2022-11-03 01:31:26 -03:00
|
|
|
"""Parser for bytecodes.inst."""
|
|
|
|
|
|
|
|
from dataclasses import dataclass, field
|
2023-07-24 13:38:23 -03:00
|
|
|
from typing import NamedTuple, Callable, TypeVar, Literal, cast
|
2022-11-03 01:31:26 -03:00
|
|
|
|
|
|
|
import lexer as lx
|
|
|
|
from plexer import PLexer
|
|
|
|
|
|
|
|
|
|
|
|
P = TypeVar("P", bound="Parser")
|
|
|
|
N = TypeVar("N", bound="Node")
|
2022-12-02 23:57:30 -04:00
|
|
|
|
|
|
|
|
|
|
|
def contextual(func: Callable[[P], N | None]) -> Callable[[P], N | None]:
|
2022-11-03 01:31:26 -03:00
|
|
|
# Decorator to wrap grammar methods.
|
|
|
|
# Resets position if `func` returns None.
|
2022-12-02 23:57:30 -04:00
|
|
|
def contextual_wrapper(self: P) -> N | None:
|
2022-11-03 01:31:26 -03:00
|
|
|
begin = self.getpos()
|
|
|
|
res = func(self)
|
|
|
|
if res is None:
|
|
|
|
self.setpos(begin)
|
2023-07-24 13:38:23 -03:00
|
|
|
return None
|
2022-11-03 01:31:26 -03:00
|
|
|
end = self.getpos()
|
|
|
|
res.context = Context(begin, end, self)
|
|
|
|
return res
|
2022-12-02 23:57:30 -04:00
|
|
|
|
2022-11-03 01:31:26 -03:00
|
|
|
return contextual_wrapper
|
|
|
|
|
|
|
|
|
|
|
|
class Context(NamedTuple):
|
|
|
|
begin: int
|
|
|
|
end: int
|
|
|
|
owner: PLexer
|
|
|
|
|
2023-08-18 10:42:45 -03:00
|
|
|
def __repr__(self) -> str:
|
2023-03-04 00:59:21 -04:00
|
|
|
return f"<{self.owner.filename}: {self.begin}-{self.end}>"
|
2022-11-03 01:31:26 -03:00
|
|
|
|
|
|
|
|
|
|
|
@dataclass
|
|
|
|
class Node:
|
2023-01-17 19:59:19 -04:00
|
|
|
context: Context | None = field(init=False, compare=False, default=None)
|
2022-11-03 01:31:26 -03:00
|
|
|
|
|
|
|
@property
|
|
|
|
def text(self) -> str:
|
2022-11-06 13:40:47 -04:00
|
|
|
return self.to_text()
|
|
|
|
|
|
|
|
def to_text(self, dedent: int = 0) -> str:
|
2022-11-03 01:31:26 -03:00
|
|
|
context = self.context
|
|
|
|
if not context:
|
|
|
|
return ""
|
2023-01-17 19:59:19 -04:00
|
|
|
return lx.to_text(self.tokens, dedent)
|
|
|
|
|
|
|
|
@property
|
|
|
|
def tokens(self) -> list[lx.Token]:
|
|
|
|
context = self.context
|
|
|
|
if not context:
|
|
|
|
return []
|
|
|
|
tokens = context.owner.tokens
|
|
|
|
begin = context.begin
|
|
|
|
end = context.end
|
|
|
|
return tokens[begin:end]
|
2022-11-03 01:31:26 -03:00
|
|
|
|
|
|
|
|
|
|
|
@dataclass
|
|
|
|
class Block(Node):
