ Block ~ (KEYWORD_else ~ (Block | IfExpr | IfVarExpr))? }
IfExpr = { KEYWORD_if ~ ExprNoStruct ~ Block ~ (KEYWORD_else ~ (Block | IfExpr | IfVarExpr))? }
IfVarExpr = { KEYWORD_if ~ KEYWORD_var ~ VarBodyNoStruct ~ Block ~ (KEYWORD_else ~ (Block | IfExpr | IfVarExpr))? }
WhileExpr = { Label? ~ KEYWORD_while ~ Expr ~ Block ~ (KEYWORD_else ~ (Block | Expr))? }
WhileExpr = { Label? ~ KEYWORD_while ~ ExprNoStruct ~ Block ~ (KEYWORD_else ~ (Block | Expr))? }
LoopExpr = { Label? ~ KEYWORD_loop ~ Block }
ForExpr = { Label? ~ KEYWORD_for ~ Expr ~ (COMMA ~ Expr)* ~ KEYWORD_as ~ BindingIdent ~ (COMMA ~ BindingIdent)* ~ Block ~ (KEYWORD_else ~ (Block | Expr))? }
ForExpr = { Label? ~ KEYWORD_for ~ Pattern ~ COLON ~ ExprNoStruct ~ (COMMA ~ ExprNoStruct)* ~ Block ~ (KEYWORD_else ~ (Block | Expr))? }
MatchExpr = { KEYWORD_match ~ Expr ~ LBRACE ~ MatchArm* ~ RBRACE }
MatchArm = { MatchPattern ~ (COMMA ~ MatchPattern)* ~ FAT_R_ARROW ~ Block }
MatchExpr = { KEYWORD_match ~ ExprNoStruct ~ LBRACE ~ MatchArm* ~ RBRACE }
MatchArm = { Pattern ~ FAT_R_ARROW ~ Block ~ COMMA? }
// ============ 其他表达式 ============
ComptimeCallExpr = { AT ~ IDENTIFIER ~ LPAREN ~ ExprList? ~ RPAREN }
ParenExpr = {
LPAREN ~ RPAREN
| LPAREN ~ Expr ~ COMMA ~ (Expr ~ COMMA)* ~ Expr? ~ RPAREN
| LPAREN ~ Expr ~ RPAREN
}
// ============ Token 定义 ============
AssignOp = { EQUAL | PLUS_EQ | MINUS_EQ | ASTERISK_EQ | SLASH_EQ | PERCENT_EQ | AMPERSAND_EQ | PIPE_EQ | CARET_EQ | L_ARROW2_EQ | R_ARROW2_EQ }
BOOL_OR = { PIPE ~ PIPE }
BOOL_AND = { AMPERSAND ~ AMPERSAND }
nob.c
+1
View File
@@ -27,6 +27,7 @@ static const char *splc0_srcs[] = {
"stage1/spl_tok.c",
"stage1/spl_lexer.c",
"stage1/spl_ast.c",
"stage1/spl_parser.c",
"stage1/spl_dumptree.c",
"stage1/spl_type.c",
"stage1/spl_lang_type.c",
stage1/spl_ast.c
+11 -2292
View File
File diff suppressed because it is too large. Load diff
stage1/spl_ast.h
+51 -78
View File
@@ -4,9 +4,6 @@
#include "../stage0/include/utils.h"
#include "spl_attr.h"
#include "spl_dbg.h"
#include "spl_diag.h"
#include "spl_lexer.h"
#include "spl_mpa.h"
#include "spl_tok.h"
/* clang-format off */
@@ -74,7 +71,9 @@
X(SPL_AST_PATTERN_UNDERSCORE, V0, pattern_underscore) \
X(SPL_AST_PATTERN_DOT, V0, pattern_dot) \
X(SPL_AST_PATTERN_TUPLE, V0, pattern_tuple) \
X(SPL_AST_PATTERN_LIST, V0, pattern_list)
X(SPL_AST_PATTERN_LIST, V0, pattern_list) \
X(SPL_AST_PATTERN_AGGREGATE, V0, pattern_aggregate) \
X(SPL_AST_PATTERN_FIELD, V0, pattern_field)
/* clang-format on */
typedef enum {
@@ -143,9 +142,6 @@ typedef enum {
SPL_UNARY_OP_DEREF,
} spl_unary_op_t;
const char *spl_binary_op_name(spl_binary_op_t op);
const char *spl_unary_op_name(spl_unary_op_t op);
struct spl_ast_node;
typedef struct spl_ast_node spl_ast_node_t;
typedef usize spl_ast_node_ref_t;
@@ -161,138 +157,113 @@ struct spl_ast_node {
spl_attr_id_vec_t attrs;
union {
/* SPL_AST_ROOT */
struct {
spl_ast_node_ref_t members; /* container_members 节点 */
spl_ast_node_ref_t members;
} root;
/* SPL_AST_CONTAINER_MEMBERS */
struct {
