| 1 | """
|
| 2 | conversion_pass.py - forward declarations, and virtuals
|
| 3 | """
|
| 4 | import mypy
|
| 5 |
|
| 6 | from mypy.nodes import (Expression, NameExpr, MemberExpr, TupleExpr, CallExpr,
|
| 7 | ClassDef, FuncDef, Argument)
|
| 8 | from mypy.types import Type, Instance, TupleType, NoneType, PartialType
|
| 9 |
|
| 10 | from mycpp import util
|
| 11 | from mycpp.util import log, SplitPyName
|
| 12 | from mycpp import pass_state
|
| 13 | from mycpp import visitor
|
| 14 | from mycpp import cppgen_pass
|
| 15 |
|
| 16 | from typing import Dict, List, Tuple, Optional, TYPE_CHECKING
|
| 17 |
|
| 18 | if TYPE_CHECKING:
|
| 19 | #from mycpp import cppgen_pass
|
| 20 | pass
|
| 21 |
|
| 22 | _ = log
|
| 23 |
|
| 24 | DotExprs = Dict[MemberExpr, pass_state.member_t]
|
| 25 |
|
| 26 |
|
| 27 | class MyTypeInfo:
|
| 28 | """Like mypy.nodes.TypeInfo"""
|
| 29 |
|
| 30 | def __init__(self, fullname: str) -> None:
|
| 31 | self.fullname = fullname
|
| 32 |
|
| 33 |
|
| 34 | class Primitive(Instance):
|
| 35 |
|
| 36 | def __init__(self, name: str, args: List[Type] = None) -> None:
|
| 37 | self.type = MyTypeInfo(name) # type: ignore
|
| 38 | self.args = args if args is not None else []
|
| 39 |
|
| 40 |
|
| 41 | MYCPP_INT = Primitive('builtins.int')
|
| 42 |
|
| 43 |
|
| 44 | class Pass(visitor.SimpleVisitor):
|
| 45 |
|
| 46 | def __init__(
|
| 47 | self,
|
| 48 | types: Dict[Expression, Type],
|
| 49 | virtual: pass_state.Virtual,
|
| 50 | forward_decls: List[str],
|
| 51 | all_member_vars: 'cppgen_pass.AllMemberVars',
|
| 52 | all_local_vars: 'cppgen_pass.AllLocalVars',
|
| 53 | module_dot_exprs: DotExprs,
|
| 54 | yield_out_params: Dict[FuncDef, Tuple[str, str]], # output
|
| 55 | dunder_exit_special: Dict[FuncDef, bool],
|
| 56 | ) -> None:
|
| 57 | visitor.SimpleVisitor.__init__(self)
|
| 58 |
|
| 59 | # Input
|
| 60 | self.types = types
|
| 61 |
|
| 62 | # These are all outputs we compute
|
| 63 | self.virtual = virtual
|
| 64 | self.forward_decls = forward_decls
|
| 65 | self.all_member_vars = all_member_vars
|
| 66 | self.all_local_vars = all_local_vars
|
| 67 | self.module_dot_exprs = module_dot_exprs
|
| 68 | # Used to add another param to definition, and
|
| 69 | # yield x --> YIELD->append(x)
|
| 70 | self.yield_out_params = yield_out_params
|
| 71 | self.dunder_exit_special = dunder_exit_special
|
| 72 |
|
| 73 | # Internal state
|
| 74 | self.inside_dunder_exit = None
|
| 75 | self.current_member_vars: Dict[str, 'cppgen_pass.MemberVar'] = {}
|
| 76 | self.current_local_vars: List[Tuple[str, Type]] = []
|
| 77 |
|
| 78 | # Where do we need to update current_local_vars?
|
| 79 | #
|
| 80 | # x = 42 # oils_visit_assignment_stmt
|
| 81 | # a, b = foo
|
| 82 |
|
| 83 | # x = [y for y in other] # oils_visit_assign_to_listcomp_:
|
| 84 | #
|
| 85 | # Special case for enumerate:
|
| 86 | # for i, x in enumerate(other):
|
| 87 | #
|
| 88 | # def f(p, q): # params are locals, _WriteFuncParams
|
| 89 | # # but only if update_locals
|
| 90 |
|
| 91 | self.imported_names = set() # MemberExpr -> module::Foo() or self->foo
|
| 92 | # HACK for conditional import inside mylib.PYTHON
|
| 93 | # in core/shell.py
|
| 94 | self.imported_names.add('help_meta')
|
| 95 |
|
| 96 | def visit_import(self, o: 'mypy.nodes.Import') -> None:
|
| 97 | for name, as_name in o.ids:
|
| 98 | if as_name is not None:
|
| 99 | # import time as time_
|
| 100 | self.imported_names.add(as_name)
|
| 101 | else:
|
| 102 | # import libc
|
| 103 | self.imported_names.add(name)
|
| 104 |
|
| 105 | def visit_import_from(self, o: 'mypy.nodes.ImportFrom') -> None:
|
| 106 | """
|
| 107 | Write C++ namespace aliases and 'using' for imports.
