but it would be pretty simple all things considered
#esoteric-python
proper vault
floral meteor
lmao
you wanna bet
although the bot cannot import custom scripts
ยฏ_(ใ)_/ยฏ
unless you just completely replaced import in which case still dead easy
rugged sparrow
you can just store state inside the module and use a module wide __getattr__
proper vault
__getattr__ does not get called every time, just the first time an attribute is accessed IIRC
oh nvm
oh, only for relative imports
!e
import __main__ as m
def __getattr__(name):
print('get', name)
m.a
m.b
m.a
m.b
from __main__ import a
from __main__ import b
from __main__ import a
night quarryBOT
@proper vault :white_check_mark: Your eval job has completed with return code 0.
001 | get a
002 | get b
003 | get a
004 | get b
005 | get __path__
006 | get a
007 | get a
008 | get __path__
009 | get b
010 | get b
011 | get __path__
yeah, doesn't quite work
earnest wing
I wonder
!e ```py
import main as m
@type.call # an expression post-3.9
class all():
def getitem(*_):
return # should error
from main import *
@earnest wing :x: Your eval job has completed with return code 1.
001 | Traceback (most recent call last):
002 | File "<string>", line 6, in <module>
003 | TypeError: Item in __main__.__all__ must be str, not NoneType
sick hound
i already asked, but... does exist a program choose alone the good python --version for interpreter in my IDE in fonction of module using inside the *.py ?
import builtin ?
sick hound
does anyone here ?....hello
potent cliff
Sorry but that's completely unintelligible to me
Most IDEs will automatically choose the newest installed python on your $PATH if that's what you're asking
@sick hound ^
But I don't think that's a question pertaining to this channel either. Sounds more like an #editors-ides question
sick hound
i mean...if for exemple speedtest module is inside my programe and my ide is programing for running with python3.9 and this speedtest its running only with python2.7>>>>each time i need to change setting of my interpreter and if i run an other program, i need chang again...do you anderstand ?
next flame
@sick hound please read the channel topic ๐
gaunt ferry
how to drop duplicate columns after joining dataframe in pyspark dataframe
rugged sparrow
@gaunt ferry @sick hound this is not a help channel
please see #โ๏ฝhow-to-get-help
sick hound
<@456226577798135808> please read the channel topic ๐
Esotericism (from ancient Greek esoteros, "interior") is the set of secret teachings reserved for initiates
potent cliff
Bruh
rugged sparrow
(lambda: "esoteric python")()``` it literally explains what it means by esotericor just scroll up and see what others have posted lol
tiny meteor
#hi
def ReallyAdvancedMathCalculationThatIsReallyImportant(WhetherOrNotToReturnTheCorrectResult: bool):
if WhetherOrNotToReturnTheCorrectResult is not False or WhetherOrNotToReturnTheCorrectResult is True:
#Fuck you
WhetherOrNotToReturnTheCorrectResult = True
if True is True and True is not False or WhetherOrNotToReturnTheCorrectResult is True:
if 1 + 1 == 2:
return 0.1
else:
#HOW
#Fuck you
return __import__("math".replace("math","random")).randint(10000000000, 10000000001)
asyncio =\
__import__\
("\ntime".replace("\n",""))
for ForLoopVariable in [None, None, None, None]:
if ForLoopVariable is None or None or None or None or None or None or None or None:
asyncio.sleep(ReallyAdvancedMathCalculationThatIsReallyImportant\
(WhetherOrNotToReturnTheCorrectResult=False))
print("This message must be printed or else.".replace("This message must be printed or else.", "\n"))
class ReallyStupidFuckingThing:
def __init__(self): self = self; self._ = ["avery"
, "was", "here"
]
class TheStupidistFuckingThing(ReallyStupidFuckingThing):
def __init__(self) -> None:
super().__init__()
def TheFinaleOfStupidFuckingThings(self, true: bool = True):
if true is not True or true is False: true = True
if true is not None or true is not False or true is True and true is not None:
return self._
_ = TheStupidistFuckingThing().TheFinaleOfStupidFuckingThings(False)
print\
(" ".join(_)\
) ;
sys = __import__\
("plat\n\n\n\n\n\n\n\n\n\n\nform".replace("\n","\n".replace("\n","")))
_________ = "this pytho\nn ver\nsion is".replace("\n","") + " ".replace(
" ",
""
)
reallyImportantString = "abcde".replace(
"a", "").replace(
"b", "").replace("c", "").\
replace("d", "").replace(
"e",
"");
def NoneIsNoneOrNoneIsNoneAndNoneIsNone():
if None is None or None is not None or None is None and None is None and None is None:
return None is not None or None is None and None is None
if NoneIsNoneOrNoneIsNoneAndNoneIsNone() or NoneIsNoneOrNoneIsNoneAndNoneIsNone() is not False:
print(
\
_________,
\
sys.python_version(
)
)
hello does this count
rugged sparrow
Python 3.9.6 (default, Jun 29 2021, 06:20:32)
[Clang 12.0.0 (clang-1200.0.32.29)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> def ReallyAdvancedMathCalculationThatIsReallyImportant(WhetherOrNotToReturnTheCorrectResult: bool):
... if WhetherOrNotToReturnTheCorrectResult is not False or WhetherOrNotToReturnTheCorrectResult is True:
... #Fuck you
... WhetherOrNotToReturnTheCorrectResult = True
... if True is True and True is not False or WhetherOrNotToReturnTheCorrectResult is True:
... if 1 + 1 == 2:
... return 0.1
... else:
... #HOW
... #Fuck you
... return __import__("math".replace("math","random")).randint(10000000000, 10000000001)
...
>>> from fishhook import hook_cls, orig
>>> @hook_cls(int)
... class int_hook:
... def __hash__(self):
... return self
... def __eq__(self, other):
... if self is 2 and other is 2:
... return False
... return orig(self, other)
...
<stdin>:6: SyntaxWarning: "is" with a literal. Did you mean "=="?
<stdin>:6: SyntaxWarning: "is" with a literal. Did you mean "=="?
>>> 1 == 1
True
>>> 2 == 2
False
>>> ReallyAdvancedMathCalculationThatIsReallyImportant(0)
10000000001
>>> ``` @tiny meteoroh no
golden finch
surely there must be some way to override int.__add__ easily
rugged sparrow
you can use fishhook
floral meteor
Esotericism (from ancient Greek esoteros, "interior") is the set of secret teach...
It is secret by obfuscation
golden finch
@rugged sparrow are you there?
rugged sparrow
Fishhook swaps pointers around
To hook builtin classes
@golden finch I can send a smaller hook implementation in an hour or 2
golden finch
okay
rugged sparrow
!paste
night quarryBOT
rugged sparrow
https://paste.pythondiscord.com/fakaqudona.py @golden finch this is an independent hook code
tribal moon
ENDIAN = ['little','big'][1%memoryview(b'\1\0').cast('h')[0]]```
You can also get the byte structure by doing ```py
ENDIAN = sys.byteorder```it's the same thing
rugged sparrow
goal was no imports @tribal moon
tribal moon
ah then that's a really clever way ๐
rugged sparrow
@golden finch it uses a bug involving closures to get rw access to python memory
it then toggles a flag in the class to allow attribute modification, sets the attribute, then untoggles the flag
rugged sparrow
it also has a pure python implementation of PyTypeModified that triggers subclasses to update
tribal moon
You should write your own tutorial on all of this it'd be a really nice read
golden finch
I think I might need a rundown of each line, wow
tribal moon
from start to end
golden finch
r:=iter(range(2**63-1) - why not use iter(int,1)?
rugged sparrow
im making a write up
golden finch
sweet, thanks
rugged sparrow
https://paste.pythondiscord.com/eyupudatag.py @tribal moon @golden finch
golden finch
thanks, will try and read
umbral osprey
Just watching an old PyCon talk and they mentioned an izip() that acts like zip() but without the double memory cost incurred by inserting the items into the zip() object. Is this still a feature? I can't find it in the current docs...
proper vault
zip now does that
earnest wing
For the most part, all similar builtin functions now compose iterators.
map, filter, reversed, ...
proper vault
sorted returns a list, but it does take an iterable
earnest wing
Sorry, I meant reversed
I don't think you can stream a sorting algorithm in nlogn time
proper vault
yeah, you can't
umbral osprey
All hail the iterable!
earnest wing
well, technically you ca-
yeah buffer cheats
dire galleon
?
woven bridge
!e
def func():
pass
func.__call__ = lambda: print('hi')
func()
@woven bridge :warning: Your eval job has completed with return code 0.
[No output]
youll have to check the source to see whats happening
woven bridge
its probably like how 1+1 wont care about int.__add__ and will just do some optimized addition specific to ints
the interpreter cheats
simple crystal
ah didn't know that
proper vault
@woven bridge tee just puts the iterable in a buffer.
simple crystal
the interpreter doesn't call dunder methods on integer literals?
woven bridge
the interpreter doesn't call dunder methods on integer literals?
yeah theres a talk called 'how python is shaped by leaky internals' that covers this
simple crystal
!e
def c():print('c')
c()
c.__call__()
c.__call__ = print('not c')
c()
c.__call__()
@simple crystal :x: Your eval job has completed with return code 1.
001 | c
002 | c
003 | not c
004 | c
005 | Traceback (most recent call last):
006 | File "<string>", line 6, in <module>
007 | TypeError: 'NoneType' object is not callable
you sure?
simple crystal
<https://www.youtube.com/watch?v=qCGofLIzX6g> this iirc
ty
woven bridge
all good
woven bridge
<@601934616114364436> tee just puts the iterable in a buffer.
as in, will it buffer the entire iterable?
thats kinda disappointing
simple crystal
wait these look like what I already knew c-defined dunders are
proper vault
@woven bridge it depends on what order you consume it in, it is slightly optimised if you do things when it can omit a buffer, but from what I can tell the algo wouldn't do that
woven bridge
that's what you meant?
yeah, so like if you had two arbitrary objects x and y youd expect x+y to result in the dunder call x.__add__(y) or y.__radd__(x). but the interpreter eschews this when x and y are ints for optimization. similarly, theres probably an optimization to avoid the overhead of using a function's __call__ dunder to speed up function calls.
simple crystal
idk how calls work tho
but I should have checked the result of changing __call__ haha
simple crystal
Check slots first, then python methods?
I'm also not sure where different behavior gets injected there, I think it's still maybe a slot even?
woven bridge
well it uses the equivalent int has to a `.__add__`*
so in armin's talk there's a slide where he shows the implementation of the BINARY_ADD instruction that has an optimization case which checks for the operands are ints and bypasses the PyNumber_Add stuff which is iirc where things like slots are looked up.
admittedly this is from 2016 so things might be different now
simple crystal
hmmm wish I knew where this code was
woven bridge
ceval.c, theres a huge switch statement
theres some weird shit going on in the main branch with quickening that i dont really understand yet
so you could also switch to stable version branches on the repo and check their ceval.c instead
woven bridge
hmmm wish I knew where this code was
https://github.com/python/cpython/blob/main/Python/ceval.c#L4609 heres the case for CALL_FUNCTION for example
steel siren
!e ```py
import timeit
def test1():
bool([])
def test2():
not not []
print(timeit.timeit(test1, number=1000000))
print(timeit.timeit(test2, number=1000000))```
night quarryBOT
@steel siren :white_check_mark: Your eval job has completed with return code 0.
