is ast stuff on-topic for this channel?
#esoteric-python
gilded orchid
distant wave
on topic enough imo
crystal mica
Aha!py [*map(a.__rshift__, range(len(bin(a)) - 2))]
Making it more exotic
and shorter
is it shorter lol
zealous widget
you can initialize prev with a dummy loop
gilded orchid
how do I execute an ast tree after I've modified it?
bitter iris
compile() it and then exec()/eval() it
gilded orchid
is it possible to use ast to replace print@"Hello" with print("Hello")?
my current attempt is:
import ast
tree = ast.parse(input())
class test(ast.NodeTransformer):
def visit_BinOp(self,node):
if node==ast.MatMult:
return ast.call(func=node.left, args=[node.right])
exec(compile(ast.fix_missing_locations(test().visit(tree)), filename="<ast>", mode="exec"))
yes
proper vault
In [46]: curse(type(print), '__matmul__', lambda s,e: s(e))
In [47]: print@"Hello"
Hello
``` you can use forbiddenfruit.curse (though you have to do the same for `type(lambda:0)`
sick hound
that's different, you're messing with the objects at runtime
not a nerd is using the ast
gilded orchid
yeah I was trying to use ast
I changed it to if type(node.op)==ast.MatMult: and it still didn't work
nvm I got it working
import ast
tree = ast.parse(input())
class test(ast.NodeTransformer):
def visit_BinOp(self,node):
if type(node.op)==ast.MatMult:
return ast.Call(func=node.left, args=[node.right], keywords=[])
exec(compile(ast.fix_missing_locations(test().visit(tree)), filename="<ast>", mode="exec"))
is it possible to make a@b@c equivelent to a(b,c) using ast?
proper vault
maybe you could do partial on function if it takes more args than one, until it only takes one. Then you run into trouble with variadic functions though
gilded orchid
actually it'd probably be better to turn it into a(b(c))
earnest wing
Is it possible to make a class that when called given N, returns either an instance of the class or a tuple with N distinct instances of the class?
What I mean is:
a = Thing(1) # a is an instance of Thing
a, b = Thing(2) # a and b are distinct instances of Thing
I think you can mess with __new__ and metaclasses to do that but a better way would be using classmethods. Like Thing.from_multiple(3) etc. and then initializing 3 different copies of Thing
crystal mica
Alternatively you can do that via __mul__
!e ```py
class Thing:
def mul(self, multiplier):
return tuple(Thing() for _ in range(multiplier))
a = Thing()
print(a)
b, c = Thing() * 2
print(b, c)
night quarryBOT
@crystal mica :white_check_mark: Your eval job has completed with return code 0.
001 | <__main__.Thing object at 0x7f8832e729d0>
002 | <__main__.Thing object at 0x7f8832e09610> <__main__.Thing object at 0x7f8832e11e50>
I second isidential's idea of having a classmethods
earnest wing
I went with a classmethod, except I overwrote __class_getitem__() for maximum jank.
Ergo,
a = Thing[1]
a, b = Thing[2]
Since this is not a very serious use case, I can afford to mess with syntax a little.
And that's about it! My project is complete:
And here's this as text:
>>> from no_strings import BFInterpreter
>>> _,__,___ = BFInterpreter[3]
>>> _.interpret(+_[+__[___<<___<+___>>___>>___]+__<-__<-__<<__<+__<++__]<<_._<++_._<++_._._+++_._<<++_._<---_._>>_._>_._+++_._------_._>-_._>>--_._)
Source code: +[+[<<<+>>>>]+<-<-<<<+<++]<<.<++.<++..+++.<<++.<---.>>.>.+++.------.>-.>>--.
Output: Hello, World!
formal sandal
I think I made a thing like that in the past 🤔
But mine uses << and >> instead of < and >, so yours is superior.
Why are you using 3 different objects, though?
low acorn
l1, indices = [a >> i for i in range(len(bin(a)) - 2)], [i for i, s in enumerate(bin(a)[2:]) if s == '1']```
@crystal mica i know its yesterday's topic but the indices list u built is wrong, may be rather something like this python from math import ceil, log a = 7 print([a >> i for i in range(ceil(log(a, 2)))], [i for i in range(ceil(log(a, 2))) if (a >> i) % 2 == 0])
out : [7, 3, 1] []
your code would out : ([7, 3, 1], [0, 1, 2])
earnest wing
@formal sandal There's a cornercase where _>_>_[_>_] is parsed as >[>>] due to the method I used to fix comparison chains breaking things
The fix for the cornercase is to have a separate object for each layer of loops
cursive plover
And here's this as text:
@earnest wing
could you explain how that worked? Like how did those symbols even output smth, and why did they output Hello, World!
earnest wing
It's a brainfuck interpreter, but instead of taking source code as a string ("+-<>[],."), it takes source code as valid python expressions.
cursive plover
well how does that expression translate to Hello World
earnest wing
+_[+__[___<<___<+___>>___>>___]+__<-__<-__<<__<+__<++__]<<_._<++_._<++_._._+++_._<<++_._<---_._>>_._>_._+++_._------_._>-_._>>--_._
=>
"+[+[<<<+>>>>]+<-<-<<<+<++]<<.<++.<++..+++.<<++.<---.>>.>.+++.------.>-.>>--."
cursive plover
umm
snow beacon
It's brainfuck code.
earnest wing
^
cursive plover
i know @snow beacon but I'd like to understand it
But how do those symbols translate to letters
snow beacon
AFAIK the first parts in the brackets set up a bunch of cells to a certain amount close to the alphabet, then the rest offsets and outputs the values.
Every . is outputting one character.
cursive plover
Is there some reference for that? like how'd one even discover that?
I'm guessing by readig about the BFInterpreter?
earnest wing
https://esolangs.org/wiki/Brainfuck Here's how the language itself works.
cursive plover
I'm like curious, the guy who wrote that code originally, how did he actually do it
snow beacon
It might be on Rosettacode or something.
cursive plover
ah
Rosettacode?
oh it's a website I just checked
ah that's a very good article
snow beacon
The source of the program, I believe: https://codegolf.stackexchange.com/questions/55422/hello-world/68494#68494
cursive plover
explains the Hello World code too
earnest wing
Since all commands for BF are valid characters in Python source code, it's possible to make an implementation like this by using a class that overrides many of the builtin methods (x.__add__(y) for x + y, etc.)
crystal mica
@low acorn aha, that's a good catch!
marsh void
Why are you all doing this brainfuck! # coding trick exists!
cursive plover
@marsh void # coding tricks?
snow beacon
When you change the encoding of a file to a custom python decoder so that you can use whatever syntax you want.
You know how you can do #coding=utf8?
Or something similar, I don't know exactly.
You can define your own encoding.
marsh void
# coding: omg_braces
def main() {
print('What a twist!');
}
main();``` you can make this workscreaming
gilded orchid
depends on what y is I guess
analog rock
print(...)
gilded orchid
do you want to print something if a condition is true?
gilded orchid
print(a*x,end='') (if x is a boolean, and a is the thing you want to print)
analog rock
I swear that's longer
analog rock
😄
gilded orchid
it's longer by 1 char I'm, pretty sure
x and print(a) is probably the best solution
analog rock
yeah
gilded orchid
although there might be a better one in specific edge cases
analog rock
e.g.?
gilded orchid
if you were already printing without a newline so you already has ,end=''
for an example
analog rock
ah
strong hollow
probably been done before but still:
class PrintDict(dict):
def __getitem__(self, item):
print(f"get {repr(item)}")
return super().__getitem__(item)
def __setitem__(self, item, value):
print(f"set {repr(item)} = {repr(value)}")
return super().__setitem__(item, value)
class PrintMeta(type):
@classmethod
def __prepare__(cls, name, bases, **kwds):
return PrintDict()
class SomeClass(metaclass = PrintMeta):
foo = "bar" # yup, this prints
my_locals = locals() # this is actually a live copy of the namespace, observe:
my_locals["foo"] = "baz"
assert foo == "baz" # runs just fine
prints
get '__name__'
set '__module__' = '__main__'
set '__qualname__' = 'SomeClass'
set 'foo' = 'bar'
get 'locals'
set 'my_locals' = {'__module__': '__main__', '__qualname__': 'SomeClass', 'foo': 'bar'}
get 'my_locals'
set 'foo' = 'baz'
get 'foo'
for those too lazy to copy / paste
thin trout
That's pretty cool
analog rock
prints a blank line tho
gilded orchid
oh yeah nvm
analog rock
also my x is actually a set
not a bool
is there any shorthand for x.lower()?