|
| 108 | We need them in the 'decl' phase for default arguments like
|
| 109 | runtime_asdl::scope_e -> scope_e
|
| 110 | """
|
| 111 | # For MemberExpr . -> module::func() or this->field. Also needed in
|
| 112 | # the decl phase for default arg values.
|
| 113 | for name, alias in o.names:
|
| 114 | if alias:
|
| 115 | self.imported_names.add(alias)
|
| 116 | else:
|
| 117 | self.imported_names.add(name)
|
| 118 |
|
| 119 | def oils_visit_member_expr(self, o: 'mypy.nodes.MemberExpr') -> None:
|
| 120 | # Why is self.types[o] missing some types? e.g. hnode.Record() call in
|
| 121 | # asdl/runtime.py, failing with KeyError NameExpr
|
| 122 | lhs_type = self.types.get(o.expr) # type: Optional[Type]
|
| 123 |
|
| 124 | is_small_str = False
|
| 125 | if util.SMALL_STR:
|
| 126 | if util.IsStr(lhs_type):
|
| 127 | is_small_str = True
|
| 128 |
|
| 129 | # This is an approximate hack that assumes that locals don't shadow
|
| 130 | # imported names. Might be a problem with names like 'word'?
|
| 131 | if is_small_str:
|
| 132 | # mystr.upper()
|
| 133 | dot = pass_state.StackObjectMember(
|
| 134 | o.expr, lhs_type, o.name) # type: pass_state.member_t
|
| 135 |
|
| 136 | elif o.name in ('CreateNull', 'Take'):
|
| 137 | # heuristic for MyType::CreateNull()
|
| 138 | # MyType::Take(other)
|
| 139 | type_name = self.types[o].ret_type.type.fullname
|
| 140 | dot = pass_state.StaticClassMember(type_name, o.name)
|
| 141 | elif (isinstance(o.expr, NameExpr) and
|
| 142 | o.expr.name in self.imported_names):
|
| 143 | # heuristic for state::Mem()
|
| 144 | module_path = SplitPyName(o.expr.fullname or o.expr.name)
|
| 145 | dot = pass_state.ModuleMember(module_path, o.name)
|
| 146 | else:
|
| 147 | # mylist->append(42)
|
| 148 | dot = pass_state.HeapObjectMember(o.expr, lhs_type, o.name)
|
| 149 |
|
| 150 | self.module_dot_exprs[o] = dot
|
| 151 |
|
| 152 | self.accept(o.expr)
|
| 153 |
|
| 154 | def oils_visit_mypy_file(self, o: 'mypy.nodes.MypyFile') -> None:
|
| 155 | mod_parts = o.fullname.split('.')
|
| 156 | comment = 'forward declare'
|
| 157 |
|
| 158 | self.write('namespace %s { // %s\n', mod_parts[-1], comment)
|
| 159 |
|
| 160 | # Do default traversal
|
| 161 | self.indent += 1
|
| 162 | super().oils_visit_mypy_file(o)
|
| 163 | self.indent -= 1
|
| 164 |
|
| 165 | self.write('}\n')
|
| 166 | self.write('\n')
|
| 167 |
|
| 168 | def oils_visit_class_def(
|
| 169 | self, o: 'mypy.nodes.ClassDef',
|
| 170 | base_class_sym: Optional[util.SymbolPath],
|
| 171 | current_class_name: Optional[util.SymbolPath]) -> None:
|
| 172 | self.write_ind('class %s;\n', o.name)
|
| 173 | if base_class_sym:
|
| 174 | self.virtual.OnSubclass(base_class_sym, current_class_name)
|
| 175 |
|
| 176 | # Do default traversal of methods, associating member vars with the
|
| 177 | # ClassDef node
|
| 178 | self.current_member_vars = {}
|
| 179 | super().oils_visit_class_def(o, base_class_sym, current_class_name)
|
| 180 | self.all_member_vars[o] = self.current_member_vars
|
| 181 |
|
| 182 | def _ValidateDefaultArg(self, arg: Argument) -> None:
|
| 183 | t = self.types[arg.initializer]
|
| 184 |
|
| 185 | valid = False
|
| 186 | if isinstance(t, NoneType):
|
| 187 | valid = True
|
| 188 | if isinstance(t, Instance):
|
| 189 | # Allowing strings since they're immutable, e.g.
