esoteric-python

[past plank]

just like how windows 10 is gonna be the last version of windows right

[sly ibex]

a,b=0,1
x=[]
for i in range(int(input())):x.append(a);a,b=b,a+b
print(*x)```

this is the fibonacci sequence, but im trying to golf it, some advices? this looks too begginer

[low lynx]

there's a code golf challenge for the fibonacci numbers that I think would be really similar to that so I'll avoid explaining too much

but it's probably better not to keep a list and just print stuff

[sly ibex]

ahh because the puzzoles is with a list

but thanks stickie

i could do it shorter

13... characters -

a,b,x=0,1,[]
exec('x+=a,;a,b=b,a+b;'*int(input()))
print(*x)```

[quartz wave]

why do you need a list

oh

ok i c

[low lynx]

is it possible to set something up to run code whenever globals is modified?

[sick hound]

Tho this is very wrong

[rugged sparrow]

#internals-and-peps message

We have done it a few times here but you can swap the globals dictionary with a subclass of it

[low lynx]

ooh nice

[fleet bridge]

accidental list(range(2*10**9) almost killed my computer

i do not recommend running that code

[astral rover]

i think itd be fine to run that 😉

[sick hound]

Let's see

!e ```py
list(range(2*10**9))

[night quarry]

@sick hound :x: Your 3.11 eval job has completed with return code 1.

001 | Traceback (most recent call last):
002 |   File "/home/main.py", line 1, in <module>
003 |     list(range(2*10**9))
004 | MemoryError

[astral rover]

!e list(range(2*10**9)

[night quarry]

@astral rover :x: Your 3.11 eval job has completed with return code 1.

001 |   File "/home/main.py", line 1
002 |     list(range(2*10**9)
003 |         ^
004 | SyntaxError: '(' was never closed

[low lynx]

@vast basin

[tough willow]

for me it automatically kills the process

[versed eagle]

ye there's a few ways

you can change the type of the globals dict to hook its __setitem__
you could use a tracing function
you could also do bytecode parsing but that gets super complicated when you're dealing with multiple versions of python

hooking the globals type is probably the best/easiest option

[meager zinc]

yeah last time I did it I used bytecode parsing and it is super portable (real)

[vast basin]

Looks good to me.

[fleet bridge]

For me at allocated around 40 gb of memory (considering i have only 16 gb of physical memory)

[arctic skiff]

!epy tree=lambda d:[f:=d,[(f:=f-1,print(f"{' '*f}{'*'*((j+1)*2-1)}"))for j in range(d)]] tree(11)

[night quarry]

@arctic skiff :white_check_mark: Your 3.11 eval job has completed with return code 0.

001 |           *
002 |          ***
003 |         *****
004 |        *******
005 |       *********
006 |      ***********
007 |     *************
008 |    ***************
009 |   *****************
010 |  *******************
011 | *********************
... (truncated - too many lines)

Full output: https://paste.pythondiscord.com/cilatijubo.txt?noredirect

[arctic skiff]

Its just 11 lines and still says too many lines

Guess it also takes \n

[last locust]

That max is set to 10, which is why you get the message, but the truncation logic is wrong and allows 11

It's a bug

bot#2492

[glass drum]

GitHub

[bot] #2492 Eval output allows too many lines

[tough willow]

92 -> 74

!e

tree=lambda d:[print(('*'*i).center(d*2))for i in range(1,d*2,2)]
tree(11)

[night quarry]

@tough willow :white_check_mark: Your 3.11 eval job has completed with return code 0.

001 |           *           
002 |          ***          
003 |         *****         
004 |        *******        
005 |       *********       
006 |      ***********      
007 |     *************     
008 |    ***************    
009 |   *****************   
010 |  *******************  
011 | ********************* 
... (truncated - too many lines)

Full output: https://paste.pythondiscord.com/ivaqiyalud.txt?noredirect

[wheat river]

tree=lambda d:[print(f"{'*'*i:^{d*2}}")for i in range(1,d*2,2)]

[quartz wave]

63 -> 62 ```py
def tree(d):
for i in range(1,d2,2):print(f"{''i:^{d2}}")

62 -> 59 ```py
def tree(d,i=1):
while i<d2:print(f"{''i:^{d2}}");i+=2

[quartz wave]

!e 74 -> 41 ```py
i=1
while i<22:print(f"{'*'*i:^22}");i+=2

[night quarry]

@quartz wave :white_check_mark: Your 3.11 eval job has completed with return code 0.

001 |           *           
002 |          ***          
003 |         *****         
004 |        *******        
005 |       *********       
006 |      ***********      
007 |     *************     
008 |    ***************    
009 |   *****************   
010 |  *******************  
011 | ********************* 
... (truncated - too many lines)

Full output: https://paste.pythondiscord.com/livaxudoba.txt?noredirect

[quartz wave]

59 -> 58 ```py
def tree(d,i=1):exec("print(f"{'*'i:^{d2}}");i+=2;"*d)

[arctic skiff]

Single line only(no use of statements and no semi colons)

[quartz wave]

since when was that a rule

[arctic skiff]

Since now

[quartz wave]

60 ```py
def tree(d,i=1):exec("print(f"{'*'i:^{d2}}")\ni+=2\n"*d)

[midnight jasper]

:0 soo cool

[arctic skiff]

Ig I should change expected indent to no use of statements

Sopy tree=lambda d,i=1:exec("print(f\"{'*'*i:^{d*2}}\")\ni+=2\n"*d) 

[quartz wave]

ok i'm making my own set of rules

least character amount wins

[gleaming linden]

*bytes

Degen solutions are dumb

[quartz wave]

ok yes

[wheat river]

def tree(d,i=1):exec("print(f\"{'*'*i:^{d*2}}\");i+=2\n"*d)

59

[quartz wave]

i already have a 58

[wheat river]

ah, ok 👍

[kindred stone]

just slapped some random things together and got tired while writing this, so i didnt completed the code

!e

print((classmethod(type(slice.__mro__[0]).__mro__[1].__doc__).__class__.__name__[8].upper()).__add__(....__repr__()[::-1][-1].lower()).__add__(type(int.from_bytes(b"\xFF\xFF", byteorder="little").__eq__(int(str("190326753248076643792647923649875".__add__("69593707450387")[::-1][-1]))-1)).__name__[-1].__mul__(memoryview(b"69").nbytes)).__add__(chr(memoryview(bytes(int.from_bytes(b"\0x", byteorder="big") & int.from_bytes(b"\0x", byteorder="big"))).nbytes - 9)).__add__(str(['x+=a,;a,b=b,a+b;', '*x']).split("'").__getitem__(int("101", 2).__mul__(int("10", 8).__sub__(int("3F", 16))).__mod__(int("11", 2)).__add__(int("101", 2)).__sub__(4))))

[night quarry]

@kindred stone :white_check_mark: Your 3.11 eval job has completed with return code 0.

Hello, 

[eager ore]

Hello,

[unique heath]

golfing challenge: invert boolean (true -> false, false -> true)

lambda b:b-1if b else b+1```

!e

print((lambda b:b-1if b else b+1)(True))

[night quarry]

@unique heath :white_check_mark: Your 3.11 eval job has completed with return code 0.

001 | /home/main.py:1: SyntaxWarning: invalid decimal literal
002 |   print((lambda b:b-1if b else b+1)(True))
003 | 0

[unique heath]

!e

print((lambda b:b-1if b else b+1)(False))

[night quarry]

@unique heath :white_check_mark: Your 3.11 eval job has completed with return code 0.

001 | /home/main.py:1: SyntaxWarning: invalid decimal literal
002 |   print((lambda b:b-1if b else b+1)(False))
003 | 1

[unique heath]

it works lfmao

im dumb lambda b:not b exists

[fleet bridge]

f = lambda b:b-1if b else b+1
f = lambda b:0if b else 1
f = lambda b:not b
f = lambda b:1-b
f = 1 .__sub__
f = 0 .__pow__
import operator as o
f = o.not_

[languid hare]

!e

NOT = 1 .__sub__

print(NOT(True))
print(NOT(False))

[night quarry]

@languid hare :white_check_mark: Your 3.11 eval job has completed with return code 0.

001 | 0
002 | 1

[languid hare]

aw a little slower than denball

[fleet bridge]

f = lambda b:b-1if b else b+1
f = lambda b:0if b else 1

[low lynx]

1-b

[unique heath]

how thfeuf kc dopes that wokr

[arctic skiff]

!epy print((lambda _:False if _ else True)(True))

[night quarry]

@arctic skiff :white_check_mark: Your 3.11 eval job has completed with return code 0.

False

[arctic skiff]

!epy print(not True)

[night quarry]

@arctic skiff :white_check_mark: Your 3.11 eval job has completed with return code 0.

False

[unique heath]

lambda b:0if b else 1 where

[arctic skiff]

It went on vacation

[unique heath]

lambda b: bool(0if b else 1) shorter

[arctic skiff]

Though you can do not b

[arctic skiff]

This is shortest

[arctic skiff]

!epy print((lambda _:~_)(True))

[night quarry]

@arctic skiff :white_check_mark: Your 3.11 eval job has completed with return code 0.

-2

[unique heath]

thats still True

[arctic skiff]

Yeah i see

Though isnt ~ bitwise NOT

Ahh i see why it doesn't work

[arctic skiff]

!e```py
print((lambda :-1if _ else~)(True))
print((lambda :-1if _ else~)(False))

[night quarry]

@arctic skiff :white_check_mark: Your 3.11 eval job has completed with return code 0.

001 | /home/main.py:1: SyntaxWarning: invalid decimal literal
002 |   print((lambda _:_-1if _ else~_)(True))
003 | /home/main.py:2: SyntaxWarning: invalid decimal literal
004 |   print((lambda _:_-1if _ else~_)(False))
005 | 0
006 | -1

[languid hare]

how didnt i think of

f = 0 .__eq__

first

its shorter and returns a boolean

!e

f = 0 .__eq__

print(f(True))
print(f(False))

[night quarry]

@languid hare :white_check_mark: Your 3.11 eval job has completed with return code 0.

001 | False
002 | True

[versed eagle]

2's complement

it flips the sign bit

[low lynx]

~ is one's complement

• is two's complement

[rough moat]

while("False")

[sick hound]

this is cursed

[rough moat]

yeah

it runs tho lmao

[sick hound]

yeah got it

[rough moat]

it's because an empty string acts like false and a well, not empty string acts like true

[sick hound]

yeah I know

imagine showing a beginner how while loops work with this

[gentle plover]

(1).__sub__

[versed eagle]

1 .__sub__ was already said

0 .__eq__ was also mentioned iirc

[dusty zodiac]

a,y=map(int,open(0))
print(sum(366 if i%4<1 and(i%100>0 or i%400<1)else 365 for i in range(y-a,y))+1)

anyone know a better way to shorten this further?

