Nim

I hope y’all like nested loops:

• Day 3, part 1
• Day 3, part 2

That was not fun to solve. Lots of ugly code. This is a less ugly second version. Language: Nim.

day_03.nim

Rust

I’ve been using Regexes for every day so far, this time it helped in finding numbers along with their start and end position in a line. For the second part I mostly went with the approach of part 1 which was to look at all numbers and then figure out if it has a part symbol around it. Only in part 2 I saved all numbers next to a gear * in a hash table that maps each gear position to a list of adjacent numbers. Then in the end I can just look at all gears with exactly 2 numbers attached.

Also it has to be said, multiplying two numbers is the exact opposite of getting their ratio!

I wrote today’s program in Python. (I am going to do a different language each day.) The only thing that gave me a little trouble was I was counting “\n” as a part label. Once I realized that I was able to get both problems done very quickly.

My code for the two parts is on github:-

• https://github.com/pngwen/advent-2023/blob/main/day-3a.py
• https://github.com/pngwen/advent-2023/blob/main/day-3b.py

Language: C#

I aimed at keeping it as simple and short as reasonably possible this time, no overbuilding here!

I even used a goto to let me break out of multiple loops at once 🤮 (I had to look up how they worked!) I would totally fail me in a code review!

internal class Day3 : IRunnable
    {
        public void Run()
        {
            var input = File.ReadAllLines("Days/Three/Day3Input.txt");
            int sum = 0;
            string numStr = "";
            var starMap = new Dictionary<(int,int),List>();
            for (int i = 0; i < input.Length; i++)           
                for (int j = 0; j < input[i].Length; j++)
                {
                    if (char.IsDigit(input[i][j]))                    
                        numStr += input[i][j];                    
                    if (numStr.Length > 0 && (j == input[i].Length - 1 || !char.IsDigit(input[i][j + 1])))
                    {
                        for (int k = Math.Max(0, i - 1); k < Math.Min(i + 2, input.Length); k++)                        
                            for (int l = Math.Max(0, j - numStr.Length); l < Math.Min(j + 2, input[i].Length); l++)                            
                                if (!char.IsDigit(input[k][l]) && input[k][l] != '.')
                                {
                                    sum += int.Parse(numStr);
                                    if (input[k][l] == '*')
                                    {
                                        if (starMap.ContainsKey((k, l)))                                        
                                            starMap[(k, l)].Add(int.Parse(numStr));                                        
                                        else
                                            starMap.Add((k,l),new List { int.Parse(numStr) });
                                    }
                                    goto endSymbSearch;
                                }                           
                    endSymbSearch:
                        numStr = "";
                    }
                }            
            Console.WriteLine("Result1:"+sum.ToString());
            Console.WriteLine("Result2:" + starMap.Where(sm => sm.Value.Count == 2).Sum(sm => sm.Value[0] * sm.Value[1]));
        }
    }

Edit: Updated now with part 2.

Managed to have a crack at this a bit earlier today, I’ve only done Part 01 so far. I’ll update with Part 02 later.

I tackled this with the personal challenge of not loading the entire puzzle input into memory, which would have made this a bit easier.

Solution in Rust 🦀

View formatted on GitLab

use std::{
    env, fs,
    io::{self, BufRead, BufReader, Read},
};

fn main() -> io::Result<()> {
    let args: Vec = env::args().collect();
    let filename = &args[1];
    let file1 = fs::File::open(filename)?;
    let file2 = fs::File::open(filename)?;
    let reader1 = BufReader::new(file1);
    let reader2 = BufReader::new(file2);

    println!("Part one: {}", process_part_one(reader1));
    println!("Part two: {}", process_part_two(reader2));
    Ok(())
}

fn process_part_one(reader: BufReader) -> u32 {
    let mut lines = reader.lines().peekable();
    let mut prev_line: Option = None;
    let mut sum = 0;
    while let Some(line) = lines.next() {
        let current_line = line.expect("line exists");
        let next_line = match lines.peek() {
            Some(Ok(line)) => Some(line),
            Some(Err(_)) => None,
            None => None,
        };
        match (prev_line, next_line) {
            (None, Some(next)) => {
                let lines = vec![¤t_line, next];
                sum += parse_lines(lines, true);
            }
            (Some(prev), Some(next)) => {
                let lines = vec![&prev, ¤t_line, next];
                sum += parse_lines(lines, false);
            }
            (Some(prev), None) => {
                let lines = vec![&prev, ¤t_line];
                sum += parse_lines(lines, false);
            }
            (None, None) => {}
        }