|
2023-01-17 19:59:19 -04:00
|
|
|
# This just holds a context which has the list of tokens.
|
|
|
|
pass
|
2022-11-03 01:31:26 -03:00
|
|
|
|
|
|
|
|
2022-11-15 23:59:19 -04:00
|
|
|
@dataclass
|
2022-11-17 21:06:07 -04:00
|
|
|
class StackEffect(Node):
|
2023-08-04 13:35:56 -03:00
|
|
|
name: str = field(compare=False) # __eq__ only uses type, cond, size
|
2023-01-17 19:59:19 -04:00
|
|
|
type: str = "" # Optional `:type`
|
2023-01-29 21:28:39 -04:00
|
|
|
cond: str = "" # Optional `if (cond)`
|
2023-01-17 19:59:19 -04:00
|
|
|
size: str = "" # Optional `[size]`
|
2023-01-29 21:28:39 -04:00
|
|
|
# Note: size cannot be combined with type or cond
|
2023-01-17 19:59:19 -04:00
|
|
|
|
2023-08-18 10:42:45 -03:00
|
|
|
def __repr__(self) -> str:
|
2023-08-04 13:35:56 -03:00
|
|
|
items = [self.name, self.type, self.cond, self.size]
|
|
|
|
while items and items[-1] == "":
|
|
|
|
del items[-1]
|
|
|
|
return f"StackEffect({', '.join(repr(item) for item in items)})"
|
|
|
|
|
2023-01-17 19:59:19 -04:00
|
|
|
|
|
|
|
@dataclass
|
2023-01-29 21:28:39 -04:00
|
|
|
class Expression(Node):
|
2023-01-17 19:59:19 -04:00
|
|
|
size: str
|
2022-11-15 23:59:19 -04:00
|
|
|
|
|
|
|
|
|
|
|
@dataclass
|
2022-11-17 21:06:07 -04:00
|
|
|
class CacheEffect(Node):
|
2022-11-15 23:59:19 -04:00
|
|
|
name: str
|
|
|
|
size: int
|
|
|
|
|
|
|
|
|
2022-12-02 23:57:30 -04:00
|
|
|
@dataclass
|
|
|
|
class OpName(Node):
|
|
|
|
name: str
|
|
|
|
|
|
|
|
|
2022-11-17 21:06:07 -04:00
|
|
|
InputEffect = StackEffect | CacheEffect
|
|
|
|
OutputEffect = StackEffect
|
2022-12-02 23:57:30 -04:00
|
|
|
UOp = OpName | CacheEffect
|
2022-11-17 21:06:07 -04:00
|
|
|
|
|
|
|
|
2022-11-03 01:31:26 -03:00
|
|
|
@dataclass
|
2022-11-08 12:22:56 -04:00
|
|
|
class InstHeader(Node):
|
2023-11-17 15:49:42 -04:00
|
|
|
annotations: list[str]
|
2023-06-12 15:19:04 -03:00
|
|
|
kind: Literal["inst", "op"]
|
2022-11-03 01:31:26 -03:00
|
|
|
name: str
|
2022-11-17 21:06:07 -04:00
|
|
|
inputs: list[InputEffect]
|
|
|
|
outputs: list[OutputEffect]
|
2022-11-08 12:22:56 -04:00
|
|
|
|
|
|
|
|
|
|
|
@dataclass
|
|
|
|
class InstDef(Node):
|
2023-11-17 15:49:42 -04:00
|
|
|
annotations: list[str]
|
2023-06-12 15:19:04 -03:00
|
|
|
kind: Literal["inst", "op"]
|
2022-12-02 23:57:30 -04:00
|
|
|
name: str
|
|
|
|
inputs: list[InputEffect]
|
|
|
|
outputs: list[OutputEffect]
|
2022-11-08 12:22:56 -04:00
|
|
|
block: Block
|
|
|
|
|
|
|
|
|
2022-11-06 13:40:47 -04:00
|
|
|
@dataclass
|
2022-12-02 23:57:30 -04:00
|
|
|
class Macro(Node):
|
2022-11-06 13:40:47 -04:00
|
|
|
name: str
|
2022-12-02 23:57:30 -04:00
|
|
|
uops: list[UOp]
|
2022-11-06 13:40:47 -04:00
|
|
|
|
|
|
|
|
2022-11-03 01:31:26 -03:00
|
|
|
@dataclass
|
|
|
|
class Family(Node):
|
|
|
|
name: str
|
2022-11-15 23:59:19 -04:00
|
|
|
size: str # Variable giving the cache size in code units
|
2022-11-03 01:31:26 -03:00
|
|
|
members: list[str]
|
|
|
|
|
2023-08-04 13:35:56 -03:00
|
|
|
|
2023-06-11 18:31:59 -03:00
|
|
|
@dataclass
|
|
|
|
class Pseudo(Node):
|
|
|
|
name: str
|
2024-05-29 06:47:56 -03:00
|
|
|
inputs: list[InputEffect]
|
|
|
|
outputs: list[OutputEffect]
|
2024-01-12 13:30:27 -04:00
|
|
|
flags: list[str] # instr flags to set on the pseudo instruction
|
|
|
|
targets: list[str] # opcodes this can be replaced by
|
|
|
|
|
2023-06-11 18:31:59 -03:00
|
|
|
|
2023-12-07 08:49:40 -04:00
|
|
|
AstNode = InstDef | Macro | Pseudo | Family
|
2022-11-03 01:31:26 -03:00
|
|
|
|
2024-01-12 13:30:27 -04:00
|
|
|
|
2022-11-03 01:31:26 -03:00
|
|
|
class Parser(PLexer):
|
2022-12-02 23:57:30 -04:00
|
|
|
@contextual
|
2023-12-07 08:49:40 -04:00
|
|
|
def definition(self) -> AstNode | None:
|
2022-12-02 23:57:30 -04:00
|
|
|
if macro := self.macro_def():
|
|
|
|
return macro
|
|
|
|
if family := self.family_def():
|
|