spl_ast_node_ref_vec_t external_tags; /* external_tag_item 节点列表 */
spl_ast_node_ref_vec_t external_tags;
spl_ast_node_ref_vec_t declarations;
} container_members;
/* SPL_AST_FN_DECL */
struct {
spl_ast_node_ref_vec_t tag_list; /* tag_item 节点 */
spl_ast_node_ref_vec_t tag_list;
bool is_comptime;
const char *name;
spl_ast_node_ref_vec_t params; /* param_decl 节点 */
spl_ast_node_ref_t ret_type; /* 类型节点，可空 */
spl_ast_node_ref_t body; /* 表达式节点，可空表示仅声明 */
spl_ast_node_ref_vec_t params;
spl_ast_node_ref_t ret_type;
spl_ast_node_ref_t body;
} fn_decl;
/* SPL_AST_VAR_DECL / SPL_AST_CONST_DECL */
struct {
spl_ast_node_ref_vec_t attr_list;
spl_ast_node_ref_t pattern; /* 模式节点 */
spl_ast_node_ref_t type; /* 类型节点，可空 */
spl_ast_node_ref_t init_expr; /* 表达式节点，可空 */
spl_ast_node_ref_t pattern;
spl_ast_node_ref_t type;
spl_ast_node_ref_t init_expr;
} var_const_decl;
/* SPL_AST_TYPE_DECL */
struct {
spl_ast_node_ref_vec_t attr_list;
const char *name;
spl_ast_node_ref_t type_expr; /* 类型节点 */
spl_ast_node_ref_t type_expr;
} type_decl;
/* SPL_AST_MEMBER_DECL */
struct {
spl_ast_node_ref_vec_t attr_list;
const char *name;
spl_ast_node_ref_t type_expr; /* 可空 */
spl_ast_node_ref_t init_expr; /* 可空 */
spl_ast_node_ref_t type_expr;
spl_ast_node_ref_t init_expr;
} member_decl;
/* SPL_AST_PARAM_DECL */
struct {
spl_ast_node_ref_vec_t attr_list;
const char *name;
spl_ast_node_ref_t type_expr;
} param_decl;
/* SPL_AST_TAG_ITEM / SPL_AST_EXTERNAL_TAG_ITEM */
struct {
bool is_at; /* true: @, false: # */
bool is_external; /* true: @!或#! */
bool is_at;
bool is_external;
const char *name;
spl_ast_node_ref_vec_t args;
} tag_item;
/* SPL_AST_COMPTIME_DECL */
struct {
spl_ast_node_ref_t body; /* block 或 expr 节点 */
spl_ast_node_ref_t body;
} comptime_decl;
/* SPL_AST_TOP_LEVEL_BLOCK */
struct {
spl_ast_node_ref_vec_t tag_list; /* tag_item 节点列表 */
spl_ast_node_ref_t block; /* 表达式节点block */
spl_ast_node_ref_vec_t tag_list;
spl_ast_node_ref_t block;
} top_level_block;
/* SPL_AST_STATEMENT_RET */
struct {
spl_ast_node_ref_t expr; /* 可空 */
spl_ast_node_ref_t expr;
} statement_ret;
/* SPL_AST_STATEMENT_BREAK */
struct {
spl_ast_node_ref_t label; /* 可空 */
spl_ast_node_ref_t expr; /* 可空 */
spl_ast_node_ref_t label;
spl_ast_node_ref_t expr;
} statement_break;
/* SPL_AST_STATEMENT_CONTINUE */
struct {
spl_ast_node_ref_t label; /* 可空 */
spl_ast_node_ref_t label;
} statement_continue;
/* SPL_AST_STATEMENT_DEFER */
struct {
spl_ast_node_ref_t body; /* block 或 statement 节点 */
spl_ast_node_ref_t body;
} statement_defer;
/* SPL_AST_STATEMENT_COMPTIME */
struct {
spl_ast_node_ref_t body; /* block 或 expr 节点 */
spl_ast_node_ref_t body;
} statement_comptime;
/* SPL_AST_STATEMENT_DECL */
struct {
spl_ast_node_ref_t decl; /* var/const/type 声明节点 */
spl_ast_node_ref_t decl;
} statement_decl;
/* SPL_AST_STATEMENT_EXPR */
struct {
spl_ast_node_ref_t expr;
} statement_expr;
/* SPL_AST_AGGREGATE_MEMBER */
struct {
spl_ast_node_ref_vec_t attr_list;
const char *name;
spl_ast_node_ref_t type_expr; /* 可空 */
spl_ast_node_ref_t init_expr; /* 可空 */
spl_ast_node_ref_t type_expr;