cause it's longg
thin trout
Nope
gilded orchid
not really
analog rock
ooof
thin trout
I've seen worst tbh
gilded orchid
although if you use it multiple times you could put it in a variable (str.lower)
analog rock
nope
only once
but it's a substantial chunk of my 78 chars
import sys is quite a few actually
gilded orchid
could you send your code?
analog rock
import sys
for i in sys.argv:{*map(chr,range(97,123))}-{*i.lower()}or print(i)
prints each arg if it contains every letter in the alphabet
gilded orchid
could you just use regular input() instead?
gilded orchid
import sys
[print(i)for i in sys.argv if {*map(chr,range(97,123))}-{*i.lower()}]
nvm that's longer
analog rock
u can save 1 char with no whitespace after if
gilded orchid
isn't string.ascii_letters in both cases?
analog rock
?
gilded orchid
the set will always have uppercase letters in it
so it'll either never print or always print
analog rock
no cause ASCII 97-122 is just lowercase
gilded orchid
oh yeah nvm
couldn't you change it to map(chr,range(65,91)) then use .upper() instead?
1 char shorter
grizzled cloak
you could try the inverse:
not 2847-sum(map(ord,{*i.lower()})) will only be true if it uses all letters
analog rock
sadly that won't work if it has any characters like . or
which it does
it works if you filter with isalpha
but that's substantially longer
anyways I think I'd rather try to shorten it by myself now
thanks for the help though!
analog rock
shortest way to reverse digits of an int?
sick hound
str(i)[::-1] is pretty short
analog rock
trouble is I need it back as an int
sick hound
hmm
analog rock
which is longg
sick hound
int(str(i)[::-1])?
analog rock
slightly long tho
sick hound
i can't really think of any way to do it shorter
zealous widget
from math import ceil,log
def reverse(n):
return sum(n//10**j%10 * 10**k for j,k in enumerate(range(ceil(log(n, 10)) - 1, -1, -1)))
who needs strings
edgy kelp
people who want it short
zealous widget
didn't need the dummy first loop
now it's short
even shorter
can't do better than that
reverse=lambda n:sum(n//10**(d-1-j)%10*10**j for j in range(d))if(d:=ceil(log(n,10)))else 0
challenge: golf a sorting algorithm (without using the builtin)
(function/program that takes an array of numbers and sorts it)
def g(l):
p=0
while p<len(l):
if p==0 or l[p]>=l[p-1]:
p+=1
else:
l[p],l[p-1]=l[p-1],l[p]
p-=1
return l
here's a terrible and very golfable gnome sort for example
gilded orchid
sure
sick hound
wait a little bit
f=lambda a:(lambda b:f(a)if b!=sorted(b)else b)(__import__("random").sample(a,len(a)))
this is it
bogosort
gilded orchid
that uses the builtin though
'(without using the builtin)'
also bogosort isn't allowed because it's kinda cheating
how about: your algorithm has to be able to work with an array of 20 elements in under 10 seconds
gilded orchid
your solution still uses sorted(b) so it's invalid
edgy kelp
from random import shuffle
sort = lambda array: [array.insert(i,array.pop(array.index(min(array[i:]))))for i in range(len(array))]
arr = [*range(100)]
shuffle(arr)
sort(arr)
print(arr)
Don't like all the methods but fairly short
proper vault
def srt(a):
o=[]
while a:o+=[min(a)];a.remove(min(a))
return o
```by no means the shortest, but quite shortwhy didnt you do it oneliner
proper vault
Because it would be longer
sudden osprey
You might be able to shorten it using .pop()
proper vault
Need [], otherwise the += will not work
proper vault
srt = lambda a:[a.pop(a.index(min(a)))for _ in a[:]]
```turns out, oneline was the wayAh, clever
zealous widget
unfortunately remove is too many letters:
srt = lambda a:[b for _ in a[:]if a.remove(b:=min(a))or 1]
srt = lambda a:[a.remove(b:=min(a))or b for _ in a[:]]
```still longer, but not as muchis there any way to write numbers more concisely?
edgy kelp
compared to?
analog rock
writing them like 1800
edgy kelp
so you want 1800 in a more concise way?
analog rock
at least for relatively large numbers it would be shorter to write them as mathematical things
maybe not as small as 1800
edgy kelp
you have the mathematical notation for floats
analog rock
which is?
edgy kelp
>>> 1.8e3
1800.0
you can write numbers greater than 999999 in hex
analog rock
I mean that works
edgy kelp
could go hex or octal for something shorter
https://docs.python.org/3/reference/lexical_analysis.html#integer-literals your options with literals
analog rock
ta
vestal solstice
j has the coolest syntax I've seen
420blazeit -> 274765352357189
base in decimal, "b", the digits 0-9a-z
should be asymptotically short af
gilded orchid
you can do int(string,b) to convert a string from base b to decimal
vestal solstice
oh yeah, same thing doesn't go past 36 though
gilded orchid
also implementing that J base thing in ast could be cool
edgy kelp
you got me very confused with the j thought you meant python complex at first
gilded orchid
haha
edgy kelp
wondering how on earth that works
gilded orchid
I didn't even know python had complex numbers until I bruteforced all valid syntax 3 char strings
is there any golfing use for complex numbers?
edgy kelp
makes sense to include it considering python's use with data analysis
zealous widget
if you ever have a challenge that needs 2-d vectors
using complex numbers is often easier
proper vault
It probably does -(0+1j) -> (-0-1j)
zealous widget
In [64]: -0j.real
Out[64]: -0.0
In [65]: -0-0j.real
Out[65]: 0.0
the floating point standard brings that in
this line looks mildly esoteric
>>> -0.*-0.
0.0
The floating point standard peaked at Float3. It allows for:
NaN
-NaN
Inf
-Inf
0.0
-0.0
1.0
-1.0
!
snow beacon
I presume you've seen NaN Gates and Flip-FLOPS?
I've got to be honest, I hate floats.
I was pretty interested in posits for representing numbers a while back; the standard allows a 2-bit posit that can represent {0, 1, -1, complex infinity}.
To be honest I can't think of any other numbers anyone could want.
earnest wing
@marsh void #coding is not very modular for the (very common) scenario in which one wishes to interpret stringless bf in the middle of an otherwise normal script
cursive plover
n = int(input())
wave = list(map(int, input().split()))
up = True
down = False
for i in range(n - 1):
if up and wave[i] > wave[i+1]:
wave[i], wave[i+1] = wave[i+1], wave[i]
elif down and wave[i] < wave[i+1]:
wave[i], wave[i+1] = wave[i+1], wave[i]
up, down = down, up
print(wave)
How would you reduce the code to baar? I know it's essentially
(lambda n, wave, up, down : Code here) (int(input()), list(map(int, input().split())), True, False)
But like how do I get the whole for loop with the 2 conditions in baar
sick hound
you can remove the variables up and down and use i%2 instead
n = int(input())
wave = list(map(int, input().split()))
for i in range(n - 1):
if i%2==0 and wave[i] > wave[i+1]:
wave[i], wave[i+1] = wave[i+1], wave[i]
elif i%2==1 and wave[i] < wave[i+1]:
wave[i], wave[i+1] = wave[i+1], wave[i]
print(wave)```since wave[i], wave[i+1] = wave[i+1], wave[i] and wave[i], wave[i+1] = wave[i+1], wave[i] are the same you can turn that if/elif into just one condition
n = int(input())
wave = list(map(int, input().split()))
for i in range(n - 1):
if (i%2==0 and wave[i] > wave[i+1]) or (i%2==1 and wave[i] < wave[i+1]):
wave[i], wave[i+1] = wave[i+1], wave[i]
print(wave)```now to 1-line it...
actually let's 1-line wave[i], wave[i+1] = wave[i+1], wave[i] first since that's the hard part
cursive plover
oh wow, yeah
sick hound
(lambda x,y:(wave.__setitem__(i,x),wave.__setitem__(i+1,y)))(wave[i+1],wave[i])
and now...