|
|
return family
|
2023-06-11 18:31:59 -03:00
|
|
|
if pseudo := self.pseudo_def():
|
|
|
|
return pseudo
|
2023-11-07 05:42:39 -04:00
|
|
|
if inst := self.inst_def():
|
|
|
|
return inst
|
2023-07-24 13:38:23 -03:00
|
|
|
return None
|
2022-11-03 01:31:26 -03:00
|
|
|
|
|
|
|
@contextual
|
|
|
|
def inst_def(self) -> InstDef | None:
|
2022-12-02 23:57:30 -04:00
|
|
|
if hdr := self.inst_header():
|
2022-11-03 01:31:26 -03:00
|
|
|
if block := self.block():
|
2023-01-05 17:01:07 -04:00
|
|
|
return InstDef(
|
2023-11-07 05:42:39 -04:00
|
|
|
hdr.annotations,
|
2023-08-04 13:35:56 -03:00
|
|
|
hdr.kind,
|
|
|
|
hdr.name,
|
|
|
|
hdr.inputs,
|
|
|
|
hdr.outputs,
|
|
|
|
block,
|
2023-01-05 17:01:07 -04:00
|
|
|
)
|
2022-11-03 01:31:26 -03:00
|
|
|
raise self.make_syntax_error("Expected block")
|
|
|
|
return None
|
|
|
|
|
|
|
|
@contextual
|
2022-11-08 12:22:56 -04:00
|
|
|
def inst_header(self) -> InstHeader | None:
|
2023-11-07 05:42:39 -04:00
|
|
|
# annotation* inst(NAME, (inputs -- outputs))
|
|
|
|
# | annotation* op(NAME, (inputs -- outputs))
|
|
|
|
annotations = []
|
|
|
|
while anno := self.expect(lx.ANNOTATION):
|
2024-02-20 06:50:59 -04:00
|
|
|
if anno.text == "replicate":
|
|
|
|
self.require(lx.LPAREN)
|
|
|
|
times = self.require(lx.NUMBER)
|
|
|
|
self.require(lx.RPAREN)
|
|
|
|
annotations.append(f"replicate({times.text})")
|
|
|
|
else:
|
|
|
|
annotations.append(anno.text)
|
2023-11-07 05:42:39 -04:00
|
|
|
tkn = self.expect(lx.INST)
|
|
|
|
if not tkn:
|
|
|
|
tkn = self.expect(lx.OP)
|
|
|
|
if tkn:
|
2023-07-24 13:38:23 -03:00
|
|
|
kind = cast(Literal["inst", "op"], tkn.text)
|
2022-12-02 23:57:30 -04:00
|
|
|
if self.expect(lx.LPAREN) and (tkn := self.expect(lx.IDENTIFIER)):
|
2022-11-03 01:31:26 -03:00
|
|
|
name = tkn.text
|
|
|
|
if self.expect(lx.COMMA):
|
GH-98831: Typed stack effects, and more instructions converted (#99764)
Stack effects can now have a type, e.g. `inst(X, (left, right -- jump/uint64_t)) { ... }`.
Instructions converted to the non-legacy format:
* COMPARE_OP
* COMPARE_OP_FLOAT_JUMP
* COMPARE_OP_INT_JUMP
* COMPARE_OP_STR_JUMP
* STORE_ATTR
* DELETE_ATTR
* STORE_GLOBAL
* STORE_ATTR_INSTANCE_VALUE
* STORE_ATTR_WITH_HINT
* STORE_ATTR_SLOT, and complete the store_attr family
* Complete the store_subscr family: STORE_SUBSCR{,DICT,LIST_INT}
(STORE_SUBSCR was alread half converted,
but wasn't using cache effects yet.)
* DELETE_SUBSCR
* PRINT_EXPR
* INTERPRETER_EXIT (a bit weird, ends in return)
* RETURN_VALUE
* GET_AITER (had to restructure it some)
The original had mysterious `SET_TOP(NULL)` before `goto error`.
I assume those just account for `obj` having been decref'ed,
so I got rid of them in favor of the cleanup implied by `ERROR_IF()`.
* LIST_APPEND (a bit unhappy with it)
* SET_ADD (also a bit unhappy with it)
Various other improvements/refactorings as well.
2022-12-08 17:31:27 -04:00
|
|
|
inp, outp = self.io_effect()
|
2022-11-08 12:22:56 -04:00
|
|
|
if self.expect(lx.RPAREN):
|
2022-12-02 23:57:30 -04:00
|
|
|
if (tkn := self.peek()) and tkn.kind == lx.LBRACE:
|
2023-11-07 05:42:39 -04:00
|
|
|
return InstHeader(annotations, kind, name, inp, outp)
|
2022-11-03 01:31:26 -03:00
|
|
|
return None
|
|
|
|
|
GH-98831: Typed stack effects, and more instructions converted (#99764)
Stack effects can now have a type, e.g. `inst(X, (left, right -- jump/uint64_t)) { ... }`.
Instructions converted to the non-legacy format:
* COMPARE_OP
* COMPARE_OP_FLOAT_JUMP
* COMPARE_OP_INT_JUMP
* COMPARE_OP_STR_JUMP
* STORE_ATTR
* DELETE_ATTR
* STORE_GLOBAL
* STORE_ATTR_INSTANCE_VALUE
* STORE_ATTR_WITH_HINT
* STORE_ATTR_SLOT, and complete the store_attr family
* Complete the store_subscr family: STORE_SUBSCR{,DICT,LIST_INT}
(STORE_SUBSCR was alread half converted,
but wasn't using cache effects yet.)
