AoC 2022 | Day 7 | Solutions & Spoilers

[queen breach]

idk why i can combine s and d assignments, but it worked

[last flower]

Does it work? That would make s and d reference the same list

[queen breach]

yeah, it somehow works

and i have no idea why

my best guess is that one of them is reassigned somewhere

[inner ingot]

you also reassign it again on line 4, so maybe you got lucky

[last flower]

I think i just got stuck in an infinite loop for my input

nvm, you're using stdin for input

[queen breach]

nvm i counted wrong

that was line 5, not line 4

[queen breach]

got lucky

nothing < 10k before my first cd ..

[high bane]

min(_ for _ in d if _>sum(s)-4e7)
min(filter((sum(s)-4e7).__gt__,d)
😔

gt or ge idk

[inner ingot]

lol

[queen breach]

your filter thing is missing )

or rather, your min is missing that

[high bane]

rip

[viscid sail]

it does work

[rancid heath]

Maybe it's input-dependent

[viscid sail]

check yours. I doubt that may be case

[royal axle]

absolute trash

[queen breach]

it's input dependent

[royal axle]

i put all paths in string form /abc/def/ghi/.../ and than build with that a dict

[queen breach]

and i think it's plausible that there's an input which doesn't work

[royal axle]

the rest was pretty much ok

[rancid heath]

Part 2 gives the correct answer but part 1 is incorrect

[queen breach]

yeah

[viscid sail]

you forgot bracket for min

[high bane]

no u

[queen breach]

gotem

[viscid sail]

and the answer comes wrong even after correcting that kek

[cunning wraith]

My solution: ```python
class File:
def init(self, name, size):
self.name = name
self._size = size

def size(self):
    return self._size

def __repr__(self):
    return f"{self.name}"

class Directory:
def init(self, name):
self.name = name
self.children = {}

def get_dir(self, dirname):
    if dirname not in self.children:
        self.children[dirname] = Directory(dirname)
        
    return self.children[dirname]

def add_file(self, file):
    self.children[file.name] = file

def walk_dirs(self):
    yield self
    for child in self.children.values():
        if isinstance(child, Directory):
            yield from child.walk_dirs()

def __repr__(self):
    return f"{self.name}/{self.children}"

def size(self):
    return sum(child.size() for child in self.children.values())

root = Directory("root")
working_dir = [root]

for instruction in data.split("\n"):
if instruction.startswith("$ cd"):
dirname = instruction.removeprefix("$ cd").strip()
if dirname == "/":
working_dir = [root]
elif dirname == "..":
working_dir.pop()
else:
working_dir.append(working_dir[-1].get_dir(dirname))
elif instruction[0].isdigit():
size, name = instruction.split()
working_dir[-1].add_file(File(name, int(size)))

total = 0
for directory in root.walk_dirs():
size = directory.size()
if size <= 100000:
total += size
print("Part 1:", total)

remaining = 70000000 - root.size()
needed = 30000000 - remaining
possible = []
for directory in root.walk_dirs():
size = directory.size()
if size >= needed:
possible.append(size)
print("Part 2:", min(possible))

[royal axle]

how long did that took u?

[cunning wraith]

not sure, I did it a few hours ago and wasn't going for speed though.

[elder bolt]

I've seen a few solutions utilizing a type of directory class for this problem, wondering if I want to do a second solution using a class

[inner ingot]

i want to make a solution that simulates it on an actual file system

[cunning wraith]

I thought of doing that haha

[viscid sail]

if you want to use filter do this :

min(filter(lambda x:x>sum(s)-4e7,d))

[wheat cradle]

why trash tho

[high bane]

if its not gt its gotta be lt

[inner ingot]

i mean the only hard part would be parsing. there's already a bunch of tools that can do the solving part pretty much

[royal axle]

Input parsing is always trash

Working with the parsed stuff was ez

[wheat cradle]

um

okay

[queen breach]

ok

imagine getting skill diffed by input parsing

[hallow ether]

I really want to do a 100% bash solution to this one

[cunning wraith]

Just gotta hope you don't get any 1tb files in your input :P

although the filesystem is 70mb (taking 1 size as 1 byte) so should be fine

[inner ingot]

i'll just put it on the NAS 🥴

[soft hatch]

I'm more worried about reading the thing

[queen breach]

isn't size stored in metadata

or am i remembering wrong

[inner ingot]

wdym?

[soft hatch]

oh wait, I thought you were talking input files

my bad

[inner ingot]

wait what

ohhh, you thought 1tb long input

also, i am woefully inexperienced with linux stuff. can you spoof file metadata to show a file size without actually storing that amount of data

[cunning wraith]

Not sure, although you could always implement a custom filesystem to let you do that 🙃

[inner ingot]

oh yeah. i've been wanting to do something like that tbh

[tough hinge]

When I got stuck, I was very seriously considering actually writing to disk and letting pathlib do the parsing

[queen breach]

💀

[tough hinge]

Was going to write the size as text inside the file and everything

[soft hatch]

you don't even need the names

[queen breach]

lol

[tough hinge]

I have to keep track of them somehow

[queen breach]

do you though?

[soft hatch]

the input follows a dfs order

[queen breach]

just stuff it into a list

[tough hinge]

Department of Family Services?

[soft hatch]

you don't actually need to keep track of names

[queen breach]

dog for shenanigans

[bronze jewel]

I cannot even remember if I did a DFS. Just remember using match

[soft hatch]

a stack to track the sizes

that's all you really need

[queen breach]

ye

[inner ingot]

i didn't want to assume that while solving though

[tough hinge]

I wrote a whole three classes

Done turned into Tom

[queen breach]

overengineering

[tough hinge]

That's my specialty!

[soft hatch]

my code would not be happy about

cd a
cd ..
cd a
cd ..
...

[viscid sail]

numpy would still destroy this. atleast the print part

[soft hatch]

(+ some files in there)

[queen breach]

my specialty is underengineering

works for golf though

[inner ingot]

i had already gotten rolled by duplicate directory names so i really didn't want to assume anything

[soft hatch]

if I wrote the task I would have been the asshole adding some duplicate directory visits to input data

[tough hinge]

I mean, if you're not editing setuptools options to get your CI to run your AoC tests, and you even really AoCing?

[queen breach]

why would i do that for lc easies?

[soft hatch]

see, names can be hurtful

[tough hinge]

Duplicate directory names broke my code
There's a vote against names

[rancid heath]

My code would be fine with that LMAO

[queen breach]

skill diffed by duplicate names 😔

[slow oriole]

me who didnt even encounter such problem

[queen breach]

my problem was the dumbest

[rancid heath]

I had double-counting issues resulting as the manifestation of a different bug

[slow oriole]

⭐ n o d e s ⭐

[rancid heath]

So I used nested dictionaries instead of just keeping track

[queen breach]

i counted leaf nodes 💀

for p1

[slow oriole]

are leaf nodes something different

[rancid heath]

Leaf nodes are files

As opposed to directories

[slow oriole]

oh

[bronze jewel]

I just used a defaultdict with the paths and total size

[slow oriole]

then what is the term for directory

[rancid heath]

That was my first solution but it didn't work kekw

[wheat cradle]

yeah this worked just fine

[inner ingot]

internal nodes

[bronze jewel]

Ooof.
First attempt for me didn't either. Then I realised that I was counting dirs etc only once (basically left out the brackets part in the question)

Anyone got a node version that I can peek at?

[slow oriole]

my node class
basically self.sum contains sum of file size in the uppermost level i.e that dir itself , and total just adds up the sum of internals

class node:
    def __init__(self,c,p):
        self.curr = c
        self.parent = p
        self.v = {}
        self.sum = 0
    @property
    def total(self):
        return self.sum + sum(i.total for i in self.v.values() if isinstance(i,node)) 

[slow oriole]

ah

okay

[rancid heath]

I just used dictionaries; when I was making my physical ElfFS I updated my code to use a type alias (to declare what it was already doing) - Folder = dict[str, typing.Union[int, "Folder"]]

(mypy doesn't like recursive type aliases but Pylance is fine with them)

[slow oriole]

my pylance is disturbing

i beat the shit out of it

[rancid heath]

If you want me to send the repo link again I can

(I posted it way earlier at the beginning of the thread too lmao)

[bronze jewel]

I can do a search

[slow oriole]

TIL you can just put a # type: ignore for ignoring pylance

[rancid heath]

You can also find my GH in my connections

[soft hatch]

I wonder what the first problem that actually requires some compute power will be

[rancid heath]

Yeah # type: ignore means 'disable all type-checking for this line'

[bronze jewel]

That is easier.... The search is being wack on my phone

[slow oriole]

oh thats pretty cool

the problem was match case

[rancid heath]

*not technically true if you're writing a multiline statement

[slow oriole]

the warnings were wrong

[slow oriole]

it will ignore the entire file if you put it on the top tho

[rancid heath]

If it's at the end of a parenthesised statement, it's disabled for the whole statement; if it's on (e.g.) a single parameter it disables only that expression

[rancid heath]

Good to know

[soft hatch]

[rancid heath]

It happens

Usually when a type-hint somewhere is wrong

Sometimes when self is involved

[soft hatch]

usually the warnings are right, but in an annoying and pedantic way

[weak mauve]

technically correct

unexpected EOF

[hollow trail]

need some help 🥲

day_input = day_input.strip().split('\n')
path = Path()
size_map = defaultdict(int)
line = 0
size_map[''] = 0

while line < len(day_input):    
    out : list[str] = day_input[line].split(' ')

    match out[1]:
        case 'cd':
            path = (path / out[2]).resolve()
        case 'ls':
            tline = line+1
            while tline < len(day_input) and not day_input[tline].startswith('$'):         
                if (size := day_input[tline].split(' ')[0]).isnumeric():
                    size_map[path.stem] += int(size)  # add size of file to parent dir                
                tline += 1
            size_map[path.parent.name] += size_map[path.stem]  # add size of file to parent of parent dir
            line = tline - 1
    
    line += 1

print(sum([i for i in size_map.values() if i <= 100000]))

[soft hatch]

I hate pylint with a passion

[hollow trail]

gives incorrect ans on p1

[slow oriole]

i hate every linter with a passion

just set a break point and use a debugger (not formatting related ofc)

[tough hinge]

OMD you're actually using pathlib

[queen breach]

formatters are cool though

black is nice when i'm not golfing

[slow oriole]

yeah

[cunning wraith]

I don't mind some linters, but pylint is way too picky imo

[slow oriole]

black is nice

[bronze jewel]

Black is great

[queen breach]

the only lints i have are type hinting and style lints

[bronze jewel]

The default linter for pycharm is.... Augh.... Painful

[queen breach]

don't use pycharm so can't relate

[soft hatch]

black is opinionated, and I don't agree with a bunch of its opinions :^)

[slow oriole]

oh my dog ?🐶

[tough hinge]

yep

[queen breach]

sure, and that's your opinion

[slow oriole]

dang

[queen breach]

otoh i like black's choices

[soft hatch]

formatters in general are good though

[soft hatch]

" for quotes is heresy

[hollow trail]

this works on sample input tho 🤔

[queen breach]

so non-python is heresy?

