stage1 新的ast结构 使用pest描述语法
This commit is contained in:
1 parent
34df303b56
commit
98b35a98e3
12 files changed
+3344
-2168
No files matched your search
+384
@@ -0,0 +1,384 @@
|
||||
// 顶层容器
|
||||
Root = { SOI ~ ContainerMembers ~ EOI }
|
||||
ContainerMembers = { ExternalHeaderList ~ (ContainerDeclaration ~ (COMMA | SEMICOLON)*)* }
|
||||
|
||||
HeaderItem = { (AT | SHARP) ~ IDENTIFIER ~ (LPAREN ~ ExprList? ~ RPAREN)? }
|
||||
HeaderList = { HeaderItem* }
|
||||
DeclHeader = { HeaderList? ~ KEYWORD_pub? }
|
||||
ExternalHeaderItem = { (AT | SHARP) ~ BANG ~ IDENTIFIER ~ (LPAREN ~ ExprList? ~ RPAREN)? }
|
||||
ExternalHeaderList = { ExternalHeaderItem* }
|
||||
|
||||
BindingIdent = { IDENTIFIER | UNDERSCORE }
|
||||
|
||||
ContainerDeclaration = {
|
||||
FnDecl
|
||||
| VarDecl
|
||||
| ConstDecl
|
||||
| TypeDecl
|
||||
| MemberDecl
|
||||
| ComptimeDecl
|
||||
| HeaderList ~ Block
|
||||
}
|
||||
|
||||
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)
|
||||
}
|
||||
|
||||
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)? }
|
||||
|
||||
// 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 = {
|
||||
UNDERSCORE
|
||||
| Expr
|
||||
| DOT ~ IDENTIFIER ~ (LBRACKET ~ MatchPattern ~ RBRACKET)?
|
||||
| LPAREN ~ MatchPattern ~ RPAREN
|
||||
}
|
||||
|
||||
// 块内语句
|
||||
Block = { LBRACE ~ Statement* ~ Expr? ~ RBRACE }
|
||||
Statement = {
|
||||
RetStatement
|
||||
| BreakStatement
|
||||
| ContinueStatement
|
||||
| DeferStatement
|
||||
| DeclStatement
|
||||
| ExprStatement
|
||||
| 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)? }
|
||||
|
||||
RetStatement = { KEYWORD_ret ~ Expr? ~ (COMMA | SEMICOLON)? }
|
||||
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)? }
|
||||
BoolOrExpr = { BoolAndExpr ~ (BOOL_OR ~ BoolAndExpr)* }
|
||||
BoolAndExpr = { BitOrExpr ~ (BOOL_AND ~ BitOrExpr)* }
|
||||
BitOrExpr = { BitXorExpr ~ (PIPE ~ BitXorExpr)* }
|
||||
BitXorExpr = { BitAndExpr ~ (CARET ~ BitAndExpr)* }
|
||||
BitAndExpr = { CmpEqExpr ~ (AMPERSAND ~ CmpEqExpr)* }
|
||||
CmpEqExpr = { CmpExpr ~ ((EQ_EQ | BANG_EQUAL) ~ CmpExpr)* }
|
||||
CmpExpr = { ArrowExpr ~ ((L_ARROW_EQ | L_ARROW | R_ARROW_EQ | R_ARROW) ~ ArrowExpr)* }
|
||||
|
||||
ArrowExpr = { RangeExpr ~ ((THIN_R_ARROW | THIN_L_ARROW | FAT_D_ARROW | THIN_D_ARROW) ~ RangeExpr)* }
|
||||
|
||||
RangeExpr = { ShiftExpr ~ (DOT2 ~ ShiftExpr?)? }
|
||||
ShiftExpr = { AddExpr ~ ((L_ARROW2_EQ | L_ARROW2 | R_ARROW2_EQ | R_ARROW2) ~ AddExpr)* }
|
||||
AddExpr = { MulExpr ~ ((PLUS | MINUS) ~ MulExpr)* }
|
||||
MulExpr = { PrefixExpr ~ ((ASTERISK | SLASH | PERCENT) ~ PrefixExpr)* }
|
||||
PrefixExpr = { PrefixOp* ~ PostfixExpr }
|
||||
PrefixOp = { MINUS | BANG | TILDE | AMPERSAND }
|
||||
|
||||
PostfixExpr = {
|
||||
