this is weird
#esoteric-python
night quarryBOT
@cloud fossil :white_check_mark: Your eval job has completed with return code 0.
False
cloud fossil
!e
print(0 and 1/0)
@cloud fossil :white_check_mark: Your eval job has completed with return code 0.
0
oh okay
proper vault
fun fact: Java does this too, to a lesser extent
jshell> boolean weird() {System.out.println("Hello World!"); return true;}
| replaced method weird()
jshell> weird();
Hello World!
$3 ==> true
jshell> weird() || weird()
Hello World!
$4 ==> true
wait isn't Java a compiled language
proper vault
yes it is
proper vault
jshell is a new feature which allows a java repl
cloud fossil
i wish for a cshell
simple crystal
if you want to evaluate both sides of an and or or no matter what in Python, make it so you're using bitwise & or |, this requires actually using 1 for True and 0 for False (which is what True and False basically are) or bitwise equivalents and they have pretty high precedence in order of operations
cloud fossil
this is a bit off-topic, but PHP has a shell thing as well, right?
simple crystal
!e
print(True|1)
@simple crystal :white_check_mark: Your eval job has completed with return code 0.
1
@simple crystal :white_check_mark: Your eval job has completed with return code 0.
1
always 1
naive roost
!e ```py
print(True|2)
@naive roost :white_check_mark: Your eval job has completed with return code 0.
3
ᓚᘏᗢecho "#include <stdio.h>`n_Bool weird() {puts(`"Hello`");return 1;}int main(){weird()||weird();}" | tcc -run -xc -
Hello
```yeah, C does this as wellexcuse the code in a string literal
cloud fossil
that does not work on objects from libraries like discord.py though
simple crystal
!e ```py
print(True|2)
```
lol that's why you need to use 1 and 0
proper vault
yeah, | will not work on non-bools
cloud fossil
it's just bitwise or
simple crystal
it will but you need to know what's going on
cloud fossil
will it even work on negatives
proper vault
it isn't a boolean operator
proper vault
of course, you can do bool(a)|bool(b)
viscid nymph
!e
print(True|int(False))
now this is more esoteric to my liking
night quarryBOT
@viscid nymph :white_check_mark: Your eval job has completed with return code 0.
1
sometimes it lets you when you edit
viscid nymph
!source e
night quarryBOT
viscid nymph
sometimes it lets you when you edit
as long as you edit within ten seconds
simple crystal
ah
cloud fossil
that's cool
rugged sparrow
!e ```py
import gc
class magic:
def length_hint(self):
return 1
def iter(self):
for obj in gc.get_objects():
if type(obj) == tuple and len(obj) == 1:
try:1 in obj
except SystemError:
yield obj
break
weird = tuple(magic())
print(weird)
print(weird[0] is weird)
@rugged sparrow :white_check_mark: Your eval job has completed with return code 0.
001 | ((...),)
002 | True
idek what i'm looking at
proper vault
so you get a reference to a tuple while python still assumes it can run PyTuple_SetItem safely?
rugged sparrow
yea
proper vault
fancy
cloud fossil
!e
apple = ["apple"]
print(apple)
apple.append(apple)
print(apple)
@cloud fossil :white_check_mark: Your eval job has completed with return code 0.
001 | ['apple']
002 | ['apple', [...]]
so that's what i'm looking at
rugged sparrow
https://github.com/python/cpython/blob/f0a6fde8827d5d4f7a1c741ab1a8b206b66ffd57/Objects/abstract.c#L1966 @proper vault it happens in this function
its because the function uses PyTuple_SET_ITEM which is a macro that assigns directly
and you need __length_hint__ because _PyTuple_Resize fails if the tuple has more than one reference
proper vault
cool
rugged sparrow
i wonder what other weird stuff can be done using gc inside of initalizers
snow beacon
Can you put two keys that are equal in a dict?
I mean, without __hash__ shenanigans.
floral meteor
Challenge: make an InfiniteList such that setitem and getitem can write to and read from any index, creating a default value there (hint: extending the data) if it would otherwise not exist or return indexerror.
Negative values do not reference the list backwards, rather extends the list leftways
hard spoke
Challenge: make an InfiniteList such that __setitem__ and __getitem__ can write ...
...infinite_list = collections.defaultdict(lambda: object())?
floral meteor
Assuming object() to be the default value?
earnest wing
is that all the spec has to say?
hard spoke
something like that, though perhaps it should be the same object and not a new one each time
earnest wing
you don't want to use a dict as an infinite tape, lol
floral meteor
Watch me.
earnest wing
indexing has less overhead
hard spoke
why not? 🙂
floral meteor
Overhead is for nerds
hard spoke
indexing has less overhead
I disagree, indexing has a ton of overhead when your indexes go to +-infinity 😉
earnest wing
if you're indices are infinite, use a dict with float keys :>
floral meteor
No
hard spoke
why float keys? int is infinite-sized
hard spoke
neither is float
floral meteor
Neither is your mother
earnest wing
float("inf")
floral meteor
!e ```py
a = import ('collections').defaultdict(int)
i = int()
BITS=(True<<True)<<(True+False**False+True)
def plus():
a[i]+=True<<int()# cause true to become int
a[i]%=(True<<True)**BITS
def minus():
a[i]-=True<<int()
a[i]%=(True<<True)**BITS
def left():i-=True<<int()
def right():i+=True<<int()
def dot(file=import('sys').stdout):file.write(chr(a[i]))
def comma(file=import('sys').stdin):a[i]=ord(file.read(True))
[plus()for I in range(100)]
minus();minus();dot();minus();dot();plus();plus();plus();dot()
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
bad
I like it
viscid nymph
!e ```py
a = __import__ ('collections').defaultdict(int)
i = int()
BITS=(True<<T...
defaultdict(int) 
floral meteor
I did int. XD
viscid nymph
😮
floral meteor
Jk pulling your leg
viscid nymph
thanks
snow beacon
!e ```py
a = __import__ ('collections').defaultdict(int)
i = int()
BITS=(True<<T...
FTFY ```py
a = import ('collections').defaultdict(int)
i = int()
BITS=((int()**int())<<(int()**int()))<<((int()**int())+(int()**int())+(int()**int()))
def plus():
a[i]+=(int()**int())<<int()
a[i]%=((int()**int())<<(int()**int()))**BITS
def minus():
a[i]-=(int()**int())<<int()
a[i]%=((int()**int())<<(int()**int()))**BITS
def left():i-=(int()**int())<<int()
def right():i+=(int()**int())<<int()
def dot(file=import('sys').stdout):
with file as f:f.write(chr(a[i]))
def comma(file=import('sys').stdin):
with file as f:a[i]=ord(f.read((int()**int())))
[plus()for _ in range(int(bin((int()**int()) << (int()**int()) + (int()**int()))[(int()**int()) + (int()**int()):]))]
minus();minus();dot();minus();dot();plus();plus();plus();dot()
What's this supposed to do
raise errors?
why do you do int()**int() instead of int()+int()
snow beacon
Because they are different numbers.
hard spoke
python has 0**0 be 1 😔
viscid nymph
wait I'm dumb
snow beacon
It also closes all the files after use. (Stdin included.)
viscid nymph
it raises errors
snow beacon
A small price to pay for file safety.
floral meteor
pffft
i make my own file class that doesn't block the file when its open
you noobs use straight file.open(filename) I make the read method open and close the file ahahahaha
snow beacon
Personally I use pathlib.
I don't need to get into the nitty-gritty when there are abstractions.
floral meteor
I am the nitty-gritty abstractions.
aight im gonna nitty gritty abstractions literally bound the entire global namespace to an object cos I can
snow beacon
I *am* the nitty-gritty abstractions.
So the abstractions are a fatal Python error. I see.
floral meteor
!e ```bf
#[
a,i=import ('collections').defaultdict(int),int()
BITS=((int()**int())<<(int()**int()))<<((int()**int())+(int()**int())+(int()**int()))
def add(*s):
a[i<<int()]+=(int()**int())<<int()
a[i>>int()]%=((int()**int())<<(int()**int()))**BITS
return s[int()]
pos=add
def sub(*s):
a[i<<int()]-=(int()**int())<<int()
a[i>>int()]%=((int()**int())<<(int()**int()))**BITS
return s[int()]
neg=sub
def lt(*s):globals()['i']-=(int()**int())<<int();return s[int()]
def gt(*s):globals()['i']+=(int()**int())<<int();return s[int()]
def lshift(*s):return lt(*s)<int()<<int()
def rshift(*s):return gt(*s)>int()>>int()
def getattr(s,o):
try:return object.getattr(s,o)
except:import('sys').stdout.write(chr(a[i<<int()]>>int()));return s
def call(self,f,a):
if a:a[i>>int()]=ord((f or import('sys').stdin).read(len(a)<<int()))
return self
bool=lambdas:not int()
def getitem(s,item):
while a[i>>int()]:(v:=item()if hasattr(item,'call')else+s if item>=int()else-s).class is int and(v<int()and-s or+s)
return s
=type("brainfucker",(int,),globals())(int())
#]
(++++++++++)[lambda:>int()and ++++++++++<int()and-int()**int()]>int()and -- and (.int-int()).int+++ and _.int(int())
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
bad
gottim
it's a polyglot with no numeric literals!
evidence the bot runs brainfuck, not python.
!e ```py
#[
a,i=import ('collections').defaultdict(int),int()
BITS=((int()**int())<<(int()**int()))<<((int()**int())+(int()**int())+(int()**int()))
def add(*s):
a[i<<int()]+=(int()**int())<<int()
a[i>>int()]%=((int()**int())<<(int()**int()))**BITS
return s[int()]
pos=add
def sub(*s):
a[i<<int()]-=(int()**int())<<int()
a[i>>int()]%=((int()**int())<<(int()**int()))**BITS
return s[int()]
neg=sub
def lt(*s):globals()['i']-=(int()**int())<<int();return s[int()]
def gt(*s):globals()['i']+=(int()**int())<<int();return s[int()]
def lshift(*s):return lt(*s)<int()<<int()
def rshift(*s):return gt(*s)>int()>>int()
def getattr(s,o):
try:return object.getattr(s,o)
except:import('sys').stdout.write(chr(a[i<<int()]>>int()));return s
def call(self,f,a):
if a:a[i>>int()]=ord((f or import('sys').stdin).read(len(a)<<int()))
return self
bool=lambdas:not int()
def getitem(s,item):
while a[i>>int()]:(v:=item()if hasattr(item,'call')else+s if item>=int()else-s).class is int and(v<int()and-s or+s)
return s
=type("brainfucker",(int,),globals())(int())
#]
(+++++++)[lambda:>_ and +++++++ and < and -]>_ and +++++ and(.+++_).int(int())
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
69
I am invincible!
shood i make a bash interpreter in a brainfuck/python polyglot generated by an obfuscated+code art c++ script and then make the result run a windows command prompt emulator written in bash?
and then use that as an IDE to write files in rust.
or matlab
XD
snow beacon
Follow your dreams, even if your dreams are... that.
simple crystal
sounds more like a gosh-darned nightmare
floral meteor
XD
unique viper
heh, well this is probably an appropriate room for this question. Is there a simpler way to override all of the dunder methods rather than just doing it for every one?
simple crystal
!e
print([0xfor x in (1, 2, 3)])
@simple crystal :white_check_mark: Your eval job has completed with return code 0.