[rancid heath]

No warning for impossible type hints though kekw
I accidentally declared a method with a self annotation that could never be satisfied and no warning

[soft hatch]

in python in particular

[weak meadow]

is everyone else ignoring ls too, or did you guys build proper parsers?

[queen breach]

i stick to " because that way it's consistent between all langs

[slow oriole]

it might be failing due to presence of duplicate names at different dir levels (in the actual input)

[bronze jewel]

turns on PC to then accidentally turn it off again....

[weak mauve]

did you add the sizes to all ancestor dirs?

[queen breach]

a proper parser could ignore ls and be fine

[soft hatch]

python itself uses ' though

[queen breach]

¯_(ツ)_/¯

i like it when my habits from 1 lang translate well to others

[teal rune]

I wound up ignoring everything at wasn't either a "cd" or size value

[queen breach]

quotes is one example of that

[rancid heath]

You can turn off quote normalisation

[slow oriole]

am i the only one who split by '$'

[hollow trail]

hmmm propagate the result the back up huh

[slow oriole]

[soft hatch]

(I can't force ')
but yeah, I know

black's inability to run on part of a file is pretty annoying

[rancid heath]

IIRC they've discussed changing their stance on this

[weak mauve]

also some directories can have the same name.

[rancid heath]

Also blacken is a thing

[tough hinge]

And blacken docs

[bronze jewel]

One thing about a linter I previously used was the removing of imports when I decided to comment out code or remove it.... And no auto import

[tough hinge]

Because black=good

[rancid heath]

Yeah I hate this about the default JS/TS formatter

[cunning wraith]

would you use ''' for multiline strings?

(or, do you?)

[weak mauve]

and for a directory there is no difference between .name and .stem

[tough hinge]

And for not directories?

[weak mauve]

it only makes a difference if the name contains a dot.

[tough hinge]

What does it actually do?

[weak mauve]

the stem is the name without the suffix

[hollow trail]

i think this is the issue

[weak mauve]

path.stem + path.suffix == path.name

"data.tar" + ".gz" == "data.tar.gz"

[hollow trail]

yh ik

[soft hatch]

I would. I'm fine with """ for docstrings because telling docstrings and multiline strings apart is actually nice, but sadly that seems to be an unpopular opinion

[rancid heath]

My opinion on quote normalisation is that I didn't used to like it, but now it makes reading things quickly easier and helped me use the right quotes in other languages

[soft hatch]

I've never had issues context switching between langs, but maybe that's just me

[rancid heath]

*helped me learn

Not context switching so much as habit while learning

[inner ingot]

when switching from python to other langs i sometimes mess up : and {} but it only takes like 1 error to full switch over

[cunning wraith]

Was working with typescript and python a couple of days ago and found myself adding a semicolon to one of my lines of Python

[rancid heath]

kekw

[soft hatch]

I find " visually noisier than ' when reading code

[queen breach]

valid

[inner ingot]

hmm. i kinda just gloss over it

[queen breach]

i find " more familiar

[soft hatch]

||the code?||

[rancid heath]

I tend to read '' as 'internal string' and "" as 'external string'

Or 'might be printed'

[queen breach]

i only use ' when inside another string, when golfing, or when writing code quickly

[inner ingot]

like, i kinda ignore the delimiter, it's just not really an issue what it is

[soft hatch]

I only use " if it makes the string have less escaping 😛

[queen breach]

oh escaping

that too

i usually don't need quotation marks in my strings though, so that's rare

[soft hatch]

💩my string💩

[rancid heath]

Honestly I just use ' because it's easier to type than shift-2 (") and then I let black normalise it

[cunning wraith]

I could probably argue that I preferred one over the other visually but then again I'm able to ignore the red squiggly lines from pylance all over my code so I can't really speak lol

[rancid heath]

It does cause me to sometimes syntax-error in languages that care though lol

[inner ingot]

well that's cause you have a weird keyboard

[queen breach]

lol

[rancid heath]

Standard UK ISO layout

[viscid sail]

😩 can't get any lesser. Have been trying since an hour and a half now. I guess 190 is the limit.

[soft hatch]

even on US ' is easier, no?

[queen breach]

yeah

[little wedge]

https://github.com/nekitdev/advent-of-code/blob/main/advent_of_code/year_2022/day_7.py even more funky code

[inner ingot]

yeah, but " is basically the same since you shift with the other hand

[little wedge]

intrusively linked structs be gaming

[queen breach]

i memed about uk stuff being esoteric in one of the code jams

i think joe had the qualifier use uk spelling and i opened a pr fixing the "esoteric spelling"

[soft hatch]

have you seen swedish layout? typing []{} is terrible

[tough hinge]

FYI, the creators have asked people not to share their inputs
https://www.reddit.com/r/adventofcode/comments/k99rod/comment/gf2ukkf/

[rancid heath]

Esoteric spelling is discarding consistency with the root word (analysing analyses, anyone?)

[little wedge]

oh

[queen breach]

i was memeing

[rancid heath]

I don't mind having a few of the inputs posted, please don't go on a quest to collect many or all of the inputs for every puzzle.
I read this as meaning 'a couple of repos is probably fine but don't go looking'

[queen breach]

(realizing that i spelt the word wrong genuinely was the most time consuming part of the quali though)

[hollow trail]

ok ty jenna
doing it with pathlib was fun

day_input = get_input(7)
day_input = day_input.strip().split('\n')
path = Path()
size_map = defaultdict(int)
dir_map = defaultdict(list)
line = 0
while line < len(day_input):    
    out : list[str] = day_input[line].split(' ')

    match out[1]:
        case 'cd':
            path = (path / out[2]).resolve()
            if out[2] != '..':
                dir_map[str(path.parent)].append(str(path))
 
        case 'ls':
            tline = line+1
            while tline < len(day_input) and not day_input[tline].startswith('$'):         
                if (size := day_input[tline].split(' ')[0]).isnumeric():
                    size_map[str(path)] += int(size)  # add size of file to parent dir                
                tline += 1
            for parent in path.parents:
                size_map[str(parent)] += size_map[str(path)]  # add size of file to parent of parent dir
            line = tline - 1
    
    line += 1
print(sum([i for i in size_map.values() if i <= 100000]))  # p1
remaining = 70000000 - size_map['/']
print(min(i for i in size_map.values() if remaining + i >= 30000000))  #p2

[rancid heath]

I forgot about .parent
Does that exist on PurePath?

[hollow trail]

yes

[queen breach]

i don't think pathlib is efficient in terms of characters

especially since you don't need the paths at all

[hollow trail]

ehhh i didnt do it for efficiency

i did for fun

[queen breach]

that is invalid, code golf is the only way

[hollow trail]

forgive me for i have sinned

i shall come back to the correct "path"

[slow oriole]

no

[queen breach]

no you don't need any path

[slow oriole]

yes

[hollow trail]

bruhhh

it was a joke

[queen breach]

saw the pun from a mile away

dw, we got it

[hollow trail]

hahahha

[slow oriole]

if i didnt sleep rn

i will again wake late

and see some super golfed code of day 8

[queen breach]

lol

i just wait for artemis to post an initial golf, then for cef to completely ignore it and post a better one

and then i snipe the last few chars

[slow oriole]

oh

so artemis is the root

thanks for the intel

/j

[teal orchid]

Could someone help me understand why this error is happening? Will probably need to show more code. Trying to recursively update the size of the parent directory when we add a new file

def set_folder_size(self, file_size):
  self.size += file_size
    if self.parent is not None:
      self.get_folder(self.parent).set_folder_size(file_size)

AttributeError: 'NoneType' object has no attribute 'set_folder_size'

Im not sure why this isnt being caught by my None check

[weak mauve]

yes, what does get_folder?

probably get_folder does only know about subfolders.

[teal orchid]

Im returning None somwhere, somehow

def get_folder(self, target):
  if self.name == target:
    return self
  elif self.subfolders:
    for f in self.subfolders:
      sf = f.get_folder(target)
      if sf: return sf
return None    

[teal orchid]

i check against itself first

[rancid heath]

It's not in a subfolder

[tough hinge]

Replace your return None with a raise ValueError(target)

Then you can start excepting [and/or printing] various crap from there
Even better with a debugger

[weak mauve]

Maybe it is in a subfolder. But not in every subfolder.

So depth-first-search doesn't work.

[teal orchid]

but my point is shouldnt if self.name == target return self return the object before it starts DFSing?

since Im directly passing it the name of the object i want

[viscid sail]

is this match-case thingy new in python3.10?
Also, what's the folder variable

[rancid heath]

How sure are you that it isn't ..?

[teal orchid]

its not

its ValueError: /

which doesnt have a parent

but that should be caught by None

[rancid heath]

Yes and I'm guessing FT

[weak mauve]

so self.name == self.parent?

[rancid heath]

That sounds like an absolute path

[viscid sail]

how is the program even running?

[teal orchid]

in this case:

def set_folder_size(self, file_size):
  self.size += file_size
    if self.parent is not None:
      self.get_folder(self.parent).set_folder_size(file_size)

I am just passing the parent of the current folder to update the parents size as well

[viscid sail]

Also, won't CDir.pop() become an empty list at some point throwing an error?

[weak mauve]

yes, you're passing the parent to self.get_folder()

[teal orchid]

Ah shit yeah

Should be self.parent.get_folder()?

[weak mauve]

That doesn't seem to make sense either. But I'm not sure.

[rancid heath]

I don't think you need get_folder at all

[weak mauve]

is target a string?

it's compared to self.name.

is self.parent a string?

[viscid sail]

422 => 190
Before and after

golfed!

[weak mauve]

much better

[teal orchid]

Is there a better way to get the parent object if I’m already in the class?