* DELETE_SUBSCR
* PRINT_EXPR
* INTERPRETER_EXIT (a bit weird, ends in return)
* RETURN_VALUE
* GET_AITER (had to restructure it some)
The original had mysterious `SET_TOP(NULL)` before `goto error`.
I assume those just account for `obj` having been decref'ed,
so I got rid of them in favor of the cleanup implied by `ERROR_IF()`.
* LIST_APPEND (a bit unhappy with it)
* SET_ADD (also a bit unhappy with it)
Various other improvements/refactorings as well.
2022-12-08 17:31:27 -04:00
|
|
|
def io_effect(self) -> tuple[list[InputEffect], list[OutputEffect]]:
|
2022-11-03 01:31:26 -03:00
|
|
|
# '(' [inputs] '--' [outputs] ')'
|
|
|
|
if self.expect(lx.LPAREN):
|
2022-11-17 21:06:07 -04:00
|
|
|
inputs = self.inputs() or []
|
2022-11-03 01:31:26 -03:00
|
|
|
if self.expect(lx.MINUSMINUS):
|
2022-11-17 21:06:07 -04:00
|
|
|
outputs = self.outputs() or []
|
2022-11-03 01:31:26 -03:00
|
|
|
if self.expect(lx.RPAREN):
|
2022-11-17 21:06:07 -04:00
|
|
|
return inputs, outputs
|
2022-11-03 01:31:26 -03:00
|
|
|
raise self.make_syntax_error("Expected stack effect")
|
|
|
|
|
2022-11-17 21:06:07 -04:00
|
|
|
def inputs(self) -> list[InputEffect] | None:
|
2022-11-15 23:59:19 -04:00
|
|
|
# input (',' input)*
|
2022-11-03 01:31:26 -03:00
|
|
|
here = self.getpos()
|
|
|
|
if inp := self.input():
|
2023-07-24 13:38:23 -03:00
|
|
|
inp = cast(InputEffect, inp)
|
2022-11-03 01:31:26 -03:00
|
|
|
near = self.getpos()
|
|
|
|
if self.expect(lx.COMMA):
|
|
|
|
if rest := self.inputs():
|
|
|
|
return [inp] + rest
|
|
|
|
self.setpos(near)
|
|
|
|
return [inp]
|
|
|
|
self.setpos(here)
|
|
|
|
return None
|
|
|
|
|
2022-11-15 23:59:19 -04:00
|
|
|
@contextual
|
2022-11-17 21:06:07 -04:00
|
|
|
def input(self) -> InputEffect | None:
|
GH-98831: Typed stack effects, and more instructions converted (#99764)
Stack effects can now have a type, e.g. `inst(X, (left, right -- jump/uint64_t)) { ... }`.
Instructions converted to the non-legacy format:
* COMPARE_OP
* COMPARE_OP_FLOAT_JUMP
* COMPARE_OP_INT_JUMP
* COMPARE_OP_STR_JUMP
* STORE_ATTR
* DELETE_ATTR
* STORE_GLOBAL
* STORE_ATTR_INSTANCE_VALUE
* STORE_ATTR_WITH_HINT
* STORE_ATTR_SLOT, and complete the store_attr family
* Complete the store_subscr family: STORE_SUBSCR{,DICT,LIST_INT}
(STORE_SUBSCR was alread half converted,
but wasn't using cache effects yet.)
* DELETE_SUBSCR
* PRINT_EXPR
* INTERPRETER_EXIT (a bit weird, ends in return)
* RETURN_VALUE
* GET_AITER (had to restructure it some)
The original had mysterious `SET_TOP(NULL)` before `goto error`.
I assume those just account for `obj` having been decref'ed,
so I got rid of them in favor of the cleanup implied by `ERROR_IF()`.
* LIST_APPEND (a bit unhappy with it)
* SET_ADD (also a bit unhappy with it)
Various other improvements/refactorings as well.
2022-12-08 17:31:27 -04:00
|
|
|
return self.cache_effect() or self.stack_effect()
|
2022-11-03 01:31:26 -03:00
|
|
|
|
2022-11-17 21:06:07 -04:00
|
|
|
def outputs(self) -> list[OutputEffect] | None:
|
2022-11-03 01:31:26 -03:00
|
|
|
# output (, output)*
|
|
|
|
here = self.getpos()
|
|
|
|
if outp := self.output():
|
|
|
|
near = self.getpos()
|
|
|
|
if self.expect(lx.COMMA):
|
|
|
|
if rest := self.outputs():
|
|
|
|
return [outp] + rest
|
|
|
|
self.setpos(near)
|
|
|
|
return [outp]
|
|
|
|
self.setpos(here)
|
|
|
|
return None
|
|
|
|
|
2022-11-15 23:59:19 -04:00
|
|
|
@contextual
|
2022-11-17 21:06:07 -04:00
|
|
|
def output(self) -> OutputEffect | None:
|
GH-98831: Typed stack effects, and more instructions converted (#99764)
Stack effects can now have a type, e.g. `inst(X, (left, right -- jump/uint64_t)) { ... }`.