spl_ast_node_ref_t init_expr;
} aggregate_member;
/* SPL_AST_STRUCT_INIT_ITEM */
struct {
const char *name;
spl_ast_node_ref_t expr; /* 可空 */
spl_ast_node_ref_t expr;
} struct_init_item;
/* SPL_AST_MATCH_ARM */
struct {
spl_ast_node_ref_t pattern; /* 模式节点 */
spl_ast_node_ref_t body; /* block 或 expr 节点 */
spl_ast_node_ref_t pattern;
spl_ast_node_ref_t body;
} match_arm;
/* SPL_AST_LABEL */
struct {
const char *name;
} label;
/* ---- 表达式子类型 ---- */
spl_mpa_int_t int_lit;
spl_mpa_float_t float_lit;
struct {
@@ -305,7 +276,6 @@ struct spl_ast_node {
struct {
bool val;
} bool_lit;
/* null_lit, undefined_lit 无数据 */
struct {
const char *name;
@@ -356,13 +326,13 @@ struct spl_ast_node {
bool has_len;
usize len;
bool is_underscore_len;
spl_ast_node_ref_t elem_type; /* 可空 匿名时为 NULL */
spl_ast_node_ref_t elem_type;
spl_ast_node_ref_vec_t elems;
} array_literal;
struct {
const char *name; /* 可空 匿名时为 NULL */
spl_ast_node_ref_vec_t fields; /* struct_init_item 节点列表 */
const char *name;
spl_ast_node_ref_vec_t fields;
} aggregate_literal;
struct {
@@ -402,14 +372,14 @@ struct spl_ast_node {
struct {
spl_ast_node_ref_t label;
spl_ast_node_ref_vec_t iter_exprs;
spl_ast_node_ref_vec_t idents;
spl_ast_node_ref_t pattern;
spl_ast_node_ref_t body;
spl_ast_node_ref_t else_branch;
} for_expr;
struct {
spl_ast_node_ref_t matched_expr;
spl_ast_node_ref_vec_t arms; /* match_arm 节点 */
spl_ast_node_ref_vec_t arms;
} match_expr;
struct {
@@ -418,7 +388,6 @@ struct spl_ast_node {
spl_ast_node_ref_t ret_type;
} fn_type;
/* ---- 类型子类型 ---- */
struct {
const char *name;
spl_ast_node_ref_vec_t generic_args;
@@ -451,7 +420,7 @@ struct spl_ast_node {
struct {
const char *kind_name;
spl_ast_node_ref_vec_t members; /* aggregate_member 节点 */
spl_ast_node_ref_vec_t members;
} type_aggregate;
struct {
@@ -459,13 +428,10 @@ struct spl_ast_node {
spl_ast_node_ref_vec_t args;
} type_comptime_call;
/* ---- 模式子类型 ---- */
struct {
const char *name;
} pattern_ident;
/* pattern_underscore 无数据 */
struct {
const char *name;
spl_ast_node_ref_t sub_pattern;
@@ -474,10 +440,18 @@ struct spl_ast_node {
struct {
spl_ast_node_ref_vec_t patterns;
} pattern_tuple;
struct {
spl_ast_node_ref_vec_t patterns;
} pattern_list;
struct {
spl_ast_node_ref_vec_t fields;
} pattern_aggregate;
struct {
const char *name;
spl_ast_node_ref_t pattern;
} pattern_field;
};
};
@@ -488,8 +462,6 @@ typedef struct spl_ast {
spl_ast_node_vec_t node_buckets;
spl_ast_cstr_ref_vec_t str_buckets;
spl_ast_node_ref_t root;
spl_diag_t *diag;
} spl_ast_t;
typedef VEC(spl_ast_t *) spl_ast_ptr_vec_t;
@@ -497,9 +469,10 @@ void spl_ast_init(spl_ast_t *ast, const spl_tok_vec_t *tok_vec /*move*/);
void spl_ast_drop(spl_ast_t *ast);
spl_ast_node_t *spl_ast_node(spl_ast_t *ast, spl_ast_node_ref_t node);
const char *spl_ast_node_name(spl_ast_node_t *node);
void spl_ast_parse(spl_ast_t *ast);
void spl_ast_dump(spl_ast_t *ast, spl_ast_node_ref_t node);
const char *spl_ast_kind_name(spl_ast_node_kind_t kind);
const char *spl_binary_op_name(spl_binary_op_t op);