[15]
__getattr__ I think it is?
unique viper
i tried that, even overriding both __getattr__ and __getattribute__
my goal is essentially to get one piece of data to look like another piece of data for all interactions
floral meteor
Would have weird consequences so I'll try to make one in the cursedutils package
snow beacon
my goal is essentially to get one piece of data to _look like_ another piece of ...
Inheritance doesn't cut it?
unique viper
let me give a little more context...
snow beacon
I made some code that copied all of the methods of boolean objects for one of the Esoteric Python Challenges: https://github.com/python-discord/esoteric-python-challenges/blob/master/challenges/06-maybe-keyword/solutions/IFcoltransG/Maybe.py
floral meteor
`__getattr__` I think it is?
Dunders are impotent unless properly bound.
unique viper
i have a library that does a lot of deferred actions. It's a web-browser automated testing library.
I wrote a thing that lets you write down a note, but the note-writing part is in the deferred action, so if you try to access that note before the deferred action is performed, the note won't be there yet.
I'm trying to make an imposter class that will just store the lookup key for the note, so when it actually needs to be used, it can go get the note instead
snow beacon
If you were to do it idiomatically, that would probably be writing a wrapper around all of the note logic, then making a different wrapper that goes and gets the note.
You could also consider overwriting the __class__ of the note proxy once it fetches the note.
unique viper
here's the actual issue in my codebase: https://github.com/perrygoy/screenpy/issues/51
it's got an example in the OP, but i'm not sure if it's easy to understand without more context
basically i don't want the rest of my codebase to have to know how to go and get the note themselves
i want the rest of my codebase to just be able to use the Imposter as if it were the stored value
snow beacon
The "right" way would be wrapping both types of note/defining a clear API for what notes have to implement. The "fun" way would be one of the other things this channel has to offer.
unique viper
i feel i might be in the right place
unique viper
this Maybe/Might_Be class is great
snow beacon
Other people made entries for the same challenge, that might also be worth looking into.
floral meteor
i would just brute force the Maybe object
snow beacon
I wrote that solution using a program, then tidied it up.
The one I ended up with is adaptable and maintainable.
There is no benefit to spelling out all umpteen hundred methods by hand.
floral meteor
watch me.
snow beacon
It's not that it's hard, it's just that it's pointless.
floral meteor
there done but it's 2600 chars
night quarryBOT
Hey @floral meteor!
verbal geode
!e ```py
#[
a,i=__import__ ('collections').defaultdict(int),int()
BITS=((int()**...
excuse me?!
snow beacon
https://paste.pythondiscord.com/atopanahur.py
real, imag, numerator and denominator are properties, not functions.
(If only there were an automated solution.)
rustic aspen
would it be possible to import python into itself?
thin trout
would it be possible to import python into itself?
It depends on your definition of "importing python"
rustic aspen
i obviously could just make a file called python.py but thats not what im looking for
im not 100% sure what importing python would be, but i think i have the vague idea that it contains all the defaults and compilers and stuff
thin trout
Hmmmm
Well, since python is a language with huge introspection, you can almost hook into anything
For example, you can use the ast module to compile code
opaque hemlock
Can anyone help me with programming a coding language
my var command is not wrking
simple crystal
if you want a slower turtle, you stack another turtle on top of the first turtle
- Albert Einstein
edgy spire
wouldnt the top turtle be twice as fast
fluid tree
wouldnt the top turtle be twice as fast
If it tried to walk it would fall off
edgy spire
hmm i guess you'd have to make the bottom turtle faster to keep up with it
simple crystal
that's how PyPy works
finite rose
night quarryBOT
@finite rose :white_check_mark: Your eval job has completed with return code 0.
001 | True
002 | True
clearly not in special relativity, and this is an einstein quote 😛
edgy spire
clearly not in special relativity, and this is an einstein quote 😛
ok but how much did einstein really know about turtles
plucky wedge
Einstein, famed physicist turtle researcher
snow beacon
Hopefully better than Freud with eels.
simple crystal
@rugged sparrow okay so I looked at new fishhook stuff and it looks like you are looking for pointers in a heap class which point to memory addresses within the heap class itself?
I wonder if it's technically possible to find other data which happen to have integer values within the memory addresses of the heap class, tho I doubt that's likely
ah okay they point to deeper in the class
byte: b'(\xb0]\x99M\x02\x00\x00' val: 2532308791336, pointer: 10 points to: 51
byte: b'@\xb0]\x99M\x02\x00\x00' val: 2532308791360, pointer: 12 points to: 54
byte: b'x\xb1]\x99M\x02\x00\x00' val: 2532308791672, pointer: 13 points to: 93
byte: b'`\xb1]\x99M\x02\x00\x00' val: 2532308791648, pointer: 14 points to: 90
byte: b'\xc8\xb1]\x99M\x02\x00\x00' val: 2532308791752, pointer: 20 points to: 103
byte: b'\x00\xb2]\x99M\x02\x00\x00' val: 2532308791808, pointer: 30 points to: 110
which makes sense cuz that's the memory allocated for it on the heap
rugged sparrow
<@!274715613115711488> okay so I looked at new fishhook stuff and it looks like ...
I am using the fact that PyHeapTypeObjects are a regular PyTypeObject followed directly by its tp_as_* structs to calculate the size of the structs (and their pointer offset inside the PyTypeObject)
simple crystal
is there any chance other data as ints will be the same as the memory addresses?
rugged sparrow
Not in the same way
There are a few other structures that are stored inside the PyHeapTypeObject struct but they aren't super useful for monkey patching
simple crystal
after a few tries I managed to find one with an int that looked like a pointer within the heap class at 120 from _start but that wouldn't change the functionality apparently
another at 118
124
points to 2 slots up
rugged sparrow
You need to look for pointers within the heap class size (between start and end)
And go by multiples of base_size
simple crystal
yah I just stuffed this in your code and looped it
if i//base_size not in (10, 12, 13, 14, 20, 30):
print(f'wtf {i//base_size} points to: {(_val-_start)//base_size}')
but running only sometimes produces 1 or 2 wtfs
simple crystal
!e
from ctypes import *
import sys
base_size = sizeof(c_ssize_t)
wrapper_type = type(int.__add__)
static_size = type.__sizeof__(type) // base_size
def mem(addr, size):
return (c_char*size).from_address(addr)
def doin():
class _scratch:...
_size = type.__sizeof__(_scratch)
_start = id(_scratch)
_end = _start + _size
_cls_mem = mem(_start, _size).raw # actual memory # mutable copy
_intermediate = []
for i in range(0, _size, base_size):
_val = int.from_bytes(_cls_mem[i:i + base_size], sys.byteorder)
if _start < _val < _end:
if i//base_size not in (10, 12, 13, 14, 20, 30):
print(f'wtf {i//base_size} points to: {(_val-_start)//base_size}')
_intermediate.append((i//base_size, _val))
_last_addr = None
_offsets, _sizes = [0], [static_size]
for _offset, _addr in sorted(_intermediate, key=lambda i:i[1]):
if _last_addr is not None:
_sizes.append((_addr - _last_addr)//base_size)
_offsets.append(_offset)
_last_addr = _addr
while True:
doin()
lol carriage return doesn't work
night quarryBOT
@simple crystal :x: Your eval job timed out or ran out of memory.
001 | wtf 130 points to: 51
002 | wtf 130 points to: 51
003 | wtf 130 points to: 51
004 | wtf 130 points to: 51
005 | wtf 130 points to: 51
006 | wtf 130 points to: 51
007 | wtf 130 points to: 51
008 | wtf 130 points to: 51
009 | wtf 130 points to: 51
010 | wtf 130 points to: 51
011 | wtf 130 points to: 51
... (truncated - too many lines)
Full output: too long to upload
simple crystal
ah this is different
when I run it fro a while with the carriage returns I get
1818237271088wtf 120 points to: 122
1818237340688wtf 124 points to: 126
1818237276608
for example
rugged sparrow
Pm me the total code I want to look at it
simple crystal
kk
rugged sparrow
I'll need to sift thru it in like an hour
simple crystal
it's usually only 1 or 2 wtfs per run if any
kk lemme know if you figure out what's happening haha
rugged sparrow
It's probably some uninstalized memory containing big values
So pure chance
I don't think my logic is wrong
simple crystal
ah okay and that's why it stops after a bit
because it overwrites uninstalized memory where heap classes are being stored
rugged sparrow
It shouldn't be overwriting anything?
In the code you sent nothing should be getting overwritten
simple crystal
I mean the heap classes are defined then garbage collected
I wonder why my computer always has them as the memory addresses 2 base_sizes away and the bot has one at 130 pointing to 51
simple crystal
why's that?
rugged sparrow
Need to look at source
simple crystal
!e
from ctypes import *
import sys
base_size = sizeof(c_ssize_t)
wrapper_type = type(int.__add__)
static_size = type.__sizeof__(type) // base_size
def mem(addr, size):
return (c_char*size).from_address(addr)
def doin():
class _scratch:...
_size = type.__sizeof__(_scratch)
_start = id(_scratch)
_end = _start + _size
_cls_mem = mem(_start, _size).raw # actual memory # mutable copy
_intermediate = []
for i in range(0, _size, base_size):
_val = int.from_bytes(_cls_mem[i:i + base_size], sys.byteorder)
if _start < _val < _end:
if i//base_size not in (10, 12, 13, 14, 20, 30, 130): # added 130
print(f'wtf {i//base_size} points to: {(_val-_start)//base_size}')
_intermediate.append((i//base_size, _val))
_last_addr = None
_offsets, _sizes = [0], [static_size]
for _offset, _addr in sorted(_intermediate, key=lambda i:i[1]):
if _last_addr is not None:
_sizes.append((_addr - _last_addr)//base_size)
_offsets.append(_offset)
_last_addr = _addr
while True:
doin()
@simple crystal :x: Your eval job timed out or ran out of memory.