Instructions converted to the non-legacy format:
* COMPARE_OP
* COMPARE_OP_FLOAT_JUMP
* COMPARE_OP_INT_JUMP
* COMPARE_OP_STR_JUMP
* STORE_ATTR
* DELETE_ATTR
* STORE_GLOBAL
* STORE_ATTR_INSTANCE_VALUE
* STORE_ATTR_WITH_HINT
* STORE_ATTR_SLOT, and complete the store_attr family
* Complete the store_subscr family: STORE_SUBSCR{,DICT,LIST_INT}
(STORE_SUBSCR was alread half converted,
but wasn't using cache effects yet.)
* DELETE_SUBSCR
* PRINT_EXPR
* INTERPRETER_EXIT (a bit weird, ends in return)
* RETURN_VALUE
* GET_AITER (had to restructure it some)
The original had mysterious `SET_TOP(NULL)` before `goto error`.
I assume those just account for `obj` having been decref'ed,
so I got rid of them in favor of the cleanup implied by `ERROR_IF()`.
* LIST_APPEND (a bit unhappy with it)
* SET_ADD (also a bit unhappy with it)
Various other improvements/refactorings as well.
2022-12-08 17:31:27 -04:00
|
|
|
return self.stack_effect()
|
|
|
|
|
|
|
|
@contextual
|
|
|
|
def cache_effect(self) -> CacheEffect | None:
|
|
|
|
# IDENTIFIER '/' NUMBER
|
|
|
|
if tkn := self.expect(lx.IDENTIFIER):
|
|
|
|
if self.expect(lx.DIVIDE):
|
|
|
|
num = self.require(lx.NUMBER).text
|
|
|
|
try:
|
|
|
|
size = int(num)
|
|
|
|
except ValueError:
|
|
|
|
raise self.make_syntax_error(f"Expected integer, got {num!r}")
|
|
|
|
else:
|
|
|
|
return CacheEffect(tkn.text, size)
|
2023-07-24 13:38:23 -03:00
|
|
|
return None
|
GH-98831: Typed stack effects, and more instructions converted (#99764)
Stack effects can now have a type, e.g. `inst(X, (left, right -- jump/uint64_t)) { ... }`.
Instructions converted to the non-legacy format:
* COMPARE_OP
* COMPARE_OP_FLOAT_JUMP
* COMPARE_OP_INT_JUMP
* COMPARE_OP_STR_JUMP
* STORE_ATTR
* DELETE_ATTR
* STORE_GLOBAL
* STORE_ATTR_INSTANCE_VALUE
* STORE_ATTR_WITH_HINT
* STORE_ATTR_SLOT, and complete the store_attr family
* Complete the store_subscr family: STORE_SUBSCR{,DICT,LIST_INT}
(STORE_SUBSCR was alread half converted,
but wasn't using cache effects yet.)
* DELETE_SUBSCR
* PRINT_EXPR
* INTERPRETER_EXIT (a bit weird, ends in return)
* RETURN_VALUE
* GET_AITER (had to restructure it some)
The original had mysterious `SET_TOP(NULL)` before `goto error`.
I assume those just account for `obj` having been decref'ed,
so I got rid of them in favor of the cleanup implied by `ERROR_IF()`.
* LIST_APPEND (a bit unhappy with it)
* SET_ADD (also a bit unhappy with it)
Various other improvements/refactorings as well.
2022-12-08 17:31:27 -04:00
|
|
|
|
|
|
|
@contextual
|
|
|
|
def stack_effect(self) -> StackEffect | None:
|
2024-02-13 09:24:48 -04:00
|
|
|
# IDENTIFIER [':' IDENTIFIER [TIMES]] ['if' '(' expression ')']
|
2023-01-29 21:28:39 -04:00
|
|
|
# | IDENTIFIER '[' expression ']'
|
2022-12-02 23:57:30 -04:00
|
|
|
if tkn := self.expect(lx.IDENTIFIER):
|
2023-01-29 21:28:39 -04:00
|
|
|
type_text = ""
|
GH-98831: Typed stack effects, and more instructions converted (#99764)
Stack effects can now have a type, e.g. `inst(X, (left, right -- jump/uint64_t)) { ... }`.
Instructions converted to the non-legacy format:
* COMPARE_OP
* COMPARE_OP_FLOAT_JUMP
* COMPARE_OP_INT_JUMP
* COMPARE_OP_STR_JUMP
* STORE_ATTR
* DELETE_ATTR
* STORE_GLOBAL
* STORE_ATTR_INSTANCE_VALUE
* STORE_ATTR_WITH_HINT
* STORE_ATTR_SLOT, and complete the store_attr family
* Complete the store_subscr family: STORE_SUBSCR{,DICT,LIST_INT}
(STORE_SUBSCR was alread half converted,
but wasn't using cache effects yet.)