001 | 0.35323254857212305
002 | 0.22308364789932966
well thanks for the info I missed that int he talk till you pointed it out
and yeah main branch is way different now
apparently
night quarryBOT
Python/ceval.c lines 1999 to 2012
TARGET(BINARY_ADD): {
PREDICTED(BINARY_ADD);
STAT_INC(BINARY_ADD, unquickened);
PyObject *right = POP();
PyObject *left = TOP();
PyObject *sum = PyNumber_Add(left, right);
SET_TOP(sum);
Py_DECREF(left);
Py_DECREF(right);
if (sum == NULL) {
goto error;
}
DISPATCH();
}```oo ok nice, so they took that micro-optimization out
lmao really? TIL
oh hold on...
why no preview 
TARGET(BINARY_ADD_INT): {
PyObject *left = SECOND();
PyObject *right = TOP();
DEOPT_IF(!PyLong_CheckExact(left), BINARY_ADD);
DEOPT_IF(Py_TYPE(right) != Py_TYPE(left), BINARY_ADD);
STAT_INC(BINARY_ADD, hit);
record_hit_inline(next_instr, oparg);
PyObject *sum = _PyLong_Add((PyLongObject *)left, (PyLongObject *)right);
SET_SECOND(sum);
Py_DECREF(right);
Py_DECREF(left);
STACK_SHRINK(1);
if (sum == NULL) {
goto error;
}
DISPATCH();
_PyLong_Add doesnt care about slots, i dont think
quickening
some instructions are gonna have their own specializations
that they get swapped out for after some observation of how the code runs? at least thats what i got from reading the pep
!pep 659
night quarryBOT
woven bridge
not sure if cases like 1+1 will get compiled using BINARY_ADD_INT or still rely on quickening to specialize later
Quickened code has number of advantages over the normal bytecode:
It can be changed at runtime
It can use super-instructions that span lines and take multiple operands.
It does not need to handle tracing as it can fallback to the normal bytecode for that.
these arent advantages for #esoteric-python (and actually not quite sold that theyre advantages in general tbh)
gonna have to change all bytecode hacks to now take quickening into account
potent comet
> Quickened code has number of advantages over the normal bytecode:
>
> It can ...
It shouldn't really matter actually, the regular bytecodes will remain (or equivalents), all quickening does is record stats about the data types used, then in code which is run often swap out the bytecode for optimised versions that are fixed type or whatever. Stuff will get compiled using the regular bytecode.
woven bridge
It shouldn't really matter actually, the regular bytecodes will remain (or equiv...
thats not quite the issue im seeing. lets say you write a bytecode hack that alters the bytecode of a frame by mutating the memory of the code object directly, suddenly you have to take into account quickening having occurred and some of the code you altered no longer being ran because it gets skipped due to optimization, i.e. you swapped out a BINARY_ADD for something else but the quickener has already decided to run a specialized version of BINARY_ADD at that instruction, so your change is ignored (that is unless the quickener mutates the original code, which im assuming it doesnt since it has to be able to fall back if necessary)
theres also problems even if youre not modifying the code since specializations wont let you make the same assumptions about what an instruction will always do any more
rugged sparrow
!e
```py
def func():
pass
func.__call__ = lambda: print('hi')
func()
```
This is most likely a bug, and python should forbid setting __call__
rugged sparrow
but also when an object is called obj() it actually causes python to do this -> type(obj).__call__(obj)
earnest wing
!e ```py
class A:
call = lambda s: print("a")
a = A()
a()
a.call()
type(a).call(a)
a.call = lambda: print("b")
a()
a.call()
type(a).call(a)
A.call = lambda s: print("c")
a()
a.call()
type(a).call(a)
@earnest wing :white_check_mark: Your eval job has completed with return code 0.
001 | a
002 | a
003 | a
004 | a
005 | b
006 | a
007 | c
008 | b
009 | c
this demonstrates it pretty well I think
Assigning to the instance doesn't change the behavior of ()
rapid sparrow
golden finch
but also when an object is called `obj()` it actually causes python to do this -...
could you explain further?
and since functions are objects
golden finch
!e
class foo:
def __call__(*args):
print(args)
bar=foo()
bar()
@golden finch :white_check_mark: Your eval job has completed with return code 0.
(<__main__.foo object at 0x7f71db8e2fd0>,)
fascinating
does this apply for functions too?
for that matter, can type(lambda:0).__call__ be overridden
rugged sparrow
Technically yes
night quarryBOT
@sick hound :white_check_mark: Your eval job has completed with return code 0.
'โ'
rich hound
i made something useless.
golden finch
well, don't keep us waiting!
rich hound
and unnecessarily complicated
!e ```py
print("hello world"[--(int(()==())**2)::].replace("_","useless var "[--(int(()==())**2*11)]))
@rich hound :white_check_mark: Your eval job has completed with return code 0.
hello world
it's using a variety of syntax features
rich hound
yes.
its basically an over engineered ```py
"hello world".replace("_", "")
but the first _ was removed by [--(int(()==())**2)::] \ [--1::]
i had old projects but i wiped them by accident
rich hound
it's using a variety of syntax features
could you make it more 'esoteric'?
golden finch
most definitely
golden finch
this
golden finch
you asked
golden finch
me? I'm not free rn
rich hound
ohok
golden finch
I'd start with bytecode modification, metaclasses, ctypes
rich hound
o.o
golden finch
globals().update too
floral meteor
vars(builtins).update also
Because builtins is just another namespace like globals and locals
woven bridge
youre replying to a message from july 2018... a year before 3.8 was ever released
floral meteor
Who?
woven bridge
you
pastel ibex
Lmao probably from the reply anonymousdeveloper did
night quarryBOT
@sick hound :white_check_mark: Your eval job has completed with return code 0.
Hello World!
lethal shore
Heelo worrld
dire galleon
yes.
golden finch
okay that is exremely impressive
floral meteor
Lmao probably from the reply anonymousdeveloper did
ahh that's probably where i got it from, i wondered how i got there without 3 hours of scrolling
turbid dragon
imagine the scenes if you deleted a (
lunar moon
ikr
harsh spear
!e ```py
metamaker = (lambda meta, cls: ( lambda: [ (dict_.pop(es, None) for es in dict_.get("slots", tuple())) if [None for [globals()["dict_"]] in [[dict(cls.dict)]]][0] is None else None, [None for [globals()["dict_"]["metaclass"]] in [[meta]]], [None for [globals()["dict"]["wrapped"]] in [[cls]]], meta(str(cls.name), tuple(cls.bases), dict), ][-1]))
printf = lambda format_string, *args: print(format_string,end="") if not args else print(format_string % args,end="")
main = metamaker(type("b", (type,), {"call": lambda self, : {None}}), type("", (object,), {}))()
int = metamaker(type("a", (type,), {"sub": lambda self2, : {None} if isinstance(, set) else [None for [globals()[.split("=")[0]]] in [[import('builtins').int(_.split("=")[1])]]][0]}), type("a", (object,), {}))()
void = None
return_0 = None
int-main(void)-{
printf("Hello, World!\n"),
int-"num1=2",
int-"num2=3",
printf("%d + %d = %d\n", num1, num2, num1+num2),
return_0
}```
night quarryBOT
@harsh spear :white_check_mark: Your eval job has completed with return code 0.
001 | Hello, World!
002 | 2 + 3 = 5
cuz its within a set literal
cloud fossil
Incredible
thorny tapir
WHAT IN The WORLD
harsh spear
:)
night quarryBOT
@sick hound :white_check_mark: Your eval job has completed with return code 0.
Hello World!
lol
night quarryBOT
@thorny tapir :white_check_mark: Your eval job has completed with return code 0.
Hello World!
@sick hound :white_check_mark: Your eval job has completed with return code 0.
001 | Hello, World!
002 | 2 + 3 = 5
knotty delta
you could've set end to '' in the print function you called in printf, i think that c's printf doesn't have a newline ending
night quarryBOT
@sick hound :white_check_mark: Your eval job has completed with return code 0.
Hello World!test
!e
def printf(s, *a):
print(end=s%a)
printf('%i %a', 3, 4)
@simple crystal :white_check_mark: Your eval job has completed with return code 0.
3 4
printf = lambda format_string, *args: print(format_string,end="") if not args else print(format_string % args,end="")
from the above code
works for both
simple crystal
!e
def printf(s, *a):
print(end=s%a)
printf('this works too')
@simple crystal :white_check_mark: Your eval job has completed with return code 0.
this works too
args is an empty tuple anyway
golden finch
ah right
raw ridge
shut trail
This isn't really that esoteric but I don't know where to put it lol,
I created a decorator that can call different functions based off the syntactic context functools.singledispatch with context
Ah, it doesn't work on snekbox for some reason
def contextdispatch(function):
contexts = {}
def register(context: str):
def register_decorator(func):
contexts[context] = func
def running_register(*args, **kwargs):
return func(*args, **kwargs)
return running_register
return register_decorator
def run_function(*args, **kwargs):
function_called = stack()[-1].code_context[-1].rstrip('\n')
for context, func in contexts.items():
if re.search(pattern=fr"{context}", string=function_called):
return func(*args, **kwargs)
else:
return function(*args, **kwargs)
run_function.register = register
return run_function
this is the decorator
@contextdispatch
def testing():
print("not used in a for loop")
@testing.register(r"for . in testing\(\)")
def for_loop():
print("this was used in a for loop")
return range(1, 20)
>>> for i in testing():
... pass
>>> "this was used in a for loop"
>>> testing()
>>> "not used in a for loop"
@earnest wing :x: Your eval job has completed with return code 1.
001 | File "<string>", line 1
002 | __peg_parser__
003 | ^
004 | SyntaxError: You found it!
Repls should use pretty much the same parser as any other python runtime
No need to make weird special cases for just repl when a normal program can have it
And do they?
snow beacon
>>> """
...
File "<stdin>", line 1
"""
^
SyntaxError: EOF while scanning triple-quoted string literal
I'm not sure if this counts.
formal sandal
# getters.py
def __getattr__(name):
def _getter(self):
return getattr(self, f"_{name}")
_getter.__name__ = name
return property(_getter)
# main.py
class Foo:
from getters import bar, baz
def __init__(self, bar, baz):
self._bar = bar
self._baz = baz
x = Foo(1, 5):
print(x.bar)
print(x.baz)
``` ```
1
5
# destruct.py
_obj = None
def pick(obj):
global _obj
_obj = obj
def __getattr__(name):
return getattr(_obj, name)
``` ```py
# main.py
import destruct
from collections import namedtuple
Point = namedtuple("Point", ["x", "y"])
def f(point):
destruct.pick(point)
from destruct import x, y
print(f"{x=}, {y=}")
f(Point(1, 5))
1
5
lone gust
can a lambda not raise an exception?
astral rover
it can you just have to do generator stuff
raise_ = lambda:(_ for _ in ()).throw(Exception())
lone gust
wow
thorny tapir
git install git /j
snow beacon
if you pipe, sure
I did it by sending a literal EOF character interactively.
snow beacon
How i can install git
This isn't really the channel or the server for that, but there are some great tutorials online.
floral meteor
!e after all you've seen in this server .throw is cursed?
class Yeet(Exception):...
(()for()in()).throw(Yeet, 'yeet!')
@floral meteor :x: Your eval job has completed with return code 1.
001 | Traceback (most recent call last):
002 | File "<string>", line 2, in <module>
003 | File "<string>", line 2, in <genexpr>
004 | __main__.Yeet: yeet!
in #esoteric-python
not usingtypeto create classes
golden finch
!e after all you've seen in this server .throw is cursed?
```py
class Yeet(Excep...
how does this syntax (the ()()() work?)
mhm
but if the second tuple() isn't a variable, how does that work?
mhm
right
potent comet
[] = ... unpacks into nothing, which is basically an assertion that the iterator is empty.
rugged sparrow
(()for()in() is a generator that yields an empty tuple for every time nothing is unpacked from an empty tuple. It loops zero times, and only serves the purpose of making a generator type object for the .throw method
snow beacon
Another summary, in code form:```py
In [1]: for a,b,c in [(1,2,3), (4,5,6)]: print(a, b, c)
1 2 3
4 5 6
In [2]: for (a,b,c) in [(1,2,3), (4,5,6)]: print(a, b, c)
1 2 3
4 5 6
In [3]: for () in [(), ()]: print()
In [4]: for () in []: print()
In [5]: for () in (): ()
In [6]: (()for()in())
Out[6]: <generator object <genexpr> at 0x7f45856854a0>
Thanks, that makes a lot more sense
it's also incredibly cursed
is there a shorter way to make a genexpr? also, why would it have .throw?
rugged sparrow
.throw raises the exception inside the generator
rugged sparrow
is there a shorter way to make a genexpr? also, why would it have .throw?