[weak mauve]

no idea how the class looks like.

or how it's used.

from your latest snippets it would be if self.parent: self.parent.set_folder_size(file_size)

[teal orchid]

class Folder(object):
    def __init__(self, name, parent=None):
        self.name = name
        self.size = 0
        self.parent = parent
        self.subfolders = []
        self.files = []

    def __repr__(self):
        return f"{self.name, self.size, self.parent, self.subfolders, self.files}"

    def add_subfolder(self, folder):
        self.subfolders.append(folder)

    def add_file(self, file):
        self.files.append(file)

    def get_folder(self, target):
        if self.name == target:
            return self
        elif self.subfolders:
            for f in self.subfolders:
                sf = f.get_folder(target)
                if sf: return sf
        raise ValueError(target)    

    def get_parent_name(self):
        return self.parent

    def set_folder_size(self, file_size):
        self.size += file_size
        if self.parent is not None:
            self.get_folder(self.parent).set_folder_size(file_size)

parent is just a string

which is why i have the get_folder function

[rancid heath]

Make parent a Folder instead

Just objectively better

[teal orchid]

would that just be

self.parent = Folder()

[rancid heath]

When constructing a folder instead of parent.name, pass parent

[weak mauve]

not just objectively better. it's the only way.™️

[rancid heath]

root = Folder('/')
tmp = Folder('tmp', root)

[rancid heath]

If the root is stored globally you could DFS for the current folder, then go up one level

Possible, just objectively worse

[faint lava]

day 7 seems like it's one you could do a lot easier if you just strip out the fun parts and don't treat it like a file structure.

[teal orchid]

so if parent is an object then i could do this

[rancid heath]

Yes

[faint lava]

I don't think my solution was optimal, but I like the way it came out, and I got to play with match/case.

[weak mauve]

hm.. yes, or store all folders globally. :think:

[queen breach]

that's exactly what it is

[clear herald]

I am currently trying and failing to create a dictionary to re-create the file structure and I think this is hurting me more than helping

[queen breach]

skill diff

[weak mauve]

or store subfolders as a dict. so you can add a .. to the parent.

[queen breach]

just store the path as a tuple or something

[rancid heath]

In hindsight I should have done this

[queen breach]

that's what i did for my first try

[rancid heath]

I had a special case for .. lmao

[queen breach]

who didn't?

[rancid heath]

Just found the relevant path every time because .. was pain

Kinda tempted to go make a proper shell and then make it runnable as a shell for like a visualisation thing

[faint lava]

I think there are 2 safe approaches: 1) a recursive dictionary, 2) a flat dictionary where each dir has the full path

[teal orchid]

So I changed the class to

class Folder(object):
    def __init__(self, name, parent=None):
        self.name = name
        self.size = 0
        self.parent = Folder()
        self.subfolders = []
        self.files = []

Traceback (most recent call last):
  File "main.py", line 49, in <module>
    root = Folder('/')
  File "main.py", line 5, in __init__
    self.parent = Folder()
TypeError: __init__() missing 1 required positional argument: 'name'

Why is it trying to make a parent if theres a default argument None?

[weak mauve]

But with a .. in the subfolders you would have to special case it when traversing the subfolders.

[queen breach]

there's also the dangerous approach of actually making the files and stuff

[faint lava]

self.parent = Folder() <--- needs the name

[rancid heath]

You mean like https://github.com/Starwort/advent-of-code-2022/tree/master/day_07_fs?

[queen breach]

Fix files being too small

[rancid heath]

Yes

[queen breach]

it's lipsum

[weak mauve]

I hope they accounted for the fs overhead.

[rancid heath]

Yes

[rancid heath]

No it's file contents size

[teal orchid]

i think im just lost now lol

[hollow trail]

hmmmm another fun way of solving it

[rancid heath]

@hollow trail see

[clear herald]

I think my approach for today will just be taking the L

[hollow trail]

beautiful

[queen breach]

it's possible to do without any dicts

it's kind of a safe approach, if you know part 2 beforehand

[clear herald]

Correct, but that is my current approach and I am not succeeding at that

[queen breach]

skill issue

[faint lava]

Nah, you got this, don't be afraid to consult with others (aks the peeps here)

[clear herald]

Also correct, if a little blunt

[hollow trail]

i did the second one

[queen breach]

be afraid, some of us are golfers

[teal orchid]

did you literally mean self.parent = Folder(name)

[faint lava]

You have written your code such that calls to Folder() must be passed a name, you can change it to make the name not required, or you can pass one. Or you can have the parent default to None until you assign one.

[rancid heath]

Where are you currently making Folders?

[faint lava]

this is probably what you want:

self.parent = None

or pass in the parent.

[rancid heath]

@teal orchid show us an example of where you're already calling Folder()

[teal orchid]

preemptively apologizing for the war crime of code you are about to witness

root = Folder('/')
for c in commands:
    if 'cd' in c.split() and c.split()[2] != '..':
        curr_dir = c.split()[2]
    elif 'cd' in c.split() and c.split()[2] == '..':
        curr_dir = root.get_folder(curr_dir).get_parent_name()
    elif not c.startswith('$'):
        if 'dir' in c.split():
            new_folder = Folder(c.split()[1], parent=root.get_folder(curr_dir))
            root.get_folder(curr_dir).add_subfolder(new_folder)
        else:
            new_file = File(c.split()[1], int(c.split()[0]))
            root.get_folder(curr_dir).add_file(new_file)
            root.get_folder(curr_dir).set_folder_size(new_file.get_size())

it gets all the data!

[rancid heath]

new_folder = Folder(c.split()[1], parent=root.get_folder(curr_dir))
Change this to new_folder = Folder(c.split()[1], parent=curr_dir)

May also need to change curr_dir to be a Folder

[teal orchid]

but curr_dir is just a string

yeah

[rancid heath]

I'm guessing that when you update curr_dir you're doing something involving .name or root?

[cedar trout]

def parse_directories(data: list[str]) -> FileSystem:
    file_system = None
    for line in data:
        match line.split():
            case ["$", "cd", dir_]:
                if file_system is None:
                    file_system = FileSystem(dir_)
                else:
                    file_system.cd(dir_)
            case ["dir", dir_name]:
                file_system.add_dir(dir_name)
            case [size, _] if size.isnumeric():
                file_system.add_content(int(size))
            case _:
                pass
    return file_system

I did it with patter matching, works really nicely for this one

[sonic lark]

this kind of repeated messaging isn't really in line with our code of conduct and the type of atmosphere we like to have in the server. Even if you mean it jokingly, it's coming across as rather mean spirited.

[faint lava]

You'll also need to update your class to allow you to pass the parent in to it.

[queen breach]

k

no more then

[faint lava]

Anyone else abuse namedtuple to make classes?

class Dir(namedtuple('Dir', 'name folders files parent')):

and later:

        case ['dir',f]   : curr.folders[f] = Dir(f, {}, [], curr)

[rancid heath]

No

Didn't really see a need for classes

Haven't at all yet so far, actually

[twin prawn]

was just thinking that i should have used named tuples, (although instead of collections.namedtuple use typing.NamedTuple and you get class like syntax)

[rancid heath]

Usually don't end up with any across the whole event

Although I did for intcode

[rich aspen]

Do you know there's a class variant of namedtuple found in the typing module?

[faint lava]

I saw NamedTuple in typing but not looked at how it works.

[rich aspen]

Oh right. It's pretty similar to a frozen dataclass, if you're familiar with dataclasses.

I mean, it's pretty similar to namedtuple too of course.

[faint lava]

I actually started with a dataclass, but the namedtuple seemes even easier to play with.

[rancid heath]

I like dataclass better because it's properly typed

[faint lava]

says the person that doesn't use classes. 🙂 🙂

[rancid heath]

Meant for AOC lol

I do use classes when they make sense for other bits of Python

Also TypedDict is quite nice

[faint lava]

I suppose my code gets a little cleaner if I treat files as directories with no children

[lucid pike]

I like attrs

[rancid heath]

attrs?

[lucid pike]

Yes, it's a great third-party package for class generation

dataclasses were in part inspired by attrs

[clear herald]

It feels close, but I think I'm throwing in the towel for the day (output based on example input):

{'/': [{'a': []}, 14848514, 8504156, {'d': []}],
 'a': [{'e': []}, 29116, 2557, 62596],
 'd': [4060174, 8033020, 5626152, 7214296],
 'e': [584]}

[rancid heath]

Why is your folder structure Folder = dict[str, list[typing.Union["Folder", int]]]?

Why use a list and not just directly nest contents?

You can do a postprocessing step to reduce each one down to a size

(||it'll also help you for part 2||)

[clear herald]

I'm currently iterating over each ls line and either creating a temporary "sub folder" or appending a file size

[cedar trout]

why is the subfolder temporary?

[clear herald]

So that I can then continue to the next command in the parsed input

[queen breach]

just track the pwd

and add directly to that

[cedar trout]

yeah, I had a FileSystem class that tracked the root directory and the pwd

[rancid heath]

IMO the easiest way of solving the puzzle is to make parsing code that generates something like {'a': {'e': {'i': 584}, 'f': 29116, 'g': 2557, 'h.lst': 62596}, 'b.txt': 14848514, 'c.dat': 8504156, 'd': {'j': 4060174, 'd.log': 8033020, 'd.ext': 5626152, 'k': 7214296}}

Which can be done without too much pain

As long as you don't make a typo in your cd code 🙃

[rancid heath]

I think you mean cwd ('current working directory'), pwd is 'print working directory'

[queen breach]

present working directory

[teal orchid]

Looking back at this, Im not really sure what this would do, because the curr_dir already exists so wouldnt this make another one?

[rancid scaffold]

How do you solve the issue of passing n number of indexes?

[teal orchid]

like if you take the first example, i start by initing root, but they that would try to make it again with parent = Folder(curr_dir)?

[rancid heath]

No, it sets the parent to the current directory

[teal orchid]

when I do Folder(curr_dir) its not calling a constructor?

[rancid heath]

root = Folder('/')
cwd = root
# some time later
# not making a new one     vvv
tmp = Folder('tmp', parent=cwd)
# some time later still
cwd = tmp

[rancid heath]

'n number of indices'?

[elder bolt]

I actually think the easiest would be something like
'/': 48381165, '/a': 94853, '/a/e': 584, '/d': 24933642

[rancid heath]

Maybe

[bronze jewel]

Yes, new in 3.10
The folder variable in the match-case I take it? That is for when there is a line for the dir <dir_name> that exists. Also, the code does crash with the popping of the empty list, but that can easily be handled with a try-catch. But is does not really matter due to the fact that is solves aoc-generated inputs. if the input was manually generated, why would you start with cd .. or go any further back than the most parent directory? (just me thinking while typing 😉 )

[rancid heath]

I had a postprocess step that transformed to that

[teal orchid]

oh makes sense lol

so I have all that, now just another stupid None AttributeError

somehow

[viscid sail]

I checked, try catch would be needed. Doesn't work for my inputs 😅

[bronze jewel]

really?

[elder bolt]

Here's my parsing. Not the cleanest, but creates the structure I show and I believe would be the easiest

def get_input():
    """Get and parse the input."""
    f = open('day7/input.txt', 'r')
    raw = f.read().strip()
    lines = [line for line in raw.split('\n')]

    dir_tree = {}
    current_dir = []

    for line in lines:
        command = line.split(' ')
        if command[0] == '$':
            if command[1] == 'cd':
                if command[2] == '..':
                    current_dir.pop()
                else:
                    current_dir.append(command[2])
        else:
            if command[0].isdigit():
                for i in range(len(current_dir)):
                    dir = '/'.join(current_dir[0:i+1])
                    if dir in dir_tree:
                        dir_tree[dir] += int(command[0])
                    else:
                        dir_tree[dir] = int(command[0])

    return dir_tree

[bronze jewel]

popping the issue?