const char *spl_unary_op_name(spl_unary_op_t op);
#endif /* __SPL_AST_H__ */
stage1/spl_lang_type.c
+31 -8
View File
@@ -645,10 +645,24 @@ spl_type_id_t spl_def_get_type_id(spl_lang_type_t *type, spl_def_id_t def_id) {
return lang ? spl_lang_type_get_ir_type(type, lang) : 0;
}
void spl_lang_type_dump(spl_lang_type_t *type, spl_lang_type_id_t id) {
static void spl_lang_type_dump_rec(spl_lang_type_t *type, spl_lang_type_id_t id,
spl_lang_type_id_vec_t *stack) {
for (usize i = 0; i < vec_size(*stack); i++) {
if (vec_at(*stack, i) == id) {
printf("<rec#%zu>", id);
return;
}
}
if (vec_size(*stack) > 32) {
printf("...");
return;
}
vec_push(*stack, id);
spl_lang_type_node_t *n = spl_lang_type_node(type, id);
if (!n) {
printf("none");
vec_pop(*stack);
return;
}
switch (n->kind) {
@@ -691,7 +705,7 @@ void spl_lang_type_dump(spl_lang_type_t *type, spl_lang_type_id_t id) {
}
case SPL_LANG_TYPE_PTR:
printf("*");
spl_lang_type_dump(type, n->extra.elem);
spl_lang_type_dump_rec(type, n->extra.elem, stack);
break;
case SPL_LANG_TYPE_ARRAY: {
spl_type_node_t *tn = spl_type_node(&type->ir, n->ir_type_id);
@@ -700,31 +714,31 @@ void spl_lang_type_dump(spl_lang_type_t *type, spl_lang_type_id_t id) {
} break;
case SPL_LANG_TYPE_SLICE:
printf("[]");
spl_lang_type_dump(type, n->extra.elem);
spl_lang_type_dump_rec(type, n->extra.elem, stack);
break;
case SPL_LANG_TYPE_TUPLE:
printf("(");
for (usize i = 0; i < vec_size(n->extra.elems); i++) {
if (i)
printf(",");
spl_lang_type_dump(type, vec_at(n->extra.elems, i));
spl_lang_type_dump_rec(type, vec_at(n->extra.elems, i), stack);
}
printf(")");
break;
case SPL_LANG_TYPE_RANGE:
spl_lang_type_dump(type, n->extra.elem);
spl_lang_type_dump_rec(type, n->extra.elem, stack);
printf("..");
spl_lang_type_dump(type, n->extra.elem);
spl_lang_type_dump_rec(type, n->extra.elem, stack);
break;
case SPL_LANG_TYPE_FN:
printf("fn(");
for (usize i = 0; i < vec_size(n->extra.fn.params); i++) {
if (i)
printf(",");
spl_lang_type_dump(type, vec_at(n->extra.fn.params, i));
spl_lang_type_dump_rec(type, vec_at(n->extra.fn.params, i), stack);
}
printf(")");
spl_lang_type_dump(type, n->extra.fn.ret);
spl_lang_type_dump_rec(type, n->extra.fn.ret, stack);
break;
case SPL_LANG_TYPE_STRUCT:
case SPL_LANG_TYPE_UNION:
@@ -733,6 +747,15 @@ void spl_lang_type_dump(spl_lang_type_t *type, spl_lang_type_id_t id) {
spl_type_pure_dump(&type->ir, n->ir_type_id);
break;
}
vec_pop(*stack);
}
void spl_lang_type_dump(spl_lang_type_t *type, spl_lang_type_id_t id) {
spl_lang_type_id_vec_t stack;
vec_init(stack);
spl_lang_type_dump_rec(type, id, &stack);
vec_free(stack);
}
void spl_type_def_dump(spl_lang_type_t *type, spl_def_id_t id) {
stage1/spl_parser.c
+2284
View File
File diff suppressed because it is too large. Load diff
stage1/spl_parser.h
+21
View File
@@ -0,0 +1,21 @@
#ifndef __SPL_PARSER_H__
#define __SPL_PARSER_H__
#include "spl_ast.h"
#include "spl_diag.h"
typedef struct spl_parser {
spl_ast_t *ast;
spl_diag_t *diag;
usize pos;
int failed;
int stop_agg;
usize error_count;
} spl_parser_t;