* DELETE_SUBSCR
* PRINT_EXPR
* INTERPRETER_EXIT (a bit weird, ends in return)
* RETURN_VALUE
* GET_AITER (had to restructure it some)
The original had mysterious `SET_TOP(NULL)` before `goto error`.
I assume those just account for `obj` having been decref'ed,
so I got rid of them in favor of the cleanup implied by `ERROR_IF()`.
* LIST_APPEND (a bit unhappy with it)
* SET_ADD (also a bit unhappy with it)
Various other improvements/refactorings as well.
2022-12-08 17:31:27 -04:00
|
|
|
if self.expect(lx.COLON):
|
2024-02-13 09:24:48 -04:00
|
|
|
type_text = self.require(lx.IDENTIFIER).text.strip()
|
|
|
|
if self.expect(lx.TIMES):
|
|
|
|
type_text += " *"
|
2023-01-29 21:28:39 -04:00
|
|
|
cond_text = ""
|
|
|
|
if self.expect(lx.IF):
|
|
|
|
self.require(lx.LPAREN)
|
|
|
|
if not (cond := self.expression()):
|
|
|
|
raise self.make_syntax_error("Expected condition")
|
|
|
|
self.require(lx.RPAREN)
|
|
|
|
cond_text = cond.text.strip()
|
|
|
|
size_text = ""
|
|
|
|
if self.expect(lx.LBRACKET):
|
|
|
|
if type_text or cond_text:
|
|
|
|
raise self.make_syntax_error("Unexpected [")
|
|
|
|
if not (size := self.expression()):
|
|
|
|
raise self.make_syntax_error("Expected expression")
|
2023-01-17 19:59:19 -04:00
|
|
|
self.require(lx.RBRACKET)
|
2024-06-26 16:10:43 -03:00
|
|
|
type_text = "_PyStackRef *"
|
2023-01-29 21:28:39 -04:00
|
|
|
size_text = size.text.strip()
|
2024-02-13 09:24:48 -04:00
|
|
|
return StackEffect(tkn.text, type_text, cond_text, size_text)
|
2023-07-24 13:38:23 -03:00
|
|
|
return None
|
2023-01-17 19:59:19 -04:00
|
|
|
|
|
|
|
@contextual
|
2023-01-29 21:28:39 -04:00
|
|
|
def expression(self) -> Expression | None:
|
2023-01-17 19:59:19 -04:00
|
|
|
tokens: list[lx.Token] = []
|
2023-02-07 21:35:55 -04:00
|
|
|
level = 1
|
|
|
|
while tkn := self.peek():
|
|
|
|
if tkn.kind in (lx.LBRACKET, lx.LPAREN):
|
|
|
|
level += 1
|
|
|
|
elif tkn.kind in (lx.RBRACKET, lx.RPAREN):
|
|
|
|
level -= 1
|
|
|
|
if level == 0:
|
|
|
|
break
|
2023-01-17 19:59:19 -04:00
|
|
|
tokens.append(tkn)
|
|
|
|
self.next()
|
|
|
|
if not tokens:
|
|
|
|
return None
|
2023-01-29 21:28:39 -04:00
|
|
|
return Expression(lx.to_text(tokens).strip())
|
2022-11-03 01:31:26 -03:00
|
|
|
|
2023-06-12 14:47:08 -03:00
|
|
|
# def ops(self) -> list[OpName] | None:
|
|
|
|
# if op := self.op():
|
|
|
|
# ops = [op]
|
|
|
|
# while self.expect(lx.PLUS):
|
|
|
|
# if op := self.op():
|
|
|
|
# ops.append(op)
|
|
|
|
# return ops
|
2022-11-06 13:40:47 -04:00
|
|
|
|
2022-12-02 23:57:30 -04:00
|
|
|
@contextual
|
|
|
|
def op(self) -> OpName | None:
|
|
|
|
if tkn := self.expect(lx.IDENTIFIER):
|
|
|
|
return OpName(tkn.text)
|
2023-07-24 13:38:23 -03:00
|
|
|
return None
|
2022-12-02 23:57:30 -04:00
|
|
|
|
|
|
|
@contextual
|
|
|
|
def macro_def(self) -> Macro | None:
|
2023-11-07 05:42:39 -04:00
|
|
|
if tkn := self.expect(lx.MACRO):
|
2022-12-02 23:57:30 -04:00
|
|
|
if self.expect(lx.LPAREN):
|
|
|
|
if tkn := self.expect(lx.IDENTIFIER):
|
|
|
|
if self.expect(lx.RPAREN):
|
|
|
|
if self.expect(lx.EQUALS):
|
|
|
|
if uops := self.uops():
|
|
|
|
self.require(lx.SEMI)
|
|
|