(lambda n,wave:[(lambda x,y:(wave.__setitem__(i,x),wave.__setitem__(i+1,y)))(wave[i+1],wave[i]) for i in range(n-1) if (i%2==0 and wave[i] > wave[i+1]) or (i%2==1 and wave[i] < wave[i+1])])``` i think this should workwait no whoops
(lambda n,wave:[(lambda x,y:(wave.__setitem__(i,x),wave.__setitem__(i+1,y)))(wave[i+1],wave[i]) for i in range(n-1) if (i%2==0 and wave[i] > wave[i+1]) or (i%2==1 and wave[i] < wave[i+1])])(int(input()),list(map(int,input().split())))```wait can you explain how that setitem thing worked please?
sick hound
a[b]=c is the same as a.__setitem__(b,c)
__setitem__ is the dunder method that gets called when you assign an item in a list
technically it's not the same if you override __getattribute__ but most sane classes, including list, don't do that so yeah
(lambda x,y:(wave.__setitem__(i,x),wave.__setitem__(i+1,y)))(wave[i+1],wave[i])
cursive plover
so then, (lambda x,y:(wave.__setitem__(i,x),wave.__setitem__(i+1,y)))(wave[i+1],wave[i]) translates to
((lambda x,y: (wave[i] = x, wave[i+1] = y))(wave[i+1], wave[i]) yes?
sick hound
yes
which is (wave[i] = wave[i+1], wave[i+1] = wave[i])
well except for the fact that both of the arguments are evaluated first so it does a proper swap instead of setting wave[i] = wave[i+1] and then wave[i+1] = wave[i+1]
cursive plover
well why wouldn't it work if we just did it like
(lambda n,wave: wave[i], wave[i+1] = wave[i], wave[i+1] for i in range(n-1) if (i%2==0 and wave[i] > wave[i+1]) or (i%2==1 and wave[i] < wave[i+1]))(int(input()),list(map(int,input().split())))
is there a reason we had to use __setitem__ or was that just for illustration?
sick hound
wave[i], wave[i+1] = wave[i+1], wave[i] is a statement, not an expression
you can't put it in a lambda
it's invalid syntax
we had to use __setitem__ because function calls are an expression, which means you can put them in a lambda
it's the same reason that you use the (lambda x: ...)(input()) trick instead of just (x = input(), ...)
the second one is invalid syntax
cursive plover
oh right, I didn't know that
also, we're not printing wave at the end though are we
sick hound
oh good point
(lambda n,wave:([(lambda x,y:(wave.__setitem__(i,x),wave.__setitem__(i+1,y)))(wave[i+1],wave[i]) for i in range(n-1) if (i%2==0 and wave[i] > wave[i+1]) or (i%2==1 and wave[i] < wave[i+1])],print(wave)))(int(input()),list(map(int,input().split())))```oh we're allowed to put commas like that
sick hound
i used a tuple for the __setitem__ stuff too
cursive plover
uhm, one sec
[(lambda x,y:(wave.__setitem__(i,x),wave.__setitem__(i+1,y)))(wave[i+1],wave[i]) for i in range(n-1) if (i%2==0 and wave[i] > wave[i+1]) or (i%2==1 and wave[i] < wave[i+1])]
Why's the whole thing a list
like, wouldn't it be just valid to have it without the enclosing [ ] as well?
Like (lambda x,y:(wave.__setitem__(i,x),wave.__setitem__(i+1,y)))(wave[i+1],wave[i]) for i in range(n-1) if (i%2==0 and wave[i] > wave[i+1]) or (i%2==1 and wave[i] < wave[i+1])
cursive plover
uhm, yeah but well why do we need a list comprehension here
sick hound
to emulate the for loop
we can't put a normal for loop inside a lambda but we can put a list comprehension in and a list comprehension can be used as a for loop
>>> [print(x) for x in range(5)]
0
1
2
3
4
[None, None, None, None, None]```this is the same as py for x in range(5): print(x) but it's an expression so you can put it in a lambda
if you remove the [ ] around it then it's not a list comprehension anymore, it's just invalid and python won't like it
cursive plover
oh, that makes it very clear, but one more thing, the placement of the brackets
lemme break the inner lambda down
sick hound
(lambda x,y:(wave.__setitem__(i,x),wave.__setitem__(i+1,y))) this one?
cursive plover
yeah
the for loop is in the second part, yet we placed the brackets around the whole thing and not just the second part, what I mean is
sick hound
(lambda x,y:
(
wave.__setitem__(i,x),
wave.__setitem__(i+1,y)
)
)```hm yeah
so the second part
where the loop is
(wave[i+1],wave[i]) for i in range(n-1) if (i%2==0 and wave[i] > wave[i+1]) or (i%2==1 and wave[i] < wave[i+1])
shouldn't just that part be inside the list comprehension
(lambda x,y:(wave.__setitem__(i,x),wave.__setitem__(i+1,y)))[(wave[i+1],wave[i]) for i in range(n-1) if (i%2==0 and wave[i] > wave[i+1]) or (i%2==1 and wave[i] < wave[i+1])]
sick hound
what??
cursive plover
um, sorry
sick hound
why wouldn't (lambda x,y:(wave.__setitem__(i,x),wave.__setitem__(i+1,y))) be inside the list comprehension?
cursive plover
I mean, it's just the second part that has the for loop
sick hound
?!
look at the original code
for i in range(n - 1):
if (i%2==0 and wave[i] > wave[i+1]) or (i%2==1 and wave[i] < wave[i+1]):
wave[i], wave[i+1] = wave[i+1], wave[i]```(lambda x,y:(wave.__setitem__(i,x),wave.__setitem__(i+1,y))) is basically wave[i], wave[i+1] =
which is in the for loop
cursive plover
it's essentially two parts right, one is
(lambda x,y:
(
wave.__setitem__(i,x),
wave.__setitem__(i+1,y)
)
)
and the second
(wave[i+1],wave[i]) for i in range(n-1) if (i%2==0 and wave[i] > wave[i+1]) or (i%2==1 and wave[i] < wave[i+1])
no
(lambda x,y:(wave.__setitem__(i,x),wave.__setitem__(i+1,y)))(wave[i+1],wave[i]) is one thing
you can't separate (lambda x,y:(wave.__setitem__(i,x),wave.__setitem__(i+1,y))) from (wave[i+1],wave[i]) and have it still make sense
imagine if you went back to the original code and did this py wave[i], wave[i+1] = for i in range(n - 1): if (i%2==0 and wave[i] > wave[i+1]) or (i%2==1 and wave[i] < wave[i+1]): wave[i+1], wave[i]
cursive plover
erm but isn't the general structure like so:
(lambda name_of_variables: code)(variables)
so here, the code is (wave.__setitem__(i,x),wave.__setitem__(i+1,y))
right?
sick hound
wait hang on a second
sick hound
ok so, yes, that proves that i'm right ._.