PrimaryExpr ~ (LPAREN ~ ExprList ~ RPAREN | DOT ~ IDENTIFIER | DOT ~ ASTERISK | LBRACKET ~ Expr ~ RBRACKET | LBRACKET ~ Expr ~ DOT2 ~ Expr? ~ RBRACKET | KEYWORD_as ~ TypeExpr)*
|
||||
}
|
||||
|
||||
ExprList = { (Expr ~ COMMA)* ~ Expr? }
|
||||
|
||||
PrimaryExpr = {
|
||||
FLOAT_LITERAL
|
||||
| INTEGER_LITERAL
|
||||
| CHAR_LITERAL
|
||||
| STRING_LITERAL
|
||||
| KEYWORD_true
|
||||
| KEYWORD_false
|
||||
| KEYWORD_null
|
||||
| TupleLiteral
|
||||
| AggregateLiteral
|
||||
| AnonymousAggregateLiteral
|
||||
| ArrayLiteral
|
||||
| AnonymousArrayLiteral
|
||||
| ComptimeCallExpr // @name(args)
|
||||
|
||||
| IDENTIFIER
|
||||
| Block
|
||||
| IfExpr
|
||||
| IfVarExpr
|
||||
| WhileExpr
|
||||
| LoopExpr
|
||||
| ForExpr
|
||||
| MatchExpr
|
||||
| FnTypeExpr
|
||||
}
|
||||
|
||||
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 ============
|
||||
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
|
||||
}
|
||||
|
||||
AnonymousArrayLiteral = {
|
||||
DOT ~ LBRACKET ~ ExprList? ~ RBRACKET
|
||||
}
|
||||
|
||||
// ============ 类型表达式(独立,限制语法) ============
|
||||
TypeExpr = { HeaderList? ~ TypePrefixExpr }
|
||||
|
||||
TypePrefixExpr = { ASTERISK* ~ TypePostfixExpr }
|
||||
|
||||
TypePostfixExpr = {
|
||||
TypePrimary ~ (LPAREN ~ TypeExprList ~ RPAREN | DOT ~ IDENTIFIER | LBRACKET ~ TypeExpr ~ RBRACKET | KEYWORD_as ~ TypeExpr)*
|
||||
}
|
||||
|
||||
TypePrimary = {
|
||||
IDENTIFIER
|
||||
| KEYWORD_i8
|
||||
| KEYWORD_u8
|
||||
| KEYWORD_i16
|
||||
| KEYWORD_u16
|
||||
| KEYWORD_i32
|
||||
| KEYWORD_u32
|
||||
| KEYWORD_i64
|
||||
| KEYWORD_u64
|
||||
| KEYWORD_f32
|
||||
| KEYWORD_f64
|
||||
| KEYWORD_isize
|
||||
| KEYWORD_usize
|
||||
| KEYWORD_bool
|
||||
| KEYWORD_void
|
||||
| KEYWORD_ptr
|
||||
| FnTypeExpr
|
||||
| AggregateTypeExpr
|
||||
| TypeParenExpr
|
||||
| ArrayTypeExpr
|
||||
}
|
||||
|
||||
ArrayTypeExpr = {
|
||||
LBRACKET ~ RBRACKET ~ TypeExpr
|
||||
| LBRACKET ~ (INTEGER_LITERAL | UNDERSCORE) ~ RBRACKET ~ TypeExpr
|
||||
}
|
||||
|
||||
TypeParenExpr = {
|
||||
LPAREN ~ RPAREN
|
||||
| LPAREN ~ TypeExpr ~ COMMA ~ (TypeExpr ~ COMMA)* ~ TypeExpr? ~ RPAREN
|
||||
| LPAREN ~ TypeExpr ~ RPAREN
|
||||
}
|
||||
|
||||
TypeExprList = { (TypeExpr ~ COMMA)* ~ TypeExpr? }
|
||||
|
||||
FnTypeExpr = { KEYWORD_comptime? ~ KEYWORD_fn ~ LPAREN ~ TypeExprList? ~ RPAREN ~ (COLON ~ TypeExpr)? }
|
||||
|
||||
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))? }
|
||||
|
||||
WhileExpr = { Label? ~ KEYWORD_while ~ Expr ~ 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))? }
|
||||
|
||||
MatchExpr = { KEYWORD_match ~ Expr ~ LBRACE ~ MatchArm* ~ RBRACE }
|
||||
MatchArm = { MatchPattern ~ (COMMA ~ MatchPattern)* ~ FAT_R_ARROW ~ Block }
|
||||
|
||||