[low lynx]

removing spaces and brackets firstly

[vague cairn]

a,y=map(int,open(0)) doesn't work for me, ValueError: invalid literal for int() with base 10: '\n'

[night quarry]

:incoming_envelope: :ok_hand: applied timeout to @queen jetty until <t:1681625403:f> (10 minutes) (reason: duplicates spam - sent 4 duplicate messages).

The <@&831776746206265384> have been alerted for review.

[next flame]

you need to pass in input without a newline

I think piping in with echo should work

[restive void]

Only if you add -n, IIRC

[wheat river]

a,y=map(int,open(0))
print(sum(365+(i%4<1and i%100>0or i%400<1)for i in range(y-a,y))+1)

[dusty zodiac]

the else 366 is necessary

oh wait

yeah that works

ty

[wheat river]

a,y=map(int,open(0))
print(sum(365+(i%16*(i%25<1)<(i%4<1))for i in range(y-a,y))+1)

credits: https://codegolf.stackexchange.com/questions/50798/is-it-a-leap-year

[quartz wave]

i think this works ```py
a,y=map(int,open(0))
print(sum(366-(i%16>i%25<1)for i in range(y-a,y))+1)

maybe not

[quiet rose]

/help

!e

[night quarry]

Missing required argument

code

[languid hare]

that >< comparison chain is one beauty

[quartz wave]

i don't think it works

[languid hare]

[quartz wave]

it's designed for the case that i is always a leap year

[urban ledge]

if i obfuscate my code with pyarmor, the other ppl that executes the code needs pyarmor aswell?

[serene stratus]

it will be included in a folder called pytransform

[leaden valley]

using fishhook how can i do this ```py
a = "Hello World!"
print(a) # Hi World!

[rugged sparrow]

You would be better off hooking globals

I sent some code to do it a bit above

[leaden valley]

i cant find it

[rugged sparrow]

#internals-and-peps message

[leaden valley]

thanks that'll work

its actually perfect

[meager zinc]

reimplementing std library

def isinstance(a, b):
    if type(b) == tuple:
        return any(c in type.mro(a.__class__) for c in b)
    else:
        return b in type.mro(a.__class__)

[sick hound]

a=type(int|str)
def i(i):
 if isinstance(i,a):return i.__args__
 if not isinstance(i,tuple):raise TypeError("isinstance() arg 2 must be a type, a tuple of types, or a union")
 i,u=[i],set()
 while i:
  for e in i.pop():isinstance(e,a)and(e:=e.__args__),i.append(e)if isinstance(e,tuple)else u.add(e)
 return u
def u(i,u):
 try:return getattr(i,"__instancecheck__")(u)
 except:0
def isinstance(e,o):return any(a in type(e).__mro__ or u(a,e)for a in([o]if type in type(o).__mro__ else i(o)))

[meager zinc]

thamks

is_even is a great module btw

[sick hound]

lol thanks

[fleet bridge]

I dont think this is correct
Types are compared by id, not by ==, so there might be a problem with metaclasses
And you are not taking into account __instancecheck__

[misty parcel]

как быстро выучить python? помогите, господа

[fleet bridge]

Wrong channel. Ask in #python-discussion

[misty parcel]

ok, sorry

[fleet bridge]

This is really hard to understand but it is very cool!

[tough willow]

!rule 4

[night quarry]

Rules

4. Use English to the best of your ability. Be polite if someone speaks English imperfectly.

[grave rover]

ok how the hell does this new impl work

[rugged sparrow]

ooh that one is a fun one. there is no verification that the tuple used by partial to store args is the same length as when the function started inside of the partial.__repr__. basically like this: py def __repr__(self): args_len = len(self.args) for i in range(args_len): add_to_str(repr(self.args[i]))

so WeirdRepr replaces the tuple with a specific pair of items (a tuple and an array's backing right after each other)

and then partial tries to call tp_repr on the items located in that array backing because the new tuple is shorter

so you can make a fake instance of Fake with a custom address in its slots and use that to fake an arbitrary object

*and get a reference to it

[grave rover]

This is incredibly cursed

I love it

[rugged sparrow]

its probably the most contrived exploit i have written

[grave rover]

Meanwhile I might have to use your getmem impl at work

[rugged sparrow]

oh god

[grave rover]

My supervisor asked me for a task that python doesn't really support out of the box
But using memhax I could accomplish it :)
Unfortunately most systems run windows so no /proc/self/memory for me to use

[rugged sparrow]

im really curious about what the task is tbh

and also i would recommend writing a C extension over using memhacks. that would be more stable

[grave rover]

Unfortunately company policy is that we cannot compile any native tools ourselves nor use native tools that aren't whitelisted

[rugged sparrow]

ah

*write the assembly yourself /s

[grave rover]

Ah yes, writing x86, my favorite

Honestly I can't wait for CPUs to switch to ARM (they won't, but I pray every night they do)

[rugged sparrow]

also if you actually need to get the memory of the entire process, I would use ctypes instead of any of my bugs. ```py
from ctypes import c_ubyte
import sys

mem = (c_ubyte * sys.maxsize).from_address(0)

that would be more stable then any bug would be

*stable is relative with memhacks tho

[grave rover]

I'm not sure how much of this is NDA but it involves being able to call pandas methods on regular python objects, e.g. list.map(...)

[rugged sparrow]

i mean fishhook would work for that but you said you can't non-whitelisted software

[grave rover]

Yup

Forbiddenfruit is also blacklisted unfortunately

[rugged sparrow]

makes sense tbh

[grave rover]

Oh actually forbiddenfruit has a funny way to bypass mappingproxy

[rugged sparrow]

does it use the __req__ trick?

[grave rover]

def patchable_builtin(klass):
    refs = gc.get_referents(klass.__dict__)
    return refs[0]``` this returns the dict backing the class' mappingproxy

[rugged sparrow]

oh yea i used to use that trick in fishhook v1

but __req__ didn't require gc

[grave rover]

req?

[rugged sparrow]

just normal methods not any dunders?

[grave rover]

A few dunders sadly which is gonna be a little tricky but not impossible

But what's this req thing you're talking about 👀

[rugged sparrow]

!e ```py
def getdict(cls, E=type('',(),{'eq':lambda s,o:o})()):
'''
Obtains a writeable dictionary of a classes namespace
Note that any modifications to this dictionary should be followed by a
call to PyType_Modified(cls)
'''
return cls.dict == E

print(getdict(int))```

[night quarry]

@rugged sparrow :white_check_mark: Your 3.11 eval job has completed with return code 0.

{'__new__': <built-in method __new__ of type object at 0x7f8c9fc579a0>, '__repr__': <slot wrapper '__repr__' of 'int' objects>, '__hash__': <slot wrapper '__hash__' of 'int' objects>, '__getattribute__': <slot wrapper '__getattribute__' of 'int' objects>, '__lt__': <slot wrapper '__lt__' of 'int' objects>, '__le__': <slot wrapper '__le__' of 'int' objects>, '__eq__': <slot wrapper '__eq__' of 'int' objects>, '__ne__': <slot wrapper '__ne__' of 'int' objects>, '__gt__': <slot wrapper '__gt__' of 'int' objects>, '__ge__': <slot wrapper '__ge__' of 'int' objects>, '__add__': <slot wrapper '__add__' of 'int' objects>, '__radd__': <slot wrapper '__radd__' of 'int' objects>, '__sub__': <slot wrapper '__sub__' of 'int' objects>, '__rsub__': <slot wrapper '__rsub__' of 'int' objects>, '__mul__': <slot wrapper '__mul__' of 'int' objects>, '__rmul__': <slot wrapper '__rmul__' of 'int' objects>, '__mod__': <slot wrapper '__mod__' of 'int' objects>, '__rmod__': <slot wrapper '__rmod__' of 'int' ob
... (truncated - too long)

Full output: https://paste.pythondiscord.com/ahasajetix.txt?noredirect

[rugged sparrow]

mappingproxy.__req__ dispatches to dict.__req__ on the wrapped dictionary

[grave rover]

Lmao

[rugged sparrow]

and dict.__req__ calls arg.__eq__(mp)

[grave rover]

This is so stupid, I love it

Thank you python for truthy values

[rugged sparrow]

if you want i can send you some sample code which is basically smaller fishhook with only attr hooking (no orig or anything else)

[grave rover]

Honestly at this point I want to just drop my work and look through the interpreter to see just how many ways we have to bypass mappingproxy and/or access raw memory

[rugged sparrow]

there are a lot, i have a few really neat ones that work on <=3.11

involves the GC being called in odd places

[grave rover]

What I'm curious about

Are there any that work on other impls than cpython

[rugged sparrow]

not that I have found, but thats because I only really look at cpython

other impls would have different bugs in different places probably

[grave rover]

I wonder if pypy allows for the extended_argument exploit

[rugged sparrow]

no idea

i can look

*but pypy does not implement id the same way so you would need to get an info leak as well

[grave rover]

dang

[rugged sparrow]

it does not, it has partial's implemented differently

[grave rover]

Oh I was talking about the 256 bytecode one from a while back

[rugged sparrow]

ohh the custom code object one

[rugged sparrow]

you can crash it

but its a different exploit process to get it stable

[grave rover]

Lol

I see

[fleet bridge]

wait, __req__ exists? 🧐

[rugged sparrow]

*it does not technically

[rugged sparrow]

but __eq__ still does reverse dispatch

[leaden valley]

could you mess with python to make the syntax like javascripts?

etc. True is true
{} for code blocks

and are there any tutorials or books to learn more about esoteric python and code golfing

im very interested in hooking

[sick hound]

ig for true you can do

true = True

But for others you need ctypes or cython

[leaden valley]

well yeah that's obvious

[next flame]

probably doable with a custom codec

[grave rover]

Custom codec, custom import hook for preprocessing, patching compiler at runtime

Just some ways you could probably do it

[leaden valley]

what about this?

[sick hound]

related https://gist.github.com/dzshn/b8de78f1d6b83e1f75d19e77a0ccd7cb

[rugged sparrow]

you could implement that via fishhook to avoid the Illegal hardware Instruction issue. And you would not have to deal with ctypes directly

[quartz wave]

no Illegal hardware instruction for me which is good
this is hell ```py

[] == []
False

[sick hound]

how wtf

[quartz wave]

nvm i tried to execute some other code and got an illegal instruction error

[leaden valley]

can you modify the code of functions from modules you imported

[quartz wave]

depends on the module

[leaden valley]

for example

colorama

if u wanted green u would do colorama.Fore.GREEN

but

nvm thats a bad example

i just want to change the code of functions in a module

[restive void]

If it's a module implemented in Python, then yes.