void spl_parser_init(spl_parser_t *p);
void spl_parser_drop(spl_parser_t *p);
void spl_ast_parse(spl_parser_t *p, spl_ast_t *ast);
#endif /* __SPL_PARSER_H__ */
stage1/spl_sema.c
+136 -25
View File
@@ -833,6 +833,67 @@ static spl_lang_type_id_t sema_block_body(spl_sema_t *sema, spl_ast_node_t *n,
return r;
}
static void sema_bind_pattern(spl_sema_t *sema, spl_ast_node_ref_t pat_ref,
spl_lang_type_id_t matched_tid) {
if (!pat_ref)
return;
spl_ast_node_t *pat = spl_ast_node(sema->ast, pat_ref);
if (!pat)
return;
switch (pat->kind) {
case SPL_AST_PATTERN_IDENT: {
const char *name = pat->pattern_ident.name;
spl_symbol_t sym = {.name = name, .kind = SPL_SYMBOL_KIND_VAR};
sym.node = spl_def_alloc(sema->type, SPL_DEF_VAR, name, pat->dbg, sema->ast, pat_ref);
spl_def_resolve(sema->type, sym.node, matched_tid);
if (!spl_scope_insert(sema->scope, sema->scope->current_scope, sym)) {
spl_diag_error(sema->diag, SPL_DIAG_KIND_DUPLICATE_SYMBOL, &pat->dbg,
"duplicate symbol `%s`", name);
}
pat->resolved_def_id = sym.node;
pat->resolved_lang_tid = matched_tid;
return;
}
case SPL_AST_PATTERN_UNDERSCORE:
return;
case SPL_AST_PATTERN_TUPLE: {
spl_lang_type_node_t *tn = spl_lang_type_node(sema->type, matched_tid);
if (!tn || tn->kind != SPL_LANG_TYPE_TUPLE) {
spl_diag_error(sema->diag, SPL_DIAG_KIND_TYPE_MISMATCH, &pat->dbg,
"tuple pattern requires a tuple value");
return;
}
usize natoms = vec_size(pat->pattern_tuple.patterns);
usize nelems = vec_size(tn->extra.elems);
if (natoms != nelems) {
spl_diag_error(sema->diag, SPL_DIAG_KIND_TYPE_MISMATCH, &pat->dbg,
"tuple pattern arity mismatch (expected %zu, got %zu)", nelems, natoms);
return;
}
/* 递归 bind 会 push lang_table，tn 会悬垂；先把元素类型拷到本地 */
spl_lang_type_id_vec_t elem_tids;
vec_init(elem_tids);
for (usize i = 0; i < nelems; ++i)
vec_push(elem_tids, vec_at(tn->extra.elems, i));
for (usize i = 0; i < natoms; ++i)
sema_bind_pattern(sema, vec_at(pat->pattern_tuple.patterns, i), vec_at(elem_tids, i));
vec_free(elem_tids);
pat->resolved_lang_tid = matched_tid;
return;
}
case SPL_AST_PATTERN_DOT:
case SPL_AST_PATTERN_AGGREGATE:
default:
spl_diag_error(sema->diag, SPL_DIAG_KIND_UNHANDLED_NODE, &pat->dbg,
"pattern kind %d not yet supported", (int)pat->kind);
return;
}
}
static spl_lang_type_id_t sema_coerce(spl_sema_t *sema, spl_ast_node_ref_t ref,
spl_lang_type_id_t expected) {
if (!ref || !expected)
@@ -1014,23 +1075,34 @@ static spl_lang_type_id_t sema_coerce(spl_sema_t *sema, spl_ast_node_ref_t ref,
}
spl_type_id_t agg_ir = spl_lang_type_get_ir_type(sema->type, expected);
/* 先把每个字段的 ir type_id 拷到本地：循环里 sema_coerce 会 push
type_table，irn 会悬垂。 */
spl_type_id_vec_t field_ir_ids;
vec_init(field_ir_ids);
{
spl_type_node_t *irn = spl_type_node(&sema->type->ir, agg_ir);
if (!irn) {
spl_diag_error(sema->diag, SPL_DIAG_KIND_TYPE_MISMATCH, &n->dbg,
"aggregate type unresolved");
return 0;
}
for (usize i = 0; i < vec_size(irn->agg_type.variants); i++)
vec_push(field_ir_ids, vec_at(irn->agg_type.variants, i).type_id);
}
if (en->kind == SPL_LANG_TYPE_ENUM) {