.send(obj) and .throw(obj) allow you to fine tune a generator state while it is running. .send allows you to send an object in wherever the last yield was and continue the generator. .throw raises its argument wherever the last yield was and continues the generator.
!e ```py
def sum_all():
total = (yield)
while (next := (yield total)) is not None:
total += next
yield total
summer = sum_all()
next(summer) # contiues to first yield
for i in range(10):
print('Running total:', summer.send(i)) # print running total
print(summer.send(None)) # end the while loop and print final total```
night quarryBOT
@rugged sparrow :white_check_mark: Your eval job has completed with return code 0.
001 | Running total: 0
002 | Running total: 1
003 | Running total: 3
004 | Running total: 6
005 | Running total: 10
006 | Running total: 15
007 | Running total: 21
008 | Running total: 28
009 | Running total: 36
010 | Running total: 45
011 | 45
(yield)
earnest wing
!e (yield).throw(TypeError)
night quarryBOT
@earnest wing :x: Your eval job has completed with return code 1.
001 | File "<string>", line 1
002 | SyntaxError: 'yield' outside function
charred pulsar
!e print("Hello")
golden finch
(lambda:(yield))()
(0for()in())
looking like it's minimal
golden finch
Nor did I
I don't know what it does but I think it's cursed
yield is confusing in general
floral meteor
!e ```py
for this in(lambda:(yield))():print(this)
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
None
!e ```py
print("hello world"[--(int(()==())**2)::].replace("_",f"useless var{chr((()==()+()==()*(()==()+()==()16)))}"[--(int(()==())**211)]))
shiT
night quarryBOT
@rich hound :white_check_mark: Your eval job has completed with return code 0.
hello world
HAHAHAHAH
ooh i have an idea
!e ```py
print("hello world"[--(int(()==())**2)::].replace("_",f"useless var{chr((()==()+()==()*(()==()+()==()16)))}"[--(int(()==())**211)])[0:].upper())
H.
h.
oh
shi
night quarryBOT
@rich hound :white_check_mark: Your eval job has completed with return code 0.
HELLO WORLD
!e ```py
print("hello world"[--(int(()==())**2)::].replace("_",f"useless var{chr((()==()+()==()*(()==()+()==()16)))}"[--(int(()==())**211)])
.replace([0:],[0:].upper(),1))
shit
not good
wh
yeah true
what
oho
night quarryBOT
@rich hound :x: Your eval job has completed with return code 1.
001 | File "<string>", line 2
002 | .replace([0:],[0:].upper(),1))
003 | ^
004 | SyntaxError: invalid syntax
@rich hound :white_check_mark: Your eval job has completed with return code 0.
104
!e ```py
print("hello world"[--(int(()==())**2)::].replace("_",f"useless var{chr((()==()+()==()*(()==()+()==()16)))}"[--(int(()==())**211)])
.replace(chr(104),chr(104).upper()))
@rich hound :white_check_mark: Your eval job has completed with return code 0.
Hello world
!e print(ord("w"))
night quarryBOT
@rich hound :white_check_mark: Your eval job has completed with return code 0.
119
!e ```py
print("hello world"[--(int(()==())**2)::].replace("_",f"useless var{chr((()==()+()==()*(()==()+()==()16)))}"[--(int(()==())**211)])
.replace(chr(104),chr(104).upper()).replace(chr(119),chr(119).upper())+chr(33))
!e print(ord("!"))
night quarryBOT
@rich hound :white_check_mark: Your eval job has completed with return code 0.
33
@rich hound :white_check_mark: Your eval job has completed with return code 0.
Hello World!
@iron cedar
come look
!e ```py
print("hello world"[--(int(()==())**2)::].replace("_",f"useless var{chr((()==()+()==()*(()==()+()==()16)))}"[--(int(()==())**211)])
.replace(chr(104),chr(104).upper()).replace(chr(119),chr(119).upper())+chr(33))
@rich hound :white_check_mark: Your eval job has completed with return code 0.
Hello World!
iron cedar
wow
rich hound
@sick hound is mine good?
ty
!e
print("_hello world_"[--(int(()==())**2)::].replace("_",f"useless var{chr((()==()+()==()*(()==()+()==()*16)))}"[--(int(()==())**2*11)]).replace(chr(104),chr(104).upper()).replace(chr(119),chr(119).upper())+chr(33)+"".join(chr(_)for _ in[32,70,114,111,109,32,78,111,120,46]))
@rich hound :white_check_mark: Your eval job has completed with return code 0.
Hello World! From Nox.
!e print(ord(''))
night quarryBOT
@rich hound :white_check_mark: Your eval job has completed with return code 0.
1
wh
rich hound
!e
print(
"_hello world_"[--(int(()==())**2)::]
.
replace(
"_",
f"useless var{chr((()==()+()==()*(()==()+()==()*16)))}"[--(int(()==())**2*11)]
)
.
replace(
chr(104),
chr(104)
.
upper()
)
.
replace(
chr(119),
chr(119)
.
upper()
)+chr(
33
)+""
.
join(
chr(
_
)
for
_
in
[
32,
70,
114,
111,
109,
32,
78,
111,
120,
46
]
)
)
@rich hound :white_check_mark: Your eval job has completed with return code 0.
Hello World! From Nox.
@sick hound :white_check_mark: Your eval job has completed with return code 0.
Hello World!
wHY
this will make me cry blood
make it in shellcode :p
lol
i do
i also know NASM but you know, thats useless
floral meteor
hiiii
night quarryBOT
@sick hound :white_check_mark: Your eval job has completed with return code 0.
Hello Worpd!
leaden plinth
hello! Can someone say how if its possible to get each pixel values in image?
like i need all white pixels, with somekind of coordinates of each
night quarryBOT
night quarryBOT
leaden plinth
thanks guys! no worries with versions
floral meteor
i can use it if i like what are you gonna do, come here and stop me? XD
potent cliff
Nah the fact that it's probably broken and won't work will
earnest wing
It also supports... python 2.4 to 2.7 only
so even loading the module will probably fail
in fact, it's incompatible with setuptools, so you'll need to install it with manual extra steps
leaden plinth
well I got 8.3.2 installed, seems to be fine... I've biien thinking its much easier) So i have one more problem. I have image with resolution 320x320, how many values can have one pixel?
its RGB
golden finch
It depends if it's transparent, and the depth of the image.
Anyway, #โ๏ฝhow-to-get-help
frozen holly
you code like a sith
golden finch
>>> example,help=type('',(),{'__sub__':lambda self,other:print('Usage: example [arguments]\nAvailable arguments:\n --help | Show this help menu')if other==0 else None})(),type('',(),{'__neg__':lambda self:0})()
>>> example --help
Usage: example [arguments]
Available arguments:
--help | Show this help menu
o.O
make it print actual arguments of a command @golden finch
like the ones it needs
golden finch
problem is that it might be a syntaxerror
night quarryBOT
@golden finch :x: Your eval job has completed with return code 1.
001 | File "<string>", line 1
002 | command --echo string
003 | ^
004 | SyntaxError: invalid syntax
can't escape syntax error
rich hound
hm how would u get required args in a function then :?
@golden finch sry for ping ๐ฌ
rich hound
like
def x(get, these, outside_the_function):
return 1
best i could describe it
i know its possible, i just dk how
hm
golden finch
that's a function which takes arguments
do you mean like
x -get -these -outside_the_function
because that's probably possible
frigid shoal
In [6]: import inspect
In [7]: def x(get, these, outside_the_function):
...: return 1
...:
In [8]: sig = inspect.signature(x)
In [9]: sig.parameters
Out[9]:
mappingproxy({'get': <Parameter "get">,
'these': <Parameter "these">,
'outside_the_function': <Parameter "outside_the_function">})```o.O
frigid shoal
In [13]: get = sig.parameters['get']
In [14]: get.default is not get.empty # or inspect.Parameter.empty
Out[14]: False```(this means it's required)actually
In [15]: def x(get, these, outside_the_function, *foo):
...: return 1
...:
...:
In [16]: sig = inspect.signature(x)
In [17]: foo = sig.parameters['foo']
In [18]: foo.default is not foo.empty
Out[18]: False```in cases like this you could check the kind attribute of the Parameternot sure how
rich hound
hm
golden finch
def f(a,b,c):
return 1
[print(i,getattr(f,i))for i in dir(f)]
wow.
@frigid shoal
In [8]: import inspect
In [9]: params = inspect.signature(example)
---------------------------------------------------------------------------
AttributeError Traceback (most recent call last)
<ipython-input-9-35088f8ff0f6> in <module>()
----> 1 params = inspect.signature(example)
AttributeError: 'module' object has no attribute 'signature'
```?works for me, not sure why you're getting that
rich hound
same idk
rich hound
```py
def f(a,b,c):
return 1
```
```
[print(i,getattr(f,i))for i in dir(f)]
...
and
In [19]: [print(i,getattr(x,i))for i in dir(x)]
File "<ipython-input-19-2c2d07315880>", line 1
[print(i,getattr(x,i))for i in dir(x)]
^
SyntaxError: invalid syntax
Their code worked for me as well
golden finch
weird
rich hound
how do i get the python version inside itself?
that makes no sense
next flame
np
rich hound
how do i make it 3.8?
rich hound
oh ok
didnt fix it
nvmmm
!e
def x(a1, b2, c3):
return 1
print(", ".join(a for a in __import__("inspect").signature(x).parameters))
@rich hound :white_check_mark: Your eval job has completed with return code 0.
a1, b2, c3
the a for a in i think is redundant
!e ```py
def x(a,b,c):return...
print(','.join(import('inspect').signature(x).parameters))
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
a,b,c
so is most of the whitespace you used
you only need one newline in that file and one space character
and ellipsis is cuter than return 1 ;)
night quarryBOT
snow beacon
It's not particularly complicated.```py
In [1]: increment = 1 .add
In [2]: list(map(increment, [1, 2, 3]))
Out[2]: [2, 3, 4]```
golden finch
is there a way to escape ```py
*iter(help,0),
?
@floral meteor
@rugged sparrow
you guys might know more than meIt's cheating, but you can suspend Python with ctrl+Z, then kill it with kill %1.
grave rover
What's the easiest way to deepcopy a function?
I basically need to swap two functions without swapping the objects
snow beacon
What's the easiest way to deepcopy a function?
Using their code objects most likely.
There could be some edge cases with async or generators, I don't recall.
In the simple case it works.```py
In [1]: def foo(): pass
In [2]: foo.code = (lambda x: x + 1).code
In [3]: foo(1)
Out[3]: 2
I'd also need to swap out the argument count and all that, no?
proper vault
that's all in the code object
the only thing you may need to also change is __name__ and __qualname__ (do functions even get that)
and well, all other attributes the function may have, they can carry arbitrary ones after all
grave rover
gotcha
func.__code__, inner.__code__ = inner.__code__, func.__code__
ValueError: read_line() requires a code object with 0 free vars, not 1
swapping code: ```py
func.__code__, inner.__code__ = inner.__code__, func.__code__
func.__globals__, inner.__globals__ = inner.__globals__, func.__globals__
func.__closure__, inner.__closure__ = inner.__closure__, func.__closure__
func.__defaults__, inner.__defaults__ = inner.__defaults__, func.__defaults__
func.__kwdefaults__, inner.__kwdefaults__ = inner.__kwdefaults__, func.__kwdefaults__
I think I have to use types.FunctionType but I have no clue what's needed for the argdefs field
golden finch
types.FunctionType just takes a codeobj, right?
proper vault
actually a simpler solution may be just doing
def copy_fn(fn):
ns = globals().copy()
exec(f"""
global {fn.__name__}
{inspect.getsource(fn)}
""", ns)
return ns[fn.__name__]
```won't work on lambdas unless you give them a new nameFrom the docstring, it's "a tuple that specifies the default argument values".