// ============ 其他表达式 ============
|
||||
ComptimeCallExpr = { AT ~ IDENTIFIER ~ LPAREN ~ ExprList? ~ RPAREN }
|
||||
ParenExpr = {
|
||||
|
||||
| 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 }
|
||||
EQ_EQ = { EQUAL ~ EQUAL }
|
||||
BANG_EQUAL = { BANG ~ EQUAL }
|
||||
|
||||
keyword_suffix = _{ !(ASCII_ALPHANUMERIC | "_") }
|
||||
KEYWORD_fn = @{ "fn" ~ keyword_suffix }
|
||||
KEYWORD_type = @{ "type" ~ keyword_suffix }
|
||||
KEYWORD_pub = @{ "pub" ~ keyword_suffix }
|
||||
KEYWORD_var = @{ "var" ~ keyword_suffix }
|
||||
KEYWORD_const = @{ "const" ~ keyword_suffix }
|
||||
KEYWORD_if = @{ "if" ~ keyword_suffix }
|
||||
KEYWORD_else = @{ "else" ~ keyword_suffix }
|
||||
KEYWORD_while = @{ "while" ~ keyword_suffix }
|
||||
KEYWORD_loop = @{ "loop" ~ keyword_suffix }
|
||||
KEYWORD_for = @{ "for" ~ keyword_suffix }
|
||||
KEYWORD_as = @{ "as" ~ keyword_suffix }
|
||||
KEYWORD_match = @{ "match" ~ keyword_suffix }
|
||||
KEYWORD_ret = @{ "ret" ~ keyword_suffix }
|
||||
KEYWORD_break = @{ "break" ~ keyword_suffix }
|
||||
KEYWORD_continue = @{ "continue" ~ keyword_suffix }
|
||||
KEYWORD_defer = @{ "defer" ~ keyword_suffix }
|
||||
KEYWORD_shape = @{ "shape" ~ keyword_suffix }
|
||||
KEYWORD_struct = @{ "struct" ~ keyword_suffix }
|
||||
KEYWORD_union = @{ "union" ~ keyword_suffix }
|
||||
KEYWORD_enum = @{ "enum" ~ keyword_suffix }
|
||||
KEYWORD_comptime = @{ "comptime" ~ keyword_suffix }
|
||||
KEYWORD_void = @{ "void" ~ keyword_suffix }
|
||||
KEYWORD_bool = @{ "bool" ~ keyword_suffix }
|
||||
KEYWORD_isize = @{ "isize" ~ keyword_suffix }
|
||||
KEYWORD_usize = @{ "usize" ~ keyword_suffix }
|
||||
KEYWORD_ptr = @{ "ptr" ~ keyword_suffix }
|
||||
KEYWORD_true = @{ "true" ~ keyword_suffix }
|
||||
KEYWORD_false = @{ "false" ~ keyword_suffix }
|
||||
KEYWORD_null = @{ "null" ~ keyword_suffix }
|
||||
KEYWORD_undefined = @{ "undefined" ~ keyword_suffix }
|
||||
KEYWORD_i8 = @{ "i8" ~ keyword_suffix }
|
||||
KEYWORD_u8 = @{ "u8" ~ keyword_suffix }
|
||||
KEYWORD_i16 = @{ "i16" ~ keyword_suffix }
|
||||
KEYWORD_u16 = @{ "u16" ~ keyword_suffix }
|
||||
KEYWORD_i32 = @{ "i32" ~ keyword_suffix }
|
||||
KEYWORD_u32 = @{ "u32" ~ keyword_suffix }
|
||||
KEYWORD_i64 = @{ "i64" ~ keyword_suffix }
|
||||
KEYWORD_u64 = @{ "u64" ~ keyword_suffix }
|
||||
KEYWORD_f32 = @{ "f32" ~ keyword_suffix }
|
||||
KEYWORD_f64 = @{ "f64" ~ keyword_suffix }
|
||||
|
||||
FAT_D_ARROW = { "<=>" }
|
||||
THIN_D_ARROW = { "<->" }
|
||||
L_ARROW2_EQ = { "<<=" }
|
||||
R_ARROW2_EQ = { ">>=" }
|
||||
DOT3 = { "..." }
|
||||
THIN_R_ARROW = { "<-" }
|
||||
THIN_L_ARROW = { "->" }
|
||||
L_ARROW_EQ = { "<=" }
|
||||
R_ARROW_EQ = { ">=" }
|
||||
L_ARROW2 = { "<<" }
|
||||
R_ARROW2 = { ">>" }
|
||||
FAT_R_ARROW = { "=>" }
|
||||