if (vec_size(n->aggregate_literal.fields) != 1) {
spl_diag_error(sema->diag, SPL_DIAG_KIND_TYPE_MISMATCH, &n->dbg,
"enum literal must have exactly one variant");
vec_free(field_ir_ids);
return 0;
}
spl_ast_node_t *item = spl_ast_node(sema->ast, vec_at(n->aggregate_literal.fields, 0));
if (!item || item->kind != SPL_AST_STRUCT_INIT_ITEM) {
spl_diag_error(sema->diag, SPL_DIAG_KIND_TYPE_MISMATCH, &n->dbg,
"invalid enum literal");
vec_free(field_ir_ids);
return 0;
}
usize idx = 0;
@@ -1038,14 +1110,22 @@ static spl_lang_type_id_t sema_coerce(spl_sema_t *sema, spl_ast_node_ref_t ref,
&idx)) {
spl_diag_error(sema->diag, SPL_DIAG_KIND_NO_FIELD, &item->dbg, "no variant `%s`",
item->struct_init_item.name);
vec_free(field_ir_ids);
return 0;
}
usize field_tid_ir = vec_at(irn->agg_type.variants, idx).type_id;
if (idx >= vec_size(field_ir_ids)) {
vec_free(field_ir_ids);
return 0;
}
usize field_tid_ir = vec_at(field_ir_ids, idx);
spl_lang_type_id_t ft = sema_lang_from_ir(sema, field_tid_ir);
if (ft && item->struct_init_item.expr) {
if (!sema_coerce(sema, item->struct_init_item.expr, ft))
if (!sema_coerce(sema, item->struct_init_item.expr, ft)) {
vec_free(field_ir_ids);
return 0;
}
}
vec_free(field_ir_ids);
n->resolved_lang_tid = expected;
return expected;
}
@@ -1061,13 +1141,18 @@ static spl_lang_type_id_t sema_coerce(spl_sema_t *sema, spl_ast_node_ref_t ref,
item->struct_init_item.name);
continue;
}
usize field_tid_ir = vec_at(irn->agg_type.variants, idx).type_id;
if (idx >= vec_size(field_ir_ids))
continue;
usize field_tid_ir = vec_at(field_ir_ids, idx);
spl_lang_type_id_t ft = sema_lang_from_ir(sema, field_tid_ir);
if (ft && item->struct_init_item.expr) {
if (!sema_coerce(sema, item->struct_init_item.expr, ft))
if (!sema_coerce(sema, item->struct_init_item.expr, ft)) {
vec_free(field_ir_ids);
return 0;
}
}
}
vec_free(field_ir_ids);
n->resolved_lang_tid = expected;
return expected;
}
@@ -1671,11 +1756,23 @@ static spl_lang_type_id_t sema_infer(spl_sema_t *sema, spl_ast_node_ref_t ref) {
return 0;
}
spl_type_id_t agg_ir = spl_lang_type_get_ir_type(sema->type, agg);
/* 循环里 sema_coerce 会 push type_table，irn 会悬垂。
先把每个字段的 ir type_id 拷到本地。 */
spl_type_id_vec_t field_ir_ids;
vec_init(field_ir_ids);
{
spl_type_node_t *irn = spl_type_node(&sema->type->ir, agg_ir);
if (irn) {
if (!irn) {
n->resolved_lang_tid = agg;
return agg;
}
for (usize i = 0; i < vec_size(irn->agg_type.variants); i++)
vec_push(field_ir_ids, vec_at(irn->agg_type.variants, i).type_id);
}
for (usize i = 0; i < vec_size(n->aggregate_literal.fields); ++i) {
spl_ast_node_t *item =
spl_ast_node(sema->ast, vec_at(n->aggregate_literal.fields, i));
spl_ast_node_t *item = spl_ast_node(sema->ast, vec_at(n->aggregate_literal.fields, i));
if (!item || item->kind != SPL_AST_STRUCT_INIT_ITEM)
continue;
usize idx = 0;
@@ -1685,12 +1782,15 @@ static spl_lang_type_id_t sema_infer(spl_sema_t *sema, spl_ast_node_ref_t ref) {
item->struct_init_item.name);
continue;
}
usize field_tid_ir = vec_at(irn->agg_type.variants, idx).type_id;