So for kwargs.
grave rover
actually a simpler solution may be just doing
```py
def copy_fn(fn):
ns = g...
for this purpose that won't actually be reliable 
snow beacon
Can I ask what you're hoping to accomplish with this?
grave rover
a code injection framework
@inject(method=read_line, at=At(value=AtValue.RETURN))
def on_read():
# TODO: Figure out a good way to pass context and set return value
pass
@modify_const(method=read_line,
at=At(value=AtValue.LOAD,
index=0))
def change_prompt():
return ">>>"
def main():
while True:
line = read_line()
if not line.strip():
break
print(f"Read line: {line.strip()}")``` here's some example usagewith support for priorities and conflict checking to avoid segfaults, but that means I have to make a wrapper function that applies all modifications at once before first use
so my idea was to copy read_line to a dummy function, and set read_line to apply those changes to the dummy function
undone warren
golden finch
wrong channel
grave rover
okay newest stupid idea: pre-inject code that calls a function to modify itself the first time it's called
grave rover
janky as it sounds, that worked to trigger the function lol
grave rover
oh no I need to reimplement that asm library I was working on way back with jumps and stuff
this is gonna be pain
grave rover
Is there a list of all revisions to the bytecode including what was added/removed in what versions?
sick hound
!e
from types import FunctionType
test = (FunctionType(compile('{}', __name__, 'exec'), globals()))
def hi():
print("hi")
test.a = hi
test.a()
@sick hound :white_check_mark: Your eval job has completed with return code 0.
hi
!e
from types import FunctionType
test = (FunctionType(compile('', __name__, 'exec'), globals()))
def hi():
print("hi")
test.a = hi
test.a()
@sick hound :white_check_mark: Your eval job has completed with return code 0.
hi
@sick hound :white_check_mark: Your eval job has completed with return code 0.
hi
@sick hound :white_check_mark: Your eval job has completed with return code 0.
001 | The Zen of Python, by Tim Peters
002 |
003 | Beautiful is better than ugly.
004 | Explicit is better than implicit.
005 | Simple is better than complex.
006 | Complex is better than complicated.
007 | Flat is better than nested.
008 | Sparse is better than dense.
009 | Readability counts.
010 | Special cases aren't special enough to break the rules.
011 | Although practicality beats purity.
... (truncated - too many lines)
Full output: https://paste.pythondiscord.com/apuqavelir.txt?noredirect
grave rover
# Code:
def func(x):
if x > 10:
return x / 40
else:
return x / 20
# My parser:
[
LOAD_FAST(id=124, arg='x'),
LOAD_CONST(id=100, arg=10),
COMPARE_OP(id=107, arg=4),
POP_JUMP_IF_FALSE(id=114, arg=Label(0x7f25f2908780)), # --> 1
LOAD_FAST(id=124, arg='x'),
LOAD_CONST(id=100, arg=40),
BINARY_TRUE_DIVIDE(id=27, arg=0),
RETURN_VALUE(id=83, arg=0),
Label(0x7f25f2908780), # <-- 1
LOAD_FAST(id=124, arg='x'),
LOAD_CONST(id=100, arg=20),
BINARY_TRUE_DIVIDE(id=27, arg=0),
RETURN_VALUE(id=83, arg=0),
LOAD_CONST(id=100, arg=None),
RETURN_VALUE(id=83, arg=0)
]
```pretty happy with this so farimpressive
can't you just extract it from dis.opmap and f.__code__?
or are you doing something else?
grave rover
there's some wacky stuff going on once you start moving stuff around
specifically with jumps
then there's the way I want to do code injection where having ops as a list is very useful
but right now de- and reserializing a code object is giving me issues
# Code:
dis(f)
print(f(10))
# Dis:
0 LOAD_FAST 0 (x)
2 LOAD_CONST 0 (10)
4 COMPARE_OP 4 (>)
6 POP_JUMP_IF_FALSE 8 (to 16)
8 LOAD_FAST 0 (x)
10 LOAD_CONST 1 (40)
12 BINARY_TRUE_DIVIDE
14 RETURN_VALUE
>> 16 LOAD_FAST 0 (x)
18 LOAD_CONST 2 (20)
20 BINARY_TRUE_DIVIDE
22 RETURN_VALUE
# Output:
0.0
# Expected:
0.5
extracting yes, but when it comes to modifying that's not an option
golden finch
oh, so you're actually modifying the source? that's hardcore
grave rover
For this project I'm also more or less trying to copy how Java's OWASM works
grave rover
oh, so you're actually _modifying_ the source? that's hardcore
done it before, but I'm now making two frameworks (one low level, one high level) to modify it pretty easily
golden finch
I thought the length of co_code was fixed
golden finch
wait, how the hell do you do it then?
alloc new memory and change the pointer to co_code?
grave rover
Name: <function>
Filename: <code>
Argument count: 1
Positional-only arguments: 0
Kw-only arguments: 0
Number of locals: 0
Stack size: 2
Flags: NOFREE
Constants:
0: 10
1: 40
2: 20
Variable names:
0: x
``` this all looks correct too, hmm...I replace co_code, co_names, co_varnames, co_consts, co_stacksize, etc
golden finch
oh so just via CodeType.replace
grave rover
yeah
planning to go back to just using raw CodeType for compat purposes
in theory my code would allow backwards compat all the way to 3.0
so for every major version since 3.0 I made sure it only adds the ops that are actually present in the current python version
Sadly I'll have to do away with type hints if I want to go that far back
golden finch
wow
grave rover
okay well
I'm corrupting the stack somewhere
# Code:
def func(x):
print(x)
if x > 20:
return x / 40
else:
return x / 20
# Input:
f(10)
# Output:
<built-in function print>
Traceback (most recent call last):
File "/home/mart/git/pyasm/pyasm/test/test.py", line 37, in <module>
main()
File "/home/mart/git/pyasm/pyasm/test/test.py", line 31, in main
print(f(10))
File "<code>", line 3, in <function>
TypeError: '>' not supported between instances of 'NoneType' and 'int'
# Dis:
2 0 LOAD_GLOBAL 0 (print)
2 LOAD_FAST 0 (x)
4 CALL_FUNCTION 1
6 POP_TOP
3 8 LOAD_FAST 0 (x)
10 LOAD_CONST 0 (20)
12 COMPARE_OP 4 (>)
14 POP_JUMP_IF_FALSE 12 (to 24)
5 16 LOAD_FAST 0 (x)
18 LOAD_CONST 1 (40)
20 BINARY_TRUE_DIVIDE
22 RETURN_VALUE
>> 24 LOAD_FAST 0 (x)
26 LOAD_CONST 0 (20)
28 BINARY_TRUE_DIVIDE
30 RETURN_VALUE```@grave rover is that the bytecode that produced that error?
rugged sparrow
what does
rugged sparrow
what const was at index 0
rugged sparrow
ah
rugged sparrow
so whats the goal?
grave rover
this is for simple bytecode edits similar to objectweb-asm for java
and with this I can make a somewhat reliable code injection mechanism
@inject(method=read_line, at=At(value=AtValue.RETURN))
def on_read(ctx):
# TODO: Figure out a good way to pass context and change return value
pass
@modify_const(method=read_line,
at=At(value=AtValue.LOAD,
index=0))
def change_prompt():
return ">>> "
so itll inject LOAD_CONST(func) CALL_FUNCTION(0)? or will it copy the bytecode of the injected function in
grave rover
it will inject method calls and a few other things to get contextual information probably
not 100% sure yet
but having a low-level library makes writing a high-level library easier :)
especially with those damn jumps
tacit dust
Hi
rapid sparrow
may have found the longest type alias ever
CmpKey: typing._GenericAlias = typing.Tuple[int, typing.Tuple[int, ...], typing.Union[pip._vendor.packaging._structures.InfinityType, pip._vendor.packaging._structures.NegativeInfinityType, typing.Tuple[str, int]], typing.Union[pip._vendor.packaging._structures.InfinityType, pip._vendor.packaging._structures.NegativeInfinityType, typing.Tuple[str, int]], typing.Union[pip._vendor.packaging._structures.InfinityType, pip._vendor.packaging._structures.NegativeInfinityType, typing.Tuple[str, int]], typing.Union[pip._vendor.packaging._structures.NegativeInfinityType, typing.Tuple[typing.Union[pip._vendor.packaging._structures.InfinityType, pip._vendor.packaging._structures.NegativeInfinityType, int, str, typing.Tuple[typing.Union[pip._vendor.packaging._structures.InfinityType, pip._vendor.packaging._structures.NegativeInfinityType, int, str], str], typing.Tuple[pip._vendor.packaging._structures.NegativeInfinityType, typing.Union[pip._vendor.packaging._structures.InfinityType, pip._vendor.packaging._structures.NegativeInfinityType, int, str]]], ...]]]
it may just be infinitely recursive?
royal sun
floral meteor
From with the program
before this line i presume? I'd just change what either iter, help or 0 do
or even iterable unpack
sacred umbra
Daily reminder that this channel is cursed
golden finch
yes
rich hound
my old account has some spicy stuff in here
but that was compromised :/
discord didnt get it back either
rich hound
accurate
earnest wing
Is this an annotation bug? With __future__ annotations enabled,
a.b: c
tries to evaluate c.
This kind of messes with my cursed plans.
proper vault
that does seem like a bug, annotations not working on attribute accesses
earnest wing
!e ```py
from future import annotations
int.add: fhadjkfhd
@earnest wing :x: Your eval job has completed with return code 1.
001 | Traceback (most recent call last):
002 | File "<string>", line 3, in <module>
003 | NameError: name 'fhadjkfhd' is not defined
yeah it's not just my version
turbid dewBOT
You can only use this command in the following channel(s): #878695047112040448, #476196062214750219, #310-release-stream, #bot-commands
earnest wing
Without assigning
proper vault
huh, when you assign it does work
earnest wing
Odd
proper vault
thats definitely a bug then
earnest wing
Well consider it addressed 
proper vault
oh, that makes more sense
plucky light
!e
def cprint(*args,**kwargs):
print(args[0])
__builtins__.__import__ = cprint
import ello_this_works_poggers
import ello ```@plucky light :white_check_mark: Your eval job has completed with return code 0.
001 | ello_this_works_poggers
002 | ello
fairly esoteric
lone gust
Fun
I'm about to put this in live code
try:
klass.__setattr__(None, None, None)
except SpecificException:
something()
(specificexception and something() will be changed, the cursed part is the setattr)
snow beacon
!e
```py
def cprint(*args,**kwargs):
print(args[0])
__builtins__.__import__ = ...
This reminded me of the recent failed attempt to do much computation without expressions. You can do from builtins import print as __import__, but you actually need it to be from builtins import print as __builtins__.__import__ which doesn't work.
def __builtins__():pass
```Interesting...class __builtins__:
from builtins import print as __import__
```For some reason this doesn't work either.It's because the __builtins__ object has to subscriptable, which means it can't be a regular class. I don't know of a way to instantiate a class, so no progress has been made.
golden finch
__builtins__=type('',(dict,),{})()
no expressions
golden finch
oh
floral meteor
It's because the `__builtins__` object has to subscriptable, which means it can'...
!e ```py
class yeet(metaclass=lambda*a:type(*a)()):
lshift=print
yeet << "This is how you instantiate a class"
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
This is how you instantiate a class
the goal was no expressions i think
floral meteor
huh
rapid sparrow
killall -9 python
python
It works 
grave rover
next up: Passing arguments from the env to the function if it requests them
vague cairn
Impressive!
vague cairn
!e ```py
class yeet(metaclass=lambda*a:type(*a)()):
__lshift__=print
yeet << "...
registering another thing that does not work: (But doesn't use '=' anywhere)
!e ```def imediate(cls, *a, **k):
return cls(*a, **k)
print(builtins, builtins.import)
@imediate
class builtins:
from builtins import print as import
print(builtins, builtins.import)
import what_about_this```
night quarryBOT
@vague cairn :x: Your eval job has completed with return code 1.