|
| 190 | # prefix='' seems OK
|
| 191 | if t.type.fullname in ('builtins.bool', 'builtins.int',
|
| 192 | 'builtins.float', 'builtins.str'):
|
| 193 | valid = True
|
| 194 |
|
| 195 | # ASDL enums lex_mode_t, scope_t, ...
|
| 196 | if t.type.fullname.endswith('_t'):
|
| 197 | valid = True
|
| 198 |
|
| 199 | # Hack for loc__Missing. Should detect the general case.
|
| 200 | if t.type.fullname.endswith('loc__Missing'):
|
| 201 | valid = True
|
| 202 |
|
| 203 | if not valid:
|
| 204 | self.report_error(
|
| 205 | arg,
|
| 206 | 'Invalid default arg %r of type %s (not None, bool, int, float, ASDL enum)'
|
| 207 | % (arg.initializer, t))
|
| 208 |
|
| 209 | def _ValidateDefaultArgs(self, func_def: FuncDef) -> None:
|
| 210 | arguments = func_def.arguments
|
| 211 |
|
| 212 | num_defaults = 0
|
| 213 | for arg in arguments:
|
| 214 | if arg.initializer:
|
| 215 | self._ValidateDefaultArg(arg)
|
| 216 | num_defaults += 1
|
| 217 |
|
| 218 | if num_defaults > 1:
|
| 219 | # Report on first arg
|
| 220 | self.report_error(
|
| 221 | arg, '%s has %d default arguments. Only 1 is allowed' %
|
| 222 | (func_def.name, num_defaults))
|
| 223 | return
|
| 224 |
|
| 225 | def oils_visit_func_def(self, o: 'mypy.nodes.FuncDef',
|
| 226 | current_class_name: Optional[util.SymbolPath],
|
| 227 | current_method_name: Optional[str]) -> None:
|
| 228 | self._ValidateDefaultArgs(o)
|
| 229 |
|
| 230 | self.virtual.OnMethod(current_class_name, o.name)
|
| 231 |
|
| 232 | self.current_local_vars = []
|
| 233 |
|
| 234 | # Add params as local vars, but only if we're NOT in a constructor.
|
| 235 | # This is borrowed from cppgen_pass -
|
| 236 | # _ConstructorImpl has update_locals=False, likewise for decl
|
| 237 | # Is this just a convention?
|
| 238 | # Counterexample: what if locals are used in __init__ after allocation?
|
| 239 | # Are we assuming we never do mylib.MaybeCollect() inside a
|
| 240 | # constructor? We can check that too.
|
| 241 |
|
| 242 | if current_method_name != '__init__':
|
| 243 | # Add function params as locals, to be rooted
|
| 244 | arg_types = o.type.arg_types
|
| 245 | arg_names = [arg.variable.name for arg in o.arguments]
|
| 246 | for name, typ in zip(arg_names, arg_types):
|
| 247 | if name == 'self':
|
| 248 | continue
|
| 249 | self.current_local_vars.append((name, typ))
|
| 250 |
|
| 251 | # Traverse to collect member variables
|
| 252 | super().oils_visit_func_def(o, current_class_name, current_method_name)
|
| 253 | self.all_local_vars[o] = self.current_local_vars
|
| 254 |
|
| 255 | # Is this function is a generator? Then associate the node with an
|
| 256 | # accumulator param (name and type).
