def infer_type(self, expr: E.Apply) -> T.Base:
if not expr.arguments:
raise Error.WrongArity(expr, 1)
if not expr.arguments[0].type.coerces(T.File(optional=True)):
if isinstance(expr.arguments[0].type, T.Array):
if expr.arguments[0].type.optional or not expr.arguments[0].type.item_type.coerces(
T.File(optional=True)
):
raise Error.StaticTypeMismatch(
expr.arguments[0], T.Array(T.File(optional=True)), expr.arguments[0].type
)
else:
raise Error.StaticTypeMismatch(
expr.arguments[0], T.File(optional=True), expr.arguments[0].type
)
if len(expr.arguments) == 2:
if expr.arguments[1].type != T.String():
raise Error.StaticTypeMismatch(
expr.arguments[1], T.String(), expr.arguments[1].type
)
elif len(expr.arguments) > 2:
raise Error.WrongArity(expr, 2)
return T.Float()
def file_type(self, items, meta):
optional = False
if items and items[0].value == "?":
optional = True
return T.File(optional)
def __init__(self):
# language built-ins
self._at = _At()
self._land = _And()
self._lor = _Or()
self._negate = StaticFunction(
"_negate", [T.Boolean()], T.Boolean(), lambda x: V.Boolean(not x.value)
)
self._add = _AddOperator()
self._sub = _ArithmeticOperator("-", lambda l, r: l - r)
self._mul = _ArithmeticOperator("*", lambda l, r: l * r)
self._div = _ArithmeticOperator("/", lambda l, r: l // r)
self._rem = StaticFunction(
"_rem", [T.Int(), T.Int()], T.Int(), lambda l, r: V.Int(l.value % r.value)
)
self._eqeq = _ComparisonOperator("==", lambda l, r: l == r)
self._neq = _ComparisonOperator("!=", lambda l, r: l != r)
self._lt = _ComparisonOperator("<", lambda l, r: l < r)
self._lte = _ComparisonOperator("<=", lambda l, r: l <= r)
self._gt = _ComparisonOperator(">", lambda l, r: l > r)
self._gte = _ComparisonOperator(">=", lambda l, r: l >= r)
# static stdlib functions
for (name, argument_types, return_type, F) in [
("floor", [T.Float()], T.Int(), lambda v: V.Int(math.floor(v.value))),
("ceil", [T.Float()], T.Int(), lambda v: V.Int(math.ceil(v.value))),
("round", [T.Float()], T.Int(), lambda v: V.Int(round(v.value))),
("length", [T.Array(T.Any())], T.Int(), lambda v: V.Int(len(v.value))),
("sub", [T.String(), T.String(), T.String()], T.String(), _sub),
("basename", [T.String(), T.String(optional=True)], T.String(), _basename),
(
"defined",
[T.Any(optional=True)],
T.Boolean(),
lambda v: V.Boolean(not isinstance(v, V.Null)),
),
# context-dependent:
("write_lines", [T.Array(T.String())], T.File(), _notimpl),
("write_tsv", [T.Array(T.Array(T.String()))], T.File(), _notimpl),
("write_map", [T.Map((T.Any(), T.Any()))], T.File(), _notimpl),
("write_json", [T.Any()], T.File(), _notimpl),
("stdout", [], T.File(), _notimpl),
("stderr", [], T.File(), _notimpl),
("glob", [T.String()], T.Array(T.File()), _notimpl),
("read_int", [T.File()], T.Int(), _notimpl),
("read_boolean", [T.File()], T.Boolean(), _notimpl),
("read_string", [T.File()], T.String(), _notimpl),
("read_float", [T.File()], T.Float(), _notimpl),
("read_array", [T.File()], T.Array(T.Any()), _notimpl),
("read_map", [T.File()], T.Map((T.Any(), T.Any())), _notimpl),
("read_lines", [T.File()], T.Array(T.Any()), _notimpl),
("read_tsv", [T.File()], T.Array(T.Array(T.String())), _notimpl),
("read_json", [T.File()], T.Any(), _notimpl),
]:
setattr(self, name, StaticFunction(name, argument_types, return_type, F))
# polymorphically typed stdlib functions which require specialized
# infer_type logic
self.range = _Range()
self.prefix = _Prefix()
self.size = _Size()
self.select_first = _SelectFirst()
self.select_all = _SelectAll()
self.zip = _Zip()
self.cross = _Zip() # FIXME
self.flatten = _Flatten()
self.transpose = _Transpose()
for arg, ty in zip(expr.arguments, self.argument_types)
]
ans: V.Base = self.F(*argument_values)
return ans.coerce(self.return_type)
def _notimpl(one: Any = None, two: Any = None) -> None:
exec("raise NotImplementedError()")
_static_functions: List[Tuple[str, List[T.Base], T.Base, Any]] = [
("_negate", [T.Boolean()], T.Boolean(),
lambda x: V.Boolean(not x.value)), # pyre-fixme
("_rem", [T.Int(), T.Int()], T.Int(),
lambda l, r: V.Int(l.value % r.value)), # pyre-fixme
("stdout", [], T.File(), _notimpl),
("basename", [T.String(), T.String(optional=True)], T.String(), _notimpl),
# note: size() can take an empty value and probably returns 0 in that case.
# e.g. https://github.com/DataBiosphere/topmed-workflows/blob/31ba8a714b36ada929044f2ba3d130936e6c740e/CRAM-no-header-md5sum/md5sum/CRAM_md5sum.wdl#L39
("size", [T.File(optional=True),
T.String(optional=True)], T.Float(), _notimpl),
("ceil", [T.Float()], T.Int(), _notimpl),
("round", [T.Float()], T.Int(), _notimpl),
("glob", [T.String()], T.Array(T.File()), _notimpl),
("read_int", [T.String()], T.Int(), _notimpl),
("read_boolean", [T.String()], T.Boolean(), _notimpl),
("read_string", [T.String()], T.String(), _notimpl),
("read_float", [T.String()], T.Float(), _notimpl),
("read_array", [T.String()], T.Array(None), _notimpl),
("read_map", [T.String()], T.Map(None), _notimpl),
("read_lines", [T.String()], T.Array(None), _notimpl),