if (idx >= vec_size(field_ir_ids))
continue;
usize field_tid_ir = vec_at(field_ir_ids, idx);
spl_lang_type_id_t ft = sema_lang_from_ir(sema, field_tid_ir);
if (ft && item->struct_init_item.expr)
sema_coerce(sema, item->struct_init_item.expr, ft);
}
}
vec_free(field_ir_ids);
n->resolved_lang_tid = agg;
return agg;
}
@@ -1793,28 +1893,39 @@ static spl_lang_type_id_t sema_infer(spl_sema_t *sema, spl_ast_node_ref_t ref) {
vec_push(sema->label_stack, lbl);
sema_push_scope(sema);
for (usize i = 0; i < vec_size(n->for_expr.iter_exprs); ++i) {
spl_lang_type_id_t container_tid = sema_infer(sema, vec_at(n->for_expr.iter_exprs, i));
usize nsrc = vec_size(n->for_expr.iter_exprs);
spl_ast_node_ref_t pat_ref = n->for_expr.pattern;
spl_ast_node_t *pat = pat_ref ? spl_ast_node(sema->ast, pat_ref) : NULL;
if (pat && pat->kind == SPL_AST_PATTERN_LIST) {
usize natoms = vec_size(pat->pattern_list.patterns);
for (usize i = 0; i < nsrc; ++i) {
spl_lang_type_id_t container_tid =
sema_infer(sema, vec_at(n->for_expr.iter_exprs, i));
spl_lang_type_id_t elem_tid = sema_for_elem_type(sema, container_tid);
if (i < vec_size(n->for_expr.idents)) {
spl_ast_node_t *id_node = spl_ast_node(sema->ast, vec_at(n->for_expr.idents, i));
if (id_node && id_node->kind == SPL_AST_EXPR_IDENT) {
const char *var_name = id_node->ident.name;
if (strcmp(var_name, "_") != 0) {
spl_symbol_t sym = {.name = var_name, .kind = SPL_SYMBOL_KIND_VAR};
sym.node = spl_def_alloc(sema->type, SPL_DEF_VAR, var_name, id_node->dbg,
sema->ast, vec_at(n->for_expr.idents, i));
spl_def_resolve(sema->type, sym.node, elem_tid);
if (!spl_scope_insert(sema->scope, sema->scope->current_scope, sym)) {
spl_diag_error(sema->diag, SPL_DIAG_KIND_DUPLICATE_SYMBOL,
&id_node->dbg, "duplicate symbol `%s`", var_name);
if (i < natoms)
sema_bind_pattern(sema, vec_at(pat->pattern_list.patterns, i), elem_tid);
}
id_node->resolved_def_id = sym.node;
id_node->resolved_lang_tid = elem_tid;
if (natoms > nsrc) {
spl_diag_error(sema->diag, SPL_DIAG_KIND_TYPE_MISMATCH, &pat->dbg,
"for: pattern has %zu atoms but only %zu sources", natoms, nsrc);
}
} else {
for (usize i = 0; i < nsrc; ++i) {
spl_lang_type_id_t container_tid =
sema_infer(sema, vec_at(n->for_expr.iter_exprs, i));
if (i == 0 && pat_ref) {
spl_lang_type_id_t elem_tid = sema_for_elem_type(sema, container_tid);
sema_bind_pattern(sema, pat_ref, elem_tid);
}
}
if (nsrc > 1 && pat_ref) {
spl_diag_error(sema->diag, SPL_DIAG_KIND_TYPE_MISMATCH, &pat->dbg,
"for: multiple sources require pattern list");
}
}
if (n->for_expr.body)
sema_infer(sema, n->for_expr.body);
if (n->for_expr.else_branch)
stage1/spl_type.c
+28 -6
View File
@@ -205,10 +205,23 @@ bool spl_type_is_subtype(const spl_type_t *type, spl_type_id_t sub, spl_type_id_
return false;
}
void spl_type_pure_dump(spl_type_t *type, spl_type_id_t id) {
static void spl_type_pure_dump_rec(spl_type_t *type, spl_type_id_t id, spl_type_id_vec_t *stack) {
for (usize i = 0; i < vec_size(*stack); i++) {
if (vec_at(*stack, i) == id) {
printf("<rec#%zu>", id);
return;
}
}
