505 lines
16 KiB
Plaintext
505 lines
16 KiB
Plaintext
import t, c
|
||
from stdint import *
|
||
import memhub
|
||
import string
|
||
import viperlib
|
||
from .tokens import TokOp
|
||
|
||
|
||
# ============================================================
|
||
# ASTKind - Node type enum (kind() virtual function return value)
|
||
# ============================================================
|
||
|
||
class ASTKind(t.CEnum):
|
||
# Core 15 nodes
|
||
Module: t.State
|
||
FunctionDef: t.State
|
||
ClassDef: t.State
|
||
Assign: t.State
|
||
If: t.State
|
||
For: t.State
|
||
While: t.State
|
||
Return: t.State
|
||
Expr: t.State
|
||
Name: t.State
|
||
Constant: t.State
|
||
BinOp: t.State
|
||
UnaryOp: t.State
|
||
Call: t.State
|
||
Compare: t.State
|
||
# Module Layer
|
||
Expression: t.State
|
||
Interactive: t.State
|
||
FunctionType: t.State
|
||
# Statement
|
||
Delete: t.State
|
||
AugAssign: t.State
|
||
AnnAssign: t.State
|
||
With: t.State
|
||
Raise: t.State
|
||
Try: t.State
|
||
Assert: t.State
|
||
Global: t.State
|
||
Nonlocal: t.State
|
||
Pass: t.State
|
||
Break: t.State
|
||
Continue: t.State
|
||
Import: t.State
|
||
ImportFrom: t.State
|
||
Match: t.State
|
||
# Expression Layer
|
||
BoolOp: t.State
|
||
Lambda: t.State
|
||
IfExp: t.State
|
||
Dict: t.State
|
||
Set: t.State
|
||
ListComp: t.State
|
||
SetComp: t.State
|
||
DictComp: t.State
|
||
GeneratorExp: t.State
|
||
Await: t.State
|
||
Yield: t.State
|
||
YieldFrom: t.State
|
||
FormattedValue: t.State
|
||
JoinedStr: t.State
|
||
Attribute: t.State
|
||
Subscript: t.State
|
||
Starred: t.State
|
||
List: t.State
|
||
Tuple: t.State
|
||
Slice: t.State
|
||
NamedExpr: t.State
|
||
# Auxiliary Nodes
|
||
ExceptHandler: t.State
|
||
Arguments: t.State
|
||
Arg: t.State
|
||
Keyword: t.State
|
||
Alias: t.State
|
||
WithItem: t.State
|
||
Comprehension: t.State
|
||
OpNode: t.State
|
||
# Pattern Matching
|
||
MatchCase: t.State
|
||
MatchValue: t.State
|
||
MatchSingleton: t.State
|
||
MatchSequence: t.State
|
||
MatchMapping: t.State
|
||
MatchClass: t.State
|
||
MatchStar: t.State
|
||
MatchAs: t.State
|
||
MatchOr: t.State
|
||
|
||
|
||
# ============================================================
|
||
# ASTCtx - Name context (Load/Store/Del)
|
||
# ============================================================
|
||
|
||
class ASTCtx(t.CEnum):
|
||
Load: t.State
|
||
Store: t.State
|
||
Del: t.State
|
||
|
||
|
||
# ============================================================
|
||
# OpKind - Operator enum (BinOp/UnaryOp/Compare)
|
||
# ============================================================
|
||
|
||
class OpKind(t.CEnum):
|
||
# Binary arithmetic operators
|
||
Add: t.State
|
||
Sub: t.State
|
||
Mult: t.State
|
||
MatMult: t.State
|
||
Div: t.State
|
||
Mod: t.State
|
||
Pow: t.State
|
||
LShift: t.State
|
||
RShift: t.State
|
||
BitOr: t.State
|
||
BitXor: t.State
|
||
BitAnd: t.State
|
||
FloorDiv: t.State
|
||
# Boolean operators
|
||
And: t.State
|
||
Or: t.State
|
||
# Comparison operators
|
||
Eq: t.State
|
||
Ne: t.State
|
||
Lt: t.State