[elder bolt]

Should update to 3.10 so I can use case

And... should not be using dir as a variable

[rancid heath]

Update to 3.11

It's just faster than 3.10 in most cases

[bronze jewel]

also, does just x.split() by default split on whitespaces?

[rancid heath]

Optimising interpreter :)

[bronze jewel]

i am too lazy to change to 3.11 now

[rancid heath]

Yes, ' \t\f\v\n' by default

[bronze jewel]

might just stick with 3.10.5

[elder bolt]

Ah, for some reason I thought it just splits all characters by default

[teal orchid]

root = Folder('/')
for c in commands:
    if 'cd' in c.split() and c.split()[2] != '..':
        curr_dir = c.split()[2]
    elif 'cd' in c.split() and c.split()[2] == '..':
        curr_dir = root.get_folder(curr_dir).get_parent_name()
    elif not c.startswith('$'):
        if 'dir' in c.split():
            new_folder = Folder(c.split()[1], parent=Folder(curr_dir))
            root.get_folder(curr_dir).add_subfolder(new_folder)
        else:
            new_file = File(c.split()[1], int(c.split()[0]))
            root.get_folder(curr_dir).add_file(new_file)
            root.get_folder(curr_dir).set_folder_size(new_file.get_size())

Traceback (most recent call last):
  File "c:\Python\AdventOfCode\day7\day.py", line 55, in <module>       
    curr_dir = root.get_folder(curr_dir).get_parent_name()
  File "c:\Python\AdventOfCode\day7\day.py", line 28, in get_parent_name
    return self.parent.name
AttributeError: 'NoneType' object has no attribute 'name'

Somehow the parent isnt getting set

this is just a mess now lol

[rancid heath]

That's what the list contructor does

Or unpacking it

[viscid sail]

yeah 😅
got intrigued and copy-pasted the code and ran it.
popping seems to be the issue

[elder bolt]

Unpacking was something new I learned by reading these chats xD

[rancid heath]

You're creating a new folder of curr_dir, that's wrong

(and that folder has no parent)

[teal orchid]

yeah

isnt that what i was confused about earlier

[rancid heath]

What you want instead of curr_dir = c.split()[2] is curr_dir = curr_dir.get_folder(c.split()[2]) (assuming that that call returns a Folder... it does, right?)

And if it's .. you just curr_dir = curr_dir.parent

Then after you define the root, just set curr_dir = root

Now curr_dir is a Folder that you can just directly interact with instead of finding it by name from the root every time

And Folder.parent is also a Folder (or None, but then you have other problems) that you can just interact with

[teal orchid]

think it would be curr_dir = root.get_folder(c.split()[2]) but yeah it returns a folder

[rancid heath]

It's only if the cd dir starts with / that you need a special case (that's an absolute path and you need to set curr_dir = root before trying to cd there)

[rancid heath]

No, because c.split()[2] isn't going to be a unique name

Only if it starts with a / is that the right behaviour

cd foo = find foo in the current directory i.e. curr_dir.get_folder('foo')

[warped kraken]

hi guys

i coded a solution for part 1

and it works on the test data

but not on my input data

it's erroring for the input data

i'll post

but it's very messy

[bronze jewel]

Also Guitar, that first if-statement is a bit redundant. Can be removed (from looking at the code, doesn't look like ls will be an issue either)

[teal orchid]

and then i just need a special case like you said for / to init curr_dur?

[bronze jewel]

mind sharing your input?

[rancid heath]

If the destination (c.split()[2]) starts with /, set curr_dir to root and then find the path, yeah

But also set curr_dir to root immediately after creating root

Don't wanna use unbound variables by accident 🙃

[warped kraken]

text = open(r"C:\Users\ollys\Documents\AOC Files\2022 #7.txt", "r").read().splitlines()

curfolder = ""
tree = {}

for line in range(len(text)):
    stepfd = 1
    if text[line][0] == "$" and text[line][2] == "c":
        #cd
        if text[line][5] != ".":
            #cd directory (not cd ..)
            curfolder = text[line][5:]
    if text[line][0] == "$" and text[line][2] == "l":
        #ls
        tree[curfolder] = []
        array = []
        while text[stepfd + line][0] != "$":
            array.append(text[stepfd + line])
            stepfd += 1
        tree[curfolder] = array

def hasnodirs(dir):
    for item in tree[dir]:
        if item[0].isdigit():
            return(dir)
        else:
            return(False)

#directories that only contain files (no other directories)

def getfilesize(folder):
    total = 0
    for file in folder:
        for char in range(len(file)):
            if file[char] == " ":
                endofnum = char
                break
        total += int(file[:endofnum])
    total = str(total) + " "
    return total


def searchandreplace():
    for dir in tree:
        for item in range(len(tree[dir])):
            if tree[dir][item].startswith("dir"):
                dirname = tree[dir][item][4:]
                for finaldir in range(len(finaldirs)):
                    if dirname == finaldirs[finaldir]:
                        tree[dir][item] = finaldirsizes[finaldir]
                        break

foldersizes = {}

while not hasnodirs("/"):
    finaldirs = []
    finaldirsizes = []

    for dir in tree:
        if hasnodirs(dir):
            finaldirs.append(dir)


    for finaldir in finaldirs:
        finaldirsizes.append(getfilesize(tree[finaldir]))

    for i in range(len(finaldirs)):
        foldersizes[finaldirs[i]] = finaldirsizes[i]

    searchandreplace()

part1tot = 0

for dirsize in foldersizes:
    if int(foldersizes[dirsize]) <= 100000:
        part1tot += int(foldersizes[dirsize])

print(part1tot)

sry for destroying chats

but does anyone know why this only works for test data?

[teal orchid]

wait so set it twice?

curr_dir = root = Folder('/')
for c in commands:
    if 'cd' in c.split() and c.split()[2] == '/':
        curr_dir = root

[viscid sail]

#1049917148186804275 message
guess that isn't allowed :/

[rancid heath]

Yeah, that works; I'd recommend making cd handling an if in an if though

if 'cd' in c.split():
  if c.split()[2] == '/': curr_dir = root
  elif c.split()[2] == '..': ...

[teal orchid]

yeah thats where im up to now

root = Folder('/')
for c in commands:
    if 'cd' in c.split():
        if c.split()[2] == '/':
            curr_dir = root
        if c.split()[2] != '..':
            curr_dir = curr_dir.get_folder(c.split()[2])
        if c.split()[2] == '..':
            curr_dir = curr_dir.parent
    elif not c.startswith('$'):
        if 'dir' in c.split():
            new_folder = Folder(c.split()[1], parent=curr_dir)
            print(f"curr_dir: {curr_dir}, new_folder: {new_folder}")
            curr_dir.add_subfolder(new_folder)
        else:
            new_file = File(c.split()[1], int(c.split()[0]))
            curr_dir.add_file(new_file)
            curr_dir.set_folder_size(new_file.get_size())
        
print(root)

[faint lava]

and followed up on my idea that files are just folders without children:

class File(namedtuple('File', 'name fsize folders files parent')):
    @property
    def size(self):
        return self.fsize + sum(d.size for d in [*self.folders.values(),*self.files])
    def find(self,n,m):
        if m>self.size>n: yield self
        yield from [f for d in self.folders.values() for f in d.find(n,m)]

curr = drive = File('/',0,{},[],None)
for line in open(filename).read().splitlines():
    match line.strip('$').split():
        case ['cd','/']  : curr = drive
        case ['cd','..'] : curr = curr.parent
        case ['cd',f]    : curr = curr.folders[f]
        case ['dir',f]   : curr.folders[f] = File(f,0,{},[],curr)
        case [n,f]       : curr.files.append(File(f,int(n),{},[],curr))

print('part1', sum(d.size for d in drive.find(0,100000)))
print('part2', min(d.size for d in drive.find(drive.size - 40000000, drive.size)))

[bronze jewel]

oooooof

[teal orchid]

now the parent just doesnt set?

curr_dir: ('/', 0, None, [], []), new_folder: ('a', 0, None, [], [])
curr_dir: ('/', 23352670, None, [('a', 0, None, [], [])], [('b.txt', 14848514), ('c.dat', 8504156)]), new_folder: ('d', 0, None, [], [])
curr_dir: ('a', 0, None, [], []), new_folder: ('e', 0, None, [], [])
Traceback (most recent call last):
  File "c:\Python\AdventOfCode\day7\day.py", line 57, in <module>
    curr_dir = curr_dir.parent
AttributeError: 'NoneType' object has no attribute 'parent'

Probably gonna take a break for now lol

[faint lava]

I don't mind having a few of the inputs posted, please don't go on a quest to collect many or all of the inputs for every puzzle. Doing so makes it that much easier for someone to clone and steal the whole site. I put tons of time and money into Advent of Code, and the many inputs are one way I prevent people from copying the content.
I think this sums it up. In general if you're helping someone and they give you their input to test, that's ok.
But don't just go posting your input on a public forum, and don't go trying to collect a lot of different inputs.
it's unkind.

[faint lava]

your curr_dir at that point is None. consider adding some prints to see how that happened, or use a debugger and follow the values as they change.

[bronze jewel]

@viscid sail it is interesting because why would the input go further back than cd /? possibly an error with your input values? But i guess I can just add in a try-catch or if-statement into there

[faint lava]

And I think I'm happy with that current version, the match/case looks clean, and the class seems a lot simpler.

And mostly, I feel preapred if this is a callback puzzle where he asks us to add more functionality.

[rancid heath]

That means you're trying to cd .. while in /

[rancid heath]

Oh god maybe

[teal orchid]

oh duh

[rancid heath]

I should do a proper rewrite tomorrow

[teal orchid]

== / is also != ..

[rancid heath]

Wait hold on wtf

        if c.split()[2] != '..':
            curr_dir = curr_dir.get_folder(c.split()[2])
        if c.split()[2] == '..':
            curr_dir = curr_dir.parent

What

[viscid sail]

possibly a list of cd .. emptying the list

[rancid heath]

Oh wait I can't read nvm lmao

But also just make that an if-elif chain lmao

I'd add some debug-prints to that logic though

[bronze jewel]

learnt a fair bit from this. thanks

[rancid heath]

Print 'finding folder [what are we looking for] in [current folder name]' at the start of the if cd bit and 'now in folder [new folder name]' at the end of the if cd bit and you should be able to find out what's wrong @teal orchid

[faint lava]

Awesome. I don't get to use match/case much so this was fun to play with.
that last case used to have an if on it too before I stripped the '$', which is handy to know you can do:

case [n,f] if n.isdigit():
    curr.files.append(File(f,int(n),{},[],curr))

[teal orchid]

finding folder / in ('/', 0, None, [], [])
now in folder ('/', 0, None, [], [])
finding folder a in ('/', 23352670, None, [('a', 0, None, [], []), ('d', 0, None, [], [])], [('b.txt', 14848514), ('c.dat', 8504156)])
now in folder ('a', 0, None, [], [])
finding folder e in ('a', 94269, None, [('e', 0, None, [], [])], [('f', 29116), ('g', 2557), ('h.lst', 62596)])
now in folder ('e', 0, None, [], [])
finding folder .. in ('e', 584, None, [], [('i', 584)])
now in folder None
finding folder .. in None

...how

[faint lava]

sounds like the parent isn't being set correctly?