if (vec_size(*stack) > 32) {
printf("...");
return;
}
vec_push(*stack, id);
spl_type_node_t *n = spl_type_node(type, id);
if (!n) {
printf("none");
vec_pop(*stack);
return;
}
switch (n->kind) {
@@ -248,11 +261,11 @@ void spl_type_pure_dump(spl_type_t *type, spl_type_id_t id) {
break;
case SPL_TYPE_PTR:
printf("*");
spl_type_pure_dump(type, n->ptr_type.pointee);
spl_type_pure_dump_rec(type, n->ptr_type.pointee, stack);
break;
case SPL_TYPE_ARRAY:
printf("[%zu]", n->array_type.len);
spl_type_pure_dump(type, n->array_type.element);
spl_type_pure_dump_rec(type, n->array_type.element, stack);
break;
case SPL_TYPE_STRUCT:
case SPL_TYPE_UNION:
@@ -271,7 +284,7 @@ void spl_type_pure_dump(spl_type_t *type, spl_type_id_t id) {
printf(", ");
spl_type_member_t *m = &n->agg_type.variants.data[i];
printf("%s:", m->name ? m->name : "?");
spl_type_pure_dump(type, m->type_id);
spl_type_pure_dump_rec(type, m->type_id, stack);
}
printf("}");
break;
@@ -281,12 +294,21 @@ void spl_type_pure_dump(spl_type_t *type, spl_type_id_t id) {
for (usize i = 0; i < vec_size(n->fn_type.params); i++) {
if (i)
printf(",");
spl_type_pure_dump(type, vec_at(n->fn_type.params, i));
spl_type_pure_dump_rec(type, vec_at(n->fn_type.params, i), stack);
}
printf(")");
spl_type_pure_dump(type, n->fn_type.ret);
spl_type_pure_dump_rec(type, n->fn_type.ret, stack);
break;
}
vec_pop(*stack);
}
void spl_type_pure_dump(spl_type_t *type, spl_type_id_t id) {
spl_type_id_vec_t stack;
vec_init(stack);
spl_type_pure_dump_rec(type, id, &stack);
vec_free(stack);
}
spl_type_id_t spl_type_fn_param_tid(spl_type_t *type, spl_type_id_t fn_tid, usize idx) {
stage1/splc0.c
+11 -4
View File
@@ -15,6 +15,7 @@
#include "spl_diag.h"
#include "spl_ir2vm.h"
#include "spl_lexer.h"
#include "spl_parser.h"
#include "spl_sema.h"
#include "spl_tok.h"
@@ -141,8 +142,11 @@ static spl_ast_t *driver_load(void *user, const char *from_fname, const char *pa
spl_tok_vec_t toks = spl_lex(src, key);
spl_ast_t *ast = malloc(sizeof *ast);
spl_ast_init(ast, &toks);
ast->diag = diag;
spl_ast_parse(ast);
spl_parser_t parser;
spl_parser_init(&parser);
parser.diag = diag;
spl_ast_parse(&parser, ast);
spl_parser_drop(&parser);
vec_at(d->modules, idx).ast = ast;
vec_at(d->modules, idx).src = src;
@@ -161,7 +165,6 @@ static int compile_spl(const char *src, const char *fname, const char *outpath,
drv.main_path = strdup(fname);
drv.main_toks = spl_lex(src, drv.main_path);
spl_ast_init(&drv.main_ast, &drv.main_toks);
drv.main_ast.diag = &diag;
map_init(drv.path_to_idx, MAP_HASH_STR, MAP_CMP_STR);
vec_init(drv.modules);
@@ -190,7 +193,11 @@ static int compile_spl(const char *src, const char *fname, const char *outpath,
int sema_ok = 0, a2ir_ok = 0, irb_ok = 0, ir2vm_ok = 0;
spl_diag_set_stage(&diag, SPL_STAGE_PARSE);
spl_ast_parse(&drv.main_ast);
spl_parser_t parser;
spl_parser_init(&parser);
parser.diag = &diag;
spl_ast_parse(&parser, &drv.main_ast);
spl_parser_drop(&parser);
if (spl_diag_has_error(&diag) || spl_diag_has_fatal(&diag)) {
spl_diag_summary(&diag);
rc = 1;