|
res = Macro(tkn.text, uops)
|
|
|
|
return res
|
2023-07-24 13:38:23 -03:00
|
|
|
return None
|
2022-12-02 23:57:30 -04:00
|
|
|
|
|
|
|
def uops(self) -> list[UOp] | None:
|
|
|
|
if uop := self.uop():
|
2023-07-24 13:38:23 -03:00
|
|
|
uop = cast(UOp, uop)
|
2022-12-02 23:57:30 -04:00
|
|
|
uops = [uop]
|
|
|
|
while self.expect(lx.PLUS):
|
|
|
|
if uop := self.uop():
|
2023-07-24 13:38:23 -03:00
|
|
|
uop = cast(UOp, uop)
|
2022-12-02 23:57:30 -04:00
|
|
|
uops.append(uop)
|
|
|
|
else:
|
|
|
|
raise self.make_syntax_error("Expected op name or cache effect")
|
|
|
|
return uops
|
2023-07-24 13:38:23 -03:00
|
|
|
return None
|
2022-12-02 23:57:30 -04:00
|
|
|
|
|
|
|
@contextual
|
|
|
|
def uop(self) -> UOp | None:
|
|
|
|
if tkn := self.expect(lx.IDENTIFIER):
|
|
|
|
if self.expect(lx.DIVIDE):
|
|
|
|
if num := self.expect(lx.NUMBER):
|
|
|
|
try:
|
|
|
|
size = int(num.text)
|
|
|
|
except ValueError:
|
|
|
|
raise self.make_syntax_error(
|
|
|
|
f"Expected integer, got {num.text!r}"
|
|
|
|
)
|
|
|
|
else:
|
|
|
|
return CacheEffect(tkn.text, size)
|
|
|
|
raise self.make_syntax_error("Expected integer")
|
|
|
|
else:
|
|
|
|
return OpName(tkn.text)
|
2023-07-24 13:38:23 -03:00
|
|
|
return None
|
2022-12-02 23:57:30 -04:00
|
|
|
|
2022-11-06 13:40:47 -04:00
|
|
|
@contextual
|
|
|
|
def family_def(self) -> Family | None:
|
2022-11-03 01:31:26 -03:00
|
|
|
if (tkn := self.expect(lx.IDENTIFIER)) and tkn.text == "family":
|
2022-11-15 23:59:19 -04:00
|
|
|
size = None
|
2022-11-03 01:31:26 -03:00
|
|
|
if self.expect(lx.LPAREN):
|
2022-12-02 23:57:30 -04:00
|
|
|
if tkn := self.expect(lx.IDENTIFIER):
|
2022-11-15 23:59:19 -04:00
|
|
|
if self.expect(lx.COMMA):
|
|
|
|
if not (size := self.expect(lx.IDENTIFIER)):
|
2023-09-07 10:39:03 -03:00
|
|
|
if not (size := self.expect(lx.NUMBER)):
|
2024-01-12 13:30:27 -04:00
|
|
|
raise self.make_syntax_error(
|
|
|
|
"Expected identifier or number"
|
|
|
|
)
|
2022-11-03 01:31:26 -03:00
|
|
|
if self.expect(lx.RPAREN):
|
|
|
|
if self.expect(lx.EQUALS):
|
2022-11-15 23:59:19 -04:00
|
|
|
if not self.expect(lx.LBRACE):
|
|
|
|
raise self.make_syntax_error("Expected {")
|
2022-11-03 01:31:26 -03:00
|
|
|
if members := self.members():
|
2022-11-15 23:59:19 -04:00
|
|
|
if self.expect(lx.RBRACE) and self.expect(lx.SEMI):
|
2022-12-02 23:57:30 -04:00
|
|
|
return Family(
|
|
|
|
tkn.text, size.text if size else "", members
|
|
|
|
)
|
2022-11-03 01:31:26 -03:00
|
|
|
return None
|
|
|
|
|
2023-11-30 07:03:30 -04:00
|
|
|
def flags(self) -> list[str]:
|
|
|
|
here = self.getpos()
|
|
|
|
if self.expect(lx.LPAREN):
|
|
|
|
if tkn := self.expect(lx.IDENTIFIER):
|
|
|
|
flags = [tkn.text]
|
|
|
|
while self.expect(lx.COMMA):
|
|
|
|
if tkn := self.expect(lx.IDENTIFIER):
|
|
|
|
flags.append(tkn.text)
|
|
|
|
else:
|
|
|
|
break
|
|
|
|
if not self.expect(lx.RPAREN):
|
|
|
|
raise self.make_syntax_error("Expected comma or right paren")
|
|
|
|
return flags
|
|
|
|
self.setpos(here)
|
|
|
|
return []
|
|
|
|
|
2023-06-11 18:31:59 -03:00
|
|
|
@contextual
|
|
|
|
def pseudo_def(self) -> Pseudo | None:
|
|
|
|
