Hi there,
I've noticed that, when a frame's __builtins__ is a subclass of dict with an
overridden __getitem__ method, this overriden method is not used by the
IMPORT_NAME instruction to lookup __import__ in the dictionary; it uses the
lookup function of normal dictionaries (via _PyDict_GetItemIdWithError). This
is contrary to the behaviour of the similar LOAD_BUILD_CLASS, as well as the
typical name lookup semantics of LOAD_GLOBAL/LOAD_NAME, which all use
PyDict_CheckExact for a "fast path" before defaulting to PyObject_GetItem,
which is unexpected.
Perhaps more seriously, if __builtins__ is not a dict at all, then it gets
erroneously passed to some internal dict functions resulting in a mysterious
SystemError ("Objects/dictobject.c:1440: bad argument to internal function")
which, to me, indicates fragile behaviour that isn't supposed to happen.
I'm not sure if this intended, so I didn't want to open an issue yet. It also
seems a highly specific use case and changing it would probably cause a bit of
a slow-down in module importing so is perhaps not worth fixing. I just wanted
to ask here in case this issue had been documented anywhere before, and to
check if it might actually be supposed to happen before opening a bug report.
I cannot find evidence that this behaviour has changed at all in recent history
and it seems to be the same on the main branch as in 3.9.6.
A short demo of these things is attached.
Links to relevant CPython code in v3.9.6:
IMPORT_NAME: https://github.com/python/cpython/blob/v3.9.6/Python/ceval.c#L5179
BUILD_CLASS: https://github.com/python/cpython/blob/v3.9.6/Python/ceval.c#L2316
LOAD_NAME: https://github.com/python/cpython/blob/v3.9.6/Python/ceval.c#L2488
LOAD_GLOBAL: https://github.com/python/cpython/blob/v3.9.6/Python/ceval.c#L2546
Thanks,
Patrick Reader
class MyDict(dict):
# keep a reference around to avoid infinite recursion
print = print
dict = dict
def __getitem__(self, key):
self.print("getting:", key)
# Can't use super here because we'd have to keep a reference around instead of looking it up
# in __builtins__ (to prevent infinite recursion), but then there's no __class__ cell which
# breaks the lookup mechanism. Instead, just refer to dict by name
return self.dict.__getitem__(self, key)
__builtins__ = MyDict(vars(__builtins__))
int # prints "getting: int"
__import__ # prints "getting: __import__"
class X: pass # prints "getting: __build_class__"
import math # does not print "getting: __import__" because it uses dictobject internal lookup
################################################################################
# try these individually in the Python shell, because they all error on their own
__builtins__ = "not a dictionary"
int # TypeError: string indices must be integers (because it's trying to do effectively `"not a dictionary"["int"]`)
__import__ # same error
class X: pass # same error (trying to load __build_class__)
import math # SystemError: Objects/dictobject.c:1440: bad argument to internal function
_______________________________________________
Python-Dev mailing list -- python-dev@python.org
To unsubscribe send an email to python-dev-le...@python.org
https://mail.python.org/mailman3/lists/python-dev.python.org/
Message archived at
https://mail.python.org/archives/list/python-dev@python.org/message/ZQMF6XC76J4APJPB3X6PGATG6CV5NN44/
Code of Conduct: http://python.org/psf/codeofconduct/