001 | wtf 118 points to: 114
002 | wtf 119 points to: 114
003 | wtf 118 points to: 114
004 | wtf 119 points to: 114
005 | wtf 118 points to: 114
006 | wtf 119 points to: 114
007 | wtf 118 points to: 114
008 | wtf 119 points to: 114
009 | wtf 118 points to: 114
010 | wtf 119 points to: 114
011 | wtf 118 points to: 114
... (truncated - too many lines)
Full output: too long to upload
simple crystal
typicla output for my computer with same script I gave bot:
wtf 122 points to: 124
or nothing
oh and the mutable copy comment was from earlier when I was comparing the array of chars to a create_string_buffer
rugged sparrow
If the type has cached keys then the just getting the mem of type.sizeof gets some random memory at the end
simple crystal
ah cool
rugged sparrow
I'll figure out a way to fix that
simple crystal
ohhh is it because I made type(__scratch__) into type?
rugged sparrow
No
simple crystal
ok forgot I did that
rugged sparrow
It shouldn't break anything in fishhook to read that memory because any values there are preceded by the pointers I am looking for
Still something I should prob fix
simple crystal
makes sense
...could you jut use static_size?
instead of _size in the first for loop
rugged sparrow
Did it fix the weird issue,
simple crystal
haven't tried it with this testing script I'll check
just was trying to change the size of scratch
seems to, yep
!e
from ctypes import *
import sys
base_size = sizeof(c_ssize_t)
wrapper_type = type(int.__add__)
static_size = type.__sizeof__(type) // base_size
def mem(addr, size):
return (c_char*size).from_address(addr)
def doin():
class _scratch:__slots__=()
_size = type.__sizeof__(_scratch)
_start = id(_scratch)
_end = _start + _size
_cls_mem = mem(_start, _size).raw # actual memory # mutable copy
_intermediate = []
for i in range(0, _size, base_size):
_val = int.from_bytes(_cls_mem[i:i + base_size], sys.byteorder)
if _start < _val < _end:
if i//base_size not in (10, 12, 13, 14, 20, 30): # added 130
print(f'wtf {i//base_size} points to: {(_val-_start)//base_size}')
_intermediate.append((i//base_size, _val))
_last_addr = None
_offsets, _sizes = [0], [static_size]
for _offset, _addr in sorted(_intermediate, key=lambda i:i[1]):
if _last_addr is not None:
_sizes.append((_addr - _last_addr)//base_size)
_offsets.append(_offset)
_last_addr = _addr
while True:
doin()
@simple crystal :warning: Your eval job timed out or ran out of memory.
[No output]
nice
rugged sparrow
Yea it causes the class to have no cached keys
I'll update that in fishhook tonight
simple crystal
ah neat
simple crystal
that also gets rid of the tp_members pointer I think
oh it just doesn't register because it points to _end
so maybe it's not tp_members
simple crystal
nice
unique glacier
Hey ,guys can you make python file find a largest amoung the five numbers. you can?After make it send the code to DISCORD.
snow beacon
Hey ,guys can you make python file find a largest amoung the five numbers. you c...
If you go to #❓｜how-to-get-help it will tell you how to get help with that.
sick hound
Without exec and eval is python turing complete if you aren't allowed to use whitespace? 
cloud fossil
well, you won't be able to define functions
nor classes
and maybe even not import anything
do newlines count as whitespace to you?
cloud fossil
that just means one line statements i think idk
i mean... one line entire code XD
and no tabbing
sick hound
!e you can use semicolons, but yeah, no indents py print(9);print(8);print(7);
night quarryBOT
@sick hound :white_check_mark: Your eval job has completed with return code 0.
001 | 9
002 | 8
003 | 7
no need for the last one btw
sick hound
true
cloud fossil
!e
import(time)
@cloud fossil :x: Your eval job has completed with return code 1.
001 | File "<string>", line 1
002 | import(time)
003 | ^
004 | SyntaxError: invalid syntax
oof
i mean... maybe you could straight up use the loader or whatever idk how it works
maybe it's not a statement so you can use it as a function or something
sick hound
I was thinking you could do some fancy stuff with __import__ but idk how 
cloud fossil
same
oh i know
!e
time=__import__("time");print("sleeping");time.sleep(2);print("slept","for","two","seconds!")
@cloud fossil :white_check_mark: Your eval job has completed with return code 0.
001 | sleeping
002 | slept for two seconds!
meh, it works
vague hearth
well, you won't be able to define functions
!e you can use type to make methods thopy print(type("main",(),{"__init__":lambda*a:(setattr(a[0],"i",a[1]),setattr(a[0],"j",a[2]))[0],"add":lambda*a:(a[0].i+a[0].j)})(1,3).add())
night quarryBOT
@vague hearth :white_check_mark: Your eval job has completed with return code 0.
4
print(type("main",(),{"add":lambda*a:(a[1]+a[2])})().add(1, 3))```(or use this if you juts want a function without instantiating with the values)
potent comet
You cal also directly instantiate types.CodeType() and then types.FunctionType() with all the correct values, to create a function that does anything.
earnest wing
Might as well just exec(your code encoded in a funny way, decoded at runtime)
snow beacon
You cal also directly instantiate `types.CodeType()` and then `types.FunctionTyp...
Or take a function and replace its bytecode.
grave rover
@rugged sparrow any clue why this wouldn't work? ```py
buffer = b"\x00" * calcsize('3P')
int_type = get_view(id(int), CPyTypeObject.instance())
int_type.tp_as_mapping = get_view(id(buffer), CPyBytesObject.instance()).ob_sval[0]._addr
# bytecode: compiled form of
# void do_x(PyLongObject* self, PyLongObject* pylong_1) {
# self->digit = pylong_1->digit;
# }
bytecode = b"\x55" \
b"\x48\x89\xe5" \
b"\x48\x89\x7d\xf8" \
b"\x48\x89\x75\xf0" \
b"\x48\x8b\x45\xf0" \
b"\x8b\x50\x14" \
b"\x48\x8b\x45\xf8" \
b"\x89\x50\x14" \
b"\x90" \
b"\x5d" \
b"\xc3"
bytecode_py_addr = get_view(id(bytecode), CPyBytesObject.instance()).ob_sval[0]._addr
int_type.tp_as_mapping.mp_subscript = bytecode_py_addr
pythonapi.PyType_Modified(py_object(int))
print(10[30]) # call mp_subscript
# Error: TypeError: 'int' object is not subscriptable
First of all you need to memory map bytecode as RWX
Also make sure to set mp_subscript and sq_item
And both of those functions return an PyObject*
So can't be void
@grave rover ^ also are you sure that your offsets are right
grave rover
offsets are 100% right
rugged sparrow
https://github.com/python/cpython/blob/76beadb8ff86eb2bb248bf918bfb20c4069932f4/Objects/abstract.c#L156 the you should be getting to here in the source
grave rover
unless something changed between 3.9 and 3.10 but isn't listed as being different on the docs
rugged sparrow
A new slot was added to tp_as_async
rugged sparrow
Yea there is a 4th slot now
@grave rover those functions actually return a PyObject
Well mp_subscript does
grave rover
right, I made a slight mistake when I moved over lol
rugged sparrow
Other than those it should work as is (with a crash due to memory execution issues)
grave rover
wait so how would I map that as executable?
rugged sparrow
 you can use libc
I have an example In my asm_hook code
On my GitHub @grave rover
https://github.com/chilaxan/pysnippets/blob/cf8a8701969adeb54bc91fa1f135061d942b95e5/asm_hook.py#L21
grave rover
I'm starting to not like how python handles negative numbers
ob_size of 1 for negative numbers becomes 0xffffffff
rugged sparrow
ob_size of 1 for negative numbers becomes 0xffffffff
That is cause ob_size is a signed 8byte number
So it actually becomes -1
grave rover
it just makes my life a little less enjoyable
it also somehow makes printing 10 an issue
rugged sparrow
It does?
grave rover
10[20] = -30 # Injected code: 10.ob_digit = 20.ob_digit + -30.ob_digit
print(10) # segfault
rugged sparrow
Also if you want to avoid having to stuff raw assembly into your program you can grab the default mp_subscript from a python class
grave rover
yeah but I was looking for a way to inject custom C code :^)
rugged sparrow
Ah
grave rover
injected code is c int do_x(PyLongObject* self, PyLongObject* pylong_1, PyLongObject* pylong_2) { self->digit[0] = pylong_1->digit[0] + pylong_2->digit[0]; return 0; }
rugged sparrow
Where are you injecting that?
rugged sparrow
` 10[20] = -30 # Injected code: 10.ob_digit = 20.ob_digit + -30.ob_digit`
` ...
take a look at the raw memory of 10 to see if what happened was expected
grave rover
it's not
when I make -30 into 30 it stays 10
on -30 the ob_digit value becomes 1
even if it doesn't go negative, ob_size becomes negative
rugged sparrow
are you sure the assembly is doing things correctly?
grave rover
well uh
I kinda forgot ob_size was ssize_t
and made it an int when it should've been a long
works perfectly fine now lol
rugged sparrow
ah nice
last locust
TIL this is a shorter way of inverting a bool (var ^= True vs var = not var)```py
var = True
var ^= True
var
False
var ^= True
var
True```
last locust
`var^=1`
That's technically not the same though
Because it gives an int
>>> var = True
>>> var ^=1
>>> var
0
>>> var^=1
>>> var
1```aww
last locust
Still the same truthiness but not strictly bool anymore
proper vault
they are different
True is a bool and 1 is an int
you can't return 1 from __bool__ for example
merry ridge
https://pastebin.ubuntu.com/p/cbn238sTmv/ hi the complexity of this code is n ^ 2 can you help me
snow beacon
You could probably get it to n^3 by wrapping everything in an extra loop.
fluid tree
you can't return `1` from `__bool__` for example
This makes me sad
simple crystal
TIL this is a shorter way of inverting a bool (`var ^= True` vs `var = not var`)...
the shortest way to invert an actual bool (1 or 0) is ~-
!e
if ~-True:
print('~-1 is 0')
if~-False:
print('~-0 is -1')
@simple crystal :white_check_mark: Your eval job has completed with return code 0.
~-0 is -1
it doesn't return a bool
but it evaluates as the opposite of the bool in the majority of cases
fluid tree
but it evaluates as the opposite of the bool in the majority of cases
When doesn't it?
1-(bool) seems easiest (although ^=1 is best for in-place)
Or if you want to be consistent use ^ for both
simple crystal
when the bool is getting passed around and having operations done to it that require it to be an actual bool or 0 or 1 I guess
last locust
the shortest way to invert an actual bool (1 or 0) is `~-`
I wanted it to be strictly bool
Besides it's not any shorter
var=~-var
var^=True
grave rover
You could probably get it to n^3 by wrapping everything in an extra loop.
why stop at n^3?
snow beacon
At a certain point, passing around the n you have to loop over (in order that you don't accidentally do a constant number of iterations or something equally bad) will just become overhead, making it complex for nothing.
earnest wing
Nontrivial complexity is way more fun
simple crystal
Besides it's not any shorter
that's only for 3 char var and there's no reason to actually use more than 1 char when golfing
floral meteor
if only you could combine augmented assignments
like
>>> a = 0
>>> b = (a +:= 3)+1
>>> b %+*= 2
>>> print(b*a)
12
I think you can with match case sorta
floral meteor
instead of
>>> a = 0
>>> a += 3
>>> b = a + 1
>>> b %= 2
>>> b = (b+2)*2
>>> print(b*a)
12
this ain't no darn ++ langy here now
walrii are not well known for their hat wearing
floral meteor
what about helmet?