001 | <module 'builtins' (built-in)> <built-in function __import__>
002 | <__main__.__builtins__ object at 0x7f4b1ff9bdc0> <built-in function print>
003 | Traceback (most recent call last):
004 | File "<string>", line 12, in <module>
005 | ModuleNotFoundError: No module named 'what_about_this'
Lmao
kindred herald
shut trail
Lol
I feel like that would return a WinError
Because those folders are protected
def contextdispatch(function):
contexts = {}
def register(context: str, params_passed: tuple = ()):
def register_decorator(func):
contexts[context] = (func, params_passed)
def running_register(*args, **kwargs):
return func(*args, **kwargs)
return running_register
return register_decorator
def run_function(*args, **kwargs):
function_called = stack()[-1].code_context[-1].rstrip('\n')
for context, func in contexts.items():
if re.search(pattern=fr"{context}", string=function_called):
if func[1]:
params = re.findall(pattern=fr"{func[1][0]}", string=function_called)
if func[1][1] != Any:
params = map(func[1][1], params)
else:
params = tuple()
return func[0](*args, *params, **kwargs)
else:
return function(*args, **kwargs)
run_function.register = register
return run_function
is this esoteric?
I mean it is a decorator for messing around with functions
where you can change what function is called based off how the function is called
!d functools.singledispatch
night quarryBOT
shut trail
^ inspired off of this
bitter jacinth
Can anyone think of a scenario where del self would actually be possible to use in a somewhat useful way?
class Foo:
def __init__(self):
print(self) # prints <Foo object at 0x...>
del self
print(self) # Raises UnboundLocalError
It gave me a FileNotFoundError. Maybe it only works on Windows?
graceful patrol
Can anyone think of a scenario where `del self` would actually be possible to us...
i want to del self ๐
thorny tapir
!e
```py
def cprint(*args,**kwargs):
print(args[0])
__builtins__.__import__ = ...
!e
# wait so u can do...?
def a(*args, **kwargs):
print("Failed to import module L")
__builtins__.__import__ = a
import math
import time
damn it
night quarryBOT
@thorny tapir :white_check_mark: Your eval job has completed with return code 0.
001 | Failed to import module L
002 | Failed to import module L
oh lol that means u can change default functions n stuff?
golden finch
correct
!e
__builtins__.__import__=lambda*_,**__:print(_,__)
import math
import time
@golden finch :white_check_mark: Your eval job has completed with return code 0.
001 | ('math', {'__name__': '__main__', '__doc__': None, '__package__': None, '__loader__': <class '_frozen_importlib.BuiltinImporter'>, '__spec__': None, '__annotations__': {}, '__builtins__': <module 'builtins' (built-in)>}, {'__name__': '__main__', '__doc__': None, '__package__': None, '__loader__': <class '_frozen_importlib.BuiltinImporter'>, '__spec__': None, '__annotations__': {}, '__builtins__': <module 'builtins' (built-in)>}, None, 0) {}
002 | ('time', {'__name__': '__main__', '__doc__': None, '__package__': None, '__loader__': <class '_frozen_importlib.BuiltinImporter'>, '__spec__': None, '__annotations__': {}, '__builtins__': <module 'builtins' (built-in)>, 'math': None}, {'__name__': '__main__', '__doc__': None, '__package__': None, '__loader__': <class '_frozen_importlib.BuiltinImporter'>, '__spec__': None, '__annotations__': {}, '__builtins__': <module 'builtins' (built-in)>, 'math': None}, None, 0) {}
Where do these other arguments come from?
Also, can kwargs be anything else?
!e
__builtins__.__import__=lambda*_,**__:print(_,__)
import math
import time
from tkinter import *
def f(x):
import ctypes
@golden finch :x: Your eval job has completed with return code 1.
001 | ('math', {'__name__': '__main__', '__doc__': None, '__package__': None, '__loader__': <class '_frozen_importlib.BuiltinImporter'>, '__spec__': None, '__annotations__': {}, '__builtins__': <module 'builtins' (built-in)>}, {'__name__': '__main__', '__doc__': None, '__package__': None, '__loader__': <class '_frozen_importlib.BuiltinImporter'>, '__spec__': None, '__annotations__': {}, '__builtins__': <module 'builtins' (built-in)>}, None, 0) {}
002 | ('time', {'__name__': '__main__', '__doc__': None, '__package__': None, '__loader__': <class '_frozen_importlib.BuiltinImporter'>, '__spec__': None, '__annotations__': {}, '__builtins__': <module 'builtins' (built-in)>, 'math': None}, {'__name__': '__main__', '__doc__': None, '__package__': None, '__loader__': <class '_frozen_importlib.BuiltinImporter'>, '__spec__': None, '__annotations__': {}, '__builtins__': <module 'builtins' (built-in)>, 'math': None}, None, 0) {}
003 | ('tkinter', {'__name__': '__main__', '__doc__': None, '__package
... (truncated - too long)
Full output: https://paste.pythondiscord.com/facibuhohe.txt?noredirect
rugged sparrow
!e py __builtins__.__import__ = lambda *a, **k: print(a, k) from foo.bar import b as c
night quarryBOT
@rugged sparrow :x: Your eval job has completed with return code 1.
001 | ('foo.bar', {'__name__': '__main__', '__doc__': None, '__package__': None, '__loader__': <class '_frozen_importlib.BuiltinImporter'>, '__spec__': None, '__annotations__': {}, '__builtins__': <module 'builtins' (built-in)>}, {'__name__': '__main__', '__doc__': None, '__package__': None, '__loader__': <class '_frozen_importlib.BuiltinImporter'>, '__spec__': None, '__annotations__': {}, '__builtins__': <module 'builtins' (built-in)>}, ('b',), 0) {}
002 | Traceback (most recent call last):
003 | File "<string>", line 2, in <module>
004 | ('io', {'__builtins__': <module 'builtins' (built-in)>}, {'__builtins__': <module 'builtins' (built-in)>}, [], 0) {}
005 | ImportError: cannot import name 'b' from '<unknown module name>' (unknown location)
dire sparrow
Oh god.
cerulean rivet
guys, is possible doing that without metaclass?```py
class _MetaModel(type):
def len(self):
return self.hack
class Model(metaclass=_MetaModel):
pass
class Player(Model):
hack = 1
len(Player) # 1```
how?
night quarryBOT
@sick hound :white_check_mark: Your eval job has completed with return code 0.
1
this don't changes nothing 
the difference is only ```py
class Model(metaclass=_MetaModel):
class Model(_MetaModel('Model',(),{})):
make a class that supports len
i'm using __init_subclass__ of Model
import typing
T = typing.TypeVar('T', bound="Model")
class _attribute:
__slots__ = ()
def __init__(self) -> None:
pass
def __get__(self, *_) -> typing.Any:
...
def __set__(*_) -> None:
raise AttributeError("you can't set this attribute")
class ModelMeta(type):
def __len__(self):
return 1
class Model(metaclass=ModelMeta):
__models__: typing.ClassVar[dict[str, type[T]]] = {}
def __init_subclass__(cls, **options) -> None:
self: T = cls.__mro__[-2]
if cls.__base__ != self:
raise RuntimeError("you can't subclass a Model")
name = options.pop("name", cls.__name__)
self.__models__[name] = cls
cls.it = []
for identifier, annotation in cls.__annotations__.items():
...
descriptor = _attribute()
setattr(cls, identifier, descriptor)
class Person(Model, name="pessoa"):
name: str
age: int
class Book(Model, name="livro"):
title: str
author: Person
typehints
lmao
golden finch
What's going on here?
!e
original=__builtins__.__build_class__
def build(*args,**kwargs):
print(args,kwargs)
return original(*args,**kwargs)
__builtins__.__build_class__=build
import re
@golden finch :white_check_mark: Your eval job has completed with return code 0.
001 | (<function _C at 0x7fc20a2c9820>, '_C') {}
002 | (<function DynamicClassAttribute at 0x7fc20a2c99d0>, 'DynamicClassAttribute') {}
003 | (<function _GeneratorWrapper at 0x7fc20a2c99d0>, '_GeneratorWrapper') {}
004 | (<function auto at 0x7fc20a2e0670>, 'auto') {}
005 | (<function _EnumDict at 0x7fc20a2e0670>, '_EnumDict', <class 'dict'>) {}
006 | (<function EnumMeta at 0x7fc20a2e0670>, 'EnumMeta', <class 'type'>) {}
007 | (<function Enum at 0x7fc20a2e0670>, 'Enum') {'metaclass': <class 'enum.EnumMeta'>}
008 | (<function IntEnum at 0x7fc20a2e0670>, 'IntEnum', <class 'int'>, <enum 'Enum'>) {}
009 | (<function Flag at 0x7fc20a2e19d0>, 'Flag', <enum 'Enum'>) {}
010 | (<function IntFlag at 0x7fc20a2e19d0>, 'IntFlag', <class 'int'>, <enum 'Flag'>) {}
011 | (<function error at 0x7fc20a2e3d30>, 'error', <class 'Exception'>) {}
... (truncated - too many lines)
Full output: https://paste.pythondiscord.com/kuvetubiyo.txt?noredirect
golden finch
so that's the original builtin
!e
original=__builtins__.__build_class__
def build(f,n,*bases,**kwargs):
old=original(f,n,*bases,**kwargs)
if 'metaclass' in kwargs:
return kwargs['metaclass'](n,bases,dict(old.__dict__))
else:
return type(n,bases,dict(old.__dict__))
__builtins__.__build_class__=build
import re
@golden finch :x: Your eval job has completed with return code 1.
001 | Traceback (most recent call last):
002 | File "<string>", line 10, in <module>
003 | File "/usr/local/lib/python3.9/re.py", line 124, in <module>
004 | import enum
005 | File "/usr/local/lib/python3.9/enum.py", line 664, in <module>
006 | class Enum(metaclass=EnumMeta):
007 | File "<string>", line 3, in build
008 | File "/usr/local/lib/python3.9/enum.py", line 169, in __prepare__
009 | enum_dict = _EnumDict()
010 | File "/usr/local/lib/python3.9/enum.py", line 82, in __init__
011 | super().__init__()
... (truncated - too many lines)
Full output: https://paste.pythondiscord.com/neyigedivo.txt?noredirect
rapid sparrow
What's going on here?
__build_class__ is showing the functions that set up the scope for each class
golden finch
I wrote the code in the first example
The thing is
Why doesn't the second work?
I use the original __build_class__ function to get the namespace
or is it a mistake in metaclasses?
!e
original=__builtins__.__build_class__
def build(f,n,*bases,**kwargs):
old=original(f,n,*bases,**kwargs)
return type(n,bases,dict(old.__dict__))
__builtins__.__build_class__=build
import re
@golden finch :x: Your eval job has completed with return code 1.
001 | Traceback (most recent call last):
002 | File "<string>", line 7, in <module>
003 | File "/usr/local/lib/python3.9/re.py", line 124, in <module>
004 | import enum
005 | File "/usr/local/lib/python3.9/enum.py", line 664, in <module>
006 | class Enum(metaclass=EnumMeta):
007 | File "<string>", line 3, in build
008 | File "/usr/local/lib/python3.9/enum.py", line 169, in __prepare__
009 | enum_dict = _EnumDict()
010 | File "/usr/local/lib/python3.9/enum.py", line 82, in __init__
011 | super().__init__()
... (truncated - too many lines)
Full output: https://paste.pythondiscord.com/uwamihelas.txt?noredirect
golden finch
same error
Traceback (most recent call last):
File "<string>", line 7, in <module>
File "/usr/local/lib/python3.9/re.py", line 124, in <module>
import enum
File "/usr/local/lib/python3.9/enum.py", line 664, in <module>
class Enum(metaclass=EnumMeta):
File "<string>", line 3, in build
File "/usr/local/lib/python3.9/enum.py", line 169, in __prepare__
enum_dict = _EnumDict()
File "/usr/local/lib/python3.9/enum.py", line 82, in __init__
super().__init__()
TypeError: super(type, obj): obj must be an instance or subtype of type
@rapid sparrow in this example, what am I doing wrong with my reimplementation? it doesn't appear to have anything obviously wrong with it
code reproduced below
original=__builtins__.__build_class__
def build(f,n,*bases,**kwargs):
old=original(f,n,*bases,**kwargs)
if 'metaclass' in kwargs:
return kwargs['metaclass'](n,bases,dict(old.__dict__))
else:
return type(n,bases,dict(old.__dict__))
__builtins__.__build_class__=build
import re
!e
original=__import__("builtins").__build_class__
def build(func, name, *bases, **kwargs):
callee = kwargs["metaclass"] if "metaclass" in kwargs else type
ns = {}
print(f"exec({func.__code__=}, {ns=}, func.__globals__=)")
global __name__
__name__ = func.__module__
exec(func.__code__, func.__globals__)
print(f"callee({name=}, {bases=}, ns=)")
cls = callee(name, bases, ns)
print(f"{cls=}")
return cls
__import__("builtins").__build_class__=build
import re
@rapid sparrow :x: Your eval job has completed with return code 1.