PLUS_EQ = { "+=" }
|
||||
MINUS_EQ = { "-=" }
|
||||
ASTERISK_EQ = { "*=" }
|
||||
SLASH_EQ = { "/=" }
|
||||
PERCENT_EQ = { "%=" }
|
||||
AMPERSAND_EQ = { "&=" }
|
||||
PIPE_EQ = { "|=" }
|
||||
CARET_EQ = { "^=" }
|
||||
DOT2 = { ".." }
|
||||
LPAREN = { "(" }
|
||||
RPAREN = { ")" }
|
||||
LBRACKET = { "[" }
|
||||
RBRACKET = { "]" }
|
||||
LBRACE = { "{" }
|
||||
RBRACE = { "}" }
|
||||
COMMA = { "," }
|
||||
SEMICOLON = { ";" }
|
||||
COLON = { ":" }
|
||||
DOT = { "." }
|
||||
AT = { "@" }
|
||||
SHARP = { "#" }
|
||||
DOLLAR = { "$" }
|
||||
EQUAL = { "=" }
|
||||
PLUS = { "+" }
|
||||
MINUS = { "-" }
|
||||
ASTERISK = { "*" }
|
||||
SLASH = { "/" }
|
||||
PERCENT = { "%" }
|
||||
AMPERSAND = { "&" }
|
||||
PIPE = { "|" }
|
||||
CARET = { "^" }
|
||||
TILDE = { "~" }
|
||||
BANG = { "!" }
|
||||
L_ARROW = { "<" }
|
||||
R_ARROW = { ">" }
|
||||
UNDERSCORE = { "_" }
|
||||
|
||||
keyword = _{
|
||||
KEYWORD_fn
|
||||
| KEYWORD_type
|
||||
| KEYWORD_var
|
||||
| KEYWORD_const
|
||||
| KEYWORD_if
|
||||
| KEYWORD_else
|
||||
| KEYWORD_while
|
||||
| KEYWORD_loop
|
||||
| KEYWORD_for
|
||||
| KEYWORD_as
|
||||
| KEYWORD_match
|
||||
| KEYWORD_ret
|
||||
| KEYWORD_break
|
||||
| KEYWORD_continue
|
||||
| KEYWORD_defer
|
||||
| KEYWORD_shape
|
||||
| KEYWORD_struct
|
||||
| KEYWORD_union
|
||||
| KEYWORD_enum
|
||||
| KEYWORD_comptime
|
||||
| KEYWORD_void
|
||||
| KEYWORD_bool
|
||||
| KEYWORD_true
|
||||
| KEYWORD_false
|
||||
| KEYWORD_null
|
||||
| KEYWORD_undefined
|
||||
| KEYWORD_i8
|
||||
| KEYWORD_u8
|
||||
| KEYWORD_i16
|
||||
| KEYWORD_u16
|
||||
| KEYWORD_i32
|
||||
| KEYWORD_u32
|
||||
| KEYWORD_i64
|
||||
| KEYWORD_u64
|
||||
| KEYWORD_f32
|
||||
| KEYWORD_f64
|
||||
| KEYWORD_isize
|
||||
| KEYWORD_usize
|
||||
| KEYWORD_ptr
|
||||
}
|
||||
IDENTIFIER = @{
|
||||
!(keyword ~ !(ASCII_ALPHANUMERIC | "_")) ~ (ASCII_ALPHA | "_" ~ (ASCII_ALPHANUMERIC | "_")) ~ (ASCII_ALPHANUMERIC | "_")*
|
||||
}
|
||||
|
||||
INTEGER_LITERAL = { DECIMAL | HEX | BINARY | OCTAL | "0" }
|
||||
DECIMAL = { ASCII_NONZERO_DIGIT ~ (ASCII_DIGIT | "_")* }
|
||||
HEX = { "0" ~ ("x" | "X") ~ ASCII_HEX_DIGIT ~ (ASCII_HEX_DIGIT | "_")* }
|
||||
BINARY = { "0" ~ ("b" | "B") ~ ASCII_BIN_DIGIT ~ (ASCII_BIN_DIGIT | "_")* }
|
||||
OCTAL = { "0" ~ ("o" | "O") ~ ASCII_OCT_DIGIT ~ (ASCII_OCT_DIGIT | "_")* }
|
||||
FLOAT_LITERAL = { ASCII_DIGIT ~ "." ~ ASCII_DIGIT ~ (ASCII_DIGIT | "_")* }
|
||||
CHAR_LITERAL = { "'" ~ (!"'" ~ ANY) ~ "'" }
|
||||
STRING_LITERAL = { "\"" ~ ANY* ~ "\"" }
|
||||
|
||||
COMMENT = _{ LINE_COMMENT | BLOCK_COMMENT }
|
||||
LINE_COMMENT = { "//" ~ (!"\n" ~ ANY)* }
|
||||
BLOCK_COMMENT = { "/*" ~ ANY* ~ "*/" }
|
||||
WHITESPACE = _{ " " | "\t" | "\r" | "\n" }
|
||||
Reference in new issue
Block a user