[leaden valley]

its a third party library

[restive void]

That doesn't matter, it just matters whether it's Python or something else (like an extension module written in C, for example)

[leaden valley]

ah i misread i thought you said implemented by python

[grave rover]

https://pypi.org/project/pymixin

[leaden valley]

thanks that looks pretty good

[leaden valley]

i get this error when importing: py from asm import Deserializer, LOAD_CONST, CALL_FUNCTION, POP_TOP, Serializer, RETURN_VALUE, LOAD_FAST ModuleNotFoundError: No module named 'asm'

fixed it but now i get this error.. py ALL_OPS = {v: eval(k) for k, v in opmap.items()} ^^^^^^^ File "<string>", line 1, in <module> NameError: name 'CACHE' is not defined

[vast wave]

which python version

[leaden valley]

py 3.11.2

[vast wave]

hm

[grave rover]

you don't have pyasm installed which it depends on
which is strange considering it's a required dependency

[leaden valley]

for some reason when i installed pymixin, no other modules were installed

[grave rover]

3.11+ isn't fully supported yet, try 3.10

[leaden valley]

aw that sucks i dont have 3.10

ill install 3.10 later

[rugged sparrow]

if you forbid __ in input, can you still eval arbitrary code? py inp = input('> ') if '__' in inp:exit('fail') else:eval(inp, {'__builtins__':{}}) I have been trying some stuff, wondering if i am missing something simple

[fleet bridge]

eval(chr(...)+chr(...)+...)

[rugged sparrow]

no builtins ^

[fleet bridge]

your eval will eval my eval

ah

[rugged sparrow]

{'__builtins__':{}} this makes it so python doesn't add in its builtins

so you can't access __subclasses__ on something to get to a useful class

[fleet bridge]

i can do something stupid like 10**10**10**10

[rugged sparrow]

i mean yea thats fair

but im trying to find code exec not just resource usage

[fleet bridge]

there is some package that can execute code safely
it basically forbids dunders and some builtins

[rugged sparrow]

what package?

[fleet bridge]

i dont remember

https://github.com/yaroslaff/evalidate

[rugged sparrow]

ah that evals ast

im more curious about bypasses for just blocking __ and all builtins

Ik there are some exotic ones

but wondering if there are any simple ones

[fleet bridge]

no, it is not the package i am thinking about
i cant find it, but i remember that it is evaluating string using eval with no builtins after checking that code has no suspicious expressions (such as dunders)

[rugged sparrow]

ill keep looking for it. can you think of any bypasses?

[fleet bridge]

no, this package has many stars on github and i dont think there were any vulnerabilities
so blocking __ and builtins is enough

[rugged sparrow]

if you can find that project can you send me a link?

[fleet bridge]

without dunders and builtins you can get only builtin objects (int, strings, lists, ...) and call theirs methods
there is no way to get reference to any class, because you cannot do type(...) and .__class__

funny way to get bytes object that contains numbers from 0 to 255: b''.maketrans(b'',b'')

!e assert b''.maketrans(b'',b'') == bytes(range(256))

[night quarry]

@fleet bridge :warning: Your 3.11 eval job has completed with return code 0.

[No output]

[rugged sparrow]

clever

[fleet bridge]

i think this is it:
https://github.com/zopefoundation/RestrictedPython
https://restrictedpython.readthedocs.io/en/latest/

[rugged sparrow]

awesome thanks

[fleet bridge]

I think it is possible to replace all literals (strings, lists, ints, ...) with our new fully compatible reimplementation of these classes, but with additional restriction on length of these objects. So 10**10**10**10 or [1]*10**10 doesn't explode but raises some exception.

You can replace literals by ast modification.
Maybe you also should wrap all calls, getitems and getattrs into wrapper that returns "restricted" objects instead of raw ints/lists. You also should disallow shadowing your wrapper function.
You should make sure that there is absolutely no way to get raw unrestricted objects.

[sick hound]

where's the fun in that smh

[vast wave]

probably not, but its still not a good idea to let things like that get into your locals

even if you delete __builtins__, your other vars are still available in that context

better to just entirely nuke python out and implement a pure math expr parser

[finite blaze]

Hey, I wonder what are you working on?

[dusty zodiac]

x={"FIRE":"WATER","WATER":"EARTH","EARTH":"FIRE","FRIEND":"HEAL","FOE":"ATTACK"}
for i in range(int(input())):t,e=input().split();print([f"{x[t]} {x[e]}",f"{x[t]} {e}"][t[1]=='R'])

[rugged sparrow]

Ik, the goal is to build a challenge where the goal is to break out

[rugged sparrow]

but if you pass globals={'__builtins__': {}} you cannot access the upper scope vars directly

[vast wave]

oh right, it overwrites

[vast wave]

hesitant to say impossible, but i don't see a way to do it

you have nothing to help you escape, just standard python syntax

no functions

[rugged sparrow]

jail.py

import builtins
# remove all the stuff we don't need
for key in vars(builtins).copy():
    if key not in {'input', 'print', 'Exception', 'repr', 'eval'}:
        del builtins.__dict__[key]

del builtins
while True:
    try:
        inp = input('>>> ')
        for blocked in {'__', ':', ' ', '\t', '\n'}:
            if blocked in inp:
                raise Exception('Invalid Input')
        else:
            print(repr(eval(inp, {'__builtins__': {}})))
    except Exception as e:
        print('Error', e)
``` this is the current jail

it is possible to break btw @vast wave

just tricky

[vast wave]

how

[rugged sparrow]

@fleet bridge @quartz wave challenge for yall ^

[vast wave]

oh wait

i have an idea

[rugged sparrow]

what's your idea?

[vast wave]

lemme test it

[rugged sparrow]

send it with spoiler tags when you do

[vast wave]

nvm

doesnt work, the other solution would be invalid

yeah no i dont see a way to do it

what's the solution

[rugged sparrow]

i'm not gonna say until more time has passed

@versed eagle @dry mirage you might also want to try this challenge ^

[fleet bridge]

it just crashed my process (maybe because of a lot of my custom REPL stuff)
i will try that challenge without this: ```py
for key in vars(builtins).copy():
if key not in {'input', 'print', 'Exception', 'repr', 'eval'}:
del builtins.dict[key]

[rugged sparrow]

you shouldnt run it in a repl. just python3 jail.py. It calls input to get user input

[fleet bridge]

yeah, in this case it works 👍

[rugged sparrow]

yea the builtins clearing messes with a lot of repl stuff

but it also makes the challenge more difficult

[fleet bridge]

Same code with some convenient features (better errors and ingoring empty lines): ```py
import builtins

remove all the stuff we don't need

for key in vars(builtins).copy():
if key not in {'input', 'print', 'Exception', 'repr', 'eval'}:
del builtins.dict[key]
del builtins
while True:
try:
inp = input('>>> ')
if not inp:
continue
for blocked in {'__', ':', ' ', '\t', '\n'}:
if blocked in inp:
raise Exception(f'Invalid input: {blocked!r}')
else:
print(repr(eval(inp, {'builtins': {}})))
except Exception as e:
print('Error:', e)

[rugged sparrow]

Nice, ill add that to mine

I can also host this somewhere so that it can be accessed via netcat

i wonder if replit would let me

[fleet bridge]

i wonder if it is even possible to create any dict except for empty one

comprehensions can be done like this: [(x)for(x)in[1,2,3]]

no statements are allowed (because of no and :)

wait, it is eval, so anyway i cannot use statements

no walrus, no lambdas, no dunders

empty set: {1}&{2}

some dicts can be made like this: ```py

''.maketrans('13','24')
{49: 50, 51: 52}

another way to create dicts: {}.fromkeys({1,2,3})

>>> {}.keys().mapping
mappingproxy({})

[rugged sparrow]

if you store the dict somehow you can add and remove items with update and pop/remove

[fleet bridge]

arbitrary dicts:```py

[(d.update([('a','b')]),d)[-1]for(d)in[{}]][0]
{'a': 'b'}

list comprehension is used for assignment

string with dunder: '_''_init_''_'

>>> type('',(),[(d.update([('_''_init_''_',[].append)]),d)[-1]for(d)in[{}]][0])()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: list.append() takes exactly one argument (0 given)
>>> type('',(),[(d.update([('_''_init_''_','{}'.format)]),d)[-1]for(d)in[{}]][0])()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
IndexError: Replacement index 0 out of range for positional args tuple
>>> type('',(),[(d.update([('_''_init_''_',''.format)]),d)[-1]for(d)in[{}]][0])()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: __init__() should return None, not 'str'
``` bound functions are not descriptors, so i dont know how to capture self (if we figure out how to get `type`)

[rugged sparrow]

like i said, its tricky

[fleet bridge]

this is hard, i like it

TIL int.from_bytes accepts any Iterable[int]: ```py

0 .from_bytes([1,2,3,4])
16_909_060

[rugged sparrow]

thats interesting i didn't know that

[fleet bridge]

i figured out how to get some code object and some frame object

[rugged sparrow]

how?

[fleet bridge]

generators

[rugged sparrow]

noice

[fleet bridge]

so i think i can construct almost any code object

AttributeError: attribute 'f_code' of 'frame' objects is not writable
😭

i also can do iter(x)

[rugged sparrow]

did you get a reference to iter? or using a listcomp?

[fleet bridge]

>>> g=((0)for(_)in[0])
>>> g.gi_frame.f_locals
{'.0': <tuple_iterator object at 0x000001875C267EB0>}
>>> g.gi_frame.f_locals['.0']
<tuple_iterator object at 0x000001875C267EB0>

[versed eagle]

oo okey

whats the goal

[rugged sparrow]

get arbitary code exec from the limited eval

[versed eagle]

okey

[fleet bridge]

i have access to globals() and to globals()['__builtins__'] so i can add arbitrary vars to global and builtin scopes

[rugged sparrow]

@fleet bridge when you solve it make sure to spoiler

[fleet bridge]

it is ok that i am posting here some thoughts?

[languid hare]

ooh is there a challenge somewhere

[rugged sparrow]

yea thats all good

[fleet bridge]

there

[rugged sparrow]

#esoteric-python message @languid hare

run it as python3 jail.py, the goal is to get full code exec from the limited eval

[languid hare]

ah

[fleet bridge]

how to create new global var: ```py

[()for(d)in[(()for()in()).gi_frame.f_globals]for(d['var'])in['value']],var
([()], 'value')

[leaden valley]

you're a genius

im definitely stealing that

[vast wave]

.replace()

[rugged sparrow]

that would give you a new var just in that one line since globals are a new dict on every run

[fleet bridge]

yes, but you can use it instead of walrus to make arbitrary complex expression with assignments

[rugged sparrow]

true, it does allow for more complex expressions

[rugged sparrow]

replace returns a new code object so you would have to find a way to assign it somewhere

[fleet bridge]

any of these things will lead to breaking out of jail: getattr, type, <global frame>, 1 allowed dunder

maybe reference to any python function, but im not sure

[rugged sparrow]

type alone would allow escaping from jail?