(lambda x,y:(wave.__setitem__(i,x),wave.__setitem__(i+1,y)))(wave[i+1],wave[i])
x,y is the name_of_variables
(wave.__setitem__(i,x),wave.__setitem__(i+1,y)) is the code
wave[i+1],wave[i] is the variables
the general structure is (lambda name_of_variables: code)(variables), not (lambda name_of_variables: code)[(variables) for i in range(n-1)]
cursive plover
AH yes, so after the whole lambda expression do you have the for loop, which is why the whole thing becomes a list comprehension
sick hound
yes
(lambda x,y:(wave.__setitem__(i,x),wave.__setitem__(i+1,y)))(wave[i+1],wave[i]) is wave[i], wave[i+1] = wave[i+1], wave[i]
cursive plover
Ah, damn, yes
sick hound
which is inside the for loop (which became a list comprehension)
cursive plover
right, now all of it makes sense
so the structure will always be
[(lambda name_of_variables: code)(variables) for i in range(n-1)]
right?
cloud fossil
i've actually used this once in my code because hey it was a lot shorter than make a function and do so
cursive plover
and hence, that whole thing breaks down like so:
(
lambda n,wave:( #this bracket is so we get a tuple, to fit in print()
[(lambda x,y:(wave.__setitem__(i,x),wave.__setitem__(i+1,y)))(wave[i+1],wave[i]) for i in range(n-1) if (i%2==0 and wave[i] > wave[i+1]) or (i%2==1 and wave[i] < wave[i+1])] #The inner lambda expression
,print(wave)) #the tuple ends here
) #the lambda thing ends here
(int(input()),list(map(int,input().split()))) #and these are the variables
range(n-1) is only because that's what it was in the original code
and you're not required to have a lambda call in the list comprehension
cursive plover
yeah, I just copy pasted your sentence from above and didn't change it but yeah I get the idea
sick hound
the general structure for a list comprehension is [thing for thing in thing], and you can also have an if thing in there ([thing for thing in thing if thing])
and they don't all have to be the same i just couldn't be bothered to come up with more interesting placeholders
cursive plover
what do you mean I don't need to have the lambda call inside the comprehension? Didn't we just discuss why we needed that inside?
why wouldn't
(lambda x,y:(wave.__setitem__(i,x),wave.__setitem__(i+1,y)))be inside the list comprehension?
here
sick hound
in this specific case you do need it
but in general, it is possible to have a list comprehension which doesn't have a lambda inside it
i showed you an example earlier
[print(x) for x in range(5)]```this list comprehension isn't particularly useful because what it does is so ridiculously simple
but yeah, not all list comprehensions have lambdas
cursive plover
oh right, I read it the other way round
you meant that not all list comprehensions require a lambda, which is, well yeah
what I misread it as there could be lambda expressions which had their code inside a list comprehension but weren't called inside it
which well, is yes again true if I look at the outer lambda expression in the solution you wrote to my original question
sick hound
well technically you can do this py [(lambda x: x) for i in range(5)]
it's just silly
cursive plover
because that is basically (lambda n, wave: [code])(variables)
hm yeah
Right, I learnt a lot of things from that talk. Thanks a lot for breaking everything down into such a simple manner, that was beautiful. ❤️
marsh void
Meanwhile if I manage to do my bytecode modifying thing, we might be able to write actually nice multi line lambdas
cursive plover
bytecode? you mean from advent of code?
sick hound
yep
the interpreter doesn't directly run the source code, it turns it into bytecode and then executes that
analog rock
How could this be optimized:
for i in range(1800,2401,4):not i%50and i%16or print(i)
in terms of chars, I mean
not speed
burnt pasture
yes, Python is executed just like Java, though people call Java compiled and Python interpreted
cursive plover
for i in range(1800,2401,4):not i%50and i%16or print(i)
how's the interpreter able to interpret it right when there's no space before and and or
analog rock
cause it knows an identifier can't start with a digit
burnt pasture
the lexical analyzer sees "5" and starts building an int. It stops when it gets to "a" because that can't be part of an int.
cursive plover
oh wow, that's very clever
analog rock
yeet
cursive plover
@burnt pasture also, java has this jvm that makes it portable, python doesn't have that sort of thing yeah?
burnt pasture
@cursive plover Python has a VM also. It's not named something like "JVM", but it's the same idea
cursive plover
yeah I get that, I'm figuring out how they're the same
analog rock
cause if i%16!=0 then i%16>0
and we can replace the 0 with i%50
cause we need that to be 0
and then we just check if the i%50 is actually 0
cursive plover
by doing a <1, ah that's very nice
analog rock
I have recently come across a code golf technique: e.g.```py
"otnweo"[n::2]
In what cases is it beneficial to use this?
proper vault
whenever you need to decide between 2 strings of similar length based on a predicate
analog rock
or more than 2?
indigo sand
you could do something like (lambda n:[(print('zottffssenenwhoiieiireoruvxvgno ere ehe e nt '[int(i)::10])) for i in n])(input()) to print all of the digits in a number as words, or similar things for different scenarios.
proper vault
seldom do you a situation where the stars align so that it works
proper vault
what would be a case where you would get 0,1,2 and need print different things based on that?
vestal solstice
oh, maybe that's how primo did it
analog rock
🤷
I found a relatively concise version using that method and it's still 30th
primo moves in mysterious ways
sweaty boi
vestal solstice
it kinda ignores my submissions, they don't get scored or saved
maybe it's a browser thing
yeah, Safari 11.1.2 is no good
frozen ice
anyone know if there's somewhere to discuss stackless python?
vestal solstice
here? makes sense to me
frozen ice
makes sense to me too, but not sure if anyone else around would have used it before. 🙂
In particular, I'm having an issue with using stacklesssocket + boto3
and online resources seem to be rather scarce... so if any of you have some experience here, ding me in DM or hit me up in channel if you could.
vestal solstice
yeah it works in chrome
vestal solstice
b who
vestal solstice
sick skills
analog rock
afaik on python he's highest apart from primo
vestal solstice
true
analog rock
primo must literally spend 90% of every day doing these lol
vestal solstice
lmao
analog rock
when u minify ur code and it's a byte longer
spare tide
Every time I come to this server I look at this room in awe thinking about what I would have to do to be able to post into esoteric Python.
analog rock
what's the shortest way to sum the digits in a string?
edgy kelp
can't sum strings to get the codepoint sum so I'm guessing just a simple sum with a genexp
sick hound
what do you mean by "digits in a string"?
sick hound
if you int the string and then do %9 then (iirc) you get the correct result modulo nine
sick hound
but that probably isn't very useful- yeah
analog rock
I need modulo 2
sick hound
...i wonder if there's a similar trick for that
edgy kelp
count the ones
sudden osprey
.count('1') would be quite short but i reckon somebody else can do better
Aah beat me😁
indigo sand
is there a maximum length of the numbers?
analog rock
6
I mean if there's a way to do that without converting to binary in the first place
indigo sand
if they're 6 digit binary numbers, int()%9%2 should work and be a bit shorter than .count('1')
indigo sand
int of whatever string you have
sudden osprey
%9%6 maybe?
analog rock
%9%2 works
sudden osprey
But doesn't work with 111111
indigo sand
it gets 0
analog rock
still open to even shorter suggestions tho
any mathsy way to check if the number of 1s in a number's binary will be even?
indigo sand
this not very nice(and very long) bit of code should tell you if a number's binary will have an even number of ones:
import math
def m(n):
return bool(sum([int(n/(2**i))%2for i in range(int(math.log2(n)))])%2)
```of course, it doesn't work for 0, but this is the most mathsy I could get it to be. I'm sure someone could make it much shorter.
int(n/(2**i))%2
n//2**i%2
sick hound
n//2**i%2 isn't a solution, i'm just golfing pie42's code
sick hound
right now we have bool(sum([n//2**i%2for i in range(int(math.log2(n)))])%2)
that doesn't need to be a list comprehension bool(sum(n//2**i%2for i in range(int(math.log2(n))))%2)
we don't need bool() either honestly sum(n//2**i%2for i in range(int(math.log2(n))))%2
difficult to golf it more than that though
sick hound
iirc the current shortest way we've found is s.count('1')%2
zealous widget
this is basically the same as converting the number to binary and counting the ones
i mean pie's code, not jan's
there's no real shortcut besides that
old flint
what're some ways to replace an if __name__ == "__main__": main() with a single function call instead
I guess one option is to directly pass along __name__
is there some other hack to climb up the stack and pull out __name__?
zealous widget
(lambda name: (main if name == "__main__" else lambda:None)())(__name__)
old flint
ah, to clarify, the goal is to have an importable function that could be called
e.g.
from lib import run_only_if_main
run_only_if_main()
or more usefully
from lib import run_only_if_main
def main():
pass
run_only_if_main(f=main)
rugged sparrow
run_only_if_main = lambda f:globals()['__name__'] == '__main__' and f()```run_only_if_main(main)
old flint
right, but you'd still have to pass in __name__
rugged sparrow
honestly i think that would work
old flint
if run_only_if_main is defined in another file, it'll still pull __name__ as the imported module name
rugged sparrow
hmm
zealous widget
decorate it
rugged sparrow
@zealous widget __name__ is still pulled as the imported module tho
zealous widget
so
rugged sparrow
so how can the decorator check if the module it was imported into is main?