[faint lava]

where do you set the parent?

[teal orchid]

So in the class def I have self.parent = None and the line is

if 'dir' in c.split():
  new_folder = Folder(c.split()[1], parent=curr_dir)
  curr_dir.add_subfolder(new_folder)
else:
  new_file = File(c.split()[1], int(c.split()[0]))
  curr_dir.add_file(new_file)
  curr_dir.set_folder_size(new_file.get_size())

[viscid sail]

@faint lava did you take a look at today's golf? take a shot at it if you like 🙂

[faint lava]

I got mine down to 17 readable lines and I'm pretty happy with it. 🙂 I've mostly given up on actual golf this year. Last year I had some fun with it. This year no time.

you want self.parent = parent and None as the default in the def.

[bronze jewel]

yep. in my case I left it in because I went the path of ignoring the dir <folder> . If it goes, then the code gets confused. but it can also be easily solved with adding another case for that

I have no idea how people can golf.... it just hurts my brain

3 characters in and I am already lost

[rancid heath]

@teal orchid this is the solution to your issue (notice in your output all the parents are null)

[faint lava]

I don't think of it as coding, more like it's puzzle solving.

[bronze jewel]

and I admire the okes that can do that stuff. Truly impressive

[teal orchid]

yeah about that

[faint lava]

It can be fun, but eeking out those last few chars can be time consuming.

[teal orchid]

or maybe it just looks weird

still fucky

[faint lava]

yes that looks better.

[rancid heath]

That looks like it's behaving, you can turn off the prints now

[faint lava]

maybe want to trim what you print when printing a folder, but the info is good.

[bronze jewel]

aight, see you guys again for day8

[rancid heath]

(that's also why I said print folder name FYI LMAO)

[faint lava]

I'm out too. ciao all.

[rancid heath]

I should sleep also

[red canyon]

parsing feelsbadman

[teal orchid]

yeah its still not working but ill see what i can do, thanks for your help

[red canyon]

i was definitely too late for this

[rancid heath]

Your output looks fine

[red canyon]

had to redo it because apparently there's duplicates

[rancid heath]

That's just what recursive structures look like when you print them like that

[teal orchid]

Traceback (most recent call last):
  File "c:\Python\AdventOfCode\day7\day.py", line 55, in <module>
    curr_dir = curr_dir.get_folder(c.split()[2])
  File "c:\Python\AdventOfCode\day7\day.py", line 23, in get_folder
    sf = f.get_folder(target)
  File "c:\Python\AdventOfCode\day7\day.py", line 23, in get_folder
    sf = f.get_folder(target)
  File "c:\Python\AdventOfCode\day7\day.py", line 25, in get_folder
    raise ValueError(target)
ValueError: d

[red canyon]

speaking of value errors, i had keyerror and had to discover all_directories = defaultdict(lambda: 0)

[faint lava]

Yeah, to catch the people that treat it like a flat file system instead of a hierarchy.

[rancid heath]

Weird

[red canyon]

sadge! it worked perfectly for the test case 😦

[rancid heath]

OH

That's because you're doing DFS instead of BFS

So you run out of possibilities in an unrelated folder before finding the real one

Just change the raise to a return None and it should be fine

[faint lava]

you can probably still do that if you include the path as part of the name, so likley not a big fix, or at least it's not a rewrite

[rancid scaffold]

I spent hours trying to do this in a python way and I just cannot work out how... so this is what I did:

[rancid heath]

Lmao

[red canyon]

yeah i tried that but ended up restarting it anyway

[rancid heath]

https://www.reddit.com/r/adventofcode/comments/zeubth/2022_day_7_i_created_a_copy_of_the_elves/

[faint lava]

nice visualization!

[red canyon]

i might definitely revisit this one

[rancid heath]

I love the number of people who all did the same 'just make the folders lol' thing

[rancid scaffold]

Aha they actually made the files the right size - mine are just text files that include the "size"

[rancid heath]

Myself included, though after I solved the puzzle

[rancid heath]

(that was me)

[rancid scaffold]

Nice!

[rancid heath]

And yeah, using a lipsum generator

[rancid scaffold]

I could not work out how to do it any other way

[red canyon]

github allowed that 20gb monstrosity?

[rancid heath]

4MB...

[red canyon]

or was it 2

oh

[cyan musk]

Any python aoc utils you recommend? To get data , submit answer from python.

[rancid scaffold]

mine only comes in at 217kb

[red canyon]

this was hella relatable

[red canyon]

there's aoc lube

[rancid heath]

If you want plain-and-simple, nothing fancy: advent-of-code-data
If you want batteries-included aoc solving, there's aoc_helper or aoc_lube

if you get aoc_helper, install 1.6.15 because 1.7 includes @young nacelle's broken testing PR and I didn't get around to reverting that yet

[cyan musk]

Thanks!

[cyan musk]

Thank you

[red canyon]

i need day 6 difficulty again

moving crates was fun too

this one.. not quite

[rancid heath]

Yeah today was kinda painful

First day >1k :(

[rancid scaffold]

Day 5 and 6 i felt like I had this

[cyan musk]

I just need access to chatGPT then I'm good

[rancid heath]

(~2.5k p1 and ~2k p2)

[red canyon]

crates, day5? was probably my favorite

[rancid scaffold]

today I feel like... I got to the answer but I feel nothing but shame

[red canyon]

i don't think it solves advent of code

[rancid scaffold]

See leaderboard

[rancid heath]

People literally used it to solve aoc with 10s times

[red canyon]

wtf

[faint lava]

LOL, someone made a github of the elves' filesystem with all of the files filled with junk text to meet the size.

[rancid heath]

If you're talking about https://www.reddit.com/r/adventofcode/comments/zeubth/2022_day_7_i_created_a_copy_of_the_elves/, that was me

[rancid scaffold]

#1049917148186804275 message

[cyan musk]

It does. You can 1:1 copy puzzle and it spits out python code. It's crazy

[faint lava]

You're bonkers.

[red canyon]

god damn

[faint lava]

I love it.

[rancid heath]

Leaderboard got cleaned I think

[red canyon]

i thought it would work for 1 or two liners but can you really throw in a whole ass question in there?

[rancid heath]

Wait no nvm

[cyan musk]

Yep

[rancid scaffold]

hmmm, not sure I agree with that

[faint lava]

they solved some of the days pretty quickly, and then got hammered the last couple of days.

[rancid scaffold]

So apparently the search is "deep first" so you can just replace the cd with open and lcose brackets and it all solves out into a nice nested list?

[red canyon]

yesterday was basically cake, i wonder how they did it today

[cyan musk]

https://twitter.com/max_sixty/status/1599270031996903424?s=46&t=AbrCTdPNL2J6u52szdjZBQ

Max (@max_sixty)

I placed 1st (out of ~100,000) in #AdventOfCode Part 1 today, again by having GPT-3 write the code

A win for GPT-3, @OpenAI, AI — and the creativity of humans!

Likes

300

Twitter · Dec 4, 2022, 5:11 AM

[faint lava]

didn't make the leaderboard for day7

neither of them

[red canyon]

oof

[rancid scaffold]

I imagine the AI solutions will either solve it instantly or never

[red canyon]

are people actually competing for the leaderboards?

not me waking up 10 hours after the problem goes live

[rancid scaffold]

Not personally

but people do

[red canyon]

to each their own i guess, feeling rewarding is one thing but taking them hella seriously is defo not healthy

[rancid scaffold]

for me AOC is about learning to write better code and picking up some new tricks. better code > fast code

[red canyon]

+1

definitely keeps my programming skills in check

[cyan musk]

Global too fast for me but private lb yes 🙂

[rancid heath]

I do

[rancid heath]

I made global once this year

[red canyon]

busts out chatgpt3

[rancid heath]

Definitely used to be better than I am now

[rancid scaffold]

its a good community thing too - plenty of help so you can get a wide range of solutions to a problem unlike basically any other problem you come accross

[rancid heath]

No

[cyan musk]

Congratulations

[rancid heath]

I'm gonna go for the night, see you all in 6 hours or so

Well, some of you

[rancid scaffold]

My private leaderboard we do based on stars, not time. We are all busy with jobs / life etc so finding time to sit down and do it might not be any time competitive.

[faint lava]

Some do. Not me. I had a good result last year. Pretty casual this year.

[red canyon]

i feel bad for people shitting on the tweet author for using the ai and yoinking leaderboard position

[rancid scaffold]

Honestly I dont mind the Ai stuff. You cant prevent it so embrace it

[red canyon]

it's not like i'm gonna get a position on the leaderboard anyway

[faint lava]

I don't mind the AI stuff, but claiming that you solved it in 10 seconds using the AI is a little disingenuous.

[red canyon]

that i can 100% get behind

[rancid scaffold]

Depends i guess, if you wrote the Ai, and the scripts to support it then I think you deserve some credit

[faint lava]

Yeah that seems fair.

[rancid scaffold]

if you just pulled it off github and sovle_aoc() then.... not really your work

[faint lava]

Like if this was google as a team posting their score I'd be like damn.

[rancid scaffold]

To be fair there is only a few people making it work

which makes me think its not that straightforward.

