Advent of Code 2023 Day 15 (BaseA Style)

Discussion thread for day 15 of the Advent of Code - https://adventofcode.com/2023/day/15

Accepted answers

All comments

Today was a good confidence booster! I started by trying to avoid an iterative macro, but found the logic was more straightforward with a macro than without (at least to me). Now I want to go back and try a non-macro solution!

Spoiler

Part 1:
The bulk of this logic was done in a simple multi-row formula tool with built-in Alteryx functions, which just felt so good. CharToInt() was so useful.

Part 2:
After rereading the instructions approximately 15 times, I finally felt like I understood the problem. It luckily sounded much more complicated than it ended up being in practice.

Re-apply the "HASHMAP" algorithm to just the lens label instead of the full string
Find the first line with a = operation and remove any records before that. My production data wasn't working in my iterative macro because the first 3 operations were removals, so this was my solve for that.
Find max iterations and use that to stop the macro
Use iterative macro to see if a label already existed in a box, and apply the operation. The workhorse of this macro is the join tool between the box map and the instruction set.

Workflow:

Macro:

AOC2023_Day15_KelseyKincaid.yxzp

CoG

This was a fun one, once I figured out the instructions (which took me longer than I'd like to admit...). The method here is pretty much just brute force, so I won't share any further details as to algorithm. I think the workflow itself is a pretty cool use-case for creating conditional logic in Alteryx (i.e. update if row exists, add row if not)

Workflow:

Spoiler

Main Workflow:
_Main Workflow.png

Hash String (iterative macro):
_Hash String.png

Fill Boxes with Lenses (iterative macro):
_Fill Boxes.png

phottovy

I probably spent more time rereading the description than actually building my workflow.

Spoiler

P1:

P2 (which also reuses the P1 macro:

DaisukeTsuchiya

P2 is hard to understand the problem.

Spoiler

I lost my way to find Box2 in P2 sample. @gawa gave me one sentence hint then I got it.
Thank you.

スクリーンショット 2023-12-15 174413.png

<Macro for P2>
スクリーンショット 2023-12-15 175709.png

Day15_Rev1.yxzp

Gawa

It may take a bit long to understand what you are asked to solve in part2, but once you get it, logic would be simpler.

Spoiler

AncientPandaman

I read part 2 for 20+ times. to find out how to allocate the box.

and today also a non macro day.

Spoiler

Part1: token > multiple tool

part2:
1. use part one method for label only (note: input more that 2 characters, waste lot of time to found this)
2. sort descending RecordID
3. take first "=" for value, and take last "=" before "-" for RecordID
4. sort RecordID back to ascending and add sequence_num (slot) by box.
5. calculate as request and sum

recordID	label	action	value	box	type	remark
10	aa	=	5	1	take	first, for value
7	aa	=	4	1	take	last, for recordID
5	aa	-		1	stop
3	aa	=	3	1	stop
1	aa	=	2	1	stop

Day 15.yxmd

cgoodman

Spoiler

I'm surprised with all of the macro based approaches, especially when there are not many macros for part 1.

Part 2 the trick is to group the values by the boxes and then you just need to work out if the operation is to add a new lens, replace an existing lens or take a lens out. So you can build out a string with the lens in order and make use of regex_replace functions to either replace or delete.

Spoiler

Multi-row for part 2:
IF !Contains([Row-1:lensBox],[label]) AND [Operation] = "=" THEN
[Row-1:lensBox]+"["+[label]+" "+tostring([FocalLength])+"]"
ELSEIF
Contains([Row-1:lensBox],[label]) AND [Operation] = "="
THEN
REGEX_Replace([Row-1:lensBox], [label]+'(\s\d)', [label]+" "+tostring([FocalLength]))
ELSEIF
[Operation] = "-" AND contains([Row-1:lensBox],[label])
THEN
REGEX_Replace([Row-1:lensBox], '(\[)'+[label]+'(\s\d\])',"")
ELSE
[Row-1:lensBox]
ENDIF

(Ignore the lack of indentation!)

AkimasaKajitani

Part 1 is straightforward. But Part 2 question text was too difficult to understand. I continued to search the box number.

Spoiler

Raj

Had to spend a good amount of time reading the situation again and again.
although the part 1 was not complicated as excepted

Overall, a good challenge so far and got to know about the ASCII value

Formula use for part 1 - Mod(([Row-1:value] + CharToInt([data]))*17,256)

example if ASCII values
a 97
b 98
c 99
d 100
e 101
f 102
= 61
- 45
1 49
5 53
t 116

alisonpitt

Like others said, this was definitely one where reading the question took the longest. Otherwise, this was a nice quick one...the calm before the storm?

Spoiler

This was the challenge where I realized that I use 2 tools in every workflow now: RegEx (tokenize to rows) and Multi-Row Formula. This was definitely a regex-heavy solution!

Day 15 - AP.yxmd

mmontgomery

P1 was straightforward, P2 in spoiler

Spoiler

P1 was an iterative macro to get the next value without much of a hassle.

For P2, I kept breaking out into separate parts to handle the various cases:
1. No box already determined
2. Match on a box
2a. Match on a fellow member in the box (split into = and -)
2b. No match on a fellow member in the box and rule written accordingly
2c. Remaining boxes with members in them
3. Bring it all together and have the macro run as long as there were rows

2023_Day15.yxzp

DataNath

Finally caught up on Day 15! Took me way too long to troubleshoot but was good to stick at this and finally get it done.

Spoiler

Day 15.yxzp

estherb47

Part 1 was so easy, but my data contained some edge cases that messed with my first approaches.

I had created a macro for the HASH process, but it became too difficult to debug when it came to part 2, so abandoned that approach. Hints to my success in the spoiler.

Spoiler

If you're just looking for "xt7" and your data contain "rxt7", you won't get the correct answer. Think in ways that define each pattern as whole patterns, not parts of patterns.

You can get away with simple text formulas and not solely on regular expressions to get this to work, which is faster in the end as I've found it.

AOC15.yxmd

ccanterman

Solved! I spent longer than I'm going to admit tracking down a mistake I made. Instead of doing a proper RegEx parse (or other RegEx function) for the lens type + power (e.g. abc=2), when I was determining if a label was already present in the box, I just used the Left function with a value of 2, which worked beautifully with the sample data.. and it does not work beautifully with the real data, since I was saying that tcp = tccm and then replacing it.

Spoiler

I ended up rebuilding all of Part 2 and got a much more beautiful and faster workflow. I picked a box and looked at everything that it was doing as a single macro iteration, which was way faster than going through all 4k "steps" one at a time. It also didn't even need to be a macro but I had already built it when I realized it was overkill. We're calling it all learning and I'm moving on

Tokimatsu

Once I understood the problem, I was able to solve it neatly.

Spoiler

Quick Links

This months top contributors

atcodedog05 19458

Qiu 15866

binu_acs 15708

MarqueeCrew 13708

apathetichell 13703