[:=
that's fine and I will draft the PEP now
I made a fake PEP to add ridiculous concatenation functionality to dict class for a technical writing class
floral meteor
actually, I was hoping 'bracket-equals' operators would exist and do something
but you can d12 = d1 | d2
simple crystal
if both dicts have same key, stuff all the values in a tuple lol
if the values are tuples concatenate
if one isn't a tuple concatenate that value in a tuple with other tuple
ridiculous
floral meteor
!e ```py
a = {'a':1, 'b':3}
b = {'b':2, 'c':3}
print(a|b)
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
{'a': 1, 'b': 2, 'c': 3}
show me some haskell
simple crystal
!e
class ConcatenationDict(dict):
"""dict which allows for concatenation."""
def __add__(self, addend):
"""Concatenate this dict with another."""
updated_self = {}
for key, value in self.items():
if key in addend:
if isinstance(value, tuple):
if isinstance(addend[key], tuple):
updated_self.update({key: value + addend[key]})
else:
updated_self.update({key: value + (addend[key],)})
else:
if isinstance(addend[key], tuple):
updated_self.update({key: (value,) + addend[key]})
else:
updated_self.update({key: (value, addend[key])})
else:
updated_self.update(key, value)
addend.update(updated_self)
return updated_self
if __name__ == '__main__':
"""Test the functionality of ConcatenationDict with example provided in PEP"""
a = ConcatenationDict({'key1': 1, 'key2': (1, 2), 'key3': (1,2)})
b = ConcatenationDict({'key1': 2, 'key2': (3, 4), 'key3': 3})
a += b
print(a)
@simple crystal :white_check_mark: Your eval job has completed with return code 0.
{'key1': (1, 2), 'key2': (1, 2, 3, 4), 'key3': (1, 2, 3)}
I will probably learn haskell in the next couple days just to have another entry for a code golf context
been itching for an excuse to learn it
I guess I should link it lol this week is very easy
floral meteor
alright, now turn this mess into dict comprehension
simple crystal
from a picture? pssh
floral meteor
it's an extract of your code
simple crystal
o
haha
seems p easy
just make ternaries with isinstance or some shortcuts for equivalent
I gotta walk and buy more beer rn tho
floral meteor
a={key:[value,(value,)][not isinstance(addend[key]),tuple)]+[addend[key],(addend[key],)][not isinstance(addend[key],tuple)]if key in addend else value for key,value in self.items()}
addend.update(a)
return a
!e ```py
print(~3 <=- 1)
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
True
I name it thus: "rocket operator"
def Switch():
"""python can't tell me I can't have a switch suite :/"""
call=lambda*args,**kwargs:lambda f:f(*args,**kwargs)
named_as=lambda n:lambda f:setattr(f,'__name__',n)or f
@call()
class case:
def __init__(self): self.state, self.value =- 1, None
def __getitem__(self, item):
if self.state <=- 1:raise SyntaxError("Cannot find matching switch") # __annotations__ hack only works in globals
if self.value == item and not self.state: self.state=1; return(lambda f,*u,**ck:f(*u,**ck))
else:return(lambda *No, **ne:None) # XD
def switch(value):case.value,case.state = value,0
@named_as('default')
def otherwise(f,*u,**ck):
if case.state <- 0:raise SyntaxError("Cannot find matching switch")
elif not case.state: case.state =- 1; f(*u,**ck)
return switch, case, otherwise #
How could I improve this?
floral meteor
nope?
we should move this channel to its own section called The Dark Side, and make specialist channels, like code art, golfing, obfuscation and polyglots
haskell look like kabloop
i don't think haskell could be made in python syntax
floral meteor
!e ```py
def disp(a):print(a)
class struct:
@classmethod
def bootstrap(c,__g):
try:assert __g
except AssertionError as e:__g=e.traceback.tb_frame.f_back.f_globals
@lambda f:setattr(import('ctypes').py_object.from_address(id(__g)+8),'value',type("dict",(dict,),{'missing':f}))
def missing(self,key):
try:return getattr(builtins,key)
except:self[key]=c()
return self[key]
def getattr(self,v):
try:return(f:=lambda:object.getattr(self,v))()
except AttributeError:setattr(self,v,self.class())
return f()
struct.bootstrap(globals())
#---
namespace.globals.data.x = 4
namespace.globals.data.y = 3
function=lambda data: (data.x * data.y) ** 0.5 // 1
namespace.locals.data.output = function(namespace.globals.data)
disp(namespace.locals.data.output)
@floral meteor :x: Your eval job has completed with return code 139 (SIGSEGV).
001 | Traceback (most recent call last):
002 | File "<string>", line 18, in <module>
003 | File "<string>", line 11, in __missing__
004 | AttributeError: __getitem__
I hope you take it as a compliment thay I've learned not to bother trying to figure out what your code does cuz it does weird stuff while looking weird lol
floral meteor
!e alright, devoid of the globals() hack, and implementing manual methodification:
class struct:
def __getattribute__(self,v):
try:return(f := lambda:object.__getattribute__(self,v))()
except AttributeError:setattr(self,v,self.__class__())
return f()
def __setattr__(self,n,v):
if isinstance(v,(lambda:None).__class__):
victim = __import__('ctypes').py_object.from_address(id(self)+8)
victim.value = type("struct",(victim.value,),{n:v})
elif isinstance(v,print.__class__):
victim = __import__('ctypes').py_object.from_address(id(self)+8)
victim.value = type("struct",(victim.value,),{n:lambda s,*a,**k:v(*a,**k)})#bound builtin behaviour
else:object.__setattr__(self,n,v)
__repr__=lambda s:str([f'{this}: {getattr(s,this)}'for this in dir(s)])[1 :- 1] # name the operator XD
namespace=struct()
namespace.globals.function.cout.__lshift__=print
namespace.globals.data.x, namespace.globals.data.y = 3, 4
namespace.locals.function.magnitude=lambda s,d:int(((d.x**2) + (d.y**2)) ** 0.5)
namespace.locals.data.output = namespace.locals.function.magnitude(namespace.globals.data)
namespace.globals.function.cout << namespace.locals.data.output
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
5
gottim
@sick hound @simple crystal this is an attempt to integrate the behaviour of the struct object from matlab into python, also making a variable manager in the process, by categorizing variables into a namespace structure.
simple crystal
ah nice
floral meteor
key behaviour
binding functions
substruct referencing
successful binding behaviour and variable categorisation
thus calculating the magnitude of a vector with x value 3 and y value 4
which is 5
ty
because break isn't an expression, rather it is a statement
!e ```py
try:
exec ("""for i in range(10):
print(i)or break
""")
except SyntaxError:
for i in range(10):
if not print(i+1):break
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
1
@simple crystal :x: Your eval job has completed with return code 1.
001 | 1
002 | Traceback (most recent call last):
003 | File "<string>", line 2, in <module>
004 | File "<string>", line 2
005 | print(i)or break
006 | ^
007 | SyntaxError: invalid syntax
008 |
009 | During handling of the above exception, another exception occurred:
010 |
011 | Traceback (most recent call last):
... (truncated - too many lines)
Full output: https://paste.pythondiscord.com/poyahehopu.txt?noredirect
floral meteor
exec does not insert the line, rather makes a subprocess
simple crystal
ah
can't break a superprocess loop
I guess allowing something to return break
would be the most implicit thing ever
floral meteor
!e you can however update a dict like such
d = {}
exec ("""for i in range(10):
globals().update({chr(i+97):i})
""",d,d)
print(d)
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
{'__builtins__': {'__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 f
... (truncated - too long)
Full output: https://paste.pythondiscord.com/otehutacag.txt?noredirect
night quarryBOT
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
{'__builtins__': {'__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 f
... (truncated - too long)
Full output: https://paste.pythondiscord.com/duxacaqilu.txt?noredirect
floral meteor
this is cursed
i like it
!e ```py
exec('',(d:={}),d)or d['builtins']'globals'.update({'a':4})
print(a)
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
4
the second arg gets processed more than once?
oh it's to set it then do stuff to it
!e
exec('',(d:={},)or d['__builtins__']['globals']().update({'a':4})
print(a)
@simple crystal :x: Your eval job has completed with return code 1.
001 | File "<string>", line 2
002 | print(a)
003 | ^
004 | SyntaxError: invalid syntax
ah
!e
exec('',d:={})
d['__builtins__']['globals']().update({'a':4})
print(a)
lol
night quarryBOT
@simple crystal :white_check_mark: Your eval job has completed with return code 0.
4
!e ```py
d, e = {}, {}
exec('locals()["a"]=4',d,e) or d['builtins']'globals'.update(e)
print(a)
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
4
EVEN MORE OBFUSCATED VERSION
simple crystal
and if you multiply the string it's a for loop lol
sick hound
!e exec('', globals().update({'a':4})),print(a)
night quarryBOT
@sick hound :white_check_mark: Your eval job has completed with return code 0.
4
b:*+%/=5
You can use an If Expression instead of a statement, if you like
Better to use a onelinerifier
frigid lava
Does anyone know a bank transaction GUI ???????
fluid tree
sudden willow
!e
```py
time=__import__("time");print("sleeping");time.sleep(2);print("slept","...
!e
(lambda t:(__import__('sys').__stdout__.write('sleeping\n'),t.sleep(2),__import__('sys').__stdout__.write(''.join(x for x in('slept ','for ','two ','seconds ')))))(__import__('time'))
@sudden willow :white_check_mark: Your eval job has completed with return code 0.
001 | sleeping
002 | slept for two seconds
Ew ;
!e [x() for x in (lambda: print("sleeping"), lambda: import("time").sleep(2), lambda: print("slept for two seconds"))]
night quarryBOT
@fluid tree :white_check_mark: Your eval job has completed with return code 0.
001 | sleeping
002 | slept for two seconds
toxic jewel
cute
sudden willow
!e `exec('', globals().update({'a':4})),print(a)`
!e
@lambda v: globals().get('c')()
@lambda v: globals().__setitem__('c', lambda: print('hello'))
class v: pass
forgot to eval pain
!e
@lambda v: globals().get('c')()
@lambda v: globals().__setitem__('c', lambda: print('hello'))
class v: pass
@sudden willow :white_check_mark: Your eval job has completed with return code 0.
hello
I just found this thing I wrote that looks like brainfuck, but can find the fattest cat in a pet adoption website.
floral meteor
!e ```py
d, e = {}, {}
exec('locals()["a"]=4',d,e) or d['__builtins__']['globals...
wait this can be abused by using dict-like custom classes
!e ```py
__ =()==()==()==()==()==()==()==()==()==()==()
+=++++++++++++++++++++
*=--------------------
<<=-//--++--++--++--**
print(__)
@floral meteor :x: Your eval job has completed with return code 1.
001 | Traceback (most recent call last):
002 | File "<string>", line 4, in <module>
003 | OverflowError: too many digits in integer
why?
timid imp
!eval
from ctypes import c_void_p
class NoMore:
def __str__(self):
return "baby don't hurt me"
__repr__ = __str__
c_void_p.from_address(id(True)+8).value = id(NoMore)
what = love = object()
what is love
@timid imp :warning: Your eval job has completed with return code 0.
[No output]
F U
rugged sparrow
!e ```py
from ctypes import c_void_p
class NoMore:
def str(self):
return "baby don't hurt me"
repr = str
c_void_p.from_address(id(True)+8).value = id(NoMore)
what = love = object()
print(what is love)```
night quarryBOT
@rugged sparrow :white_check_mark: Your eval job has completed with return code 0.
baby don't hurt me
!e ~~~~True
night quarryBOT
@sick hound :warning: Your eval job has completed with return code 0.