[fleet bridge]

hmm

you can create any type...

[fleet bridge]

like this

[rugged sparrow]

you don't have dunders tho so you can't get to useful ones like FrozenImporter

[fleet bridge]

ah, you can get frametype, generatortype and construct new generator with arbitrary code

[rugged sparrow]

and you can't make methods

[earnest wing]

gonna try this hold on

[rugged sparrow]

!e py type(()for()in())()

[night quarry]

@rugged sparrow :x: Your 3.11 eval job has completed with return code 1.

001 | Traceback (most recent call last):
002 |   File "/home/main.py", line 1, in <module>
003 |     type(()for()in())()
004 | TypeError: cannot create 'generator' instances

[earnest wing]

which version

[rugged sparrow]

3.11 works

#esoteric-python message for this challenge

[leaden valley]

have you bypassed it? @rugged sparrow

[rugged sparrow]

yes

[fleet bridge]

reference to almost any module would also be very useful

[rugged sparrow]

it is possible to solve it

I am also working on a harder one now 🙂

[leaden valley]

could i know it secretly

[rugged sparrow]

nope, you gotta solve it

[leaden valley]

it was worth a shot

[earnest wing]

just to be clear, is the goal builtin retrieval

[fleet bridge]

i think the goal is to do os.system('...') or eval('...') or exec('...')

[earnest wing]

kay

[rugged sparrow]

yea that is the goal

my POC does os.system

actually I would say that the goal is just os.system or something similar

[fleet bridge]

How to get any char if you know its ord():

>>> ord(':')
58
>>> '\0'.translate([(d.update([(0,58)]),d)[-1]for(d)in[{}]][0])
':'

wait, it can be done in a lot easier way: ```py

'\x3a'
':'

[rugged sparrow]

that will help once you have eval in order to get imports working again

[fleet bridge]

i'm giving up, i'll try it tomorrow

[rugged sparrow]

@earnest wing lmk how far you get

[sudden osprey]

well I think I know a way of getting dunder methods

yeah, should be pretty doable from there

[rugged sparrow]

What's your strategy?

[sudden osprey]

||_＿dunder＿_|| ||second character is not an underscore||

[rugged sparrow]

oh that is not an intended solution

wtf

but fair enough

(those will be blocked in the next one tho)

[lunar marsh]

classic pyjail cheese lmao

[sudden osprey]

haha, did sort of feel like cheating :P

[rugged sparrow]

or it doesnt work?

odd

[rugged sparrow]

||this doesn't work (but should, so im looking into that now)||

[leaden valley]

you can do 1 .__dunder__ to save a char

[earnest wing]

nope

read the challenge

no spaces

[leaden valley]

ah

[rugged sparrow]

that trick also does not work after the builtin clearing

||i assume because it tries to import some encoding corrector (but that fails due to lost __import__)||

[sudden osprey]

Ah interesting

[leaden valley]

that makes sense

[rugged sparrow]

yea it attempts to import unicodedata

so the cheese does not work thats good

but ill explicitly block that on the next one

[meager zinc]

I think I have a solution

but I gtg and I haven't written the full thing out

[rugged sparrow]

@sudden osprey thanks tho, the next one will have this: inp = normalize('NFKD', input('>>> '))

to avoid the odd error, this is the new challenge code

import builtins
from unicodedata import normalize
# remove all the stuff we don't need
for key in vars(builtins).copy():
    if key not in {'input', 'print', 'Exception', 'repr', 'eval'}:
        del builtins.__dict__[key]

del builtins
while True:
    try:
        inp = normalize('NFKD', input('>>> '))
        if not inp:
            continue
        for blocked in {'__', ':', ' ', '\t', '\n'}:
            if blocked in inp:
                raise Exception(f'Invalid Input {blocked!r}')
        else:
            print(repr(eval(inp, {'__builtins__': {}})))
    except Exception as e:
        print('Error', e)
``` @sudden osprey

[earnest wing]

i got distracted

[rugged sparrow]

fair enough

[earnest wing]

but not far, mostly toyed around with ||generator frames / code objects||

as well as the funny ||.0 local||

[low lynx]

[rugged sparrow]

|| youre close there ||

[leaden valley]

im trying to think of a way to bypass " "

in my string

starting to think its impossible

[rugged sparrow]

i guarantee it is not impossible

[lunar marsh]

i got rce

[rugged sparrow]

send me your impl, Im curious about your exploit path

you can dm it or just spoilertag it here

[gentle plover]

"\x20"

[leaden valley]

yeah i got it

||my attempt was to use load_module but of course that uses __import__ so didnt work||

[gentle plover]

WAIT
I FOUND SOMETHING

||"<{a.b} encoded using \x-stuff>".format(a)||

[wide crest]

what is this channel for?

[rugged sparrow]

python weirdness

[wide crest]

hm

yea i have something i guess

Ive shown this before in #python-discussion and in #1035199133436354600 but im wondering if the people who frequent here can help because i cant get W H and all the other letters

its only using class conversion so

i cant use and functions

[quartz wave]

what's "break" btw

[rugged sparrow]

like call os.system or something like that

get out of the python jail

[quartz wave]

ok

i figured out how to get type but it's stuck in a string

[rugged sparrow]

out of curiousity, how would type help you if you got it out of the string?

[quartz wave]

i forgot

[wide crest]

https://tenor.com/view/kirby-i-forgot-i-forgor-gif-22449575

[rugged owl]

[restive void]

||do type({}) to get dict, then you can make a class and assign it "arbitrary" members (except it wouldn't help you create functions either..)||

[rugged sparrow]

yea i am making another jail where you can make classes off the bat (i provide a function for it) but it doesn't make it any easier (hopefully, this new one is supposed to be more difficult)

[restive void]

Okay, I figured out how to set global variables. Not sure if that helps me yet :D

[fleet bridge]

can you share your way for setting globals?

[fleet bridge]

this is my

[fleet bridge]

||```py

'{x.class.base.subclasses}'.format(x=())
<built-in method subclasses of type object at 0x00007FFED0B78DF0>

[rugged sparrow]

probably custom terminal stuff?

[night quarry]

@half escarp :x: Your 3.11 eval job has completed with return code 143 (SIGTERM).

001 | Error name 'getattr' is not defined
002 | Error name 'getattr' is not defined
003 | Error name 'getattr' is not defined
004 | Error name 'getattr' is not defined
005 | Error name 'getattr' is not defined
006 | Error name 'getattr' is not defined
007 | Error name 'getattr' is not defined
008 | Error name 'getattr' is not defined
009 | Error name 'getattr' is not defined
010 | Error name 'getattr' is not defined
011 | Error name 'getattr' is not defined
... (truncated - too many lines)

Full output: too long to upload

[rugged sparrow]

probably not

ill check it

its an exception raised by the module repr

[rugged sparrow]

Try to spoiler tag solutions

there ya go

that solution uses the intended strategy of || using generator frames to step out to the main frame||

[quartz wave]

||.f_back||?

is it supposed to be ran on a file to make it work?

this is a very version specific solution

doesn't work in 3.11

||had to adjust offset by 1 backwards||

[rugged sparrow]

Yea you run the code as jail.py

[quartz wave]

i put it in the REPL

[rugged sparrow]

Ah yea that won't quite work

I mean it's still escapable but the repl makes it a degree easier

[quartz wave]

how?

[rugged sparrow]

Certain stuff is initialized upon entering the repl that puts a lot more things into local scope

[quartz wave]

aren't all those wiped out by the 2nd argument to eval()

||no-offset|| way ||```py
[*([x.append(((x[0].gi_frame.f_back.f_back.f_globals)for()in((),)))or(x[0])for(x)in[[]]][0])][0]["\x5f\x5fbuiltins\x5f\x5f"].eval('((x.{0}call{0}.{0}globals{0})for(x)in"".{0}class{0}.{0}bases{0}[0].{0}subclasses{0}()if"{0}globals{0}"in(x).{0}call{0}.{0}dir{0}()).{0}next{0}()["sys"].modules["os"]'.format('\x5f\x5f')).system

[rugged sparrow]

||Part of the challenge is getting os.system after breaking out of the eval||

The one I am working on right now will definitely not work in the repl and should be way more difficult then this one

[quartz wave]

eh idk

[restive void]

Sorry, I meant builtins.
||.setdefault with the globals from the frame of a generator comprehension||

[fleet bridge]

my code also can set builtins

[vast wave]

what the fuck

i was gone for one day

[leaden valley]

is there a way to obfuscate function kwarg names

func(OBFUSCATED_NAME="Hi")
``` OBFUSCATED_NAME will be interpreted as another name inside func

thats my goal

[low lynx]

just pass in **{}

[leaden valley]

worked thanks

[rugged sparrow]

yea people here work fast

[leaden valley]

could you call __import__ by string using generator frames, like in here with builtins (i could probably use this method and getattr but there could be a better way)

it doesnt need to be using generator frames

[kindred stone]

# .env
SENSITIVE="sensitive data" #\n
MORE="more data"

read_dotenv=lambda k: [v[1]for l in open(".env", "r")if (v:=l.replace("\n", "").split("="))[0]==k]or None

assert read_dotenv("SENSITIVE") == "sensitive data"
assert read_dotenv("MORE") == "more data"
assert read_dotenv("NON_EXISTENT_KEY") is None

looking to shorten this

[rugged sparrow]

__import__ is removed from builtins so you would have to find a frame that stored it in globals

[leaden valley]

hm; got it 🙂

[last locust]

assert all((read_dotenv(x)==y for x,y in zip(("SENSITIVE","MORE","NON_EXISTENT_KEY"),("sensitive data","more data",None)))```

Might actually be shorter without the zip, not sure

Not strictly the same, but would give the same result (one assert instead of 3, and does ==None instead of None)

[kindred stone]

these asserts where there just to tell what are the expected results of the function, i was looking to the actual implementation of read_dotenv being shorter

[last locust]

Well you've got a space after the if that can be removed to begin with

You also don't need the , "r" in the open() since r mode is the default

[arctic elm]

your code fails your asserts :/

because of comments

[kindred stone]

oh bruh

need to ignore #

[arctic elm]

:^)

exec(open('.env').read())                                                   read_dotenv=lambda x:globals().get(x,None)

[kindred stone]

sure

[arctic elm]

actually

exec(open('.env').read())                                                   read_dotenv=globals().get

not in the spirit of what you want, but passes your tests

[kindred stone]

none of my tests passes on my own function🤦‍♂️

i wrote them right before sending and didnt tested them, mb

only the last one that checks for None works 😆

[kindred stone]

this works pretty well, nice catch

[arctic elm]

"works" is debatable

it happens to do what you tested

[kindred stone]

yeah

if you dont set the variables with quotes it wont work

but its nice for what was proposed

[arctic elm]

just don't use it in any real code...