|
| 257 | # This is info is consumed by both the Decl and Impl passes
|
| 258 | _, _, c_iter_list_type = cppgen_pass.GetCReturnType(o.type.ret_type)
|
| 259 | if c_iter_list_type is not None:
|
| 260 | self.yield_out_params[o] = ('YIELD', c_iter_list_type)
|
| 261 |
|
| 262 | def oils_visit_dunder_exit(self, o: ClassDef, stmt: FuncDef,
|
| 263 | base_class_sym: util.SymbolPath) -> None:
|
| 264 | self.inside_dunder_exit = o
|
| 265 | super().oils_visit_dunder_exit(o, stmt, base_class_sym)
|
| 266 | self.inside_dunder_exit = None
|
| 267 |
|
| 268 | def visit_return_stmt(self, o: 'mypy.nodes.ReturnStmt') -> None:
|
| 269 | # Mark special destructors
|
| 270 | if self.inside_dunder_exit:
|
| 271 | self.dunder_exit_special[self.inside_dunder_exit] = True
|
| 272 | super().visit_return_stmt(o)
|
| 273 |
|
| 274 | def visit_raise_stmt(self, o: 'mypy.nodes.RaiseStmt') -> None:
|
| 275 | if self.inside_dunder_exit:
|
| 276 | # Note: this doesn't check function calls that raise, but it's
|
| 277 | # better than nothing
|
| 278 | self.report_error(
|
| 279 | o, "raise not allowed within __exit__ (C++ doesn't allow it)")
|
| 280 | return
|
| 281 | super().visit_raise_stmt(o)
|
| 282 |
|
| 283 | def oils_visit_assign_to_listcomp(self, lval: NameExpr,
|
| 284 | left_expr: Expression,
|
| 285 | index_expr: Expression, seq: Expression,
|
| 286 | cond: Expression) -> None:
|
| 287 | # We need to consider 'result' a local var:
|
| 288 | # result = [x for x in other]
|
| 289 |
|
| 290 | # what about yield accumulator, like
|
| 291 | # it_g = g(n)
|
| 292 | self.current_local_vars.append((lval.name, self.types[lval]))
|
| 293 |
|
| 294 | super().oils_visit_assign_to_listcomp(lval, left_expr, index_expr, seq,
|
| 295 | cond)
|
| 296 |
|
| 297 | def _MaybeAddMember(self, lval: MemberExpr,
|
| 298 | current_method_name: Optional[str],
|
| 299 | at_global_scope: bool) -> None:
|
| 300 | assert not at_global_scope, "Members shouldn't be assigned at the top level"
|
| 301 |
|
| 302 | # Collect statements that look like self.foo = 1
|
| 303 | # Only do this in __init__ so that a derived class mutating a field
|
| 304 | # from the base class doesn't cause duplicate C++ fields. (C++
|
| 305 | # allows two fields of the same name!)
|
| 306 | #
|
| 307 | # HACK for WordParser: also include Reset(). We could change them
|
| 308 | # all up front but I kinda like this.
|
| 309 | if current_method_name not in ('__init__', 'Reset'):
|
| 310 | return
|
| 311 |
|
| 312 | if isinstance(lval.expr, NameExpr) and lval.expr.name == 'self':
|
| 313 | #log(' lval.name %s', lval.name)
|
| 314 | lval_type = self.types[lval]
|
| 315 | c_type = cppgen_pass.GetCType(lval_type)
|
| 316 | is_managed = cppgen_pass.CTypeIsManaged(c_type)
|
| 317 | self.current_member_vars[lval.name] = (lval_type, c_type,
|
| 318 | is_managed)
|
| 319 |
|
| 320 | def oils_visit_assignment_stmt(self, o: 'mypy.nodes.AssignmentStmt',
|
| 321 | lval: Expression, rval: Expression,
|
| 322 | current_method_name: Optional[str],
|
| 323 | at_global_scope: bool) -> None:
|
| 324 |
|
| 325 | if isinstance(lval, MemberExpr):
|
| 326 | self._MaybeAddMember(lval, current_method_name, at_global_scope)
|
| 327 |
|
| 328 | if lval in self.types and isinstance(self.types[lval], PartialType):
|
| 329 | t = self.types[lval]
|
| 330 | self.report_error(
|
| 331 | o,
|
| 332 | "Mismatched types: trying to assign expression of type '%s' to "
|
| 333 | "a PartialType variable '%s' (was likely assigned None before).\n"
|
| 334 | "Tip: If your type translates to a heap-allocated type (e.g. str "
|
| 335 | "or class), you can annotate it with '# type: Optional[T]'. "
|
| 336 | "If your type is allocated on the stack (int), then you can use "
|
| 337 | "-1 or similar as an in-band null value" %
|
| 338 | (t.var.type.type.name, t.var.name))
|
| 339 | return
|
| 340 |
|
| 341 | # Handle:
|
| 342 | # x = y
|
| 343 | # These two are special cases in cppgen_pass, but not here
|
| 344 | # x = NewDict()
|
| 345 | # x = cast(T, y)
|
| 346 | #
|
| 347 | # Note: this has duplicates: the 'done' set in visit_block() handles
|
| 348 | # it. Could make it a Dict.