zealous widget
the decorator isn't imported
rugged sparrow
the goal is something that can be imported by itself
cursive plover
for i in range(51):int(f'{i:b}')%9%2or print(i)
@analog rock why do we have a %2 there?
cursive plover
Also, the %9 trick to get sum of digits in a number would fail if the number is a multiple of 9, no?
low acorn
if
run_only_if_mainis defined in another file, it'll still pull__name__as the imported module name
@old flint```python
#in your lib utils
run_only_if_main = lambda f,g:g['name'] == 'main' and f()
#in the caller
from lib.utils import run_only_if_main
def main():
print("i am the main code")
run_only_if_main()(f=main,g=globals())
nothing magical but at least you avoid typing the annoying '__name__'
apparently you could make use of inspect but its dependent of the python implementaion so not suitable for production
old flint
yea, I guess that's what I'd be looking at
wondering if there's some hacky solution to look for __name__ where I want it
probably it'll be too finicky for me to use in any actual code but was curious if there was any obvious strategy I'm missing
brazen geyser
walk up the stack til you find the calling frame, check __name__ in its f_globals
in this case, you'd only need to go 1 level up the stack i suppose
if sys._getframe(1).f_globals.get('__name__', None) == '__main__':
@old flint
@low acorn how often do you see python builds that cpython based in production?
that aren't*
old flint
haha that might be it. Yea, I'll be using the assumption that I'm exactly one import away
analog rock
can I do +=1 on a string?
to make it go to the next ascii value?
cause that'd be niceee
brazen geyser
no
analog rock
didn't think so tbh
analog rock
probably hackery that is too long for my golf
distant wave
You'd be able to save ~9 chars by taking advantage of lower chars, probably
but that requires the chr(ord( so saves nothing
analog rock
yeah that's very longg
cursive plover
So I had this horrendously long code
redacted code (link here: https://rextester.com/ORGD62976)
which I managed to reduce to
x = bin(int(input()))[2:]
even = x[::2].count('1') if len(x)%2 == 0 else x[1::2].count('1')
odd = x[1::2].count('1') if len(x)%2 == 0 else x[::2].count('1')
y = ""
for i in range(len(x)):
y += ("0" if i%2 == 0 else x[i]) if ((odd < even and len(x)%2 != 0) or (odd >= even and len(x)%2 == 0)) else (x[i] if i%2 == 0 else "0")
print(int(y,2))
now if someone could help me baar this:
y = ""
for i in range(len(x)):
y += ("0" if i%2 == 0 else x[i]) if ((odd < even and len(x)%2 != 0) or (odd >= even and len(x)%2 == 0)) else (x[i] if i%2 == 0 else "0")
then, I could get the whole program to be a baar thing
(lambda x, even, odd: (The baar code, print(int(y,2)))(input stuff)
so how do I get those three statements into a baar? :/
thin trout
A list comp and a join?
cursive plover
oh wait yeah, lemme try that
thin trout
"".join(("0" if i%2 == 0 else x[i]) if ((odd < even and len(x)%2 != 0) or (odd >= even and len(x)%2 == 0)) else (x[i] if i%2 == 0 else "0") for i in range(len(x))) ```y = [("0" if i%2 == 0 else x[i]) if ((odd < even and len(x)%2 != 0) or (odd >= even and len(x)%2 == 0)) else (x[i] if i%2 == 0 else "0") for i in range(len(x))]
yep!
x = bin(int(input()))[2:]
even = x[::2].count('1') if len(x)%2 == 0 else x[1::2].count('1')
odd = x[1::2].count('1') if len(x)%2 == 0 else x[::2].count('1')
y = [("0" if i%2 == 0 else x[i]) if ((odd < even and len(x)%2 != 0) or (odd >= even and len(x)%2 == 0)) else (x[i] if i%2 == 0 else "0") for i in range(len(x))]
print(int(("".join(y)),2))
bingo! thank you! so my final code would be this:
python (lambda x, even, odd: print(int(("".join([("0" if i%2 == 0 else x[i]) if ((odd < even and len(x)%2 != 0) or (odd >= even and len(x)%2 == 0)) else (x[i] if i%2 == 0 else "0") for i in range(len(x))])),2))) (bin(int(input()))[2:], x[::2].count('1') if len(x)%2 == 0 else x[1::2].count('1'), x[1::2].count('1') if len(x)%2 == 0 else x[::2].count('1'))
wait that gives me an error: NameError: name 'x' is not defined
Nevermind, that lambda conversion failed.
How do i convert the code into a lambda expression
Okay I got it, I got rid of even and odd and did it like this:
(lambda x: print(int(("".join([("0" if i%2 == 0 else x[i]) if (((x[1::2].count('1') if len(x)%2 == 0 else x[::2].count('1')) < (x[::2].count('1') if len(x)%2 == 0 else x[1::2].count('1')) and len(x)%2 != 0) or ((x[1::2].count('1') if len(x)%2 == 0 else x[::2].count('1')) >= (x[::2].count('1') if len(x)%2 == 0 else x[1::2].count('1')) and len(x)%2 == 0)) else (x[i] if i%2 == 0 else "0") for i in range(len(x))])),2)))(bin(int(input()))[2:])
I wonder if there's a better way to write that :/
sick hound
x = bin(int(input()))[2:]
even = x[::2].count('1') if len(x)%2 == 0 else x[1::2].count('1')
odd = x[1::2].count('1') if len(x)%2 == 0 else x[::2].count('1')```you were using x to create even and odd
sick hound
so you would need to do py (lambda x: (lambda even, odd: ...)(<define even>, <define odd>))(<define x>)
cursive plover
@sick hound well, then I did
(lambda x:
(lambda even, odd:
print(int(("".join([("0" if i%2 == 0 else x[i]) if ((odd < even and len(x)%2 != 0) or (odd >= even and len(x)%2 == 0)) else (x[i] if i%2 == 0 else "0") for i in range(len(x))])),2))))
(x[::2].count('1') if len(x)%2 == 0 else x[1::2].count('1'), x[1::2].count('1') if len(x)%2 == 0 else x[::2].count('1'))
(bin(int(input()))[2:])
@sick hound it's not fixed but I still get name error
wait
(lambda x:
(lambda even, odd:
print(int(("".join([("0" if i%2 == 0 else x[i]) if ((odd < even and len(x)%2 != 0) or (odd >= even and len(x)%2 == 0)) else (x[i] if i%2 == 0 else "0") for i in range(len(x))])),2)))
(x[::2].count('1') if len(x)%2 == 0 else x[1::2].count('1'), x[1::2].count('1') if len(x)%2 == 0 else x[::2].count('1')))
(bin(int(input()))[2:])
is the fixed version
AND THAT WORKS!
Oh yeah! thank you
not pretty but yeah xd
x = f'{int(input()):b}'
@analog rock I know the f is for f string, but I googled that whole thing and I couldn't find much. How's that managing to convert the input to binary? What would I have to change b to get it into other bases? What should be the keywords I should be googling for this? I didn't find it in the docs either: https://docs.python.org/3/reference/lexical_analysis.html#f-strings
cursive plover
what is the :b thing called
analog rock
no idea I'm afraid
cursive plover
ah, alright
cursive plover
oh great, thank you ! ❤️
low acorn
@low acorn how often do you see python builds that cpython based in production? that aren't*
@brazen geyser no idea, i just repeated the piece of advice i found on stackoverflow about inspect or even sys.get_frame , the guy seemed to have a solid stacko reputation so i trusted him as i have yet to use python in a professional context for now .
low acorn
from the doc https://docs.python.org/3/library/inspect.html#inspect.currentframe or https://docs.python.org/3/library/sys.html?highlight=_getframe#sys._getframe CPython implementation detail: This function should be used for internal and specialized purposes only. It is not guaranteed to exist in all implementations of Python.
proper vault
Thing is, very few companies use python that is not CPython
low acorn
i see
bitter iris
The bad thing about implementation details isn't only avability but also backwards compability. It is easy to do backwards-incompatible changes on implementation details rather then language specs (which is really rare)
brazen geyser
re: availability i think jython, ironpython and stackless all have _getframe and inspect. micropython and circuitpython probably dont
oh yeah pypy has them i think
marsh void
Oh that’s neat
I don’t think they have gc tho
since like, PyPy does not even have a GC
gilded orchid
This page is about pypy-stm, a special in-development version of PyPy which can run multiple independent CPU-hungry threads in the same process in parallel. It is a solution to what is known in the Python world as the “global interpreter lock (GIL)” problem — it is an implementation of Python without the GIL.
wow that's kinda insane
bitter iris
there is bunch of europython talks about that subject
it is a pretty cool concept
analog rock
is there a more concise way to say x>y<1
gilded orchid
what's y? (can it be a float or negative?)
cursive plover
lol I wrote that
with jan's and akarys' help
I think I posted the normal long program before it which is baar here?
proper vault
You start with a more reasonable program and work your way down
Following some basic rules along with some creativity
analog rock
x and y are ints
vocal nest
@cursive plover what does that function do?
cursive plover
@vocal nest
okay so what that program does is:
You're given a number, which you are to convert to binary.