[kindred stone]

obviously

[leaden valley]

is hexadecimal good enough string obfuscation for scanners

[tough willow]

What Scanners

[leaden valley]

there are no specific ones, im just trying to avoid regex patterns by obfuscating strings (in this case im obfuscating "__builtins__")

yet i have no clue what is good enough

[tough willow]

Why do you want to do that

[leaden valley]

to avoid regex patterns

im learning evasion techniques

[tough willow]

Evasion for what

[leaden valley]

...

just in general

[tough willow]

I don't know what you are talking about, sorry

[leaden valley]

idk the proper term for it

[unique heath]

im bored

def print(*args, sep=" ", end="\n", file=None, flush=False):
    if not file:
        file = open(1, "w")

    file.write(sep.join(map(str, args))
    file.write(end)
    
    if flush:
        file.flush()


def input(prompt=''):
    return open(0).readlines(1).strip("\n")

[versed eagle]

now implement builtins.open

and whatever _io classes it uses

[fleet bridge]

implement type and object

[versed eagle]

from posix import * is your friend

[fleet bridge]

I guess that wouldn't work on windows

🧐

[rugged sparrow]

!e @fleet bridge researching building functions without functions as I build another challenge: || py print([ [ SumArgs.append((1,2,3,4))or(SumCall(None)), SumArgs.append((5,6,7,8))or(SumCall(None)), ] for(SumArgs)in[[]] for(SumCall)in[( [ temp.insert(0,temp.pop()+arg) for(arg)in(args) ]and(temp[0]) for(args)in(SumArgs) for(temp)in[[0]] ).send] ]) ||

[night quarry]

@rugged sparrow :white_check_mark: Your 3.11 eval job has completed with return code 0.

[[10, 26]]

[versed eagle]

iirc there's implementations of it for windows (wrappers on whatever library windows uses)

or a vm

[rugged sparrow]

had to delete and resend to run !e

[versed eagle]

✨

[fleet bridge]

I hate that. My brain melts

[rugged sparrow]

abusing generators for fun and profit 🙂

for the next one you will have to build functional classes using that strategy

[fleet bridge]

Now i understand how your code does what it does. It is very tricky

[rugged sparrow]

Right now I'm trying to research how to build a wrapper so that arguments and execution can happen at the same time like a regular function

[fleet bridge]

SumArgs is changing size, and at the same time you are iterating over it. Why it is not causing any exceptions?
Or changing container size is forbidden only for dicts?

[rugged sparrow]

Changing the container size is only forbidden on types where it's dangerous

[fleet bridge]

So, dicts + sets + frozensets?

[rugged sparrow]

Yes

Although there's a trick where you can actually change the values of a set while iterating over it

But you run into implementation quirks of the underlying Hash table

Like you can only swap out eight items

I'll dig around and see if I can find the code

[leaden valley]

can someone help with this

[quartz wave]

are you obfuscating the string or the thing itself

[leaden valley]

string

[versed eagle]

whats the regex you're trying to avoid matching

pretty hard to avoid matching a regex if you don't know what it is

[fleet bridge]

Docs says that objects equality should imply hashes equality. What if our object is returning different hashes?
I can put two objects that returns hashes 1 and 2, then do 🪄 and then their hashes are now 2 and 1 (instead of 1 and 2). And if i try using one object as key i will get value corresponding to other key.

I think y'all can make a lot of fun with that

[leaden valley]

i dont have the regex but i know that it checks for dunders, hexadecimal string and getattr

i have a way to get __import__ but i cant access the attributes because getattr is blocked

and the attribute itself is blocked

[quartz wave]

use "\x5f\x5fbuiltins\x5f\x5f"?

[versed eagle]

vars(obj)[attrstr]

[leaden valley]

thats hexadecimal

[fleet bridge]

You can scroll up a bit and find code that can execute arbitrary code and has no dunders or getattr

I posted a small code that constructs one char string from its ord

You can combine these ideas

[leaden valley]

i'll have a look and thanks @versed eagle that will work i think

[leaden valley]

does it also work for subprocess?

i need Popen

[versed eagle]

you probably don't

[fleet bridge]

You can do os.system(py -c 'import subprocess') or something

[versed eagle]

from posix import * is your friend
or nt if you're on windows ig

actually why not just import builtins

[leaden valley]

ill try other ways first

doing that seems lazy imo

octal is, not sure about binary

doesnt octal use hex?

[fleet bridge]

Wdym by "hex"?

[leaden valley]

actually nvm

[rough moat]

hexadecimal

[spiral merlin]

how do I code a bonzi buddy

but one I can set to any picture I want

you know

like a desktop stripper

[vast wave]

octal is base 8, hex is 16

[unique heath]

esopy challenge:

golf an accurate version of print

def print(*a,s='',e='\n',p=0,o=0):(f:=open(p,'w'),f.write(s.join(map(str,a))+e),f.flush()if o else ...)
``` 1liner

102 chars

[quartz wave]

103*

s should be ' '

[low lynx]

isn't that 104

nor is that accurate since the kwargs are different

[quartz wave]

yes

[unique heath]

fine

def print(*a,sep='',end='\n',file=0,flush=0):(f:=open(file,'w'),f.write(sep.join(map(str,a))+end),f.flush()if flush else ...)

[arctic skiff]

Why ```py
sep.join(map(str,a))

Cant we just do
```py
sep.join(a)```

[unique heath]

!e

print(''.join([1,2,3,4]))```

[night quarry]

@unique heath :x: Your 3.11 eval job has completed with return code 1.

001 | Traceback (most recent call last):
002 |   File "/home/main.py", line 1, in <module>
003 |     print(''.join([1,2,3,4]))
004 |           ^^^^^^^^^^^^^^^^^^
005 | TypeError: sequence item 0: expected str instance, int found

[arctic skiff]

Ok

[quartz wave]

!e ```py
def print(*a,sep='',end='\n',file=0,flush=0):(f:=open(file,'w'),f.write(sep.join(map(str,a))+end),f.flush()if flush else ...)
print(5, 6, 7)

[night quarry]

@quartz wave :warning: Your 3.11 eval job has completed with return code 0.

[No output]

[quartz wave]

wth

!e ```py
def print(*a,sep='',end='\n',file=0,flush=0):(f:=open(file,'w'),f.write(sep.join(map(str,a))+end),f.flush()if flush else ...)
print(5, 6, 7, flush=1)

[night quarry]

@quartz wave :warning: Your 3.11 eval job has completed with return code 0.

[No output]

[quartz wave]

ok it doesn't work in the sandbox

[arctic skiff]

!epy import sys sys.stdout.write('blah')

[night quarry]

@arctic skiff :white_check_mark: Your 3.11 eval job has completed with return code 0.

blah

[arctic skiff]

!epy print(open(0))

[night quarry]

@arctic skiff :white_check_mark: Your 3.11 eval job has completed with return code 0.

<_io.TextIOWrapper name=0 mode='r' encoding='utf-8'>

[quartz wave]

oh they have the fd wrong

[quartz wave]

corrected (not golfed) ```py
def print(*a,sep=' ',end='\n',file=1,flush=0):(f:=open(file,'w'),f.write(sep.join(map(str,a))+end),f.flush()if flush else ...)

[arctic skiff]

1 and 0? Where do they lead?

[quartz wave]

golfed ```py
def print(*a,sep=' ',end='\n',file=1,flush=0):(f:=open(file,'w')).write(sep.join(map(str,a))+end),flush and f.flush()

[quartz wave]

0 - stdin
1 - stdout
2 - stderr

[arctic skiff]

I thought 0 leads to stdout

[arctic skiff]

Though you forgot to close tuple

[quartz wave]

what tuple

if i add a parenthesis at the end then i forgot to start a tuple

[arctic skiff]

Nevermind got it

[low lynx]

can't you do f=open(...);f.write(...)

[quartz wave]

yes

[rugged sparrow]

Arguably using 1 for the file would not be correct. Default print always uses whatever file is in sys.stdout

[sudden osprey]

Also, the file argument is meant to take a file object, not a path

[fleet bridge]

Why there is list[::] but there is no dict[::]?

Also why slices are unhashable?

[proper vault]

I don't think anyone really thought too hard about hashing slices, especially since you can just use a tuple and seq[slice(*tup)]

[restive void]

my_dict["key_a":"key_b"] would make a copy with just the items from key a to b (in insertion order)?

[rugged sparrow]

!e you can make slices with mutable items in them: print(slice([],[],[]))

[night quarry]

@rugged sparrow :white_check_mark: Your 3.11 eval job has completed with return code 0.

slice([], [], [])

[astral rover]

wouldnt it make sense though if the underlying items are hashable to have them be hashable much like MappingProxy

[fleet bridge]

You also can make tuples with mutable items, but tuples are hashable if all items are hashable. Slices are always unhashable

There was some discussion about slice hashability in #internals-and-peps

[rugged sparrow]

fair enough

[rugged sparrow]

!e @fleet bridge figured out that multiple loop comprehensions can modify prior variables so Sum is neater py print([ [ SumCall(SumArgs.append((1,2,3,4))), SumCall(SumArgs.append((5,6,7,8))) ] for(SumArgs)in[[]] for(SumCall)in[( [ (sum) for(sum)in[0] for(arg)in(args) for(sum)in[sum+arg] ][-1] for(args)in(SumArgs) ).send] ])

[night quarry]

@rugged sparrow :white_check_mark: Your 3.11 eval job has completed with return code 0.

[[10, 26]]

[earnest wing]

specifically to prevent dict slicing I would assume, since that can be a nasty footgun

a slice is just an object so it should have no special treatment there, but also there may be an implicit assumption that slicing "works" on dicts

this can create silently incorrect, innocuous seeming programs which is not super nice

[rough moat]

how would dict slicing even work lmao

like ‘i want everything from ‘banana’ to 13.12’

[rugged sparrow]

since dicts are ordered now you could theoretically slice like that consistently, but still it would be unintuitive

[rough moat]

yeah i guess but you could simply slice through the keys or values tbh

[rugged sparrow]

yea it would not be a smart feature

[rough moat]

now imagine set slicing lmao

[fleet bridge]

Imagine integer slicing

And slice slicing

[rugged sparrow]

!e ```py
from fishhook import hook

@hook(int)
def getitem(self, idx):
return int(str(self)[idx])

print(12345[2:4])```

[night quarry]

@rugged sparrow :white_check_mark: Your 3.11 eval job has completed with return code 0.