[red canyon]

it is straightforward for obvious problems, i just gave the bot "solve advent of code 2022 day 7" and the bot was like oops sorry can't pull questions from there yet™️

[rancid scaffold]

It seems like microsoft have exclusive use of GPT-3 so I don't know how they are feeding it custom data

[elder bolt]

Just one private leaderboard and aiming to end top 100 on the server leaderboard

[rancid scaffold]

I get up too late to be on a leaderboard

[crystal pier]

Tried to do it in Rust, but couldn't manage the input parsing, so I gave up and did it in python. Code is too long for discord, so splitting it up into parsing and problem-solving sections

from dataclasses import dataclass
from functools import cache


class Node:
    def __init__(self, parent, data):
        self.parent = parent
        self.data = data
        self.children = dict()

    def root(self):
        node = self
        while node.parent:
            node = node.parent
        return node

    def add_child(self, data):
        item = Node(self, data)
        self.children[data] = item


@dataclass(frozen=True)
class Directory:
    name: str


@dataclass(frozen=True)
class File:
    name: str
    size: int


Item = Directory | File


def parse(base: str) -> Node:
    lines = (r.split() for r in base.split("\n") if r)
    name = Directory(next(lines)[2])
    root = Node(None, name)
    current_node = root
    for line in lines:
        # print(line)
        if line == ["$", "ls"]:
            continue
        elif line == ["$", "cd", ".."]:
            current_node = current_node.parent
        elif line == ["$", "cd", "/"]:
            current_node = current_node.root()
        elif line[1] == "cd":
            desired = Directory(line[2])
            current_node = current_node.children[desired]

        elif line[0] == "dir":
            name = Directory(line[1])
            current_node.add_child(name)

        else:
            [size, name] = [int(line[0]), line[1]]
            value = File(name, size)
            current_node.add_child(value)

    return current_node.root()

[soft hatch]

@inner ingot maybe you'll get a kick out of this mess

rm -rf root
mkdir root
grep -v '\bls\b' | perl -p -e 's/^dir/mkdir/;s#^([0-9]+) (.*)#dd if=/dev/zero of=\2 bs=\1 count=1#;s/^\$ //;s#cd /#cd root#' |
while read cmd; do
  $cmd
done

size() {
  wc -c $@ | tail -1 | grep -o '[0-9]*'
}

all_sizes() {
  find root -type d |
  while read f; do
    size $(find $f -type f)
  done
}

all_sizes | grep -P '^([0-9]{1,5}|100000)$' | xargs | tr ' ' + | bc
total=$(size $(find root -type f))
torem=$((30000000 - (70000000-total)))
all_sizes | awk "\$1 >= $torem" | sort -g | head -1
```running it
```sh
algmyr: [git] ~/tmp
> bash do.sh < 07input 2> /dev/null
1543140
1117448
algmyr: [git] ~/tmp
> ll root
total 240K
drwxr-xr-x  3 algmyr algmyr 4.0K Dec  7 23:29 dcvzbqf
drwxr-xr-x  9 algmyr algmyr 4.0K Dec  7 23:29 qdtw
drwxr-xr-x 10 algmyr algmyr 4.0K Dec  7 23:29 qmfvph
-rw-r--r--  1 algmyr algmyr  24K Dec  7 23:29 gsdpmrq.bsz
-rw-r--r--  1 algmyr algmyr  25K Dec  7 23:29 nfngbl.mcn
-rw-r--r--  1 algmyr algmyr 175K Dec  7 23:29 plw.frm

[crystal pier]

@cache
def size(node: Node) -> int:
    assert isinstance(node.data, File) != bool(node.children)
    if isinstance(node.data, File):
        return node.data.size
    else:
        return sum(map(size, node.children.values()))


def flatten(c):
    return (a for b in c for a in b)


def sizes(node: Node) -> list[tuple[Item, int]]:
    return [(node.data, size(node))] + list(flatten(map(sizes, node.children.values())))


def part1(base: str) -> int:
    data = parse(base)
    # print(data)
    # print(size(data))
    # print(sizes(data))
    valids = [
        size
        for (item, size) in sizes(data)
        if isinstance(item, Directory) and size <= 100000
    ]
    return sum(valids)


def part2(base: str) -> int:
    data = parse(base)
    total = 70000000
    desired = 30000000
    available = total - size(data)
    required = desired - available
    posses = [
        (item, size)
        for (item, size) in sizes(data)
        if isinstance(item, Directory) and size >= required
    ]
    val = min(posses, key=lambda x: x[1])
    print(val)
    return val[1]


def main():
    file = "data/07.txt"
    with open(file) as f:
        words = f.read()
    print(part1(words))
    print(part2(words))


if __name__ == "__main__":
    main()

[soft hatch]

yes, this is rewriting the input to commands that actually navigate around and create files, and then the operations is done on the actual files

[teal orchid]

ah shit it works on the test but not the real input

[crystal pier]

Misread it as needing the folder name, not the folder size, so printed/kept more info than required.

[teal orchid]

Wait a lot of this repeats, so I have to check if a folder exists yeah?

I didnt realize we go in and out

[heavy frigate]

any cd into a directory is always valid

but there's likely something in the real input that AoC sneakily didn't do with the example

[teal orchid]

im just thinking because everytime i see a dir i try to add it

if 'dir' in c.split():
  new_folder = Folder(c.split()[1], parent=curr_dir)
  curr_dir.add_subfolder(new_folder)

and i got a

Traceback (most recent call last):
  File "c:\Python\AdventOfCode\day7\day.py", line 77, in <module>
    curr_dir.add_file(new_file)
AttributeError: 'NoneType' object has no attribute 'add_file'

thats all i can think of rn

[soft hatch]

@heavy frigate rate my terrible thing

[narrow sail]

you only have to count files that are cd'd

[teal orchid]

we dont look in every directory?

[lone schooner]

I've been busy, but I managed to get day 7 done this morning.

[soft hatch]

I'm pretty sure you do

[lone schooner]

yay trees

[teal orchid]

oh ok

[lone schooner]

https://github.com/killjoy1221/aoc22/blob/main/day7.py#L99-L110

[heavy frigate]

You built out the actual filesystem? 😄

[tame grove]

day7.py lines 99 to 110

# part 1
print("Part 1:", sum(d.size for d in root.dirs() if d.size < 100000))

# part 2
total_storage = 70_000_000
required_storage = 30_000_000

storage_used = root.size
storage_free = total_storage - storage_used
to_free = required_storage - storage_free

print("Part 2:", min(c.size for c in root.dirs() if c.size > to_free))```

[teal orchid]

yeah im not entirely sure whats up because it worked with the test input

[soft hatch]

indeed

[heavy frigate]

I did contemplate that at one point when I hit a snag 😄

[soft hatch]

it's actually not that slow ~1.1s on an HDD

[red canyon]

rust moment, this looks neat as hell with classes

[lone schooner]

that device must be old. Only 70Mib of space.

[soft hatch]

so, one very annoying thing is that directories have sizes

[heavy frigate]

lol imagine if AoC snuck in a few 10G files just to mess with people taking that approach 😛

[soft hatch]

70MB

actually base 10 for once

[red canyon]

was it not like that? 👀 i thought i saw 70gb in one of the viz. on subreddit

[heavy frigate]

the device can only hold 70,000,000 bits

[lone schooner]

maybe they assumed it was the size in kilobytes

[heavy frigate]

so not likely

[lone schooner]

!e print(70_000_000 / 8, "bytes")

[tame grove]

@lone schooner :white_check_mark: Your 3.11 eval job has completed with return code 0.

8750000.0 bytes

[narrow sail]

bits?

[soft hatch]

bytes

[narrow sail]

isn't disk space measured in bytes

[soft hatch]

actually, what unit are they even using?

I assumed bytes

[lone schooner]

they don't say

[narrow sail]

elvites

[heavy frigate]

to be fair I don't think the description actually says what the units are

[red canyon]

aye there's no description

probably 70mb then

[soft hatch]

millibits?

😛

[red canyon]

what are elves doing with files anyway

don't they have gifts to arrange

[rancid scaffold]

#1049917148186804275 message

[heavy frigate]

Aren't you paying attention, they need those yummy stars

[lone schooner]

apparently nothing because all the folders are equally eligible for deletion without risking stability

[narrow sail]

gifts are now in folders because they've run out of materials due to the pandemic and they're going digital

[heavy frigate]

It's the classic "I don't know what this is so I will delete it" school of making space on your computer 😄

[narrow sail]

or they just don't want to do forced labor any more so they just waste time on making files

[soft hatch]

using regex to build commands to make the files felt so right somehow

[red canyon]

what did i help them move those damn crates 👿

[soft hatch]

so dumb, yet so right

[lone schooner]

Why is /usr/lib taking up so much space?

Better delete it

[soft hatch]

true

[rancid scaffold]

I feel like this might be possible with mostly replace commands

[lone schooner]

inb4 the system runs on ROM

[heavy frigate]

Just send it to /dev/null. You can email Linus Torvalds to get it back later.

[red canyon]

tries to install update
it's popos
it's popos

[soft hatch]

who ever needed anything from /usr/lib?

[lone schooner]

sudo mv /dev/* /dev/null

[soft hatch]

I'm curious what that will actually end up doing

[lone schooner]

nothing because /dev doesn't have any regular files

probably

[soft hatch]

right

that's what I'm guessing

but I don't feel like trying to mv one of my drives in /dev to test

[red canyon]

time to docker

[lone schooner]

If you move /dev/core to /dev/null, you get errors about the same file. moving any other device to /dev/null fails because "Device or resource busy"

[last flower]

Hmm, I think there's too much being stored in /dev/nvme0. Best I move the folder to my hard drive.

[narrow sail]

bets on day 8 6h 6m 18s later?

[soft hatch]

which one is busy?

the disk?

[lone schooner]

probably

[soft hatch]

I guess you could try to umount it first

[last flower]

what happens if I add /dev/null to my fstab

[soft hatch]

🥴

[lone schooner]

root@49b586774add:/# umount /
umount: /: must be superuser to unmount.

root is not superuser

[soft hatch]

lol what

[rancid scaffold]

Hmmm, could you regex it ?

cd (.*?)\$ ls(.*?)\$

[soft hatch]

yes, grep and then slapping some regex on it does the job to create runnable commands

[rancid scaffold]

https://regex101.com/r/htPnJN/1

[teal orchid]

so i did some tracing to where my error starts, and its around here

# Printing current directory:
...
('nblfzrb', 160378, [], [('jhc', 160378)])
('nblfzrb', 249412, [], [('jhc', 160378), ('fst', 89034)])
('nblfzrb', 587555, [], [('jhc', 160378), ('fst', 89034), ('sdwnsgwv.mjm', 338143)])
('nblfzrb', 718216, [], [('jhc', 160378), ('fst', 89034), ('sdwnsgwv.mjm', 338143), ('vmpgqbcd', 130661)])
None
Traceback (most recent call last):
  File "c:\Python\AdventOfCode\day7\day.py", line 74, in <module>
    curr_dir.add_file(new_file)
AttributeError: 'NoneType' object has no attribute 'add_file'

I think my cd is trying to back out too far or something? not 100% sure

[rancid scaffold]

I think this might be a quick solution to be honest : https://regex101.com/r/hz8ECy/1

Then all you have to do is parse the matches

[queen breach]

any better golfs yet?