Then in the binary form, you need to check number of 1s at 'even' and 'odd' positions(the rightmost digit is at position 1, then the digit towards the left of it at position 2, so on..)
if number of ones at the 'odd' positions are less than those at 'even' positions, then make the numbers at odd positions 0.
Otherwise, make the numbers at even positions 0.
after that, print the resultant number
cursive plover
@vocal nest it would be 5
vocal nest
because its an odd right?"
cursive plover
at position 1 you have a 1,
at position 2 you have a 0
at position 3 you have a 1
vocal nest
yeayea
cursive plover
so 1s at odd positions are greater than those at even positions, so make everything at even positions 0
(in this case, everything is already 0 at even positions)
so it remains 101, which is 5
if it was 12
cursive plover
it would bw 1100, which would become 0100 which is 4
cursive plover
yup
vocal nest
Very cool, how are you using it?
cursive plover
erm, just came across it somewhere
someone was solving these algorithmic practice questions and she showed me this
vocal nest
oh right
cursive plover
i thought it would be good practice to hone my baar coding skills
vocal nest
yea thats awesome
cursive plover
since I'm very new to it
cursive plover
lakmatiol here introduced it to me
vocal nest
awesome
proper vault
someone made this wonderful factorial function
assert (lambda f: (lambda x: x(x))(lambda y: f(lambda x: y(y)(x))))(
lambda f: lambda n: (
lambda n: n(lambda f: lambda x: lambda y: y())(lambda x: lambda y: x())
)(n)(lambda: lambda f: lambda x: f(x))(
lambda: (lambda x: lambda y: lambda f: y(x(f)))(n)(
f(
(
lambda n: (lambda p: p(lambda x: lambda y: y))(
n(
(
lambda p: (lambda a: lambda b: lambda s: s(a)(b))(
(lambda n: lambda f: lambda x: f(n(f)(x)))(
(lambda p: p(lambda x: lambda y: x))(p)
)
)((lambda p: p(lambda x: lambda y: x))(p))
)
)(
(lambda a: lambda b: lambda s: s(a)(b))(
lambda f: lambda x: x
)(lambda f: lambda x: x)
)
)
)(n)
)
)
)
)(lambda f: lambda x: f(f(f(x))))(lambda x: x + 1)(0) == 6
#---------------------^ input here as a church numeral
Lambdas are amazing
But I dont rly get them
never used them although I know perfectly how they work
proper vault
they are not that useful in actual code
cursive plover
what's x(x)
proper vault
I mostly use them to do dumb things like integer math without integers
call x with x as an argument
cursive plover
did the person who write that also explain any parts of it?
I don't quite get how that works
proper vault
it is just lambda calculus
cursive plover
I've never heard of that term before
proper vault
I would suggest reading this https://arxiv.org/pdf/1503.09060.pdf
vocal nest
it is abstracted calculus*
proper vault
it is quite unrelated to calc stuff like derivatives and intergrals
y'know how you have turing machines as the basis for computing. Lambda calculus is an alternative description of such machines, sometimes nice. e.g. erlang, haskell are based on it
cursive plover
right, thank you for the resource! I'll get onto it
proper vault
I generally use operator.itemgetter for this
but key kwargs are one of the uses for lambdas
vocal nest
😄
calm rampart
i wish python had arrow lambdas
analog rock
what's the difference?
calm rampart
shorter
proper vault
it would make me not feel like a failure for using a lambda
edgy kelp
That's exactly what guido wanted you to feel 😄
edgy kelp
realised we can do this
In [2]: from types import SimpleNamespace
In [3]: a=SimpleNamespace()
In [4]: setattr(a, "1()", "Magic string")
In [5]: "{0.1()}".format(a)
Out[5]: 'Magic string'
privat-er vars
snow beacon
It'll be in there somewhere.
cursive plover
@edgy kelp what is the 1() doing there in setattr
snow beacon
Does SimpleNamespace have a custom __setattr__?
edgy kelp
Just an object that can have instance vars.
Ought to work on them all, but it's a var only accessible through that, getattr and if there's anything else that fetches from strings
bitter iris
my little bot learned how to parse python code, lol
bitter iris
yea ast.parse + ast.dump with new indent = 4
sick hound
a = lambda n: for i in range(n): print(f"You have been funny for {i} times!");a(1000)
Condense this
Please
proper vault
for i in range(1000):print('You have been funny for',i,'times!')
wide depot
a = lambda n: for i in range(n):print('You have been funny for',i,'times!');a(100)
(lambda n: [*map(print, map("You have been funny for {} times!".format,range(n)))])(1000)```hmm
marsh void
(lambda n: any(map(print, map("You have been funny for {} times!".format, range(n)))))(1000)``` @crystal mica(lambda n:[0 for i in range(n)if print('You have been funny for',i,'times!'])(1000)```(f := lambda n: (n > 0 and print(f"You have been funny for {n} seconds") or f(n-1)))(100)
(lambda n:[0for i in range(n)if print('You have been funny for',i,'times!'])(1000)```lakmatiol left a space in
formal sandal
Why not just
(lambda n:[print('You have been funny for',i,'times!')for i in range(n)])(1000)
Who needs a list of 1000 0's?
proper vault
the list is empty, because if print(... is always false
formal sandal
oh
Well, this is 1 character shorter:
(lambda n:any(print('You have been funny for',i,'times!')for i in range(n)))(1000)
If you fix the unmatched (
sick hound
why not py any(print('You have been funny for',i,'times!')for i in range(1000))
or py for i in range(1000):print('You have been funny for',i,'times!') if we're allowing it to not be an expression
or py [print('You have been funny for',i,'times!')for i in range(1000)]
proper vault
I have no idea
distant wave
(lambda n:print(('You have been funny for {} times!\n'*n).format(*range(n))))(1000)
potent fulcrum
Hey on the topic of esoteric python / code-gore. What's anyone's favorite abuse of __init_subclass__ and metaclasses (or shudder using both at once)?
bitter iris
I used it to re-initalize mutable class defaults once
distant wave
derp
formal sandal
(lambda n:print('You have been funny for %s times!\n'*n%(*range(n),)))(1000)
(lambda n:print(('You have been funny for %s times!\n'*n%(*range(n),))[:-1]))(1000)
for correct new lines
robust palm
i=0;exec("print(f'You have been funny for {i} times');i+=1;"*1000)```exec and ; are cheating!
brisk zenith
shush! their wives might hear if you scream that too loudly.
formal sandal
Check this out.
https://repl.it/@int6h/DataDrivenTesting
It's my new data-driven unit testing framework!
Suppose you have some code that you need to test, and that code consists of modules that export something.
Then you can create a 'mirrored' structure of folders that will be used to test these modules.