001 | exec(func.__code__=<code object _C at 0x7fdc6884bbe0, file "/usr/local/lib/python3.9/types.py", line 39>, ns={}, func.__globals__=)
002 | callee(name='_C', bases=(), ns=)
003 | cls=<class 'types._C'>
004 | Traceback (most recent call last):
005 | File "<string>", line 14, in <module>
006 | File "/usr/local/lib/python3.9/re.py", line 124, in <module>
007 | import enum
008 | File "/usr/local/lib/python3.9/enum.py", line 2, in <module>
009 | from types import MappingProxyType, DynamicClassAttribute
010 | File "/usr/local/lib/python3.9/types.py", line 41, in <module>
011 | MethodType = type(_C()._m)
... (truncated - too many lines)
Full output: https://paste.pythondiscord.com/umulehusup.txt?noredirect
rapid sparrow
!e
import sys
def build(func, name, *bases, **kwargs):
callee = kwargs["metaclass"] if "metaclass" in kwargs else type
ns = {}
print(f"exec({func.__code__=}, {ns=}, func.__globals__=)")
global __name__
__name__ = func.__module__
print(f"callee({name=}, {bases=}, {ns=})")
exec(func.__code__, sys.modules[func.__module__].__dict__, ns)
print(f"callee({name=}, {bases=}, {ns=})")
cls = callee(name, bases, ns)
sys.modules[func.__module__].__dict__[name] = cls
print(f"{cls=}")
return cls
__builtins__.__build_class__=build
import re
import inspect
@rapid sparrow :x: Your eval job has completed with return code 1.
001 | exec(func.__code__=<code object _C at 0x7f48dcaa7b30, file "/usr/local/lib/python3.9/types.py", line 39>, ns={}, func.__globals__=)
002 | callee(name='_C', bases=(), ns={})
003 | callee(name='_C', bases=(), ns={'__module__': 'types', '__qualname__': '_C', '_m': <function _C._m at 0x7f48dca9b940>})
004 | cls=<class 'types._C'>
005 | exec(func.__code__=<code object DynamicClassAttribute at 0x7f48dcaa9500, file "/usr/local/lib/python3.9/types.py", line 149>, ns={}, func.__globals__=)
006 | callee(name='DynamicClassAttribute', bases=(), ns={})
007 | callee(name='DynamicClassAttribute', bases=(), ns={'__module__': 'types', '__qualname__': 'DynamicClassAttribute', '__doc__': "Route attribute access on a class to __getattr__.\n\n This is a descriptor, used to define attributes that act differently when\n accessed through an instance and through a class. Instance access remains\n normal, but access to an attribute through a class will be routed to the\n class's __getattr__ m
... (truncated - too long, too many lines)
Full output: https://paste.pythondiscord.com/hacitoxena.txt?noredirect
rapid sparrow
<@!856229099952144464> in this example, what am I doing wrong with my reimplemen...
i think something is wrong with metaclass version?
golden finch
?
rapid sparrow
in my version it dies in some code that's running while EnumMeta is set up (code trying to use .attr syntax on 'classdict')
setting up Enum with metaclass=EnumMeta, I mean
rapid sparrow
<@!856229099952144464> in this example, what am I doing wrong with my reimplemen...
it looks like you can't make a working __build_class__ in pure python, sadly
grave rover
2 0 LOAD_CONST 1 (<function on_read at 0x7f41840e4dc0>)
2 LOAD_CONST 2 (<class 'mixin.callback.CallbackInfo'>)
4 CALL_FUNCTION 0
6 CALL_FUNCTION 1
8 POP_TOP
10 LOAD_GLOBAL 0 (input)
12 LOAD_CONST 3 (<function change_prompt at 0x7f41840e4ee0>)
14 LOAD_FAST 0 (p)
16 CALL_FUNCTION 1
18 CALL_FUNCTION 1
20 RETURN_VALUE
Traceback (most recent call last):
File "/home/mart/git/pyasm/pymixin/test/test.py", line 28, in <module>
main()
File "/home/mart/git/pyasm/pymixin/test/test.py", line 21, in main
line = read_line()
File "/home/mart/git/pyasm/pymixin/test/test_lib.py", line -1, in read_line
TypeError: 'int' object is not callable
I dont even know anymore
rapid sparrow
lmao
golden finch
it looks like you can't make a working `__build_class__` in pure python, sadly
dang, why not?
rapid sparrow
dang, why not?
AC: cannot convert yet, waiting for *args support
oh, nevernind, that comment is about argument clinic (was looking at https://github.com/python/cpython/blob/main/Python/bltinmodule.c#L105)
night quarryBOT
Python/bltinmodule.c line 105
builtin___build_class__(PyObject *self, PyObject *const *args, Py_ssize_t nargs,```I'm not, or at least I don't think I am
no clue where in the bytecode that happens
an int
which is a parameter to change_prompt
that's the output from dis, yes
it's also 1:1 with what I expect the bytecode to be
well uh
code injection
kinda hard to debug
wait I think I know what the issue is
hmm
grave rover
so the wacky thing is:
- I initially inject some code
- That code applies all modifications to itself
- When it returns the bytecode will have changed and it'll be calling an int
Before injection:py 2 0 LOAD_CONST 2 (<function get_wrapper.<locals>.inject at 0x7f42013ace50>) 2 CALL_FUNCTION 0 4 POP_TOP 6 LOAD_CONST 3 (<function read_line at 0x7f42013ac8b0>) 8 LOAD_FAST 0 (p) 10 CALL_FUNCTION 0 # Call modified self 12 RETURN_VALUE 14 LOAD_GLOBAL 0 (input) 16 LOAD_CONST 1 ('> ') 18 CALL_FUNCTION 1 20 RETURN_VALUE After injection: ```py
2 0 LOAD_CONST 1 (False)
2 POP_JUMP_IF_FALSE 7 (to 14) # Jump to the function itself
4 LOAD_CONST 1 (False)
6 NOP
8 NOP
10 NOP
12 RETURN_VALUE # Match RETURN_VALUE in previous function
>> 14 LOAD_CONST 2 (<function on_read at 0x7f42013acdc0>)
16 LOAD_CONST 3 (<class 'mixin.callback.CallbackInfo'>)
18 CALL_FUNCTION 0
20 CALL_FUNCTION 1
22 POP_TOP
24 LOAD_GLOBAL 0 (input)
26 LOAD_CONST 4 (<function change_prompt at 0x7f42013acee0>)
28 LOAD_FAST 0 (p)
30 CALL_FUNCTION 1
32 CALL_FUNCTION 1
34 RETURN_VALUE
yet it still happens...
golden finch
maybe one of those functions is pushing something to the stack that you're not expecting
rapid sparrow
dang, why not?
im wondering this CO_NEWLOCALS flag might be part of the issue:
https://stackoverflow.com/questions/63536320/build-class-in-python-3-5
grave rover
maybe one of those functions is pushing something to the stack that you're not e...
it shouldn't though...
because it'd have its own stack
rapid sparrow
ah
maybe the problem might be that you call the metaclass without calling __prepare__ first?
rapid sparrow
thanks, will test that out
def build_class(func, name, *bases, metaclass=None, **kwds):
if metaclass is None:
metaclass = _extract_metaclass(bases) # may raise an exception
prepare = getattr(metaclass, "__prepare__", None)
if prepare:
namespace = prepare(name, bases, **kwds)
else:
namespace = {}
newnamespace = func(namespace)
if newnamespace is not None:
namespace = newnamespace
return metaclass(name, bases, namespace, **kwds)
def _extract_metaclass(bases):
if not bases:
return type
# XXX Should we do the "winner" calculation here?
return bases[0].__class__ # or type(bases[0])?
Guido himself wrote this one and it gets the same error, lol
golden finch
okay so it's a bug
floral meteor
what's a bug?
rapid sparrow
what's a bug?
implementing __build_class__ in python doesn't work
in 3.9 and 3.11 at least
i'd like to know if there's a way to intercept the creation of BaseException instances so I can print all the exceptions happening
possible?
there's a metric ton of them, judging by running python to the prompt and setting a breakpoint on
int PyException_SetTraceback(PyObject *, PyObject *)
184 exceptions with full tracebacks to get to >>>
mostly KeyError (looking in sys.modules) and various FileNotFoundError from the import system
grave rover
implementing `__build_class__` in python doesn't work
what's the issue?
rapid sparrow
what's the issue?
just seems like a lot for the happy path, no?
i don't think it's particular to my system or anything, just wondered what others thought
floral meteor
what's wrong with __build_class__?
golden finch
the example implementation doesn't even work
grave rover
sort of, read_line is the function I inject code into, and the resulting bytecode is there
grave rover
there's a metric ton of them, judging by running python to the prompt and settin...
How are you setting a breakpoint there btw?
rapid sparrow
How are you setting a breakpoint there btw?
$ gdb --args python3
(gdb) set $p = 0
(gdb) b PyException_SetTraceback
(gdb) commands
> print ($p = $p + 1)
> c
> end
(gdb) r
you have to be able to set a breakpoint in C, i don't know if it's possible in PyCharm?
grave rover
in pycharm you just click on the left of some python code lol
rapid sparrow
in pycharm you just click on the left of some python code lol
it's apparently possible with vscode, visual studio, dunno what else
well any native debugger really i guess, that function is exported so even like perf, valgrind, etc should see it
it would be great if there's a way to hook into similar in python
maybe pdb lets you break on BaseException.__init__ or on setting the __traceback__ attribute, but I don't know PDB very well
rapid sparrow
in pycharm you just click on the left of some python code lol
actually, I just remembered you can have pycharm break on exceptions i think
rapid sparrow
theres a Notifications box too in PyCharm that defaults to just "Terminate", you probably want "Always" also checked
rugged sparrow
maybe pdb lets you break on `BaseException.__init__` or on setting the `__traceb...
You could do that with fishhook
rapid sparrow
You could do that with fishhook
oh, that looks awesome
omg it worked
>>> import fishhook
>>> @fishhook.hook(BaseException)
... def __init__(self, *args, **kwargs):
... print(f"Got exception: {self}({args})")
... return fishhook.orig(self, *args, **kwargs)
...