[narrow sail]

what's the best one and its length

[queen breach]

i believe this one

[narrow sail]

here's a 188 then ```py
s=[];d=()
for r in open(0):
a,,b=r.split()
if'/'>b:*s,v=s;d+=v,;s[-1]+=v
elif'$ d'>r:s+=0,
if'/'<a<':':s[-1]+=int(a)
print(sum((1e5>)for _ in d),min( for _ in d if _>sum(s)-4e7))

[queen breach]

oh nice

[cosmic depot]

I didn't know you could do that

[narrow sail]

behaviour of x += y in lists are different from x = x + y in the way that you can put any iterable as y in the first one
(also true for dictionaries but using |=)

[narrow sail]

i don't know why i converted d into a tuple though

[narrow sail]

updated ```py
m=[]d=[]s.xscanall("(%B+)%B+(.+)",(a,b):{ifzip'/'>b:'/'<a<':'{@m,v=m d+=v,;m[-1]+=v}m[-1]+=int(a)elif'$ d'>r m+=0,})print(sum(x*(1e5>x):x~>d),min(x:x~>d:x>d[()]-4e7))

might merge ifzip and if into one

[vivid patrol]

only optimized based off of characters, not bytes smh

[narrow sail]

120 degen ```py
exec(bytes('㵳嵛搻⠽਩潦⁲⁲湩漠数⡮⤰਺愠⨬ⱟ㵢⹲灳楬⡴਩椠❦✯戾⨺ⱳ㵶㭳⭤瘽㬬孳ㄭ⭝瘽 汥晩␧搠㸧㩲⭳〽ਬ椠❦✯愼✼✺猺ⵛ崱㴫湩⡴⥡瀊楲瑮猨浵弨⠪攱㸵⥟潦⁲ 湩搠Ⱙ業⡮ 潦⁲ 湩搠椠⁦㹟畳⡭⥳㐭㝥⤩','u16')[2:])

[lone schooner]

📉 Not stonks! 😠

[vivid patrol]

where's the optimized for bytes solution 😠

[queen breach]

here

but not in python

[vivid patrol]

💀

[narrow sail]

the ruby expert from TCCPP made this ```ruby
$stdin = open 'day7.txt'

A=[];D=proc{s=0;$<.map{_1['..']&&break;s+=_1["$ c"]?D[]:+_1.to_i};s.tap{A<<+_1}};C=D[];p([A.select{_1<1e5}.sum,A.select{C-("|\s";4e7)<=_1}.min]);/#/

openai chatbot managed to explain it

[vivid patrol]

wow

[narrow sail]

yeah i'm going to add some parts of it into my WIP language

[royal axle]

can someone explain the question in simple English please

I cannot understand the question

for example what does this mean 14848514 b.txt?

[summer herald]

where would the place be to ask for help on this?

I've beat myself against this for 12 hours

[inner ingot]

here

[teal orchid]

the size of the file and the name of it

[royal axle]

12 hours

[summer herald]

I did yesterday's in 10 minutes -_-

[royal axle]

would making a class using list be a good idea?

or I'm just complicating the problem

[teal orchid]

I used a class, but lots of people did not

[summer herald]

is it alright to post the code I have so far here?

[royal axle]

yes

[summer herald]

import re
from itertools import chain


class Directory:

    def __init__(self, identity, parent=None):
        self.identity = identity
        self.parent = parent
        self.contents = []
        self.size = 0

    def add(self, identity, type, parent, size=None):
        if type == 'file':
            self.contents.append(File(identity, size))
        else:
            self.contents.append(Directory(identity, self))
        self.parent = parent
        size = self.get_size()

    def get_size(self):
        size = 0
        for item in self.contents:
            if isinstance(item, File):
                size += int(item.size)
            else:
                size += int(item.get_size())
        return size

    def iterable_of_sizes(self):
        return chain([self.size],
                     *[child.iterable_of_sizes() for child in self.contents])

    def cd(self, command):
        if command == '..':
            return self.parent
        else:
            return Directory(command)


class File:

    def __init__(self, name, size):
        self.name = name
        self.size = size

    def iterable_of_sizes(self):
        return chain([self.size])


with open("input_7.txt", "r+")as input_file:
    blocks = [x.rstrip().split("\n") for x in re.split(r"(?=\$)", input_file.read())]
    location = None
    for block in blocks:
        match block[0].split():
            case ["$", "ls"]:
                for entry in block[1:]:
                    if entry.split()[0].isnumeric():
                        location.add(entry.split()[-1], 'file', location, entry.split()[0])
                    else:
                        location.add(entry.split()[-1], location, 'directory')
            case ["$", "cd", dir]:
                if not location:
                    location = Directory(dir)
                location = location.cd(dir)
    print([item for item in location.iterable_of_sizes()])```

This is the second attempt. My first attempt I used treelib and I got the directory structure just fine but then could not work out how to extract the information from it

so I tried making my own classes and throwing in some bits of code from other places and am very unsure about what I am doing. This is a type of problem I just simply haven't encountered before

[royal axle]

if I solved this tomorrow, would I got 2 star? or I should solve it before the next question?

I really want to sleep

[final anchor]

you get both stars

[storm epoch]

yeah you can solve it whenever you want

[final anchor]

your score will be lower, but the stars always remain constant

[storm epoch]

even next year

[royal axle]

Thanks, have a good night

[summer herald]

It's so frustrating. I was breezing through to this point. I was travelling and did the first five on my phone using Pydroid

[teal orchid]

mine is just being weird

it works up to a certain point then dies

[vivid patrol]

?

[teal orchid]

and i know what causes it, but not why

[vivid patrol]

rips

[cosmic depot]

If you are storing just the directory name, ||there may be directories in other places that have the same name||

[teal orchid]

Yeah, I was thinking maybe that but probably not. I have some prints and I know I read in a directory in a dir, but when I cd into it, I just get a NoneType attribute error

def get_folder(self, target):
        if self.name == target:
            return self
        elif self.subfolders:
            for f in self.subfolders:
                sf = f.get_folder(target)
                if sf: return sf
        return None

Traceback (most recent call last):
  File "c:\Python\AdventOfCode\day7\day.py", line 77, in <module>
    curr_dir.add_file(new_file)
AttributeError: 'NoneType' object has no attribute 'add_file'

it works with the demo, and my input through the first 200 lines and a certain something just kills it

[vivid patrol]

directory names aren't unique

i.e.

you can have
cvf/abc
and
dlf/asw/abc

[teal orchid]

so how would I know which one to get? check the parent?

[summer herald]

when I tried my libtree approach I just added an incrementing number tag to each folder name

[vivid patrol]

well you know everything in relation to the other

🤷‍♂️

[summer herald]

│   ├── 13 bnl
│   │   ├── 19 dhw
│   │   │   └── file 237421 vccwmhl
│   │   ├── 23 dmpsnhdh
│   │   │   ├── 28 chf
│   │   │   │   ├── 33 zfgznbz
│   │   │   │   │   ├── file 171574 vglqg
│   │   │   │   │   └── file 179409 cnj.gdn```
etc.

[vivid patrol]

an approach you can use is thru cwd

but i think others just hard coded paths or smth

[teal orchid]

dont think i could implement this

because we cant really know the parent of the target without ... finding it

[young nacelle]

i also used a class ^

[vivid patrol]

you always know the parent of a directory (if you store it)

[vivid patrol]

wtfrick pro

[young nacelle]

lemme most my implementation

post

[vivid patrol]

i think most people with classes did it slower though

[young nacelle]

it's rlly m essy tho

class Node:
    def __init__(self, parent: "Node" = None):
        self._parent = parent
        self.nodes = {}
        self.files = {}

    def create_directory(self, name):
        self.nodes[name] = Node(self)

        return self.nodes[name]

    def go_to_outermost(self):
        if self._parent is None:
            return self
        else:
            return self._parent.go_to_outermost()

    def go_back(self):
        return self._parent

    def add_file(self, size, name):
        self.files[name] = size

    def get_size(self, directory_sizes):
        if not self.nodes.values():
            directory_sizes.append(self._total_size())
            return directory_sizes[-1]
        else:
            result = self._total_size()

            for node_name, node in self.nodes.items():
                result += node.get_size(directory_sizes)

            directory_sizes.append(result)
            return result

    def _total_size(self):
        return sum(self.files.values())

i did it fast so it's kinda ugly

[vivid patrol]

good enouhg

🤷‍♂️

[vivid patrol]

you even added typing(s) wtf

[young nacelle]

oh yeah lmao

[teal orchid]

yeah i have it in the class, but given the name of the directory from cd we need to find the class object first to get the parent?

[summer herald]

any ideas on mine?

[vivid patrol]

what

[summer herald]

I'm just trying to get the total size of all the directories out at the moment but even that's not working. I'm completely confused.

[vivid patrol]

current implementation? @summer herald

[summer herald]

here

would it be better to put it in a paste or something?

[vivid patrol]

immediate thing is that i'm seeing dir and not "dir"

[summer herald]

the dir from the match case is a string

[vivid patrol]

dir is a python func? unless match is weird

[last flower]

you can reassign the value of dir

[summer herald]

like a reserved word?

[vivid patrol]

!e print(dir([]))

[tame grove]

@vivid patrol :white_check_mark: Your 3.11 eval job has completed with return code 0.

['__add__', '__class__', '__class_getitem__', '__contains__', '__delattr__', '__delitem__', '__dir__', '__doc__', '__eq__', '__format__', '__ge__', '__getattribute__', '__getitem__', '__getstate__', '__gt__', '__hash__', '__iadd__', '__imul__', '__init__', '__init_subclass__', '__iter__', '__le__', '__len__', '__lt__', '__mul__', '__ne__', '__new__', '__reduce__', '__reduce_ex__', '__repr__', '__reversed__', '__rmul__', '__setattr__', '__setitem__', '__sizeof__', '__str__', '__subclasshook__', 'append', 'clear', 'copy', 'count', 'extend', 'index', 'insert', 'pop', 'remove', 'reverse', 'sort']

[summer herald]

🤷‍♀️ I can change that name, I don't think it's the issue tbh

[last flower]

!e ```py
dir = "hello"
print(dir)

[tame grove]

@last flower :white_check_mark: Your 3.11 eval job has completed with return code 0.

hello

[vivid patrol]

they don't do that

[summer herald]

yeah, changed that name, doesn't change how it runs

[vivid patrol]

doesn't really seem like you're maintaining uh

relationship between parent and child directories

[summer herald]

almost certainly!

[last flower]

they do. the match statement changes the value of dir, at least under that scope iirc

[summer herald]

I have pretty much just been moving things around at random for the last four hours or so hoping something happens

[vivid patrol]

that is so weird ok then

[vivid patrol]

    def cd(self, command):
        if command == '..':
            return self.parent
        else:
            return Directory(command)```

probably should maintain a list of child directories

and when you call cd search that list

[teal orchid]

My point is that I am storing the parent of the directory, but I have no way to access it unless I have the object, which is what the get_folder function is trying to do

[summer herald]

I hope this doesn't sound petulant, but how?

Like a list internal to the Directory object?

