[Help] [2023 Day #1 (Part 2)] Attempted solution - answer not accepted
I'm using this starter kit for the scaffolding so I can simply run mix test for the example and mix d01.p2 to run the solution for my specific data set.
I've got the following code that passes p1 completely and the p2 example but not my specific data set. I can't find any bugs but the result fails for being "too high" and I'm really not sure why this is not being accepted. I have the debug output for each step and running through it manually everything seems right to me.
Is anyone able to point me in the right direction on what I'm missing?
https://pastebin.com/US8ikNLx
defmodule AdventOfCode.Day01 do
def part1(args) do
args
|> String.split(~r/\n/, trim: true)
|> Enum.map(&line_calibration_value/1)
|> Enum.sum()
end
def part2(args) do
args
|> String.split(~r/\n/, trim: true)
|> Enum.map(&words_to_numbers/1)
|> Enum.map(&line_calibration_value/1)
|> Enum.sum()
end
defp words_to_numbers(string) do
numbers = ["one", "two", "three", "four", "five", "six", "seven", "eight", "nine"]
String.replace(string, numbers, fn x ->
(Enum.find_index(numbers, &(&1 == x)) + 1)
|> Integer.to_string()
end)
end
# TODO: doesn't work if the string has no numbers
defp line_calibration_value(string) do
# remove any non-numeric characters
number_from_string = String.replace(string, ~r/[^\d]/, "")
calibration_number = [
# first number
String.at(number_from_string, 0),
# last number, or first number is string has length of 1
String.at(number_from_string, -1)
]
calibration_number
|> List.to_string()
|> String.to_integer()
end
end
Thank you for the test case! I went out for dinner but I'll work on this tomorrow. I strongly suspect this is the exact issue but I'm going to have to rework some things so I want to tackle it with a fresh mind.
I think it's related to the replacement of words with digits. There are some overlapping words, for example in "eightwothree" the "t" is used for both "eighT" and "Two". In this case the order of replacement differs your result. It either becomes "8wo3" or "eigh23".
I think you may be right but the problem is at the end of the string. I'll add some test cases and rewrite the code. I think I'll have to ditch the regex replacements and scan through instead so I don't clobber the string in the wrong place.
Thanks, I managed to find the culprit in the end however - I was using a forward-look of 5 characters for finding if it matches a word so in a string of a3two for example, it would register the two before the 3. It was an easy fix once i found the issue.