[No output]
!e print(~~~~True)
night quarryBOT
@sick hound :white_check_mark: Your eval job has completed with return code 0.
1
i love python
night quarryBOT
@sick hound :white_check_mark: Your eval job has completed with return code 0.
1
@sick hound :white_check_mark: Your eval job has completed with return code 0.
-1
!e print(~~~True)
night quarryBOT
@sick hound :white_check_mark: Your eval job has completed with return code 0.
-2
!e print(~~~~-True)
night quarryBOT
@sick hound :white_check_mark: Your eval job has completed with return code 0.
-1
!e print(~~~~[True])
night quarryBOT
@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 | TypeError: bad operand type for unary ~: 'list'
!e print(~~~~bool([True]))
night quarryBOT
@sick hound :white_check_mark: Your eval job has completed with return code 0.
1
@sick hound :white_check_mark: Your eval job has completed with return code 0.
1
!e print(~~~~bool(..))
night quarryBOT
@sick hound :x: Your eval job has completed with return code 1.
001 | File "<string>", line 1
002 | print(~~~~bool(..))
003 | ^
004 | SyntaxError: invalid syntax
!e print(~bool('False'))
night quarryBOT
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
-2
!e print(~bool(False))
night quarryBOT
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
-1
!e @sick hound
a ,= [0x00for x in input('Hello World!')]
print(a)
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
15
Is that using hex? 0x0f
deft rock
So sneaky. Is that just a tuple of one on the LHS?
floral meteor
that's what the comma does, yes
deft rock
Python. The language where whitespace matters, until it doesn't..
floral meteor
pfft, whitespace can go die
Python. The language where whitespace doesn't matter, until it does.
snow beacon
refrains from advertising their guide on esoteric whitespace
viscid nymph
!e
from ctypes import py_object
class NoMore:
def __str__(self):
return "baby don't hurt me"
py_object.from_address(id(True)+8).value = NoMore
what = love = object()
print(what is love)
@viscid nymph :white_check_mark: Your eval job has completed with return code 0.
baby don't hurt me
cerulean osprey
wtf?
terse mortar
Ive never seen such a helpful error
sick hound
Lol
sick hound
Is there some esoteric trick to make an if statement without using “if”
proper vault
print(('more than 1', 'less than or equal to 1')[int(input())<=1])
proper vault
functions are probably the easiest for that
sick hound
If 5==5:
Print hi
Print help
Somedef()
proper vault
or tuple literals
sick hound
Can’t do it well on mobile
proper vault
there is also always
while condition:
code
break
else:
code
for _ in iter(lambda: bool(condition), False):
code
Bet ty ty
stmts = 'print("some")\nprint("statements")'
exec(stmts * bool(condition))
``` works I thinkbut not as elegant
!e ```py
for _ in iter(lambda: bool(1==1), False):
print(“hi”)
@sick hound :x: Your eval job has completed with return code 1.
001 | File "<string>", line 2
002 | print(“hi”)
003 | ^
004 | SyntaxError: invalid character '“' (U+201C)
@sick hound :x: Your eval job has completed with return code 143 (SIGTERM).
001 | hi
002 | hi
003 | hi
004 | hi
005 | hi
006 | hi
007 | hi
008 | hi
009 | hi
010 | hi
011 | hi
... (truncated - too many lines)
Full output: too long to upload
sick hound
!e ```py
for _ in iter(lambda: bool(1==1), False):
print("hi")
break
@sick hound :white_check_mark: Your eval job has completed with return code 0.
hi
you can use dict to a similar effect, unless you need fallthrough
toxic jewel
Are switch statements possibly in py?
yes
sick hound
Only in dicts though right
sick hound
Ah ok
toxic jewel
!e
def switch(obj):
v = {'owo': lambda: print('owo')}
v.get(obj)
switch('owo')()
@toxic jewel :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: 'NoneType' object is not callable
oh wait
!e
def switch(obj):
v = {'owo': lambda: print('owo')}
v.get(obj)()
switch('owo')
@toxic jewel :white_check_mark: Your eval job has completed with return code 0.
owo
called at wrong place
simple crystal
Now how do I do while without using while
Another way to use exec for it
!e
not_if = False
exec(not_if*"print('conditional 1')")
not_if = True
exec(not_if*"print('conditional 2')")
@simple crystal :white_check_mark: Your eval job has completed with return code 0.
conditional 2
oh someone did that
simple crystal
now that I sorta know haskell I rly can't wait for match case
sick hound
Without a dictionary ?
floral meteor
without a dictionary
floral meteor
although dictionary is considered the pythonic switch
but without dictionary
switch, case, default syntax
you call switch with whatever you want to switch, then use case and default as wrappers
you want me to write one up for you?
I'll just clipboard dump...
Oh yeah i get AttributeError for _File__p when i try to __p
alrighty then i'll write this up from memory and i'll try to refrain from ctypes hackery
sick hound
Ty
floral meteor
writing for test case:
switch(5)
case[4]((1).__floordiv__,0)
case[5](print,'success')
@case[6]
def do():raise SyntaxError
@otherwise
def do():return NotImplemented
try:case[5](print,'WARNING')
except:print("pass")
else:raise RuntimeError("WRONG")
@floral meteor that is because obj.__{attr_name} is mangled to obj.__{type(obj).__name__}_{attr_name}
when inside the class definition
floral meteor
mangled? I like mangled
rugged sparrow
!e ```py
class a:
def __test(self):
print('a')
print(dir(a))
a()._a__test()
@rugged sparrow :white_check_mark: Your eval job has completed with return code 0.
001 | ['__class__', '__delattr__', '__dict__', '__dir__', '__doc__', '__eq__', '__format__', '__ge__', '__getattribute__', '__gt__', '__hash__', '__init__', '__init_subclass__', '__le__', '__lt__', '__module__', '__ne__', '__new__', '__reduce__', '__reduce_ex__', '__repr__', '__setattr__', '__sizeof__', '__str__', '__subclasshook__', '__weakref__', '_a__test']
002 | a
!e ```py
def Switch():#pretend class
@lambda c:c() # instantiate this mf on the spot
class case:#pretend function
"""This is a docstring lol"""
mul=True # <- change this to disable multiple case matching. -<
def init(s):s.state,s.value =- 1,None
def getitem(s,v):
if s.state <- 0:raise RuntimeError('matching 'switch' not found')
elif(not s.state or s.mul)and v==s.value:return setattr(s,'state',1)or(lambda f,*u,**ck:f(u,**ck))
else:return(lambdaNo,**ne:None)
call=getitem
def switch(v):case.state,case.value =- 0,v
@lambda f:setattr(f,'name','default')or f
def otherwise(f,*u,**ck):
if case.state <- 0:raise RuntimeError('matching 'switch' not found')
if not case.state:case.state =- 1;return f(u,**ck)
else:case.state =- 1
return(lambdaNo,**ne:None)
return switch,case,otherwise
switch,case,otherwise = Switch()
switch(5)
case4
case5
@case[6]
def do():raise SyntaxError
@otherwise
def do():return NotImplemented
try:case5
except:print("pass")
else:raise RuntimeError("WRONG")
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
001 | success
002 | pass
SyntaxError: invalid character '' (U+20E3)
sick hound
YAASS
Holy that is stuff that I don’t even understand lmao
I got so much more py to learn
floral meteor
`SyntaxError: invalid character '' (U+20E3)`
wat
floral meteor
<@!599422361166413824> :white_check_mark: Your eval job has completed with retur...
what's this, chopped liver?
@snow beacon
It ran perfectly there is no syntaxerror
snow beacon
No, not that code. The one I replied to.
viscid nymph
<@!599422361166413824> that is because `obj.__{attr_name}` is mangled to `obj.__...
actually it's mangled to obj._{type(obj).__name__}__{attr_name}
rugged sparrow
Ah I flipped it
cerulean osprey
h="h"
def e(s):return s+"e"
def l(s):return s+"l"
def o(s):return s+"o"
def w(s):return s+"w"
def r(s):return s+"r"
def d(s):return s+"d"
def _(s):return s+" "
print(d(l(r(o(w(_(o(l(l(e(h)))))))))))
!e
night quarryBOT
Command Help
floral meteor
!e ```py
#final class
@lambda _0:_0()
class cout:
rgt=lt=print
rrshift=lshift=lambda _0,_1:print(end=_1)or _0
rge=le=lambda _0,_1:print(end=_1)or _1
def write(_0,_1):_1>>_0;return len(_1)
def test(_0):[
_0 << "Hello "<< "World"<'!',
"Hello ">> _0< "World!",
(("AAAAAAA" >= _0 )>> _0 <= "AAAAAA" )> _0,
cout.write("This works!")]
cout.test()
!e
h="h"
def e(s):return s+"e"
def l(s):return s+"l"
def o(s):return s+"o"
def w(s):return s+"w"
def r(s):return s+"r"
def d(s):return s+"d"
def _(s):return s+" "
print(d(l(r(o(w(_(o(l(l(e(h)))))))))))
@cerulean osprey :white_check_mark: Your eval job has completed with return code 0.
hello world
!e
h="h"
def e(s):return s+"e"
def l(s):return s+"l"
def o(s):return s+"o"
def w(s):return s+"w"
def r(s):return s+"r"
def d(s):return s+"d"
def _(s):return s+" "
print(d(l(r(o(w(_(o(l(l(e(h)))))))))))
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
001 | Hello World!
002 | Hello World!
003 | AAAAAAAAAAAAAAAAAAAAAAAAAA
004 | This works!
@cerulean osprey :white_check_mark: Your eval job has completed with return code 0.
hello world
!e
d="d"
def e(s):return "e"+s
def l(s):return "l"+s
def o(s):return "o"+s
def w(s):return "w"+s
def r(s):return "r"+s
def h(s):return "h"+s
def _(s):return " "+s
print(h(e(l(l(o(_(w(o(r(l(d)))))))))))
@cerulean osprey :white_check_mark: Your eval job has completed with return code 0.
hello world
#include<iostream>
int main(){std::cout<<"SSSS.DYNAZENON";return 0;}
!e import time
night quarryBOT
@cerulean osprey :warning: Your eval job has completed with return code 0.
[No output]
!e
import time
print("OwO");time.sleep(1);print("OwO")
@cerulean osprey :white_check_mark: Your eval job has completed with return code 0.
001 | OwO
002 | OwO
!e print(":pypi:")
night quarryBOT
@cerulean osprey :white_check_mark: Your eval job has completed with return code 0.
:pypi:
!e print("
")
night quarryBOT
@cerulean osprey :white_check_mark: Your eval job has completed with return code 0.
!e print("")
night quarryBOT
@cerulean osprey :white_check_mark: Your eval job has completed with return code 0.
@cerulean osprey this is starting to look like #bot-commands territory
cerulean osprey
<@!411831877985370114> this is starting to look like <#267659945086812160> terri...