Because it already has one

[vivid patrol]

!e

class Directory:
  def __init__(self):
    self.child_dirs = {}
  def cd(self, cmd):
    if cmd == "..":
      pass
    else:
      return self.child_dirs[cmd]
  def add_dir(self, dir_name):
    self.child_dirs[dir_name] = Directory()

a = Directory()
a.add_dir("blah")
print(a.cd("blah"))

oops

[tame grove]

@vivid patrol :white_check_mark: Your 3.11 eval job has completed with return code 0.

<__main__.Directory object at 0x7f7218570710>

[vivid patrol]

then keep a variable pointing to the current directory 🤷‍♂️

[teal orchid]

or I guess since we are cding in from the same level, the parent would be the same as the parent of the current directory

yep

[vivid patrol]

lowkey it would actually be fun if they did cd to realpaths

instead of relatives

[summer herald]

sorry, it might just be tiredness/frazzledness but I don't see how to integrate that into what I have, or are you saying what I have is unfixable

[vivid patrol]

yours probably will look like

    def cd(self, command):
        if command == '..':
            return self.parent
        else:
            for i in self.contents:
              if i.identity == command: 
                return i

idk

also i have little clue how match works 🤷‍♂️

[summer herald]

it's just mostly a neater way of doing if/else

case ["$", "cd", directory]: is triggered if the input is $ cd * and it passes the third word in as 'directory'

[vivid patrol]

it's different from other language's switches

and i never really felt like figuring it out kek

[last flower]

I've had people in python general yell at me after saying match is like a switch statement 😔

[vivid patrol]

😔

basic basic it's just a switch statement

but it's a lot more powerful

[last flower]

yes

[summer herald]

python's the only language I can do anything beyond toy programming with

[narrow sail]

match statement

[summer herald]

yeah it can do a lot of parsing

that's more what it's for really as far as I can understand. the tutorial uses the example of a classic text adventure game

[vivid patrol]

🤷‍♂️

[tough hinge]

Have you seen this?
https://realpython.com/python310-new-features/#structural-pattern-matching

[vivid patrol]

😔

[tough hinge]

Incognito?

[vivid patrol]

no

[tough hinge]

[vivid patrol]

better 👍

[vivid patrol]

not this article exactly but i've heard bits and pieces of match

[cyan wadi]

im still stuck in part 1, it works in the example but not in the actual input. i feel sad for getting stuck in day 7...

from dataclasses import dataclass, field


@dataclass
class Directory:
    size : int = 0
    subdirs : list = field(default_factory=list) 


f = open("./puzzle_input.txt", "r")
get_next_line = lambda file: file.readline().rstrip("\n")
max_dir_sum = 100000
dirs : dict[str, Directory] = {}

line = get_next_line(f)

while(line != ""):
    
    _, cmd, dir = line.split(" ") 

    if cmd == "cd" and dir != "..":
        line = get_next_line(f)
        _, cmd = line.split(" ")

        if cmd == "ls":
            if dir not in dirs:
                dirs[dir] = Directory()

            while((line := get_next_line(f)) != "" and not line.startswith('$')):
                val, name = line.split(" ")

                if(val == "dir"): # directory
                    dirs[dir].subdirs.append(name)
                    
                else: # file
                   dirs[dir].size += int(val)

            continue
    line = get_next_line(f)

f.close()
result = 0
for dir in dirs.values():
    dir_sum = dir.size + sum(dirs[sub].size for sub in dir.subdirs)
    if dir_sum <= max_dir_sum:
        result += dir_sum


print(result)

[vivid patrol]

just never actually did a dive into it

[vivid patrol]

directories don't necessary have unique names

i.e.
ads/asd/foo
and
bcd/foo
can both exist

[cyan wadi]

true

so i must store the entire path for each directory?

[vivid patrol]

probably one way of doing it

[cyan wadi]

okay, thanks a lot ^^

[summer herald]

I've tried this and several variations and it keeps giving me errors indicating that it's returning None

[vivid patrol]

i probably misred something or another on how you stored stuff

tldr is that you're trying to find the folder name that matches the input

[summer herald]

I get that

[slender bane]

ohh i didn’t think of doing that

[summer herald]

I suspect it's getting called before the contents list is populated

[vivid patrol]

it shouldn't

[slender bane]

i just went straight for the tree approach with a node class

[tough hinge]

Have you posted your solution?

[vivid patrol]

check to see if you're actually populating your list, and print the steps out when ur running on the test input

[summer herald]

Yeah. Did some print-debugging. It's the root

[slender bane]

https://paste.pythondiscord.com/anelafecum

[teal orchid]

yeah, ive just confirmed thats 100% my problem, I made a MWE and you can see the problem from printing the resulting tree:

$ cd /
$ ls
dir a
1 test.txt
dir b
$ cd a
$ ls
dir b
2 test2.txt
$ cd ..
$ cd b
$ ls
100 hund.txt

name: / size: 103 subdirs: [name: a size: 102 subdirs: [name: b size: 100 subdirs: [] files: [100]] files: [2], name: b size: 0 subdirs: [] files: []] files: [1]

Looks like b is going into the subdirectory of A instead of /

[summer herald]

the cd is called before any ls

[cyan wadi]

must be harder right?

[slender bane]

¯_(ツ)_/¯

seemed fine to me

[vivid patrol]

i

paths

is kinda weird

[summer herald]

I'll try and implicate a special case for the root

[slender bane]

see here if you'd like to take a peek at my solution @cyan wadi

[vivid patrol]

!e

dirs = {"/": [(1234, "abcd.txt")], "/a": [(3, "test.txt")], "/a/b": [(1111, "abcc.txt")], "/c": []}
for i in dirs:
  size = 0
  for j in dirs:
    if j.startswith(i):
      size += sum(map(lambda x: x[0], dirs[j]))
  print(f"Size of directory {i} is {size}")

interesting

[cyan wadi]

okay

[vivid patrol]

yeah that works

n^2 though 💀

[summer herald]

goes further but None's out again

[cyan wadi]

looks good but i dont think its really needed to store the directory tree with nodes, just iterating the file line by line after 'ls' knowing which directory was cd'ed

but thanks anyway 🙂

[vivid patrol]

pain to calculate directory sizes if you use paths though

[cyan wadi]

having a bit of trouble storing the whole path for each directory

maybe recursion helps

[cyan wadi]

yup

[summer herald]

https://pastebin.com/e2qSYpBC

[vivid patrol]

solution but it's slow af

[vivid patrol]

probably better to do

                if not location:
                    location = Directory(directory)
                else:
                    location = location.cd(directory)```

also uh

[summer herald]

Hmm, maybe I should go back to my treelib approach. I got a multidimensional dict out of it but could not work out how to unpack it at all

[vivid patrol]

                for entry in block[1:]:
                    if entry.split()[0].isnumeric():
                        location.add(entry.split()[-1], 'file', location, entry.split()[0])
                    else:
                        location.add(entry.split()[-1], location, 'directory')```

not sure what this is doing

[summer herald]

adding a file or a directory. but yeah the variables are in the wrong order

[narrow sail]

i'm gonna try to have fun with classes for once in this year's AoC

[tough hinge]

Ah, see, that's what I wanted to do, but I didn't want to make a bunch of searching for the node with the right name, so I wrapped them in dicts recording the names.
In retrospect I don't actually need the names for the solutions though.
Nice job, very clean
https://github.com/shenanigansd/scratchpad/blob/main/events/advent_of_code/2022/7/python/script_day7.py

[slender bane]

ahh i see

[summer herald]

It's more like 14 hours now 😅 gone through demoralisation to just tired bewilderment. I do not understand what I am trying to do.

[slender bane]

yeah, could've used a dict for the children attr instead

[vivid patrol]

oh uh line 18

    self.parent = parent

can you delete that

just need a sanity check

[summer herald]

no change in behaviour

[slender bane]

nice nice, good use of dataclasses as well

[summer herald]

yesterday's was 13 lines. I did it while my coffee was brewing 😭

[vivid patrol]

i recoded it for py 3.9

with if's

didn't break for me

[finite tundra]

I started doing this one and gave up :/ for a relatively new programmer it was almost impossible, idek how to use classes lmao

[vivid patrol]

changed add a little bit too

[vivid patrol]

you don't need classess!

[tough hinge]

Embrace classes!!

[vivid patrol]

classes are slow to write >:((

[finite tundra]

Ok well I have no idea how to do it regardless lmao

[tough hinge]

If you need speed then embrace Rust BTW

[teal orchid]

holy hell, i know exactly what is wrong with my code and what to do to fix it, but i have no idea how

this is so stupid

[inner ingot]

rust would be even slower to write, wouldn't it?

[vivid patrol]

they sent

before i edited

[summer herald]

this completes without an AttributeError?

[tough hinge]

I started typing before they edited in the [to write]

Yeah - Rust BTW takes way longer to write

[vivid patrol]

embrace javascript and approach everything as a dict(ish)

[tough hinge]

oh no

[last flower]

everything is a dict

classes are glorified dicts

[narrow sail]

a dict is a glorified hash array

or it is a hash array

[vivid patrol]

i think you mean an array is just a glorified dict

[summer herald]

a dict is [n] tuples in a trenchcoat

[narrow sail]

no i meant it's just a glorified address

[vivid patrol]

!e

arr = {0: "a", 1: "b"}
print(arr[0])```

[tame grove]

@vivid patrol :white_check_mark: Your 3.11 eval job has completed with return code 0.

a

[vivid patrol]

🙏

no one can tell the difference

it's genius

[bronze leaf]

stack-based solution for nim:

include aoc
import fusion/matching
{.experimental: "caseStmtMacros".}

let sizes = block:
  var sizes, stack: seq[int]

  proc stack_pop =
    sizes.add stack.pop
    stack[^1] += sizes[^1]

  for line in fetch(2022, 7).splitlines.mapIt(it.split):
    case line:
      of ["$", "cd", ".."]: stack_pop()
      of ["$", "cd", _]: stack.add 0
      of ["$", "ls"]: discard
      of ["dir", _]: discard
      of [@size, _]: stack[^1] += size.parseInt

  while stack.len > 1: stack_pop()
  sizes.add stack.pop
  sizes

part 1: sum sizes.filterIt(it <= 100_000)

part 2: min sizes.filterIt(it > sizes[^1] - 40_000_000)

[tough hinge]

Wait - how does that add the sizes to the parents?
Oh - The procedure dumps the previous total back onto the stack?

Dog that looks wierd

Does the input never backtrack or repeat?

[teal orchid]

Just going to throw this here if anyone has any last ideas, I added a check to compare the parent directories when searching to make sure I found the "correct one". I guess its possible that they could still be the same at multiple folders deep? Worked with my test that was broken earlier, nothing on the real input.
https://paste.pythondiscord.com/mewukiqeti