|
||
Le: t.State
|
||
Gt: t.State
|
||
Ge: t.State
|
||
Is: t.State
|
||
IsNot: t.State
|
||
In: t.State
|
||
NotIn: t.State
|
||
# Unary operators
|
||
Not: t.State
|
||
UAdd: t.State
|
||
USub: t.State
|
||
Invert: t.State
|
||
# None
|
||
NoneOp: t.State
|
||
|
||
|
||
# Constant subtype identifiers
|
||
CONST_INT: t.CDefine = 1
|
||
CONST_FLOAT: t.CDefine = 2
|
||
CONST_STR: t.CDefine = 3
|
||
CONST_BOOL: t.CDefine = 4
|
||
CONST_NONE: t.CDefine = 5
|
||
CONST_CHAR: t.CDefine = 6
|
||
|
||
# Node flags
|
||
FLAG_IS_ASYNC: t.CDefine = 1
|
||
FLAG_SIMPLE: t.CDefine = 2
|
||
FLAG_HAS_STAR: t.CDefine = 4
|
||
FLAG_IS_EXTERN: t.CDefine = 8
|
||
|
||
|
||
# ============================================================
|
||
# ASTFlag - Node flags enum
|
||
# ============================================================
|
||
|
||
class ASTFlag(t.CEnum):
|
||
IsAsync: t.State
|
||
Simple: t.State
|
||
HasStar: t.State
|
||
|
||
|
||
# ============================================================
|
||
# Helper Functions
|
||
# ============================================================
|
||
|
||
def _copy_str(pool: memhub.MemManager | t.CPtr, src: str) -> str:
|
||
"""Allocate and copy a C string from pool."""
|
||
if src is None: return None
|
||
slen: t.CSizeT = string.strlen(src)
|
||
buf: str = pool.alloc(slen + 1)
|
||
if buf is None: return None
|
||
string.memcpy(buf, src, slen + 1)
|
||
return buf
|
||
|
||
|
||
def _set_pos(node: AST | t.CPtr, lineno: t.CInt, col_offset: t.CInt,
|
||
end_lineno: t.CInt, end_col_offset: t.CInt):
|
||
if node is None: return
|
||
node.lineno = lineno
|
||
node.col_offset = col_offset
|
||
node.end_lineno = end_lineno
|
||
node.end_col_offset = end_col_offset
|
||
|
||
|
||
def _inherit_pos(node: AST | t.CPtr, ref: AST | t.CPtr):
|
||
if node is None or ref is None: return
|
||
node.lineno = ref.lineno
|
||
node.col_offset = ref.col_offset
|
||
node.end_lineno = ref.end_lineno
|
||
node.end_col_offset = ref.end_col_offset
|
||
|
||
|
||
def _init_ast(node: AST | t.CPtr, pool: memhub.MemManager | t.CPtr,
|
||
lineno: t.CInt, col: t.CInt):
|
||
"""Initialize AST base fields."""
|
||
node.parent = None
|
||
node.pool = pool
|
||
node.lineno = lineno
|
||
node.col_offset = col
|
||
node.end_lineno = lineno
|
||
node.end_col_offset = col
|
||
node.children = None
|
||
|
||
|
||
def _set_parent_list(lst: list[AST | t.CPtr] | t.CPtr, parent: AST | t.CPtr):
|
||
"""Set list children parent pointers."""
|
||
if lst is None: return
|
||
n: t.CSizeT = lst.__len__()
|
||
i: t.CSizeT = 0
|
||
while i < n:
|
||
child: AST | t.CPtr = lst.get(i)
|
||
if child is not None:
|
||
child.parent = parent
|
||
i += 1
|
||
|
||
|
||
def _append_child(lst: list[AST | t.CPtr] | t.CPtr, node: AST | t.CPtr,
|
||
pool: memhub.MemManager | t.CPtr):
|
||
"""Append child nodes to a list (module-level function, to avoid conflicts with the AST.append method name).
|
||
|
||
Call this function within the AST.append method to bypass the compiler mistaking list.append for a virtual function call to AST.append when doing type inference."""