001 | /home/main.py:7: SyntaxWarning: 'int' object is not subscriptable; perhaps you missed a comma?
002 |   print(12345[2:4])
003 | 34

[versed eagle]

it might be more useful to check what representation the int used

since presumably if the user does 0b01010100[::-1] (84), they want 0b00101010 (24) as a result, rather than 48

[sick hound]

!e ```py
from ctypes import CFUNCTYPE as y,cast as x,c_int as n,c_ssize_t as e,c_void_p as k,py_object as b;a=k16;z=a.from_address;i=z(z(id(list))[13]);u=x(i[5],y(n,b,e,b));e=x(i[3],y(b,b,e));i=a();z(id(int))[14]=x(i,k);i[:3]=0,x(y(b,b,b)(lambdaa:e(*a[::~0])),k),x(y(n,b,b,b)(lambda a,i,e:u(i,a,e)),k)

arr = [2, 3, 4, 5]
1[arr] = 2[arr]

print(f"{arr = }")
print(f"{0[arr] = }")

[night quarry]

@sick hound :white_check_mark: Your 3.11 eval job has completed with return code 0.

001 | /home/main.py:4: SyntaxWarning: 'int' object is not subscriptable; perhaps you missed a comma?
002 |   1[arr] = 2[arr]
003 | /home/main.py:7: SyntaxWarning: 'int' object is not subscriptable; perhaps you missed a comma?
004 |   print(f"{0[arr] = }")
005 | arr = [2, 4, 4, 5]
006 | 0[arr] = 2

[sick hound]

real C

[dry mirage]

yeah honestly seems to be no other reason than they didn't want slices in dicts, but there are other ways to implement that in dict, I don't see why the implementation is on slice

[rugged sparrow]

@fleet bridge if you want another challenge, I submitted one to hack the box: https://app.hackthebox.com/challenges/fake-snake

@dry mirage ^ you also might like this one

[dry mirage]

thanks defender 😔

[meager zinc]

what

huh I didn't know windows flagged hack tools

[arctic skiff]

It seriously flagged nc?

[meager zinc]

here's a different style of problem with a pretty obvious solution if you know the trick

while 1:
    try:
        for c in(i:=input('>>> ')):
            if ord(c)<256:raise Exception(f'Invalid Input {i!r}')
        print(repr(eval(i)))
    except Exception as e:print('Error: ', e)

(get arbritrary os.system)

[languid hare]

neat

[fleet bridge]

Can you post challenge description here please?

[tepid matrix]

how do you get builtins using "__builtins__" string?

generator frames arent allowed

[arctic skiff]

!epy print(globals()['__builtins__'])

[night quarry]

@arctic skiff :white_check_mark: Your 3.11 eval job has completed with return code 0.

<module 'builtins' (built-in)>

[tepid matrix]

i shouldve mentioned this too but globals is also not allowed

[arctic skiff]

Ok

!epy print(locals()['__builtins__'])

[night quarry]

@arctic skiff :white_check_mark: Your 3.11 eval job has completed with return code 0.

<module 'builtins' (built-in)>

[tepid matrix]

that too

[tough willow]

eval('__builtins__')?

[tepid matrix]

eval/exec is blocked

theres a lot of things ive tried

[arctic skiff]

You are randomly creating rules

[tepid matrix]

?

[tough willow]

say every rule

you didn't say

• locals
• globals
• eval/exec

[tepid matrix]

those were pretty obvious

thats why i didnt say

[tough willow]

no, they weren't

[tepid matrix]

u think eval isnt obvious?

[tough willow]

yes

[tepid matrix]

okay.

[tough willow]

i think i found a solution

!e

import random
print(
  getattr(random, '__builtins__').keys()
)

[night quarry]

@tough willow :white_check_mark: Your 3.11 eval job has completed with return code 0.

{'__name__': 'builtins', '__doc__': "Built-in functions, exceptions, and other objects.\n\nNoteworthy: None is the `nil' object; Ellipsis represents `...' in slices.", '__package__': '', '__loader__': <class '_frozen_importlib.BuiltinImporter'>, '__spec__': ModuleSpec(name='builtins', loader=<class '_frozen_importlib.BuiltinImporter'>, origin='built-in'), '__build_class__': <built-in function __build_class__>, '__import__': <built-in function __import__>, 'abs': <built-in function abs>, 'all': <built-in function all>, 'any': <built-in function any>, 'ascii': <built-in function ascii>, 'bin': <built-in function bin>, 'breakpoint': <built-in function breakpoint>, 'callable': <built-in function callable>, 'chr': <built-in function chr>, 'compile': <built-in function compile>, 'delattr': <built-in function delattr>, 'dir': <built-in function dir>, 'divmod': <built-in function divmod>, 'eval': <built-in function eval>, 'exec': <built-in function exec>, 'format': <built-in function format>, 
... (truncated - too long)

Full output: https://paste.pythondiscord.com/nipajoduva.txt?noredirect

[tough willow]

!e

import random
print(
  getattr(random, '__builtins__').keys()
)

[night quarry]

@tough willow :white_check_mark: Your 3.11 eval job has completed with return code 0.

dict_keys(['__name__', '__doc__', '__package__', '__loader__', '__spec__', '__build_class__', '__import__', 'abs', 'all', 'any', 'ascii', 'bin', 'breakpoint', 'callable', 'chr', 'compile', 'delattr', 'dir', 'divmod', 'eval', 'exec', 'format', 'getattr', 'globals', 'hasattr', 'hash', 'hex', 'id', 'input', 'isinstance', 'issubclass', 'iter', 'aiter', 'len', 'locals', 'max', 'min', 'next', 'anext', 'oct', 'ord', 'pow', 'print', 'repr', 'round', 'setattr', 'sorted', 'sum', 'vars', 'None', 'Ellipsis', 'NotImplemented', 'False', 'True', 'bool', 'memoryview', 'bytearray', 'bytes', 'classmethod', 'complex', 'dict', 'enumerate', 'filter', 'float', 'frozenset', 'property', 'int', 'list', 'map', 'object', 'range', 'reversed', 'set', 'slice', 'staticmethod', 'str', 'super', 'tuple', 'type', 'zip', '__debug__', 'BaseException', 'BaseExceptionGroup', 'Exception', 'GeneratorExit', 'KeyboardInterrupt', 'SystemExit', 'ArithmeticError', 'AssertionError', 'AttributeError', 'BufferError', 'EOFError', 'Imp
... (truncated - too long)

Full output: https://paste.pythondiscord.com/apixeroteh.txt?noredirect

[arctic skiff]

!epy print(__import__('__hello__').__getattribute__('__builtins__'))

[night quarry]

@arctic skiff :white_check_mark: Your 3.11 eval job has completed with return code 0.

{'__name__': 'builtins', '__doc__': "Built-in functions, exceptions, and other objects.\n\nNoteworthy: None is the `nil' object; Ellipsis represents `...' in slices.", '__package__': '', '__loader__': <class '_frozen_importlib.BuiltinImporter'>, '__spec__': ModuleSpec(name='builtins', loader=<class '_frozen_importlib.BuiltinImporter'>, origin='built-in'), '__build_class__': <built-in function __build_class__>, '__import__': <built-in function __import__>, 'abs': <built-in function abs>, 'all': <built-in function all>, 'any': <built-in function any>, 'ascii': <built-in function ascii>, 'bin': <built-in function bin>, 'breakpoint': <built-in function breakpoint>, 'callable': <built-in function callable>, 'chr': <built-in function chr>, 'compile': <built-in function compile>, 'delattr': <built-in function delattr>, 'dir': <built-in function dir>, 'divmod': <built-in function divmod>, 'eval': <built-in function eval>, 'exec': <built-in function exec>, 'format': <built-in function format>, 
... (truncated - too long)

Full output: https://paste.pythondiscord.com/yeruwehuhi.txt?noredirect

[tough willow]

works too

@tepid matrix

using getattr on a module

[arctic skiff]

Incoming text from @tepid matrix

No getattr

[tough willow]

true

[tepid matrix]

getattr isnt blocked

thanks

[tough willow]

np

should work on any module that doesn't overwrite _builtins_ and doesnt have __all__ i think

oh

[arctic skiff]

!epy print(__file__)

[night quarry]

@arctic skiff :white_check_mark: Your 3.11 eval job has completed with return code 0.

/home/main.py

[arctic skiff]

!epy import os with open(__file__) as f: a=f.read() if not os.path.isfile('/home/_'): with open('/home/_', 'x'):... exec(a) print('hi')

[night quarry]

@arctic skiff :white_check_mark: Your 3.11 eval job has completed with return code 0.

001 | hi
002 | hi

[rugged sparrow]

I can't because of their tos for submitted challenges

[sour panther]

#game-development

[low lynx]

!e @unique heath

import sys
def print(*a,sep=" ",end="\n",file=sys.stdout,flush=0):file.write(sep.join(map(str,a))+end);flush and file.flush()

print(123, 'abc', print, sep='sep')
with open('test.txt', 'w') as file:
    print(123, 456, sep='abc', end='a', file=file)

[night quarry]

@low lynx :white_check_mark: Your 3.11 eval job has completed with return code 0.

123sepabcsep<function print at 0x7f7089e844a0>

test.txt

123abc456a

[unique heath]

!e

print(len('''import sys
def print(*a,sep=" ",end="\n",file=sys.stdout,flush=0):file.write(sep.join(map(str,a))+end);flush and file.flush()'''))````

[night quarry]

@unique heath :white_check_mark: Your 3.11 eval job has completed with return code 0.

124

[unique heath]

huh

[tepid matrix]

strange

im getting 125

[winter basin]

Is there anyone who wants to donate me the nitro

?

[tepid matrix]

speak in #python-discussion

[languid hare]

should be 125, the \n is being unescaped into 1, but the source would have it as 2

[fleet bridge]

yeah, it should be "\\n" in this code that counts length of other code

[low lynx]

in actual code it would just be the one character wouldn't it?

[fleet bridge]

yes, but in source code it is two characters long

[low lynx]

in golfing we've always counted newlines as one character haven't we?

[rugged sparrow]

I typically wrap the code in a raw string so that it doesn't perform escapes

[fleet bridge]

it is not a newline, it is a string literal which contains newline symbol
so it requires two chars in code: "\n"

[quartz wave]

that is why use r when using len()

[serene garnet]

hmmm

can't i use walrus operator like

[a,b,c]:=[1,2,3]

[sick hound]

you cant

[languid hare]

mhm

not allowed

!e

([a, b, c] := [1, 2, 3])

[night quarry]

@languid hare :x: Your 3.11 eval job has completed with return code 1.

001 |   File "/home/main.py", line 1
002 |     ([a, b, c] := [1, 2, 3])
003 |      ^^^^^^^^^
004 | SyntaxError: cannot use assignment expressions with list

[serene garnet]

i have to golf my code

but it fails

l=lambda:sorted(map(int,input().split()));a,b,c=d=l()
print('YNEOS'[d!=l()or a*a+b*b!=c*c::2])

it's a code that inputs each side length of two triangles and determine if it is possible to produce two triangles of given side length, by cutting a rectangle.