>>> raise BaseException("hello")
Got exception: hello(('hello',))
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
BaseException: hello
>>>
>>> raise RuntimeError("hello")
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
RuntimeError: hello
i guess you would need to hook the specific class?
rugged sparrow
yea one sec imma write a demo
yea you need to hook them specifically
im gonna try using my lowlevel hook to hook the C function tho
rapid sparrow
im gonna try using my lowlevel hook to hook the C function tho
oh you wrote that package? that's awesome
i was just gonna try to create the hooks in a loop, since most of the exception types are in builtins
rugged sparrow
that will work for most of them, but some are used for general control flow so you run into fatal recursion issues
and yea i wrote fishhook but im gonna use my asm_hook code to try and hook the underlying C functions
rapid sparrow
this worked:
import fishhook
def __init__(self, *args, **kwargs):
print(f"Got exception: {self}({args})")
return fishhook.orig(self, *args, **kwargs)
for _, exc_type in inspect.getmembers(builtins, lambda m: inspect.isclass(m) and issubclass(m,BaseException)):
fishhook.hook(exc_type)(__init__)
quick and dirty good enough for me, the only surprise is some of the exceptions:
Got exception: hello(('hello',))
Got exception: 'ModuleSpec' object has no attribute '_initializing'(("'ModuleSpec' object has no attribute '_initializing'",))
is there a way to tell which classes would work well?
i didnt think that would be possible without a debugger, very neat
already revealed an AttributeError that was unintentionally suppressed in some code I was working on
rugged sparrow
In theory with my asm_hook code you could do a breakpoint like gdb
grave rover
okay figured out my issue
rapid sparrow
In theory with my asm_hook code you could do a breakpoint like gdb
you want to share the code?
does it rewrite instructions?
rapid sparrow
nice
grave rover
def read_line() -> str:
return input("> ")
@inject(method=read_line, at=At(value=AtValue.HEAD))
def on_read(ci: CallbackInfo):
print("Hello Mixin :)")
@modify_const(method=read_line,
at=At(value=AtValue.LOAD,
target="> "))
def change_prompt():
return ">>> "
i = 0
@redirect(method=read_line,
at=At(value=AtValue.INVOKE,
target=input))
def redirected_input(prompt: str):
global i
i = i + 1
return input("{0} {1}".format(i, prompt))
``` ```
Hello Mixin :)
1 >>> a
Read line: a
Hello Mixin :)
2 >>> b
Read line: b
Hello Mixin :)
3 >>> c
Read line: c
Hello Mixin :)
4 >>>
```Pretty happy so faris it posted anywhere?
floral meteor
lol iostream hacks
grave rover
is it posted anywhere?
Not yet, pushing later this weekend
Then I'll need to make a package name request
grave rover
How do freevars/cellvars work?
rapid sparrow
How do freevars/cellvars work?
in the case of cellvars, it's basically a pointer to a variable used in multiple scopes (like a function foo and nested_foo). The variables get dereferenced by extra sequences of bytecode the compiler emits, instead of accessed directly like a normal variable
freevars are unresolved names afaik, that can work if an attribute is added by that name to a parent scope before the function with the freevar gets run
you can find the cell objects if you look at __closure__ on such a function
grave rover
so e.g.
def make_func(prompt):
def get_input() -> str:
return input(prompt)
return get_input```here prompt is a freevar accessed by the *_DEREF ops
what would an example of a cellvar be? (and what opcodes relate to cellvars?)
sick hound
My vs code terminal isn't showing any output can anyone plz help?
grave rover
My vs code terminal isn't showing any output can anyone plz help?
wrong channel, try #โ๏ฝhow-to-get-help or #editors-ides
rapid sparrow
>>> def foo(a):
... def nested():
... return a
... return nested
...
>>> n = foo(70); n
<function foo.<locals>.nested at 0x7ff797d670>
>>> n.__closure__
(<cell at 0x7ff7b04670: int object at 0x7ff7c8a210>,)
>>> c = n.__closure__[0]
>>> c.cell_contents
70
>>> foo.__code__.co_cellvars
('a',)
>>> n.__code__.co_freevars
('a',)
>>> n.__code__.co_cellvars
()
its a freevar in the nested func and a cellvar in the outer one, looks like
but it seems like nested must have special opcodes to access ut via the cell
>>> dis.dis(foo)
2 0 LOAD_CLOSURE 0 (a)
2 BUILD_TUPLE 1
4 LOAD_CONST 1 (<code object nested at 0x7ff7a13500, file "<stdin>", line 2>)
6 LOAD_CONST 2 ('foo.<locals>.nested')
8 MAKE_FUNCTION 8 (closure)
10 STORE_FAST 1 (nested)
4 12 LOAD_FAST 1 (nested)
14 RETURN_VALUE
Disassembly of <code object nested at 0x7ff7a13500, file "<stdin>", line 2>:
3 0 LOAD_DEREF 0 (a)
2 RETURN_VALUE
LOAD_DEREF, seems to be the opcode you're asking about?
and LOAD_CLOSURE for cellvars maybe
same deal for your make_func:
>>> dis.dis(make_func)
2 0 LOAD_GLOBAL 0 (str)
2 LOAD_CONST 1 (('return',))
4 BUILD_CONST_KEY_MAP 1
6 LOAD_CLOSURE 0 (prompt)
8 BUILD_TUPLE 1
10 LOAD_CONST 2 (<code object get_input at 0x7ff766a240, file "<stdin>", line 2>)
12 LOAD_CONST 3 ('make_func.<locals>.get_input')
14 MAKE_FUNCTION 12 (annotations, closure)
16 STORE_FAST 1 (get_input)
4 18 LOAD_FAST 1 (get_input)
20 RETURN_VALUE
Disassembly of <code object get_input at 0x7ff766a240, file "<stdin>", line 2>:
3 0 LOAD_GLOBAL 0 (input)
2 LOAD_DEREF 0 (prompt)
4 CALL_FUNCTION 1
6 RETURN_VALUE
I see
floral meteor
im learning java and i feel like im back to coding like a noob, only i already know these algorithms, just not the java code for them
it's like in path of exile when you gotta go through the whole map again from the start, but it's all different, and much harder
grave rover
should I make un-wrapping wraps a feature in this or leave that to the user
snow beacon
im learning java and i feel like im back to coding like a noob, only i already k...
Have you learnt any esoteric Java yet?
bronze sorrel
tf is this
tribal moon
Java is already esoteric
rapid sparrow
its hard to think what would be esoteric java
class AbstractClassAwareSecurityBeanUtils
rapid sparrow
im learning java and i feel like im back to coding like a noob, only i already k...
you need the "Java for people who aren't new to programming" one
next flame
one * at the beginning I think?
instead of **
yes
I am responding to your error
because ** is for dict-like mappings
which map() is not
night quarryBOT
@sick hound :white_check_mark: Your eval job has completed with return code 0.
001 | 0
002 | 1
003 | 2
004 | 3
005 | 4
006 | 5
007 | 6
008 | 7
009 | 8
010 | 9
011 | 10
... (truncated - too many lines)
Full output: https://paste.pythondiscord.com/ocemeseter.txt?noredirect
golden finch
hmm yes (){}[]
floral meteor
Have you learnt any esoteric Java yet?
Ive already figured out i can write up junk, then compile and decompile to get better written code.
It converts a = a + 1 to ++a
I tried using finalize to hang java but i couldnt find the relevant implements object
golden finch
<function <lambda>.<locals>.<lambda> at 0x0000018677E7E040>
this is unpleasant
night quarryBOT
@sick hound :white_check_mark: Your eval job has completed with return code 0.
<function <lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.<lambda>.<locals>.
... (truncated - too long)
Full output: https://paste.pythondiscord.com/oyujuhuvaz.txt?noredirect
golden finch
operator=lambda _:lambda*__:_(*_(*__))
this can be used to segfault... what?
if you do it with 10k
!e
unpleasant=lambda x:exec("raise BaseException('no')")if x<1else eval('('+'lambda:'*(x+1)+'0)'+'()'*x)
print(unpleasant(100000))
@golden finch :warning: Your eval job has completed with return code 139 (SIGSEGV).
[No output]
@golden finch :warning: Your eval job timed out or ran out of memory.
[No output]
no, it's definitely a segfault
late quail
frigid wharf
no, it's definitely a segfault
it's a recursion error
!e py unpleasant=lambda x:exec("raise BaseException('no')")if x<1else eval('('+'lambda:'*(x+1)+'0)'+'()'*x) print(unpleasant(1500))
night quarryBOT
@frigid wharf :x: Your eval job has completed with return code 1.
001 | Traceback (most recent call last):
002 | File "<string>", line 2, in <module>
003 | File "<string>", line 1, in <lambda>
004 | RecursionError: maximum recursion depth exceeded during compilation
Unsafe, Classloader, and probably more
rapid sparrow
rapid sparrow
if you do it with 10k
3.10 doesnt crash, building 3.9 to see if it does
rapid sparrow
Unsafe, Classloader, and probably more
i mean they took out Unsafe completely and made reflection basically illegal.. glad I don't have to write anything in Java at the moment
grave rover
i mean they took out `Unsafe` completely and made reflection basically illegal.....
Really? it's still usable in J16
rapid sparrow
crashes during tokenization, looks like?
10,000 it could handle but not 100,000 ๐
rapid sparrow
Really? it's still usable in J16
which?
looks like they moved Unsafe to the jdk.unsupported module in J11, so I guess if it's still there now, they effectively won't be removing it
lmao
rapid sparrow
I'm using OpenJDK if that helps
no problem, glad to hear it works
grave rover
@rugged sparrow I think you were the one who asked:
https://github.com/martmists-gh/pyasm
https://github.com/martmists-gh/pymixin
Gonna see about pypi releases later
rugged sparrow
Sweet I'll take a look
grave rover
well uh
I already identified some issues with it
jumps in pyasm are not 100% correct yet
and it's a mess
rapid sparrow
and it's a mess
couldn't get the code objects to actually work, myself
golden finch
3.10 doesnt crash, building 3.9 to see if it does
just add more zeroes
night quarryBOT
@sick hound :white_check_mark: Your eval job has completed with return code 0.
<class '__main__.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A
... (truncated - too long)
Full output: https://paste.pythondiscord.com/nuhahabipo.txt?noredirect
rapid sparrow
just add more zeroes
thanks, that helped ๐
tokenizer bug bounty
here I come
jk
night quarryBOT
@sick hound :warning: Your eval job timed out or ran out of memory.
[No output]
You can only use this command in the following channel(s): #878695047112040448, #476196062214750219, #310-release-stream, #bot-commands
golden finch
waht?
dire galleon
?
rapid sparrow
whats it not doing right?
fiery hare
!e
#include <stdint.h>
#include <stdio.h>
#/* pre-declare types and functions */
void: (type) = (type) ("(void *)0"),
int16_t: (type) = (type) (int(16))(),
int32_t: (type) = (type) (int(32))(),
puts = (print),
printf: ((str) := void) = (puts),
#/* confirm function and variable declarations */
[[void]] = (void),
[[total]] = (int32_t),
[[i]] = (int32_t),
[[puts]] = (puts),
[[printf]] = (printf),
#/* ensure forwards and backwards compatibility */
import __future__ as __past__
#define FUTURE PAST
#define PAST FUTURE
#/* output a formatted string to stdout */
def printf(s, __VA_ARGS__) -> (void) :{
(void) (puts((void).__mod__(s, __VA_ARGS__), end=((void).__new__)(void))),
}
#/*******************************************************
# * SUMMATION SCRIPT *
# *******************************************************
# * Sum up all the natural numbers up to 6. The result *
# * should be 0 + 1 + 2 + 3 + 4 + 5 + 6 = 21. *
# *******************************************************/
#/* print numbers from 0 to 6 */
int32_t: i = 0,
while (i <= 6) :{
(void) (printf(" %d\n", i)),
(i := i + 1),
}
(void) (puts("__")),
#/* sum up numbers from 0 to 6 */
int32_t: total = 0,
int32_t: i = 0,
for (int32_t) in (i <- 0, i <= 6, ++i) :{
(total := total + i),
}
(void) (printf("%d\n", total)),
@fiery hare :x: Your eval job has completed with return code 1.