|
||
if lst is None or node is None:
|
||
return
|
||
lst.append(node)
|
||
|
||
|
||
# ============================================================
|
||
# Operator Mapping: TokOp -> OpKind
|
||
# ============================================================
|
||
|
||
def _binop_from_op(op_id: t.CInt) -> t.CInt:
|
||
if op_id == TokOp.Plus: return OpKind.Add
|
||
if op_id == TokOp.Minus: return OpKind.Sub
|
||
if op_id == TokOp.Star: return OpKind.Mult
|
||
if op_id == TokOp.At: return OpKind.MatMult
|
||
if op_id == TokOp.Slash: return OpKind.Div
|
||
if op_id == TokOp.Percent: return OpKind.Mod
|
||
if op_id == TokOp.StarStar: return OpKind.Pow
|
||
if op_id == TokOp.LtLt: return OpKind.LShift
|
||
if op_id == TokOp.GtGt: return OpKind.RShift
|
||
if op_id == TokOp.VBar: return OpKind.BitOr
|
||
if op_id == TokOp.Caret: return OpKind.BitXor
|
||
if op_id == TokOp.Amp: return OpKind.BitAnd
|
||
if op_id == TokOp.DSlash: return OpKind.FloorDiv
|
||
return OpKind.NoneOp
|
||
|
||
|
||
def _augop_from_op(op_id: t.CInt) -> t.CInt:
|
||
if op_id == TokOp.PlusEq: return OpKind.Add
|
||
if op_id == TokOp.MinusEq: return OpKind.Sub
|
||
if op_id == TokOp.StarEq: return OpKind.Mult
|
||
if op_id == TokOp.AtEq: return OpKind.MatMult
|
||
if op_id == TokOp.SlashEq: return OpKind.Div
|
||
if op_id == TokOp.PercentEq: return OpKind.Mod
|
||
if op_id == TokOp.StarEqEq: return OpKind.Pow
|
||
if op_id == TokOp.LtLtEq: return OpKind.LShift
|
||
if op_id == TokOp.GtGtEq: return OpKind.RShift
|
||
if op_id == TokOp.VBarEq: return OpKind.BitOr
|
||
if op_id == TokOp.CaretEq: return OpKind.BitXor
|
||
if op_id == TokOp.AmpEq: return OpKind.BitAnd
|
||
if op_id == TokOp.DSlashEq: return OpKind.FloorDiv
|
||
return OpKind.NoneOp
|
||
|
||
|
||
def _cmpop_from_op(op_id: t.CInt) -> t.CInt:
|
||
if op_id == TokOp.EqEq: return OpKind.Eq
|
||
if op_id == TokOp.ExclaimEq: return OpKind.Ne
|
||
if op_id == TokOp.Less: return OpKind.Lt
|
||
if op_id == TokOp.LessEq: return OpKind.Le
|
||
if op_id == TokOp.Greater: return OpKind.Gt
|
||
if op_id == TokOp.GreaterEq: return OpKind.Ge
|
||
return OpKind.NoneOp
|
||
|
||
|
||
# ============================================================
|
||
# dump Helper Functions
|
||
# ============================================================
|
||
|
||
def _emit(buf: t.CChar | t.CPtr, size: t.CSizeT, pos: t.CSizeT,
|
||
text: str) -> t.CSizeT:
|
||
"""Append literal to buf+pos."""
|
||
if buf is None or pos >= size:
|
||
return pos
|
||
cur: t.CChar | t.CPtr = (t.CVoid | t.CPtr)(t.CUInt64T(buf) + pos)
|
||
rem: t.CSizeT = size - pos
|
||
n: t.CInt = viperlib.snprintf(cur, rem, "%s", text)
|
||
if n < 0:
|
||
return pos
|
||
return pos + t.CSizeT(n)
|
||
|
||
|
||
def _emit_str(buf: t.CChar | t.CPtr, size: t.CSizeT, pos: t.CSizeT,
|
||
text: str) -> t.CSizeT:
|
||
"""Append a string value to buf pos (using %s format)"""
|
||
if buf is None or pos >= size:
|
||
return pos
|
||
cur: t.CChar | t.CPtr = (t.CVoid | t.CPtr)(t.CUInt64T(buf) + pos)
|
||
rem: t.CSizeT = size - pos
|
||
n: t.CInt = viperlib.snprintf(cur, rem, "%s", text)
|
||
if n < 0:
|
||
return pos
|
||
return pos + t.CSizeT(n)
|
||
|
||
|
||
def _emit_int(buf: t.CChar | t.CPtr, size: t.CSizeT, pos: t.CSizeT,
|
||
val: t.CInt64T) -> t.CSizeT:
|
||
"""Append an integer value to buf pos (using %lld format)."""
|
||
if buf is None or pos >= size:
|
||
return pos
|
||
cur: t.CChar | t.CPtr = (t.CVoid | t.CPtr)(t.CUInt64T(buf) + pos)
|
||
rem: t.CSizeT = size - pos
|
||
n: t.CInt = viperlib.snprintf(cur, rem, "%lld", val)
|
||
if n < 0:
|
||
return pos
|
||
return pos + t.CSizeT(n)
|
||
|
||
|
||
def _op_name(op: t.CInt) -> str:
|
||
"""Return the string name of an OpKind."""