[fleet bridge]

i think this: d!=l() will always be True
because d is an int and l() is a list of ints

oh wait

it is a,b,c=d=l() not a,b,c,d=l()

[fleet bridge]

are there any plans to remove these restrictions from walrus?
i think it can work as regular assignment, but it also returns whatever left right side evaluates to

[languid hare]

https://mail.python.org/pipermail/python-dev/2018-July/154628.html

In the future, if such
use-cases are found, the grammar can be expanded.

personally im not too sure why they seemed to be against iterable unpacking in the first place

in the example they gave the issue to me seems to be more of an ambiguity in the grammar, whether
x, y := 3, 4 should be a 3-tuple that also assigns y, or should it be a 2-tuple which performs a destructuring assignment

of course that goes away when you use parenthesis like a sane person
(x, y) := (3, 4)

[next flame]

works for me

>>> l=lambda:sorted(map(int,input().split()));a,b,c=d=l();print('YNEOS'[d!=l()or a*a+b*b!=c*c::2])
3 4 5
3 4 5
YES
>>> l=lambda:sorted(map(int,input().split()));a,b,c=d=l();print('YNEOS'[d!=l()or a*a+b*b!=c*c::2])
1 2 3
1 2 3
NO

[rugged sparrow]

!e really weird way of running code on interpreter exit: ```py
def onexit(func, cache=[]):
def gen():
try:yield
except:
func()
raise
g = gen()
g.send(None)
cache.append(g)

@onexit
def _():
print('this ran when the process exited')

print('I ran first')

[night quarry]

@rugged sparrow :white_check_mark: Your 3.11 eval job has completed with return code 0.

001 | I ran first
002 | this ran when the process exited

[quartz wave]

!e ```py
def gen(f):
try:yield
except:f();raise
def onexit(f,c=[]):g=gen(f);g.send(c.append(g))

onexit(lambda:print(globals()))
print('a')

[night quarry]

@quartz wave :white_check_mark: Your 3.11 eval job has completed with return code 0.

001 | a
002 | {'__name__': '__main__', '__doc__': None, '__package__': None, '__loader__': <_frozen_importlib_external.SourceFileLoader object at 0x7f880d33f750>, '__spec__': None, '__annotations__': {}, '__builtins__': <module 'builtins'>, 'gen': <function gen at 0x7f880d2e44a0>, 'onexit': <function onexit at 0x7f880d2e4680>}

[quartz wave]

nice

[rugged sparrow]

ye

it works because tp_finalize for generators calls gen_close, which in turn sets PyExc_GeneratorExit as the active exception and then calls gen_send. so if you have a bare try catch you catch the GeneratorExit which lets you run code as the generator is freed

(just have to make sure to re-raise the GeneratorExit otherwise you get a RuntimeError)

[tepid matrix]

do u know how this works cuz all im seeing is a bunch of gibberish

[tepid matrix]

this explanation doesnt make sense to my little brain

[quartz wave]

it uses like details a normal python programmer isn't supposed to know

[tepid matrix]

ah

i need to learn more esopy

[sick hound]

!e print("Hello!")

[tough willow]

#bot-commands @sick hound

[next flame]

whats the g.send(c.append(g)) for? preventing gc?

[quartz wave]

append g to c and send None to g

[low lynx]

why do you need to send something to g?

[rugged sparrow]

It's to prevent gc, the generator completes when it is collected

[low lynx]

!e

def onexit(func, cache=[]):
 def gen():
   try:yield
   except:
     func()
     raise
 g = gen()
 cache.append(g)

@onexit
def _():
 print('this ran when the process exited')

print('I ran first')

[night quarry]

@low lynx :white_check_mark: Your 3.11 eval job has completed with return code 0.

I ran first

[low lynx]

huh

[rugged sparrow]

The generator has to be in the middle of executing. That's what the .send is for

If a generator is never started, it doesn't have to be closed

*technically this is not quite true, gen_close is still called, but if the generator is not currently inside a try block it cannot catch the GeneratorExit exception, and if it has never been used (by .send or otherwise) then it cannot be inside a try block

[rugged sparrow]

gen returns a generator when called, and you can sort of imagine that to be a function that can be paused. Calling .send executes everything until the first yield. Then the generator is stored in the cache list. When the generator is garage collected (at the end of the program or if the cache is cleared), the interpreter raises a GeneratorExit exception inside of it at the current point of execution. Because it is sitting at the first yield (inside a try/except block) that GeneratorExit exception is caught, func is called, and then the GeneratorExit is re-raised by the argument-less raise.

[dry mirage]

doesn't that still get picked up by the cyclic GC?

[rugged sparrow]

As long as cache is held by onexit it won't get picked up. But on interpreter exit it gets cleared out

[low lynx]

hmm

!e

def onexit(func, cache=[]):
 def gen():
   try:yield
   except:
     func()
     raise
 g = gen()
 next(g)
 cache.append(g)

@onexit
def _():
 print('this ran when the process exited')

print('I ran first')

[night quarry]

@low lynx :white_check_mark: Your 3.11 eval job has completed with return code 0.

001 | I ran first
002 | this ran when the process exited

[sick hound]

sys._exit

[astral rover]

I thought it was os._exit

[sick hound]

oh it seems to be

[tepid matrix]

that actually makes a bit of sense

[rugged sparrow]

glad i could explain it a bit, feel free to ping me if you have any other questions

[tepid matrix]

i have a question, where the fuck did you learn this shit

i try to find books on this topic and i cant

[rugged sparrow]

i spent a lot of time reading the cpython source code and looking for edge cases

also lots of trial and error

[tepid matrix]

ah

it would help if i knew c

[rugged sparrow]

yea, C is def a good language to know for messing with the low level stuff like this

[gentle plover]

Python really is fun ```py
def onexit(func, /, *, _cache=[]):
cache.append(type("", (), {"del": lambda s, *, **__: func()})())

[rugged sparrow]

yea that will work too, i wrote mine because I wanted to see if i could get one working without any builtins

[gentle plover]

replace type with [].__class__.__class__

[rugged sparrow]

*builtins or dunders

[gentle plover]

ah

[sly ibex]

hey guys am i wrong or there is a way using f string to avoid math.ceil

[rugged sparrow]

!e you could just print(1.3.__ceil__())

[night quarry]

@rugged sparrow :white_check_mark: Your 3.11 eval job has completed with return code 0.

2

[sly ibex]

nooo haa

but

i was saying something.... shorter

using the f string thereisn't a way?

[rugged sparrow]

fstrings round to nearest

[sly ibex]

f{number:d}

something like that

oh

nearest

!e
print(int(f'{1.8:.0f}'))

[night quarry]

@sly ibex :white_check_mark: Your 3.11 eval job has completed with return code 0.

2

[rugged sparrow]

!e py print(f'{1.3 = :.0f}') print(f'{1.8 = :.0f}')

[night quarry]

@rugged sparrow :white_check_mark: Your 3.11 eval job has completed with return code 0.

001 | 1.3 = 1
002 | 1.8 = 2

[sly ibex]

!e
import math
print(math.ceil(1.3))
print(math.ceil(1.8))

[night quarry]

@sly ibex :white_check_mark: Your 3.11 eval job has completed with return code 0.

001 | 2
002 | 2

[sly ibex]

ohhh

@rugged sparrow

watch this way
int(a+.99)

[fleet bridge]

yeah this method works (if you somehow got 1-1/inf)

it also works with modular arithmetics
for example, you want to round 7 up modulo 3. You can do (7+(3-1))//3*3 and get 9

[rugged sparrow]

!e that will only work if your decimal is only 2 points: py print(int(1.001+0.99))

[night quarry]

@rugged sparrow :white_check_mark: Your 3.11 eval job has completed with return code 0.

1

[fleet bridge]

idea of 3-1 is the same as 1 - eps

what is the shortest way to get smallest positive value?

>>> import sys;eps=sys.float_info.epsilon
>>> ceil = lambda x: int(x + 1 - eps)
>>> ceil(1.0)
1
>>> ceil(1.1)
2
>>> ceil(1.00000001)
2
>>> ceil(1.000000000001)
2
>>> ceil(1.00000000000000001)
1

idk if this will work with big values

[earnest wing]

float_info.epsilon is slightly different semantically but should accidentally give the correct result for numbers close to 1

math.nextafter is closest to what you want I think

[meager zinc]

!e ```py
import sys;eps=sys.float_info.epsilon
print(f"{1e-128:.128f}")
print(f"{eps:.128f}")
print(1e-128 > 0)
print(eps > 0)

[night quarry]

@meager zinc :white_check_mark: Your 3.11 eval job has completed with return code 0.

001 | 0.00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000001
002 | 0.00000000000000022204460492503130808472633361816406250000000000000000000000000000000000000000000000000000000000000000000000000000
003 | True
004 | True

[meager zinc]

!e noting this:

for i in range(500,1,-1):
    if (v := eval(f"1e-{i}")) != 0:
        print(v, v > 0)
        break

[night quarry]

@meager zinc :white_check_mark: Your 3.11 eval job has completed with return code 0.

1e-323 True

[vast wave]

!e pro tip: ```py
print(10**-10)

[night quarry]

@vast wave :white_check_mark: Your 3.11 eval job has completed with return code 0.

1e-10

[unique heath]

!e dropping this here to help the obfooscators

print(__builtins__.eval(__builtins__.__dir__().__getitem__(__import__.__dir__().__len__().__sub__(______:=(___:=(_:=__name__.__getitem__((__:=__name__.__class__().__len__()))).__add__(_).__len__()).__mul__((____:=___.__mul__(___))).__add__((_____:=(_:=__name__.__getitem__(__)).__add__(_).__add__(_).__len__()))))))```

[night quarry]

@unique heath :white_check_mark: Your 3.11 eval job has completed with return code 0.

<built-in function eval>

[low lynx]

ah yes

first thing to do when getting eval

is to get eval from builtins

[vast wave]

!e golfed ```py
print(builtins.eval(builtins.dir()[len(dir(import))-((:=len((:=name[(:=len(name.class()))])+))***2+len(name[]++))]))

[night quarry]

@vast wave :white_check_mark: Your 3.11 eval job has completed with return code 0.

<built-in function eval>

[unique heath]

!e golfed

print(eval)```

[night quarry]

@unique heath :white_check_mark: Your 3.11 eval job has completed with return code 0.