In that case, let's switch to #bot-commands.
floral meteor
!e ```py
@lambda _0 :+ _0()
class cout:
rgt=lt=print
rrshift=lshift=lambda _0,_1:print(end=str(_1))or _0
rge=le=lambda _0,_1:print(end=str(_1))or _1
str=repr=lambda _:'std::cout'
pos=neg=invert=lambda _0:_0
_1 = lambda:'H'>=-cout
_2 = lambda:'e'>=-cout
_3 = lambda:'l'>=-cout
_4 = lambda:'o'>=-cout
_5 = lambda:' '>=-cout
_6 = lambda:'W'>=-cout
_7 = lambda:'r'>=-cout
_8 = lambda:'d'>>~cout<'!'
(*((lambda:None).class.call(_0)for _0 in(_1,_2,_3,_3,_4,_5,_6,_4,_7,_3,_8)),)
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
Hello World!
yas i made a new operator
sick hound
making an operator?
floral meteor
what happens when you bind an already bound function to another class?
!e ```py
@lambda _0:_0()
class _0:
rgt=lt=print
rrshift=lshift=lambda _0,_1:print(end=str(_1))or _0
rge=le=lambda _0,_1:print(end=str(_1))or _1
str=repr=lambda _:'std::cout'
pos=neg=invert=lambda _0:_0
@lambda _0:_0(1)
class _1(int):add=lambdaa:import('functools').reduce(int.add,a)
@lambda _0:_0(2)
class _2(int):add=_1.add
@int
@lambda _0:_0()
class _3:int=lambdaa:3
_0 <_2 .add(_3) # is it 5 or 6?
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
4
looks like i'll have to manually bind it
!e ```py
@lambda _0:_0()
class _0:
rgt=lt=print
rrshift=lshift=lambda _0,_1:print(end=str(_1))or _0
rge=le=lambda _0,_1:print(end=str(_1))or _1
str=repr=lambda _:'std::cout'
pos=neg=invert=lambda _0:_0
@lambda _0:_0(1)
class _1(int):add=lambdaa:import('functools').reduce(int.add,a)
@lambda _0:_0(2)
class _2(int):add=(lambda:None).class(_1.add.code,globals())
@int
@lambda _0:_0()
class _3:int=lambdaa:3
_0 <_2 .add(_3) # 6
@floral meteor :x: Your eval job has completed with return code 1.
001 | Traceback (most recent call last):
002 | File "<string>", line 11, in <module>
003 | File "<string>", line 11, in _2
004 | AttributeError: 'method-wrapper' object has no attribute '__code__'
!e ```py
print((int.mul.call.text_signature))
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
($self, /, *args, **kwargs)
dafuq is the $self?
simple crystal
dafuq is the $self?
__text_signature__ of a builtin method, which should
be in the following format: '($param1, ...)'.```how do you bind to multiple inputs?
i.e. ```py
def some_func($param1, $param2, $param3, /, *args, **kwargs_):
...
so why does a method not have a code attribute?
rugged sparrow
so why does a method not have a __code__ attribute?
Because only functions defined in python have code
floral meteor
But before it was bound, it wAS a function defined in python
rugged sparrow
No it wasn't
viscid nymph
!e
```py
d="d"
def e(s):return "e"+s
def l(s):return "l"+s
def o(s):return "o"+s...
not enough lambdas and decorators
fluid tree
yas i made a new operator
Annotations with unary plus?
sudden willow
@floral meteor how to custom operator owo
opaque hemlock
yea
sudden willow
!e
@lambda c: 1()
@lambda c: globals().__setitem__(1, lambda: print(2))
class c: ...
@sudden willow :x: Your eval job has completed with return code 1.
001 | <string>:1: SyntaxWarning: 'int' object is not callable; perhaps you missed a comma?
002 | Traceback (most recent call last):
003 | File "<string>", line 3, in <module>
004 | File "<string>", line 1, in <lambda>
005 | TypeError: 'int' object is not callable
:^(
astral rover
Gotta do a globals lookup
sick hound
Weirdest way to assign a string to a variable ?
viscid nymph
some way you would code it in another language
viscid nymph
!e
```py
@lambda c: 1()
@lambda c: globals().__setitem__(1, lambda: print(2))
cl...
!e
import ctypes
@lambda c: 1()
@lambda c: setattr(ctypes.py_object.from_address(id(1)+8),'value',type('int',(int,),{'__call__':lambda s:print(2)}))
class c: ...
@viscid nymph :white_check_mark: Your eval job has completed with return code 0.
001 | <string>:2: SyntaxWarning: 'int' object is not callable; perhaps you missed a comma?
002 | 2
before it was bound:
add=lambda*a:__import__('functools').reduce(int.__add__,a)
after it was bound:
~~```py
add=(lambda:None).class(_1.add.code,globals())
!e my point is ```py
x='code'in dir((a:=lambda s:None))
y='code'in dir(type("",(),{'a':a})().a)
print(x,y)
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
True False
i.e. a is a function with code, then i bind it to an object and now it's mangled
essentially meaning i can't abuse its code to double bind a function
without.... functionising it
here comes some beautiful hackery...
!e ```py
flooravg=lambda a,b:(a+b)//2
a = type("foo",(int,),dict(flooravg=flooravg))(5)
hack = lambda b:a.flooravg(b)
b = type("bar",(int,),dict(flooravg=hack))(9)
print(b.flooravg())#7
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
7
!e hol up ```py
flooravg = lambdaa:sum(a)//len(a)
a = type("foo",(int,),dict(matmul=flooravg))(5)
print(a@9)
b = type("bar",(int,),dict(matmul=lambdab:a.matmul(*b)))(9)
print(b.matmul(),b@5)
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
001 | 7
002 | 7 6
!e ```py
ex=lambda*a,**k:lambda c:c(a)
funcshinator=lambda m:lambdaa,**k:m(a,**k)
@ex()
def bind():
def decerater(bound_method,name=''):
def wraper(victim):
setattr(victim,name or bound_method.name,funcshinator(bound_method))
return victim
return wraper
return decerater
flooravg = lambdaa:sum(a)//len(a)
@ex(5)
@bind(flooravg,'matmul')
class derp(int):...
@ex(9)
@bind(derp.matmul,'matmul')
class woop(int):...
print(woop@5)#6
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
6
so essentially im passing 5 to a method bound to both derp and woop, which are instantiated with 5 and 9 respectively
snow beacon
Weirdest way to assign a string to a variable ?
class var(metaclass=lambda*a:str(a)):0
!e class hello_world(metaclass=lambda*a:str(a)):...
print(hello_world)
night quarryBOT
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
('hello_world', (), {'__module__': '__main__', '__qualname__': 'hello_world'})
It's a string, you see.
floral meteor
perhapses we could control it
!e ```py
@print
class var("XD",metaclass=lambda*a:str(a[1][0])):0
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
XD
Now that is cursed i like it
hmm
im gonna try too bootstrapinator this struct class
floral meteor
!e ```py
def matlabinator(__g={}):
class struct:
name='<struct>'
def getattribute(self,v):
try:return(f := lambda:object.getattribute(self,v))()
except AttributeError:setattr(self,v,self.class())
return f()
def setattr(self,n,v):
if isinstance(v,(lambda:None).class):
victim = import('ctypes').py_object.from_address(id(self)+8)
victim.value = type("struct",(victim.value,),{n:v})
elif isinstance(v,print.class):
victim = import('ctypes').py_object.from_address(id(self)+8)
victim.value = type("struct",(victim.value,),{n:lambda s,*a,**k:v(a,**k)})#bound builtin behaviour
else:object.setattr(self,n,v)
repr=lambda s:f"<struct object '{s.name}' len={len(dir(s))}>"
global gulag
try:g=__g or 1/0
except ZeroDivisionError as e:g=e.traceback.tb_frame.f_back.f_globals
hack=lambda o:import('ctypes').py_object.from_address(id(o)+8)
prisoner = hack(g).value
gulag = type('atexit',(),{'del':lambda s:setattr(hack(g),'value',prisoner)})()
rename=lambda name:lambda o:setattr(o,'name',name)or o
@lambda f:setattr(hack(g),'value',type("matlabinated_globals",(prisoner,),{f.name:f}))
def missing(self, key):
try:return getattr(self['builtins'],key)
except AttributeError:return(lambda s:self.setitem(self,key,s))(rename(key)(struct()))
g.update({'struct':struct})
return g
matlabinator(globals())
data.x, data.y = 5,6
std.cout.lshift=print
std.cout << data.xdata.y
@floral meteor :warning: Your eval job has completed with return code 139 (SIGSEGV).
[No output]
!e ```py
print(dir(type("",(),{})()))
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
['__class__', '__delattr__', '__dict__', '__dir__', '__doc__', '__eq__', '__format__', '__ge__', '__getattribute__', '__gt__', '__hash__', '__init__', '__init_subclass__', '__le__', '__lt__', '__module__', '__ne__', '__new__', '__reduce__', '__reduce_ex__', '__repr__', '__setattr__', '__sizeof__', '__str__', '__subclasshook__', '__weakref__']
!e ```py
class cout(metaclass=lambda*a:type(*a)()):lt=print
cout < "lol"
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
lol
!e ```py
def foo(n,a,d):
print(n)
return(*[defor e in a if e in d],)
class bar('four','two','zero',metaclass=foo):
four=lambda:4
two=lambda:2
zero=lambda:0
print(*bar,sep='')
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
001 | bar
002 | 420
!e ```py
class wat('wheeeeeee', 69, metaclass=print):self=print
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
wat ('wheeeeeee', 69) {'__module__': '__main__', '__qualname__': 'wat', 'self': <built-in function print>}
now i can make fake classes
floral meteor
!e class wat('Hello World',metaclass=lambda n,a,d:print(*a)):...
night quarryBOT
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
Hello World
!e ```py
def echo(a):
class wat(metaclass=lambda_:print(*a)):...
echo(420)
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
420
!e ```py
def execute(*a,**k):
def wraper(f):
class wat(*a,metaclass=lambda n,b,d:f(*b,**k)):...
return wat
return wraper
@print
@execute(69)
class n(int):...
@floral meteor :white_check_mark: Your eval job has completed with return code 0.
69
@hexed tangle exit dusint need to be called
sick hound
Can you write a valid python program that contains every printable ascii character except # exactly once 
floral meteor
no but i can make python exit without calling anything
floral meteor
🌶️
quaint hound
floral meteor
Python is difficult to golf, and makes it hard to onelinerise in python3
Other languages you can use the end keyword but python you need either newline or dedent
earnest wing
Can you write a valid python program that contains every printable ascii charact...