001 | File "<string>", line 8
002 | puts: ((str) := void) = (print),
003 | ^
004 | SyntaxError: cannot use assignment expressions with name
!e
#include <stdint.h>
#include <stdio.h>
#/* pre-declare types and functions */
void: (type) = (type) ("(void *)0"),
int16_t: (type) = (type) (int(16))(),
int32_t: (type) = (type) (int(32))(),
puts = (print),
printf = (puts),
#/* confirm function and variable declarations */
[[void]] = (void),
[[total]] = (int32_t),
[[i]] = (int32_t),
[[puts]] = (puts),
[[printf]] = (printf),
#/* ensure forwards and backwards compatibility */
import __future__ as __past__
#define FUTURE PAST
#define PAST FUTURE
#/* output a formatted string to stdout */
def printf(s, __VA_ARGS__) -> (void) :{
(void) (puts((void).__mod__(s, __VA_ARGS__), end=((void).__new__)(void))),
}
#/*******************************************************
# * SUMMATION SCRIPT *
# *******************************************************
# * Sum up all the natural numbers up to 6. The result *
# * should be 0 + 1 + 2 + 3 + 4 + 5 + 6 = 21. *
# *******************************************************/
#/* print numbers from 0 to 6 */
int32_t: i = 0,
while (i <= 6) :{
(void) (printf(" %d\n", i)),
(i := i + 1),
}
(void) (puts("__")),
#/* sum up numbers from 0 to 6 */
int32_t: total = 0,
int32_t: i = 0,
for (int32_t) in (i <- 0, i <= 6, ++i) :{
(total := total + i),
}
(void) (printf("%d\n", total)),
@fiery hare :white_check_mark: Your eval job has completed with return code 0.
001 | 0
002 | 1
003 | 2
004 | 3
005 | 4
006 | 5
007 | 6
008 | __
009 | 21
@terse mortar this is the worst thing I've ever experienced.
terse mortar
That is amazing
fiery hare
no it's not
terse mortar
Yes it is
fiery hare
I'm so fucking upset right now.
#/* ensure forwards and backwards compatibility */
import __future__ as __past__
#define FUTURE PAST
#define PAST FUTURE
sly root
!e
```py
#include <stdint.h>
#include <stdio.h>
#/* pre-declare types and funct...
How this works? I mean why #includes are needed
night quarryBOT
@sick hound :x: Your eval job has completed with return code 1.
001 | Traceback (most recent call last):
002 | File "<string>", line 3, in <module>
003 | MemoryError
@sick hound :warning: Your eval job timed out or ran out of memory.
[No output]
@sick hound :x: Your eval job has completed with return code 1.
001 | Traceback (most recent call last):
002 | File "<string>", line 1, in <module>
003 | ModuleNotFoundError: No module named 'psutil'
@sick hound :warning: Your eval job has completed with return code 0.
[No output]
Stelercus stop making me cry :(((
grave rover
not sure if I should be glad or worried that someone has plans for my code injection framework
I did finally make a somewhat decent readme though
golden finch
!e
```py
#include <stdint.h>
#include <stdio.h>
#/* pre-declare types and funct...
good god how does this work
grave rover
oh god you're using commas to secretly make tuples
golden finch
so that's what the commas do?
grave rover
yeah
grave rover
high-tier black magic
how black magic is it compared to this ```py
from mixin import *
def return_n_squared(n):
... return n * n
...
@inject(method=return_n_squared, at=At(value=AtValue.RETURN), cancellable=True)
... def return_n_cubed_instead(callback_info: CallbackInfo):
... n = (callback_info.return_value0.5)
... callback_info.set_return(n3)
...
return_n_squared(10)
1000.0
missing a )
night quarryBOT
@sick hound :white_check_mark: Your eval job has completed with return code 0.
001 | 0
002 | 1
003 | 2
004 | 3
005 | 4
006 | 5
007 | 6
008 | __
009 | 21
steep herald
!e print("geh")
night quarryBOT
@steep herald :white_check_mark: Your eval job has completed with return code 0.
geh
!e
S = "a"
key = 1097
print(chr(ord(S)+key))
!e
S = "ag"
key = 1097
print(chr(ord(S)+key))
sick hound
Can you write a "Hello, World!" Python program where no characters with holes in them are used? 
so #%DOo09qaA etc not allowed
sick hound
correct
night quarryBOT
@sick hound :white_check_mark: Your eval job has completed with return code 0.
Hello, World!
p
sick hound
If it's easier try only characters with holes in them 
comeon, you've got in with try is if chr, uhh
yeah, it might be impossible 
earnest wing
!e ```py
{}["H"+chr(1111//11)+"ll"+chr(111)+", W"+chr(111)+"rl"+chr(73+27)+"!"]
@earnest wing :x: Your eval job has completed with return code 1.
001 | Traceback (most recent call last):
002 | File "<string>", line 1, in <module>
003 | KeyError: 'Hello, World!'
!e
print(["H"+chr(1111//11)+"ll"+chr(111)+", W"+chr(111)+"rl"+chr(73+27)+"!"])
@knotty delta :white_check_mark: Your eval job has completed with return code 0.
['Hello, World!']
!e ```py
print(bytes([0x18*int(float('0.0'+str(int('0x249e',16)+0x1))*0x20),int(float(str(0x5)+'.'+str(int('0x216ff33c00f0f3',16)))*0x11),int(float(str(0x1)+'.'+str(int('0x32dc227a421af',16)))*int(float(str(0x4)+'.0'+str(int('0x40f6c2a076db',16)))*0xe)),int(float(str(0x1)+'.'+str(int('0x32dc227a421af',16)))*int(float(str(0x4)+'.0'+str(int('0x40f6c2a076db',16)))*0xe)),int(float(str(0x0)+'.'+str(0x0)+str(0x0)+str(int('0x299b48154dfd3e',16)))*int(float(str(0x46)+'.'+str(int('0x16ba2e55f4a',16)))*0x547))]))
@sly root :white_check_mark: Your eval job has completed with return code 0.
b'Hello'
why this work???py ๐น๐ป๐ฒ๐ท๐ฝ(๐ฌ๐ฑ๐ป(97)+๐ฌ๐ฑ๐ป(98)+๐ฌ๐ฑ๐ป(99)+๐ฌ๐ฑ๐ป(100)+๐ฌ๐ฑ๐ป(101)+๐ฌ๐ฑ๐ป(102)+๐ฌ๐ฑ๐ป(103)+๐ฌ๐ฑ๐ป(104)+๐ฌ๐ฑ๐ป(105)+๐ฌ๐ฑ๐ป(106)+๐ฌ๐ฑ๐ป(107)+๐ฌ๐ฑ๐ป(108)+๐ฌ๐ฑ๐ป(109)+๐ฌ๐ฑ๐ป(110)+๐ฌ๐ฑ๐ป(111)+๐ฌ๐ฑ๐ป(112)+๐ฌ๐ฑ๐ป(113)+๐ฌ๐ฑ๐ป(114)+๐ฌ๐ฑ๐ป(115)+๐ฌ๐ฑ๐ป(116)+๐ฌ๐ฑ๐ป(117)+๐ฌ๐ฑ๐ป(118)+๐ฌ๐ฑ๐ป(119)+๐ฌ๐ฑ๐ป(120)+๐ฌ๐ฑ๐ป(121)+๐ฌ๐ฑ๐ป(122)+๐ฌ๐ฑ๐ป(65)+๐ฌ๐ฑ๐ป(66)+๐ฌ๐ฑ๐ป(67)+๐ฌ๐ฑ๐ป(68)+๐ฌ๐ฑ๐ป(69)+๐ฌ๐ฑ๐ป(70)+๐ฌ๐ฑ๐ป(71)+๐ฌ๐ฑ๐ป(72)+๐ฌ๐ฑ๐ป(73)+๐ฌ๐ฑ๐ป(74)+๐ฌ๐ฑ๐ป(75)+๐ฌ๐ฑ๐ป(76)+๐ฌ๐ฑ๐ป(77)+๐ฌ๐ฑ๐ป(78)+๐ฌ๐ฑ๐ป(79)+๐ฌ๐ฑ๐ป(80)+๐ฌ๐ฑ๐ป(81)+๐ฌ๐ฑ๐ป(82)+๐ฌ๐ฑ๐ป(83)+๐ฌ๐ฑ๐ป(84)+๐ฌ๐ฑ๐ป(85)+๐ฌ๐ฑ๐ป(86)+๐ฌ๐ฑ๐ป(87)+๐ฌ๐ฑ๐ป(88)+๐ฌ๐ฑ๐ป(89)+๐ฌ๐ฑ๐ป(90)+๐ฌ๐ฑ๐ป(33)+๐ฌ๐ฑ๐ป(34)+๐ฌ๐ฑ๐ป(35)+๐ฌ๐ฑ๐ป(36)+๐ฌ๐ฑ๐ป(37)+๐ฌ๐ฑ๐ป(38)+๐ฌ๐ฑ๐ป(39)+๐ฌ๐ฑ๐ป(42)+๐ฌ๐ฑ๐ป(44)+๐ฌ๐ฑ๐ป(46)+๐ฌ๐ฑ๐ป(47)+๐ฌ๐ฑ๐ป(58)+๐ฌ๐ฑ๐ป(59)+๐ฌ๐ฑ๐ป(60)+๐ฌ๐ฑ๐ป(61)+๐ฌ๐ฑ๐ป(62)+๐ฌ๐ฑ๐ป(63)+๐ฌ๐ฑ๐ป(64)+๐ฌ๐ฑ๐ป(91)+๐ฌ๐ฑ๐ป(93)+๐ฌ๐ฑ๐ป(94)+๐ฌ๐ฑ๐ป(95)+๐ฌ๐ฑ๐ป(96)+๐ฌ๐ฑ๐ป(123)+๐ฌ๐ฑ๐ป(124)+๐ฌ๐ฑ๐ป(125)+๐ฌ๐ฑ๐ป(126))
last locust
why this work???```py
๐น๐ป๐ฒ๐ท๐ฝ(๐ฌ๐ฑ๐ป(97)+๐ฌ๐ฑ๐ป(98)+๐ฌ๐ฑ๐ป(99)+๐ฌ๐ฑ๐ป(100)+๐ฌ...
Python supports most unicode letters, so it compiles to be the same as just print(chr(...)+...)
cerulean rivet
hmmm
urban cradle
while True:
print(c:=(lambda x:x[::-1])(input()))
if input()=='c':__import__("clpbrd").c(c);
when you golf too much and dont understand your own program anymore
rugged sparrow
seems like you're taking in a line of input, reversing it and storing it into c
then getting input again, if its c you copy the reversed text to the clipboard?
urban cradle
its a program to reverse text
you need to take an input, modify it, then print it
true ig
i can get rid of the print and have it automatically copy to my keyboard
but that would count as removing features so eh
### Reverse text to fix RTL text in outdated programs###
### Or you can use it for other things I do not care ###
import pyperclip
# Function to reverse input text, input text = x #
def reverse(x):
return x[::-1]
# Clear junk, only works in Windows command prompt #
clear = lambda: os.system('cls')
# Loop #
active = True
while active == True:
# Print a barrier for aesthetic reasons #
print("\n##############################################################\n CHIPS REVERSER NEW INPUT \n##############################################################\n")
# Asks for input then creates barrier (asthetic) and a new line to write input text in #
valueentered = input(str("Please enter your string to reverse \n ---------------------------------------------- \n"))
# Once entered, print a new barrier and a new line for the output #
print(" ---------------------------------------------- \n")
# Use function to reverse text #
valueoutput = reverse(valueentered)
# Print output aesthetically #
print("Output \n---------------------------------------------- \n", valueoutput, " \n---------------------------------------------- \n")
# Ask for command #
cask = input(str("Enter ""c"" to copy or ""e"" to end program : "))
# Check input for command, ignore if no command received #
if cask == ("c"):
pyperclip.copy(valueoutput) #Copy output to clipboard
print("Copied to clipboard!")
original
@rugged sparrow
rich hound
make this more cryptic?
!e
z = ", ".join(map(str, [_**2 for _ in range(10)]))
print(z)
@rich hound :white_check_mark: Your eval job has completed with return code 0.
0, 1, 4, 9, 16, 25, 36, 49, 64, 81
wdym cryptic
rich hound
exactly what i said
urban cradle
just change plain characters to chr(random equation with dumbass shir) and get rid of plain numbers and do some stuff