<built-in function eval>

[vast wave]

most epic golf of all time

[quartz wave]

#python-discussion message i think ```py
for p in[eval('[f"{1950+%d*10}s",%%d,[]]'%i%x)for i,x in enumerate("5x5NExB9GEpydZwCtGa2BP 1kjfmZ8JGrDA5YrBVVZLRx 0eMhi5F148eRenCW2C48nL 4DgEvihGmE3r5wUHUS7Wwv 2eh6kA7m2vUSalEBnKs5yC 1GOVol2EIX1sPDjspcHOtG 5RI8VCGELpG38SCVZBiGs0 4L6k3WNfixzwFI92ba0dqr".split())]:
R=requests.get(f"https://api.spotify.com/v1/playlists/{p[1]}/tracks?limit=50",headers={"Authorization":f"Bearer {access_token}"});print("Playlist: "+p[0])
for t in R.json()["items"]:s=t["track"];a=s["album"];z=a["release_date"];s["name"]in p[2]and int(z.split("-")[0])<2020and a["release_date_precision"]in{"year","month","day"}and print(f"{p[0]}: {s['name']} ({z})")

[sick hound]

i am so glad i don't see code like this everydya

[sick hound]

that's some nice code wym

[rugged owl]

beautiful

[jolly sky]

wtf is "code gore"

[sick hound]

no clue myself

[rugged owl]

Basically just terrible looking code in simple terms

[sick hound]

lol

[rugged owl]

[sick hound]

I wonder if there's code porn

like food porn

when the food looks extremely good

is there one for codd

code

[eager ore]

what 😭

[kindred raft]

It's how I felt about the code I cleaned up earlier

[sharp glacier]

Good functional python

[pearl yarrow]

Pretty sure there's code porn somewhere

[cloud fossil]

Don't go look for it, it might be unsafe code

[rugged owl]

[cloud fossil]

Bad practices everywhere

Man why do so many professional unadvised technical terms apply for this

[sick hound]

just curious

[sick hound]

real

[wheat river]

x64 asm

properly aligned

[versed eagle]

the source for GNU Hello

[sharp glacier]

I wonder if there used to be a proper fp channel on this cord ? 🤔

[sick hound]

lol probably

[sick hound]

as that's goo

good

[keen thicket]

Hello my esoterism enthusiasts, I am happy to announce that for the first time ever, the International Obfuscated Python Code Competition is accepting submissions. This contest is similar to the IOCCC (International Obfuscated C Code Contest), and is intended to reward creative and perverse code style. A short bio about the judges and more details about the logistics of the competition will be released on April 30th, with submissions ending August 30th, 2023. Answers to FAQs as well as more information about the rules can be found on the website: pyobfusc.com

[sharp kestrel]

this is an approved post
https://pyobfusc.com

[versed eagle]

✨

can that get pinned

so it doesn't get buried

[keen thicket]

@sharp kestrel if you could pin it that would be great--this channel seems to have a decent amount of activity

[karmic pumice]

def J():A='Safe at last';B='Run away';C='\n';D='';E='doo-doo';F=['Baby','Mommy','Daddy','Grandma','Grandpa',"Let's go hunt",B,A,"It's the end"];G={B:'(ah!)',A:'(phew)'};H='Shark';I='';D=[f"{3*f'{A} {H if B<=4 else I}, {E}, {E}{C}'}{A} {H if B<=4 else G[A]if(A:=F[B])in G.keys()else I}{C*2}"for B in range(len(F))];return D;print(J())

[viscid latch]

What?

[finite blaze]

first time here?

[earnest wing]

Amazing! I am looking forward to stressing the judges

[keen thicket]

If you can make a judge quit judging the competition on the spot because of your submission you get ultimate notoriety

[keen thicket]

this doesn't produce any output when I run it?

is this the baby shark song lyrics generator?

[karmic pumice]

I guess this is a small problem either on the bot side or on my side. For me it works, but the bot doesnt run it, maybe because the call to the functions is on the same line as declaration 🤷
editing it to

def J():A='Safe at last';B='Run away';C='\n';D='';E='doo-doo';F=['Baby','Mommy','Daddy','Grandma','Grandpa',"Let's go hunt",B,A,"It's the end"];G={B:'(ah!)',A:'(phew)'};H='Shark';I='';D=[f"{3*f'{A} {H if B<=4 else I}, {E}, {E}{C}'}{A} {H if B<=4 else G[A]if(A:=F[B])in G.keys()else I}{C*2}"for B in range(len(F))];return D;
print(J())

actually makes it work on the bot

[keen thicket]

Oh yep! putting it on the new line makes it work for me in repl.it

[karmic pumice]

well its weird af for me now, i'm trying to run it by myself but it generates the wrong thing for some reason lmao

havent used python in a long time and just came back for some simple scripting so decided to do some funny esoteric

[keen thicket]

yeah I feel you haha

I think you just need to strip some newlines

or use ''.join()

[karmic pumice]

well the thing is before it actually printed the correct lines, now even when i backup to older versions it does.. this

[keen thicket]

I'm getting more than that

hmm

[karmic pumice]

oh, actually this was caused by the fact that cls doesnt really fully clear the terminal lmao

[keen thicket]

oh yeah lol that's an old footgun

[karmic pumice]

def a():
    b = "\n"
    c = ""
    d = "doo-doo"
    e = ["Baby", "Mommy", "Daddy", "Grandma", "Grandpa", "Let's go hunt", "Run away", "Safe at last", "It's the end"]
    f = {"Run away": "(ah!)", "Safe at last": "(phew)"}
    i = "Shark"
    j = ""
    c = [f"{3 * f'{e[g]} {i if g <= 4 else j}, {d}, {d}{b}'}{e[g]} {i if g <= 4 else (f[e[g]] if (h:=e[g]) in f.keys() else j)}{b*2}" for g in range(len(e))]
    return "".join(c)

print(a())

so thats the more ,,readable" version

[keen thicket]

you can also change the "in f.keys()" to just "in f"

also if you wanted to change all those assignments to walrus operators you could use a lambda and save a few chars overall I think haha

[karmic pumice]

!e

def A():I='Safe at last';H='Run away';C='\n';A='';D='doo-doo';B=['Baby','Mommy','Daddy','Grandma','Grandpa',"Let's go hunt",H,I,"It's the end"];E={H:'(ah!)',I:'(phew)'};F='Shark';G='';A=[f"{3*f'{B[A]} {F if A<=4 else G}, {D}, {D}{C}'}{B[A]} {F if A<=4 else E[B[A]]if(H:=B[A])in E else G}{C*2}"for A in range(len(B))];return ''.join(A)
print(A())

[night quarry]

@karmic pumice :white_check_mark: Your 3.11 eval job has completed with return code 0.

001 | Baby Shark, doo-doo, doo-doo
002 | Baby Shark, doo-doo, doo-doo
003 | Baby Shark, doo-doo, doo-doo
004 | Baby Shark
005 | 
006 | Mommy Shark, doo-doo, doo-doo
007 | Mommy Shark, doo-doo, doo-doo
008 | Mommy Shark, doo-doo, doo-doo
009 | Mommy Shark
010 | 
011 | Daddy Shark, doo-doo, doo-doo
... (truncated - too many lines)

Full output: https://paste.pythondiscord.com/bipojalupo.txt?noredirect

[karmic pumice]

fixed :d

[keen thicket]

hell yea

[karmic pumice]

fun way to get back into python lmao

[keen thicket]

yeah python is really fun for stuff like this

https://gist.github.com/juliusgeo/661cb84d04d906ca9f08b8769b85da5e

https://gist.github.com/juliusgeo/1da759d07af0b447a78d0ccb14162c57

https://gist.github.com/juliusgeo/1da759d07af0b447a78d0ccb14162c57

how do u get rid of those damn massive previews

[karmic pumice]

woah, these ones always impress me

[karmic pumice]

<link>

nah, not like html

just https://gist.github.com/juliusgeo/661cb84d04d906ca9f08b8769b85da5e

[keen thicket]

wait it works tho

hahah

it got rid of the preview

[karmic pumice]

well this is supposed to be the correct way

[keen thicket]

OH

hahaha

I was confused

also this is my fav one I've made https://gist.github.com/juliusgeo/1da759d07af0b447a78d0ccb14162c57

[karmic pumice]

yeah it looks impressive

[keen thicket]

thank god for the walrus operator hahah

I also made something that takes code that is just lambdas or basic expressions and reflows it into ascii art

https://github.com/juliusgeo/pyflate

it results in some really weird coding style

it can reflow itself so the main program kinda shows

how awkward it is to write like that

but you need to do that anyway for really intense whitespace transformation

[karmic pumice]

all of the time i wasnt using python i was learning godot and the gdscript language doesnt really have that much features yet and its not really made for funky esoteric stuff to work on it, so i kinda missed it XD

it doesnt even have list comps

[keen thicket]

ahh yeah that makes sense

damn wtf

list comps are like

the best thing about python haha

[karmic pumice]

yeah it was sad to not have list comprehensions in a language that even states on the docs that its similar to python syntax-wise

[keen thicket]

I mean most of the time when a language says they have python syntax they just mean it's easy to read

but most of the time they don't make it easy to write

[karmic pumice]

1 sec, ill share a small snippet

func Jump(move_axis: Vector2) -> void:
    if cur_jumps > 0:
        cur_jumps -= 1
        velocity.y = jump_velocity; velocity.x = move_axis.x * speed_limit_h
        var particle = preload("res://Scenes/JP.tscn").instantiate()
        particle.global_position = self.global_position - Vector2(0, -10)
        particle.rotation = PI / 4 * move_axis.x
        $"/root/World/Particles".add_child(particle)

func _physics_process(delta: float) -> void:
    var move_axis: Vector2 = Vector2(Input.get_axis("ui_left", "ui_right"), 0)
    
    if (is_on_floor()): cur_jumps = max_jumps
    if Input.is_action_just_pressed("ui_up"): Jump(move_axis)
    set_collision_mask_value(2, !Input.is_action_pressed("ui_down"))
    if !(is_on_floor()): velocity.y += gravity * delta
    
    if velocity.x == 0: velocity.x = move_axis.x * starting_velocity_h
    else:
        var accel_h = acceleration_h * move_axis.x * delta
        if sign(velocity.x) != sign(accel_h): accel_h *= 5.0;
        velocity.x = clamp(velocity.x + accel_h, -speed_limit_h, speed_limit_h)
    
    if !move_axis.x: velocity.x = move_toward(velocity.x, 0, delta * 100.0)
    move_and_slide()

(it isnt even supported in markdown yet, sadly)

[keen thicket]

wtf and they also have var?

[karmic pumice]

also, i was quite surprised that a language for a game engine doesnt have static typing for dictionary key and value types and for nested collection element types 💀

[karmic pumice]

yeah

[keen thicket]

to be fair, type hinting in python is really fucking hard

but they could surely implement that on the engine side

[karmic pumice]

well atleast its doable

[keen thicket]

I mean I worked on PyMongo last year when it switched over to being fully type annotated

and it was a nightmare

easily 9 months of work

mypy is really opinionated, but also has a ton of weird edge cases and straight up doesn't support some things

[karmic pumice]

here you get this

[keen thicket]

you constantly have to sprinkle in #type:ignore after everything

[keen thicket]

LMFAOOOOO