No, since you can't match ' or "
earnest wing
Python is difficult to golf, and makes it hard to onelinerise in python3
agree to disagree there
sick hound
No, since you can't match ' or "
good point
fluid tree
now i can make fake classes
Old but yes
fluid tree
Python is difficult to golf, and makes it hard to onelinerise in python3
It's easy lol (excluding generators and async)
If anyone has a oneline generator tell me
last locust
Form = namedtuple('Form', ['id', 'tutor_id', 'name', 'room'])
data = []
while query.next():
data.append(
Form(query.value(0), query.value(1), query.value(2), query.value(3))
)
```can anyone think of how to shorten this (aside from making the `data.append()` one line)?Was thinking of a list-comp type thing but don't think that's possible with while loops
earnest wing
two-arg iter might come in handy
something like [... for x in iter(query.next, falsey-value)]
last locust
data = [Form(x[0], x[1]...) for x in iter(query.next, 0)]```?Wellpy forms = [Form(x[0], x[1], x[2], x[3]) for x in iter(query.next, 0)]gives an error saying bool isn't subscriptable
earnest wing
You're not originally accessing the result of next, though?
so why would you do that here
last locust
something like `[... for x in iter(query.next, falsey-value)]`
What's the ... supposed to be here?
earnest wing
I was lazy to put something there
last locust
But what would it be
earnest wing
It'd be a Form(...) like in your snippet
last locust
Because I don't understand what you're suggesting
last locust
It'd be a Form(...) like in your snippet
But doing that gives an error
earnest wing
query.value(0), not x[0]
last locust
Oh yea 
forms = [Form(query.value(0), query.value(1), query.value(2), query.value(3)) for x in iter(query.next, 0)]```right so this works 👍forms = [Form(*(query.value(x) for x in range(4))) for _ in iter(query.next, 0)]```slightly shorterI think I'll stick with that
Thanks @earnest wing 👍
simple crystal
!e
import ctypes
for i in range (10):
print(a:=ctypes.pythonapi.PyType_GetSlot(ctypes.py_object(True), i))
@simple crystal :x: Your eval job has completed with return code 1.
001 | Traceback (most recent call last):
002 | File "<string>", line 3, in <module>
003 | SystemError: Objects/typeobject.c:3111: bad argument to internal function
I get:
28928
1296791640
0
0
0
743949472
1295814972
0
1298644416
1295474560
why is my True a heap object? 🤔
or at least being treated as one
False gives me that exception
rugged sparrow
@simple crystal PyType_GetSlot does not do any checking to ensure its argument is actually a Type object
simple crystal
oh, it says exception gets raised if uts not heap type
in docs
but somehow my implementation doesn't for True
rugged sparrow
ah 3.10 changed it
simple crystal
usin 3.9 for this
snow beacon
```py
forms = [Form(*(query.value(x) for x in range(4))) for _ in iter(query.nex...
I'm not sure you need the *( ).
last locust
Right yea I did do that originally and got this errorpy forms = [Form(query.value(x) for x in range(4)) for _ in iter(query.next, 0)] TypeError: <lambda>() missing 3 required positional arguments: 'tutor_id', 'name', and 'room'
It tries to convert to a namedtuple with just one value rather than waiting to get all 4
snow beacon
You could still probably do *map(query.value, range(4)) like Olivia suggested.
snow beacon
If that one doesn't work I'm more tired than I thought.
last locust
Right no that does work
It didn't work back when I was doing the wrong thing
forms = [Form(*(query.value(x) for x in range(4))) for _ in iter(query.next, 0)]
``````py
forms = [Form(*map(query.value, range(4))) for _ in iter(query.next, 0)]```shorter too 👍Thanks
drifting grove
With statements but they assign temporary variables
class __replacement__:__enter__=__exit__=lambda*a,**k:0
__replacement__=__replacement__()
def get_indent(line):
try:
if len(line)==0:return-1
indent=0
while line[indent]==' ':indent+=1
return indent-1
except:return 0
def with_shenanigans(func):
_next=False
queue=[]
code=__import__("inspect").getsource(func).split('\n')
header=[]
while code[0].strip().startswith('@'):
if not code[0].strip().startswith('@with_shenanigans'):header.append(code[0])
del code[0]
header.append(code[0])
del code[0]
final_code=header
j=0
for line in code:
if line.strip().startswith('with'):
x=list(line.strip()[5:])
del x[-1]
x=list(''.join(x).strip())
i=len(x)-1
var_name=[]
found_var_name=False
chars_checked=0
while i>=0:
if x[i]!=' ':
var_name.insert(0,x[i])
found_var_name=True
else:
if found_var_name:break
i +=~ 0
chars_checked -=~ 0
for i in range(chars_checked):del x[-1]
try:
while x[-1]in" as":del x[-1]
except:pass
expr=''.join(x)
var_name=''.join(var_name)
final_code.append(f'{" "*(get_indent(line)+1)}{var_name}={expr}')
final_code.append((' '*(get_indent(line)+1))+'with __replacement__:')
queue.append((get_indent(line),var_name))
else:
if _next:
final_code.append(_next)
_next=False
final_code.append(line)
try:
if get_indent(code[j+1])<get_indent(line):
_next=((queue[-1][0]+1)*' ')+'del '+queue[-1][1]
del queue[-1]
except:0
j-=~0
final_code.append("globals()['x']=x".replace('x',func.__name__))
final_code='\n'.join(final_code)
exec(final_code)
return globals().get(func.__name__)
@with_shenanigans
def test():
with "cake" as lie:
print(lie)
print(lie)# should give error. "lie" doesn't exist outside of with statement
test()
Link to try it (snekbox won't run it): https://trinket.io/python3/b59ea59b8d?runOption=run&showInstructions=true
rugged sparrow
that is technically doable without inspect using some trickery
rugged sparrow
gimme a sec, drafting it up
it works by adding an __exit__ and __enter__ method to object
drifting grove
that sounds like magic
rugged sparrow
!e ```py
from ctypes import *
import sys, dis
PyType_Modified = pythonapi.PyType_Modified
PyType_Modified.argtypes = [py_object]
queue = {}
def getclsdict(cls):
mapping = cls.dict
if isinstance(mapping, dict):
return mapping
return py_object.from_address(id(mapping) + 2 * sizeof(c_void_p)).value
def object_enter(self, *args):
frame = sys._getframe(1)
lasti = frame.f_lasti
f_code = frame.f_code
co_code = f_code.co_code
if co_code[lasti + 2] == dis.opmap['STORE_NAME']:
name = f_code.co_names[co_code[lasti + 3]]
queue[name] = self
return self
def object_exit(self, *args):
frame = sys._getframe(1)
loc = frame.f_locals
for name, value in queue.copy().items():
if value is self and name in loc and loc[name] is self:
del loc[name]
del queue[name]
obj_dct = getclsdict(object)
obj_dct['enter'] = object_enter
obj_dct['exit'] = object_exit
PyType_Modified(object)
with [1,2,3] as x:
print(x)```
night quarryBOT
@rugged sparrow :white_check_mark: Your eval job has completed with return code 0.
[1, 2, 3]
@drifting grove ^
doesnt quite work inside functions (deleting locals is weird)
itll store the variable but wont delete it after
rugged sparrow
it starts by getting the dictionary that supports object
then i define an enter and an exit func
then those are set inside the dictionary (because object.__enter__ = ... will fail)
PyType_Modified is a C function that informs all of object's subclasses to reinherit from its namespace
the result is that every object now has an __enter__ and an __exit__ method (and so we can use with on them)
simple crystal
nice
u can also just implement a context manager that takes any obj
!e
from contextlib import contextmanager
@contextmanager
def withy(obj):
yield obj
with withy([1,2,3]) as a:
print(a)
@simple crystal :white_check_mark: Your eval job has completed with return code 0.
[1, 2, 3]
oh it's not temporary
drifting grove
yeah
but that is shorter so it is probably better if you don't need it to be temporary
drifting grove
!e ```py
from ctypes import *
import sys, dis
PyType_Modified = pythonapi.PyTyp...
Oh
I just realized
I never properly thanked you for showing me that or for going to the trouble of making it
Thanks so much, and sorry about that
rugged sparrow
Nah you're all good, happy to show you
flat notch
!e ```py
from ctypes import *
import sys, dis
PyType_Modified = pythonapi.PyTyp...
Interesting stuff
flat notch
Nah you're all good, happy to show you
How did you come to know about this being possible, just curious about what you looked into that gave you this idea
rugged sparrow
I wanted to learn how python worked
so i read a lot of the source code
from there i learned how python code was excecuted by the C code
and how to modify it
flat notch
Ahh, I see
rugged sparrow
there are lots of interesting quirks of cpython that you can use to do tricky things
!pypi fishhook
night quarryBOT
rugged sparrow
^ that is my module for hooking into C classes
!e ```py
import gc
class magic:
def length_hint(self):
return 1
def iter(self):
for obj in gc.get_objects():
if type(obj) == tuple and len(obj) == 1:
try:1 in obj
except SystemError:
yield obj
break
weird = tuple(magic())
print(weird, weird[0] is weird)```
night quarryBOT
@rugged sparrow :white_check_mark: Your eval job has completed with return code 0.
((...),) True
^ that uses a bug in how tuples are made from iterators
(and how the cpython garbage collector works)
flat notch
from there i learned how python code was excecuted by the C code
Ooh, I suppose I'll start by looking into cpython got a better idea of what I'm dealing with
rugged sparrow
Ooh, I suppose I'll start by looking into cpython got a better idea of what I'm ...
https://github.com/chilaxan/pysnippets this repo has a bunch of my random experimentation (some of it has some comments)
flat notch
Aye, I'll check it out
forest harbor
import inspect
from unittest.mock import MagicMock
class LambdaVariable(MagicMock):
pass
class Λ:
def __getattribute__(self, name):
def make_lambda(expr):
code = inspect.stack()[1].code_context[0]
frame = inspect.stack()[1].frame
_, code = code.split(f'.{name}(', 1)
code = code[:find_first_unmatched_bracket(code)]
return eval(f"lambda {name}: {code}", frame.f_globals, frame.f_locals)
return make_lambda
def find_first_unmatched_bracket(s):
stack = []
for j,i in enumerate(s):
if i == ')' and not stack:
return j
elif i == '(':
stack.append(i)
elif i == ')':
stack.pop()
raise ValueError
λ = Λ()
x = LambdaVariable()
f = λ.x(x + 1)
print(f"{f(1) = }")
f = λ.x([x, x+1])
print(f"{f(1) = }")
f = λ.x(x + x.upper())
print(f"{f('asd') = }")
no more typing lambda
earnest wing
that's a good one!
forest harbor
wont run on the bot because of dynamic bullshit, but the output is the expected
f(1) = 2
f(1) = [1, 2]
f('asd') = 'asdASD'
I wonder whether it's possible to intercept a nameerror, avoiding needing a LambdaVariable
forest harbor
My first idea was to not even need teh λ.x to start with, just doing
x = LambdaVariable()
f = x + 1
assert f(1) == 2
But that's limited.
You can't resolve [x] into lambda x: [x], for example.
earnest wing
Unless you could hook into tuple/set/list/dict literal creation
Most likely through some c tricks
forest harbor
Yeah, but even if you could it would generate a lot of ambiguity.
x = LambdaVariable()
f = x + 1
g = x * 2
h = [f, g]
Is h meant to be a list of lambdas or a lambda that returns a list?
earnest wing
that's a design decision though, isn't it
forest harbor
I don't think so. Python is a dynamic language so both should be possible, but the language (or the expanded language i'm proposing here) offers no means of expressing both.