For example, if you want to test this file:
You will create a structure like this:
So each export corresponds to a test file
Tests can be 'imperative'
Or you can specify them as a data structure
And it automagically tests everything
ripe depot
Has anyone here used the usbserial4a package?
marsh void
Soo? :)
bitter iris
first hint: PyRun_InteractiveOneObjectEx and the oenc
marsh void
cursive plover
So I'm trying to bar this piece of code now, but it's got try catch in it so idk how I'd lambda that:
~~```python
a = input()
num = int(input())
ans = []
i = 0
j = i
while True:
try:
form = [int(char) for char in a[j:i+1]]
if sum(form) == num:
ans.append("".join(str(char) for char in form))
if (sum(form) + int(a[i+1])) <= num:
i += 1
elif (sum(form) - int(a[i])) <= num:
j += 1
else:
j = i
i += 1
except:
break
print(*ans)
What the program does btw is, given a string say `9390215671237812390`, and a number `9`, find all substrings whose sum equals `9`. The output in this case would be `9 9 90 81 9 90`actually, I got rid of the try catch, but it's uglier now
a = input()
num = int(input())
ans = []
i = 0
j = i
while i < len(a):
form = [int(char) for char in a[j:i+1]]
if not form:
i += 1
continue
if sum(form) == num:
ans.append("".join(str(char) for char in form))
if i+1 < len(a) and (sum(form) + int(a[i+1])) <= num:
i += 1
elif (sum(form) - int(a[i])) <= num:
j += 1
else:
j = i
i += 1
print(*ans)
still unsure how I would baar this, especially with that while thing in it
Any help? :/
crystal mica
Well, try to simplify it first imo, then you can do it
The simplest, straightforward is like this
!e ```py
string = '9390215671237812390'
def get_sum(s: str) -> int:
return sum(map(int, s))
def generate_all_substrings(s: str) -> list:
substrings = []
for length in range(1, len(s)):
for start in range(0, len(s) - length):
substrings.append(s[start:start+length])
return substrings
print([substr for substr in generate_all_substrings(string) if get_sum(substr) == 9])```
night quarryBOT
@crystal mica :white_check_mark: Your eval job has completed with return code 0.
['9', '9', '9', '90', '81']
This is easy to put into one line
!e py (lambda s,total:print([_s for _s in(s[i:i+l]for l in range(1,len(s))for i in range(len(s)-l))if sum(map(int,_s))==total]))('9390215671237812390',9)
night quarryBOT
@crystal mica :white_check_mark: Your eval job has completed with return code 0.
['9', '9', '9', '90', '81']
There we go
The idea of how you approach it is very important
Missing a 1
(lambda s,t:print([_s for _s in(s[i:i+l]for l in range(1,len(s))for i in range(len(s)-l+1))if sum(map(int,_s))==t]))('9390215671237812390',9)```@crystal mica :white_check_mark: Your eval job has completed with return code 0.
['9', '9', '9', '90', '81', '90']
There we go
marsh void
May I question why
Ah and also
@crystal mica I am currently working on braces in python
cursive plover
@crystal mica hm yeah, that's nice
but
isn't that O(n^2) while mine is essentially O(n)?
also
this does not maintain the order in which the substrings occur in the string, I forgot to mention that is a requirement too
how would you alter the code for that :/
crystal mica
Ah I see, there are adjustment that you can do to prevent that, for example stop when substr is greater than the t
@marsh void that sounds very bracely!
marsh void
Indeed it does!
cursive plover
you mean if sum(substring) > total, don't append?
marsh void
(lambda x,y:{for x in range(x){for y in range(y){print(x,y)}})(10,10)``` @crystal micahehe
Ah, AND! Top-level awaits on < 3.8!
crystal mica
I mean when sum(substring) > t then exit the inner loop, takewhile is a good candiate for that
!e py (lambda s,t:print([_s for _s in(s[i:j+1]for i in range(len(s))for j in range(i,len(s)))if sum(map(int,_s))==t]))('9390215671237812390',9)
night quarryBOT
@crystal mica :white_check_mark: Your eval job has completed with return code 0.
['9', '9', '90', '81', '9', '90']
Changing the order of the loop -> preserve the order of finding
!e py (lambda s,t:print([_s for _s in(s[i:j+1]for i in range(len(s))for j in __import__('itertools').takewhile(lambda j:sum(map(int,s[i:j+1]))<=t,range(i,len(s))))if sum(map(int,_s))==t]))('9390215671237812390',9)
night quarryBOT
@crystal mica :white_check_mark: Your eval job has completed with return code 0.
['9', '9', '90', '81', '9', '90']
It's very handy!
cursive plover
thanks, I'm reading through it at the moment!
I'll ask you again when I don't understand smth in that solution, or in takewhile
❤️
thin trout
I was wondering, is there any way to make this true?
(obj1 == obj2) and (obj1 != obj2) ```Eh, by modifying ojb1.__eq__ I guess?
cursive plover
!e
py (lambda s,t:print([_s for _s in(s[i:j+1]for i in range(len(s))for j in __import__('itertools').takewhile(lambda j:sum(map(int,s[i:j+1]))<=t,range(i,len(s))))if sum(map(int,_s))==t]))('9390215671237812390',9)
@crystal mica
ok so 2 things
One, is the takewhile(lambda j: ...) assigning j a list or something? because it's for j in .... , so that either is a list, or a generator? Right?
Second, how did you just do the __import__ thing? How's it different from import ? Does this thing have a name?
And why is the itertools in inverted commas, 'itertools' and not itertools ?
crystal mica
when you dopy import stringyou basically dopy string = __import__('string')
And py import string as built_in_stringispy built_in_string = __import__('string')
In this case we're importing itertools without doing a separate import line
And assign it to a variable
takewhile returns a generator
cursive plover
oh okay
crystal mica
Basically it's the same as py for things in my_iterable: if condition: yield things else: break
cursive plover
ah, it's a generator cuz it has yield, right?
crystal mica
You can certainly check what it is
!e py import itertools a = itertools.takewhile(bool, range(100)) print(type(a), a)
night quarryBOT
@crystal mica :white_check_mark: Your eval job has completed with return code 0.
<class 'itertools.takewhile'> <itertools.takewhile object at 0x7f3b5fb9df40>
It should be an iterator, so unless you unpack it
cursive plover
unless I unpack it, I won't be able to print its content
crystal mica
Yep, unless you actually iterate over it and consume it
marsh void
itertools' objects are essentially iterable classes, not generators
crystal mica
!e py import itertools a = itertools.takewhile(bool, range(1, 10)) print(type(a)) print(list(a)) print(list(a))
night quarryBOT
@crystal mica :white_check_mark: Your eval job has completed with return code 0.
001 | <class 'itertools.takewhile'>
002 | [1, 2, 3, 4, 5, 6, 7, 8, 9]
003 | []
built-in != made in C
cursive plover
one more thing
(lambda s,t:print([_s for _s in(s[i:j+1]for i in range(len(s))for j in __import__('itertools').takewhile(lambda j:sum(map(int,s[i:j+1]))<=t,range(i,len(s))))if sum(map(int,_s))==t]))('9390215671237812390',9)```
why did the whole thing have to be inside brackets, for in
ie for _s in(...)
when I remove the brackets the code gives errors
when we do a for i in range(..) we don't need the brackets right? We don't do a for i in(range(...)), so why here
vestal solstice
because s[i:j+1]for i in range(len(s) isn't syntactiaclly anything
it needs [] or () around it
cursive plover
um don't we do that sort of thing in a list comprehension though?
like x = [i for i in range(something) for j in range(something)]
oh wait, that's what you just said
my bad
right, got it. thanks lol
vestal solstice
you can build it into the comprehension, so you'll have s[i:j+1] everywhere you have _s
which is two places
let me read it...
bitter iris
Yea, there is no yield in C level, so if you are going to define a generator it should be a class with iter and next
cursive plover
actually wait
print([_s for _s in(s[i:j+1]for i in range(len(s))for j in __import__('itertools').takewhile(lambda j:sum(map(int,s[i:j+1]))<=t,range(i,len(s))))if sum(map(int,_s))==t])
isn't the whole thing already in []
so
print([_s for _s in s[i:j+1] for i in range(len(s)) for j in __import__('itertools').takewhile(lambda j:sum(map(int,s[i:j+1]))<=t,range(i,len(s))) if sum(map(int,_s))==t])
no
vestal solstice
it's now in the wrong order
thin trout
!= isn't testing for __eq__ to return False?