|
||
if op == OpKind.Add: return "Add"
|
||
if op == OpKind.Sub: return "Sub"
|
||
if op == OpKind.Mult: return "Mult"
|
||
if op == OpKind.MatMult: return "MatMult"
|
||
if op == OpKind.Div: return "Div"
|
||
if op == OpKind.Mod: return "Mod"
|
||
if op == OpKind.Pow: return "Pow"
|
||
if op == OpKind.LShift: return "LShift"
|
||
if op == OpKind.RShift: return "RShift"
|
||
if op == OpKind.BitOr: return "BitOr"
|
||
if op == OpKind.BitXor: return "BitXor"
|
||
if op == OpKind.BitAnd: return "BitAnd"
|
||
if op == OpKind.FloorDiv: return "FloorDiv"
|
||
if op == OpKind.And: return "And"
|
||
if op == OpKind.Or: return "Or"
|
||
if op == OpKind.Eq: return "Eq"
|
||
if op == OpKind.Ne: return "Ne"
|
||
if op == OpKind.Lt: return "Lt"
|
||
if op == OpKind.Le: return "Le"
|
||
if op == OpKind.Gt: return "Gt"
|
||
if op == OpKind.Ge: return "Ge"
|
||
if op == OpKind.Is: return "Is"
|
||
if op == OpKind.IsNot: return "IsNot"
|
||
if op == OpKind.In: return "In"
|
||
if op == OpKind.NotIn: return "NotIn"
|
||
if op == OpKind.Not: return "Not"
|
||
if op == OpKind.UAdd: return "UAdd"
|
||
if op == OpKind.USub: return "USub"
|
||
if op == OpKind.Invert: return "Invert"
|
||
return "NoneOp"
|
||
|
||
|
||
def _dump_list(lst: list[AST | t.CPtr] | t.CPtr, buf: t.CChar | t.CPtr,
|
||
size: t.CSizeT, pos: t.CSizeT) -> t.CSizeT:
|
||
"""Traverse a list[AST | CPtr] container, polymorphically dump each element."""
|
||
if lst is None:
|
||
pos = _emit(buf, size, pos, "[]")
|
||
return pos
|
||
pos = _emit(buf, size, pos, "[")
|
||
n: t.CSizeT = lst.__len__()
|
||
i: t.CSizeT = 0
|
||
while i < n:
|
||
if i > 0:
|
||
pos = _emit(buf, size, pos, ", ")
|
||
child: AST | t.CPtr = lst.get(i)
|
||
if child is not None:
|
||
pos = child.dump(buf, size, pos)
|
||
i += 1
|
||
pos = _emit(buf, size, pos, "]")
|
||
return pos
|
||
|
||
|
||
def _dump_op_list(lst: t.CPtr, buf: t.CChar | t.CPtr,
|
||
size: t.CSizeT, pos: t.CSizeT) -> t.CSizeT:
|
||
"""Traverse a list[CInt] (OpKind values), output a list of op names."""
|
||
if lst is None:
|
||
pos = _emit(buf, size, pos, "[]")
|
||
return pos
|
||
ops_list: list[t.CInt] | t.CPtr = (list[t.CInt] | t.CPtr)(lst)
|
||
pos = _emit(buf, size, pos, "[")
|
||
n: t.CSizeT = ops_list.__len__()
|
||
i: t.CSizeT = 0
|
||
while i < n:
|
||
if i > 0:
|
||
pos = _emit(buf, size, pos, ", ")
|
||
op: t.CInt = ops_list.get(i)
|
||
pos = _emit(buf, size, pos, _op_name(op))
|
||
i += 1
|
||
pos = _emit(buf, size, pos, "]")
|
||
return pos
|
||
|
||
|
||
# ============================================================
|
||
# AST - Polymorphic Base Class
|
||
|
||
# @t.CVTable enables vtable, supporting virtual functions like kind()/type_name()/dump()/accept()
|
||
# Child nodes are stored in a list[AST | CPtr] container (O(1) random access, more efficient than linked lists)
|
||
# The parent field maintains a pointer to the parent (used for upward traversal)
|
||
# ============================================================
|
||
|
||
@t.CVTable
|
||
class AST:
|
||
"""AST node base class. All specific node classes inherit from this.