if (tkn := self.expect(lx.IDENTIFIER)) and tkn.text == "pseudo":
|
|
|
|
size = None
|
|
|
|
if self.expect(lx.LPAREN):
|
|
|
|
if tkn := self.expect(lx.IDENTIFIER):
|
2023-11-30 07:03:30 -04:00
|
|
|
if self.expect(lx.COMMA):
|
2024-05-29 06:47:56 -03:00
|
|
|
inp, outp = self.io_effect()
|
|
|
|
if self.expect(lx.COMMA):
|
|
|
|
flags = self.flags()
|
|
|
|
else:
|
|
|
|
flags = []
|
|
|
|
if self.expect(lx.RPAREN):
|
|
|
|
if self.expect(lx.EQUALS):
|
|
|
|
if not self.expect(lx.LBRACE):
|
|
|
|
raise self.make_syntax_error("Expected {")
|
|
|
|
if members := self.members():
|
|
|
|
if self.expect(lx.RBRACE) and self.expect(lx.SEMI):
|
|
|
|
return Pseudo(tkn.text, inp, outp, flags, members)
|
2023-06-11 18:31:59 -03:00
|
|
|
return None
|
|
|
|
|
2022-11-08 12:22:56 -04:00
|
|
|
def members(self) -> list[str] | None:
|
2022-11-03 01:31:26 -03:00
|
|
|
here = self.getpos()
|
|
|
|
if tkn := self.expect(lx.IDENTIFIER):
|
2022-11-15 23:59:19 -04:00
|
|
|
members = [tkn.text]
|
|
|
|
while self.expect(lx.COMMA):
|
|
|
|
if tkn := self.expect(lx.IDENTIFIER):
|
|
|
|
members.append(tkn.text)
|
|
|
|
else:
|
|
|
|
break
|
|
|
|
peek = self.peek()
|
|
|
|
if not peek or peek.kind != lx.RBRACE:
|
|
|
|
raise self.make_syntax_error("Expected comma or right paren")
|
|
|
|
return members
|
2022-11-03 01:31:26 -03:00
|
|
|
self.setpos(here)
|
|
|
|
return None
|
|
|
|
|
|
|
|
@contextual
|
2023-01-29 21:28:39 -04:00
|
|
|
def block(self) -> Block | None:
|
2023-01-17 19:59:19 -04:00
|
|
|
if self.c_blob():
|
|
|
|
return Block()
|
2023-07-24 13:38:23 -03:00
|
|
|
return None
|
2022-11-03 01:31:26 -03:00
|
|
|
|
2022-11-08 12:22:56 -04:00
|
|
|
def c_blob(self) -> list[lx.Token]:
|
|
|
|
tokens: list[lx.Token] = []
|
2022-11-03 01:31:26 -03:00
|
|
|
level = 0
|
|
|
|
while tkn := self.next(raw=True):
|
2023-01-13 21:06:45 -04:00
|
|
|
tokens.append(tkn)
|
2022-11-03 01:31:26 -03:00
|
|
|
if tkn.kind in (lx.LBRACE, lx.LPAREN, lx.LBRACKET):
|
|
|
|
level += 1
|
|
|
|
elif tkn.kind in (lx.RBRACE, lx.RPAREN, lx.RBRACKET):
|
|
|
|
level -= 1
|
|
|
|
if level <= 0:
|
|
|
|
break
|
|
|
|
return tokens
|
|
|
|
|
|
|
|
|
|
|
|
if __name__ == "__main__":
|
|
|
|
import sys
|
2022-12-02 23:57:30 -04:00
|
|
|
|
2022-11-03 01:31:26 -03:00
|
|
|
if sys.argv[1:]:
|
|
|
|
filename = sys.argv[1]
|
|
|
|
if filename == "-c" and sys.argv[2:]:
|
|
|
|
src = sys.argv[2]
|
2022-11-22 20:04:57 -04:00
|
|
|
filename = "<string>"
|
2022-11-03 01:31:26 -03:00
|
|
|
else:
|
2023-06-12 14:47:08 -03:00
|
|
|
with open(filename, "r") as f:
|
2022-11-03 01:31:26 -03:00
|
|
|
src = f.read()
|
|
|
|
srclines = src.splitlines()
|
|
|
|
begin = srclines.index("// BEGIN BYTECODES //")
|
|
|
|
end = srclines.index("// END BYTECODES //")
|
2022-12-02 23:57:30 -04:00
|
|
|
src = "\n".join(srclines[begin + 1 : end])
|
2022-11-03 01:31:26 -03:00
|
|
|
else:
|
2022-11-22 20:04:57 -04:00
|
|
|
filename = "<default>"
|
2022-11-03 01:31:26 -03:00
|
|
|
src = "if (x) { x.foo; // comment\n}"
|
|
|
|
parser = Parser(src, filename)
|
2022-12-02 23:57:30 -04:00
|
|
|
x = parser.definition()
|
2022-11-03 01:31:26 -03:00
|
|
|
print(x)
|