vestal solstice
i and j donn't exist at that point
@thin trout there's __ne__ I think, it probably falls back on eq but it doesn't need it
cursive plover
oh yeah, so I'll have to change the order
so it would be
hm nah, that looks like the only way to do it
vestal solstice
print([s[i:j+1]for i in range(len(s))for j in __import__('itertools').takewhile(lambda j:sum(map(int,s[i:j+1]))<=t,range(i,len(s)))if sum(map(int,s[i:j+1]))==t])
here
you replace _s with s[i:j+1] in two places and the tricky part is to find a closing ) to remove 🙂
cursive plover
oh yeah that makes sense
so the earlier one had _s being picked from a generator, and being added to a list, which was finally printed
here, we've got a list which we append a substring directly to, and print the list out
yeah?
vestal solstice
sure
vestal solstice
and you can do this to "rename" it to _s anyway ```py
print([_s for i in range(len(s))for j in import('itertools').takewhile(lambda j:sum(map(int,s[i:j+1]))<=t,range(i,len(s))) for _s in [s[i:j+1]] if sum(map(int,_s)==t])
for _s in [s[i:j+1]]
make a list of one element to unpack it back under a different name
that's the problem with python comprehensions, that isn't there in haskell
where you can ..., let _s = s[i:j+1], ... , as a step
cursive plover
you mean assigning a list to a variable can be done inside a lambda expression in haskell?
vestal solstice
inside a comprehension
like there are two things you can do in a comprehension, a "guard" with if and an iteration with for
you can interleave them in any order
in haskell they are separated with commas:
[x + 3 | x <- nums, mod x 2 == 0] this means
[x + 3 for x in nums if x % 2 == 0]
python omits | and ,
vestal solstice
idk
cursive plover
also, when you say any order
I thought it's always [xyz for something if something] ?
vestal solstice
you can have ifs interleaved with fors
cursive plover
but never an if before all the fors, yeah?
vestal solstice
sure, you have no bound variables before the fors, so an if has nothing to work with
cursive plover
no I mean
cursive plover
something like
if x>y:
for i in range(something):
blah blah
i just checked, that;s invalid
cursive plover
yea
cursive plover
yeah
formal sandal
!e
a = 4
b = 9
print([d for d in[1,2,3]if a > b])
@formal sandal :white_check_mark: Your eval job has completed with return code 0.
[]
!e```
print([1 for x in "a" if False for y in 5])
!e
print([1for x in"a"for y in"b"if""])
@formal sandal :white_check_mark: Your eval job has completed with return code 0.
[]
basically you try to iterate through an integer and it doesn;t crash
because it's after an if
so doesn't happen
formal sandal
👀
bitter iris
never noticed this ambiguity if""
lament ibex
On a related note, i.e. breaking out of comprehensions, this list(next(iter([])) if x == 'a' else x for x in 'blah') works in py2 but doesn't in py3 (list doesn't absorb the stopiteration). Anyone know a workaround?
lament ibex
ah in my head you've either got prehistoric python (py2) or current python (3.7+) lol
sick hound
!e
print([1 for x in "a" if False for y in 5])
night quarryBOT
@sick hound :white_check_mark: Your eval job has completed with return code 0.
[]
I guess
formal sandal
Maybe I went too far with my toy type checking...
I can create new 'types' with tools like either, product, record etc. using simple data structures
Then I can use these 'types' as function annotations. The @tc makes them work as argument filters, and @mdtc allows overloading functions with help of this infrastructure.
I can't believe I even managed to make it all work together.
bitter iris
looks great except I guess a bit implicit @formal sandal
formal sandal
What do you mean?
formal sandal
Maybe change tc => template_check and md => multiple_dispatch or overload
But I also have mdtc which is their composition. It's kinda hard to find a name for it...
Maybe I shouldn't use md and rename mdtc to overload
bitter iris
that totally makes sense and IMHO looks better
overload might be conflict with typing.overload
formal sandal
Hm...
Well, it would be strange to use both typechecking systems at the same time.
So to avoid confusion when using both libraries, explicitly state which library you're using
formal sandal
Maybe you can use these 'templates' to do some additional checks. A 'template' doesn't necessarily mean 'type'.
This will throw an "Invalid UID" error because UID is a result of msg.
tribal moon
[*(lambda _____: _____((lambda _: _([-~True.__rmul__(abs(False))**-~True.__rmul__(abs(False)) for __ in range((-~True.__rmul__(abs(False))**-~True.__rmul__(abs(False)).__rmul__(-~True.__rmul__(abs(False))**-~True.__rmul__(abs(False))).__rmul__(-~True.__rmul__(abs(False))**-~True.__rmul__(abs(False))) + 8))]))(lambda ___: [____ for ____ in range(1, sum(___).__add__(1))])))(lambda ______: [__import__('builtins').__dict__['print'](''.join(list(map(str, _______)))) for _______ in __import__('itertools').__dict__['product'](______, repeat=len(______)) if not None])]```prints 111111111 and goes up from there
yewah
cursive plover
nice
is there a reason you named your variables as _, __, ____, instead of like a,b,c lol
I wanna fractionate the code down to understand it, but I'm scared I'll end up breaking it by replacing those vars incorrectly lol
snow beacon
They're just to confuse people.
tribal moon
snow beacon
@tribal moon You don't need to use abs because you can multiply by booleans as if they were integers. And if you did need integers, you could use +False to be slightly shorter.
tribal moon
Yeah this is really messy
@cursive plover if you want, you can replace the underscores with your own variable names
You can use notepad++ to do it
cursive plover
yeah I'm gonna do that
shell quiver
what does it mean 'obfuscation'
proper vault
making it hard to tell the actual purpose of a given part of code
shell quiver
[*(lambda _____: _____((lambda _: _([-~True.__rmul__(abs(False))**-~True.__rmul__(abs(False)) for __ in range((-~True.__rmul__(abs(False))**-~True.__rmul__(abs(False)).__rmul__(-~True.__rmul__(abs(False))**-~True.__rmul__(abs(False))).__rmul__(-~True.__rmul__(abs(False))**-~True.__rmul__(abs(False))) + 8))]))(lambda ___: [____ for ____ in range(1, sum(___).__add__(1))])))(lambda ______: [__import__('builtins').__dict__['print'](''.join(list(map(str, _______)))) for _______ in __import__('itertools').__dict__['product'](______, repeat=len(______)) if not None])]```@tribal moon
tf
formal sandal
Deliberately making it hard to tell the actual purpose of a given part of code*
formal sandal
Involuntary obfuscation is something I wish happened to my code more rarely...
zealous widget
bunch of multiplications with 0
formal sandal
!e
def is_syntax_error(s):
try: compile(s,'','eval')
except SyntaxError: print(f"'{s}' = Syntax error")
else: print(f"'{s}' = OK")
is_syntax_error("2.__add__")
is_syntax_error("(2).__add__")
is_syntax_error("2. __add__")
is_syntax_error("2 .__add__")
is_syntax_error("2..__add__")
is_syntax_error("2. .__add__")
is_syntax_error("2.. __add__")
@formal sandal :white_check_mark: Your eval job has completed with return code 0.
001 | '2.__add__' = Syntax error
002 | '(2).__add__' = OK
003 | '2. __add__' = Syntax error
004 | '2 .__add__' = OK
005 | '2..__add__' = OK
006 | '2. .__add__' = OK
007 | '2.. __add__' = OK
!e
def catcher(handler, e=Exception):
def catch(f):
def _(*args, **kwargs):
try:
return f(*args, **kwargs)
except e as _e:
return handler(*args, _e=_e, **kwargs)
return _
return catch
# Now you can write crap like this:
v = catcher(lambda _e:eval("1 .__add__"))(lambda:eval("1. __add__"))()
print(v(41))
@formal sandal :white_check_mark: Your eval job has completed with return code 0.
42
class NumberAttributes(TokenTransformer):
@pattern("number", "name")
def fix_number_attributes(self, number, attribute):
if number.string.endswith("."):
return number._replace(string = f"{number.string[:-1]} ."), attribute