|
||
|
||
Fields:
|
||
parent: Pointer to the parent node (for upward traversal)
|
||
pool: Allocator (used for allocating child nodes/strings)
|
||
lineno/col_offset: Starting position
|
||
end_lineno/end_col_offset: Ending position
|
||
children: List of child nodes (body block, added via append)
|
||
"""
|
||
parent: AST | t.CPtr
|
||
pool: memhub.MemManager | t.CPtr
|
||
lineno: t.CInt
|
||
col_offset: t.CInt
|
||
end_lineno: t.CInt
|
||
end_col_offset: t.CInt
|
||
children: list[AST | t.CPtr] | t.CPtr
|
||
|
||
def kind(self) -> t.CInt:
|
||
"""Returns the node type (ASTKind value), overridden by subclasses"""
|
||
return 0
|
||
|
||
def type_name(self) -> str:
|
||
"""Return the node type name, subclasses override"""
|
||
return "AST"
|
||
|
||
def dump(self, buf: t.CChar | t.CPtr, size: t.CSizeT,
|
||
pos: t.CSizeT) -> t.CSizeT:
|
||
"""Polymorphic dump: subclasses override to define custom format. Returns the written pos"""
|
||
return pos
|
||
|
||
def accept(self, visitor: t.CPtr):
|
||
"""Visitor pattern entry point: subclasses override to dispatch to ASTVisitor in the visitor pattern.
|
||
Current implementation is empty. override this method to dispatch to visitor.visit_X(self)"""
|
||
pass
|
||
|
||
def append(self, node: AST | t.CPtr):
|
||
"""Append node to the children list.
|
||
|
||
Used for parser to build statement blocks: module.append(stmt), if_node.append(stmt), etc.
|
||
The children list is lazily initialized on first call.
|
||
|
||
Note: list[AST|CPtr] is currently stored by value (compiler limitation), so you must set
|
||
node.parent before appending, so the parent in the copy is correct."""
|
||
if self is None or node is None:
|
||
return
|
||
if self.children is None:
|
||
self.children = list[AST | t.CPtr](self.pool, 8)
|
||
node.parent = self
|
||
_append_child(self.children, node, self.pool)
|
||
|
||
|
||
# ============================================================
|
||
# Stmt - Statement Base Class
|
||
#
|
||
# 所有语句节点（Assign/If/For/While/Return/...）继承 Stmt。
|
||
# isinstance(node, ast.Stmt) 可在运行时区分"语句"与"表达式"
|
||
# （依赖 TPV isinstance 子类集合判定 + 三层 vtable 继承）。
|
||
# 不 override kind()（继承 AST 的 0），具体节点各自返回 ASTKind 值。
|
||
# ============================================================
|
||
|
||
@t.CVTable
|
||
class Stmt(AST):
|
||
"""语句节点基类（Assign/If/For/While/Return/...）"""
|
||
|
||
|
||
# ============================================================
|
||
# ExprNode - Expression Base Class
|
||
#
|
||
# 所有表达式节点（Name/Constant/Call/BinOp/...）继承 ExprNode。
|
||
# isinstance(node, ast.ExprNode) 可在运行时区分"表达式"与"语句"。
|
||
#
|
||
# 命名说明：不能叫 Expr——ASTKind.Expr（base.vp L23）与 stmts.vp 的
|
||
# Expr 类已是"表达式语句"（Expression Statement）的既有语义。
|
||
# ============================================================
|
||
|
||
@t.CVTable
|
||
class ExprNode(AST):
|
||
"""表达式节点基类（Name/Constant/Call/BinOp/...）"""
|
||
|
||
|
||
# ============================================================
|
||
# Module dump function (for ast.dump(node, buf, size))
|
||
# ============================================================
|
||
|
||
def dump(node: AST | t.CPtr, buf: t.CChar | t.CPtr, size: t.CSizeT):
|
||
"""Serialize AST node to buf (with capacity size)"""
|
||
if node is None or buf is None or size == 0:
|
||
return
|
||
buf[0] = '\0'
|
||
final_pos: t.CSizeT = node.dump(buf, size, 0)
|
||
# NUL-terminate to prevent printf overflow
|
||
if final_pos < size:
|
||
buf[final_pos] = '\0'
|
||
else:
|
||
buf[size - 1] = '\0'
|