Still the limited one has it's uses. I do a lot of pandas with method chaining so it ends up having lambda sprinkled all over.
With some haskellisms I managed to turn
df = get_data()
df = df[df['A'] > 0]
df['E'] = df['B'] * df['C']
df['F'] = df['E'] > 0
df = df.groupby('F').max()
plot = df.plot()
df = df.sum()
plot2 = df.plot()
df = df * 2
into
df = LambdaVariable()
(
StateContainer({}, get_data())
>>_ (Query, "A > 0")
>>_ (Assign, E=df['B'] * df['C'])
>>_ (Assign, F=df['E'] > 0)
>>_ (Groupby, "F")
>>_ (GB.max)
<<_ (λ.df(StateContainer({"plot": df.plot()}, df)))
>>_ (Sum)
<<_ (λ.df(StateContainer({"plot2": df.plot()}, df)))
>>_ (df * 2)
)
function composition using __getattr__ is also possible
forest harbor
Yeah, but i'm not composing functions, i'm mapping functors. The functions Query, Assign, etc are just the regular DataFrame methods that I populated into the globals() for the lols, but otherwise unmodified.
earnest wing
yeah
forest harbor
the >>_ "operator" applies a function to the object inside the container, not to the container itselfd.
earnest wing
I admit I haven't done a lot with monads
that being said, bending python's syntax to your will like this is the kind of ideal I strive for
(for personal projects at least)
sick hound
Can someone get an item through DataFrame using a multi index? My df looks like user_id | item_id | rating . If I set the index as df.set_index(['user_id','item_id']) how can I access the respective rating? In a 2d array you d do something like df[user_id][item_id], what is the dataframe equivalent of that?
forest harbor
use a tuple df.loc[(uid, iid)]
sick hound
thanks
viscid nymph
!e ```py
from ctypes import *
import sys, dis
PyType_Modified = pythonapi.PyTyp...
How would you do this to make ints callable by the way 🤔
like
>>> 1(2)
2
near umbra
just getting into making weird python things
exec("""import pygame as A;\nZ=255;R=True;A.init();D=A.display;B=D.set_mode((Z,Z));\nwhile R:\n if A.event.get(eventtype=A.QUIT): exit()\n B.blit(A.font.SysFont('arial',25).render('hello world',R,(Z,Z,Z)),(0,0));D.flip()""")
``` how can I make this smaller?less spaces
viscid nymph
exec("""import pygame as A;\nZ=255;R=True;A.init();D=A.display;B=D.set_mode((Z,Z));\nwhile R:\n\tif A.event.get(eventtype=A.QUIT):exit()\n\tB.blit(A.font.SysFont('arial',25).render('hello world',R,(Z,Z,Z)),(0,0));D.flip()""")
slightly smaller
near umbra
ah, i didnt even think about using \t, thanks
forest harbor
can you from pygame import *? So then you can get rid of all A.
near umbra
I dont think the .display and .event would work then
forest harbor
I don't use pygame, but you should be able to use just display and event without A..
Unless they did some shananigans to avoid importing *
near umbra
i would assume exec would allow it, yes
forest harbor
Yeah, but what's the goal, make the string the smallest possible? Make the whole program small as possible?
You can probably get rid of R and replace it with 1 also.
near umbra
yeah pretty much just make it as small as possible
yeah but the render function takes a bool and idk if 1 will work
forest harbor
Unless it gets propagated all the way down to C, it should work.
near umbra
okay ill try that
forest harbor
But again, I don't use pygame, so the only way of knowing is trying.
near umbra
yep it works
forest harbor
replacing (z,z,z) with [z]*3 saves two chars, might not work though if it realy needs a tuple, so (z,)*3 saves one char.
near umbra
Yeah it needs to be a tuple
forest harbor
do it doesnt, i just tried
exec("""from pygame import *;\nZ=255;init();D=display;B=D.set_mode((Z,Z));\nwhile 1:\n\tif event.get(eventtype=QUIT):exit()\n\tB.blit(font.SysFont('arial',25).render('hello world',1,[Z]*3),(0,0));D.flip()""")
this works
rugged sparrow
How would you do this to make ints callable by the way 🤔
You would need to insert a function pointer into the tp_call slot on int
fishhook can do that all automatically
near umbra
```py
exec("""from pygame import *;\nZ=255;init();D=display;B=D.set_mode((Z,Z));...
Oh cool
late talon
what is lambda
late talon
!d lambda
night quarryBOT
ebon ingot
it's a way of making a function without naming it or using def
late talon
nope
well actually technically you can use ```py
class Hello:
print("hello World")
!e```py
class Hello:
print("hello World")
Hello
oh
rugged sparrow
well actually technically you can use ```py
class Hello:
print("hello World")
...
This is just running code inside the class creation scope
late talon
This is just running code inside the class creation scope
I didn't know that you could define a function without def
viscid nymph
I didn't know that you could define a function without def
you can define a function without def and without lambda
astral rover
not if you're faint of heart
viscid nymph
!e
from types import FunctionType
print(FunctionType(compile('print("Hello, World")', __name__, 'exec'), globals()))
@viscid nymph :white_check_mark: Your eval job has completed with return code 0.
<function <module> at 0x7fd8cbe90280>
!e
from types import FunctionType
my_func = FunctionType(compile('print("Hello, World")', __name__, 'exec'), globals())
my_func()
@viscid nymph :white_check_mark: Your eval job has completed with return code 0.
Hello, World
could do my_func = type(lambda:...)(...) but I guess that is with lambda
late talon
!e
```py
from types import FunctionType
print(FunctionType(compile('print("Hello...
this is cursed code
viscid nymph
well that's kinda how functions get initialized...
late talon
well that's kinda how functions get initialized...
interesting
viscid nymph
class function(code, globals, name=None, argdefs=None, closure=None)
!e
from types import FunctionType
print(FunctionType(compile('print(a)', __name__, 'exec'), globals(), 'my_function', ('a',)))
@viscid nymph :white_check_mark: Your eval job has completed with return code 0.
<function <module> at 0x7f60720e5280>
interesting
simple crystal
```py
exec("""import pygame as A;\nZ=255;R=True;A.init();D=A.display;B=D.set_mod...
take it out of exec, the only reason to use exec to make stuff smaller is if you're using an f string, concatenating a named long repeated command, or multiplying the string to loop it
fluid tree
from pygame import *
Z=255
init()
D=display
B=D.set_mode((Z,Z))
while 1:
if event.get(eventtype=QUIT):exit()
B.blit(font.SysFont('arial',25).render('hello world',1,[Z]*3),(0,0))
D.flip()
Much shorter when you remove exec and the escapes
near umbra
yeah but im trying to make it one line
chilly grotto
This is probably not the right room to ask but would it be proper to give a function a default argument of None so that I can do an if/else inside the function depending on whether or not an argument was passed?
Like this:
def my_function(x=None):
if !x:
do this
else:
print(x)
!d not
night quarryBOT
chilly grotto
So it would be fine?
snow beacon
The conventional way to check if something is None is using x is None.
viscid nymph
>>> not None
True
That's probably fine too.
snow beacon
Be wary of people passing in empty strings, 0, empty lists, and things like that.
viscid nymph
or even False
chilly grotto
damn. Is there a better way to implement that same idea? I feel like having the default argument as none makes the code confusing
oh
You just told me
I'm an idiot
Thank you
viscid nymph
lol
snow beacon
If you want to let people pass None as well, you'd create an object() and compare equality to that specific object.
rugged sparrow
yeah but im trying to make it one line
(lambda p,Z:(p.init(),d:=(p.display),(b:=d.set_mode((z,z))),*iter(lambda:exit() if p.event.get(eventtype=p.QUIT)else [b.blit(p.font.SysFont('arial',25).render('hello world',1,[z]*3),(0,0)),d.flip()],0)))(__import__('pygame'),255)```im not sure what the code is supposed to achieve tho
sick hound
Is there any function.__code__ attribute to check for variadic parameters?
def test(*args): ...
If I do test.__code__.co_argount it'll return 0, and I want to be able to detect the *args
snow beacon
The 64 bit of .co_flags, it seems.
snow beacon
!e ```py
h = lambdakwargs: 0
e = (lambda x: lambda: x)(None)
l = (lambda: lambdakwargs: 0)()
@lambda f: f()
def o():
def inner(**kwargs): yield inner
return inner
@lambda f: f()
def w():
def inner(*args): yield inner
return inner
def r(*args): yield from args
d = (lambda: lambda: 0)()
def reverse_bits(n): return int(f"{n:08b}"[::-1], 2)
def to_text(*fs): return bytes(reverse_bits(f.code.co_flags) >> 1 ^ 1 for f in fs).decode()
print(to_text(h,e,l,l,o), to_text(w,o,r,l,d))
@snow beacon :white_check_mark: Your eval job has completed with return code 0.
hello world
(Don't ask me to do capitals or punctuation.)
thin trout
(Don't ask me to do capitals or punctuation.)
Dewit
fluid tree
damn. Is there a better way to implement that same idea? I feel like having the ...
I use ...
acoustic gust
Do I recognize Arabic numbers?
distant wave
Perhaps
!e ```py
print(*(chr(int(a+b, 16)) for a, b in zip("٤٧٦f٦f٦٤٢٠٤d٦f٧٢٦e٦٩٦e٦٧٢c٢٠٤٦٦٥٦c٦c٦f٧٧٢٠٤٣٦١٧٤٧٣"[::2], "٤٧٦f٦f٦٤٢٠٤d٦f٧٢٦e٦٩٦e٦٧٢c٢٠٤٦٦٥٦c٦c٦f٧٧٢٠٤٣٦١٧٤٧٣"[1::2])))
@distant wave :white_check_mark: Your eval job has completed with return code 0.
G o o d M o r n i n g , F e l l o w C a t s
Discord, stop screwing with my rtl
fluid tree
yeah but im trying to make it one line
Well to begin with don't use """ or \t, and avoid \n. Do indent with only a single space.
terse mortar
I must share this beauty I found
!e
@__import__
@getattr('','',lambda f:f.__name__)
class __hello__:...
@terse mortar :white_check_mark: Your eval job has completed with return code 0.
Hello world!
oh, getattr as an identity function is smart
viscid nymph
!e
```py
@__import__
@getattr('','',lambda f:f.__name__)
class __hello__:...
```
this looks a lot like my code 
actually not sure
did you do that so it works before 3.9?
terse mortar
this looks a lot like my code <:thonk:785637987480371220>
Well i didn't say I made it lol
I just said I found it
viscid nymph
lol
thin trout
!e
```py
@__import__
@getattr('','',lambda f:f.__name__)
class __hello__:...
```
The getattr trick isn't needed in 3.9 anymore, right
terse mortar
The `getattr` trick isn't needed in 3.9 anymore, right
Probably not.
I could probably just do this with @glacial rampart
Whoops poor person lol
viscid nymph
I did the eval on a python3.8 bot so it was needed lol