|
| 349 | if isinstance(lval, NameExpr):
|
| 350 | rval_type = self.types[rval]
|
| 351 |
|
| 352 | # Two pieces of logic adapted from cppgen_pass: is_iterator and is_cast.
|
| 353 | # Can we simplify them?
|
| 354 |
|
| 355 | is_iterator = (isinstance(rval_type, Instance) and
|
| 356 | rval_type.type.fullname == 'typing.Iterator')
|
| 357 |
|
| 358 | # Downcasted vars are BLOCK-scoped, not FUNCTION-scoped, so they
|
| 359 | # don't become local vars. They are also ALIASED, so they don't
|
| 360 | # need to be rooted.
|
| 361 | is_downcast_and_shadow = False
|
| 362 | if (isinstance(rval, CallExpr) and
|
| 363 | isinstance(rval.callee, NameExpr) and
|
| 364 | rval.callee.name == 'cast'):
|
| 365 | to_cast = rval.args[1]
|
| 366 | if (isinstance(to_cast, NameExpr) and
|
| 367 | to_cast.name.startswith('UP_')):
|
| 368 | is_downcast_and_shadow = True
|
| 369 |
|
| 370 | if (not at_global_scope and not is_iterator and
|
| 371 | not is_downcast_and_shadow):
|
| 372 | self.current_local_vars.append((lval.name, self.types[lval]))
|
| 373 |
|
| 374 | # Handle local vars, like _write_tuple_unpacking
|
| 375 |
|
| 376 | # This handles:
|
| 377 | # a, b = func_that_returns_tuple()
|
| 378 | if isinstance(lval, TupleExpr):
|
| 379 | if isinstance(rval, TupleExpr):
|
| 380 | self.report_error(
|
| 381 | o, "mycpp currently does not handle code like "
|
| 382 | "'a, b = x, y'. You can move each assignment to its own line "
|
| 383 | "or return (x, y) from a function instead.")
|
| 384 | return
|
| 385 |
|
| 386 | rval_type = self.types[rval]
|
| 387 | assert isinstance(rval_type, TupleType), rval_type
|
| 388 |
|
| 389 | for i, (lval_item,
|
| 390 | item_type) in enumerate(zip(lval.items, rval_type.items)):
|
| 391 | #self.log('*** %s :: %s', lval_item, item_type)
|
| 392 | if isinstance(lval_item, NameExpr):
|
| 393 | if util.SkipAssignment(lval_item.name):
|
| 394 | continue
|
| 395 | self.current_local_vars.append((lval_item.name, item_type))
|
| 396 |
|
| 397 | # self.a, self.b = foo()
|
| 398 | if isinstance(lval_item, MemberExpr):
|
| 399 | self._MaybeAddMember(lval_item, current_method_name,
|
| 400 | at_global_scope)
|
| 401 |
|
| 402 | super().oils_visit_assignment_stmt(o, lval, rval, current_method_name,
|
| 403 | at_global_scope)
|
| 404 |
|
| 405 | def oils_visit_for_stmt(self, o: 'mypy.nodes.ForStmt',
|
| 406 | func_name: Optional[str]) -> None:
|
| 407 | # TODO: this variable should be BLOCK scoped, not function scoped, like
|
| 408 | # the tuple variables for i, x
|
| 409 | index0_name: Optional[str] = None
|
| 410 | if func_name == 'enumerate':
|
| 411 | assert isinstance(o.index, TupleExpr), o.index
|
| 412 | index0 = o.index.items[0]
|
| 413 | assert isinstance(index0, NameExpr), index0
|
| 414 | index0_name = index0.name # generate int i = 0; ; ++i
|
| 415 |
|
| 416 | if index0_name:
|
| 417 | # can't initialize two things in a for loop, so do it on a separate line
|
| 418 | self.current_local_vars.append((index0_name, MYCPP_INT))
|
| 419 |
|
| 420 | super().oils_visit_for_stmt(o, func_name)
|