        prev_line = Some(current_line);
    }
    sum
}

fn process_part_two(reader: BufReader) -> u32 {
    let mut lines = reader.lines().peekable();
    let mut prev_line: Option = None;
    let mut sum = 0;
    while let Some(line) = lines.next() {
        let current_line = line.expect("line exists");
        let next_line = match lines.peek() {
            Some(Ok(line)) => Some(line),
            Some(Err(_)) => None,
            None => None,
        };
        match (prev_line, next_line) {
            (None, Some(next)) => {
                let lines = vec![¤t_line, next];
                sum += parse_lines_for_gears(lines, true);
            }
            (Some(prev), Some(next)) => {
                let lines = vec![&prev, ¤t_line, next];
                sum += parse_lines_for_gears(lines, false);
            }
            (Some(prev), None) => {
                let lines = vec![&prev, ¤t_line];
                sum += parse_lines_for_gears(lines, false);
            }
            (None, None) => {}
        }

        prev_line = Some(current_line);
    }

    sum
}

fn parse_lines(lines: Vec<&String>, first_line: bool) -> u32 {
    let mut sum = 0;
    let mut num = 0;
    let mut valid = false;
    let mut char_vec: Vec> = Vec::new();
    for line in lines {
        char_vec.push(line.chars().collect());
    }
    let chars = match first_line {
        true => &char_vec[0],
        false => &char_vec[1],
    };
    for i in 0..chars.len() {
        if chars[i].is_digit(10) {
            // Add the digit to the number
            num = num * 10 + chars[i].to_digit(10).expect("is digit");

            // Check the surrounding character for non-period symbols
            for &x in &[-1, 0, 1] {
                for chars in &char_vec {
                    if (i as isize + x).is_positive() && ((i as isize + x) as usize) < chars.len() {
                        let index = (i as isize + x) as usize;
                        if !chars[index].is_digit(10) && chars[index] != '.' {
                            valid = true;
                        }
                    }
                }
            }
        } else {
            if valid {
                sum += num;
            }
            valid = false;
            num = 0;
        }
    }
    if valid {
        sum += num;
    }
    sum
}

fn parse_lines_for_gears(lines: Vec<&String>, first_line: bool) -> u32 {
    let mut sum = 0;
    let mut char_vec: Vec> = Vec::new();
    for line in &lines {
        char_vec.push(line.chars().collect());
    }
    let chars = match first_line {
        true => &char_vec[0],
        false => &char_vec[1],
    };
    for i in 0..chars.len() {
        if chars[i] == '*' {
            let surrounding_nums = get_surrounding_numbers(&lines, i);
            let product = match surrounding_nums.len() {
                0 | 1 => 0,
                _ => surrounding_nums.iter().product(),
            };
            sum += product;
        }
    }
    sum
}

fn get_surrounding_numbers(lines: &Vec<&String>, gear_pos: usize) -> Vec {
    let mut nums: Vec = Vec::new();
    let mut num: u32 = 0;
    let mut valid = false;
    for line in lines {
        for (i, char) in line.chars().enumerate() {
            if char.is_digit(10) {
                num = num * 10 + char.to_digit(10).expect("is digit");
                if [gear_pos - 1, gear_pos, gear_pos + 1].contains(&i) {
                    valid = true;
                }
            } else if num > 0 && valid {
                nums.push(num);
                num = 0;
                valid = false;
            } else {
                num = 0;
                valid = false;
            }
        }
        if num > 0 && valid {
            nums.push(num);
        }
        num = 0;
        valid = false;
    }
    nums
}

#[cfg(test)]
mod tests {
    use super::*;

    const INPUT: &str = "467..114..
...*......
..35..633.
......#...
617*......
.....+.58.
..592.....
......755.
...$.*....
.664.598..";

    #[test]
    fn test_process_part_one() {
        let input_bytes = INPUT.as_bytes();
        assert_eq!(4361, process_part_one(BufReader::new(input_bytes)));
    }

    #[test]
    fn test_process_part_two() {
        let input_bytes = INPUT.as_bytes();
        assert_eq!(467835, process_part_two(BufReader::new(input_bytes)));
    }
}

Python

Questions and comments welcome!

import collections
import re

from .solver import Solver

class Day03(Solver):
  def __init__(self):
    super().__init__(3)
    self.lines = []

  def presolve(self, input: str):
    self.lines = input.rstrip().split('\n')

  def solve_first_star(self):
    adjacent_to_symbols = set()
    for i, line in enumerate(self.lines):
      for j, sym in enumerate(line):
        if sym in ('0', '1', '2', '3', '4', '5', '6', '7', '8', '9', '.'):
          continue
        for di in (-1, 0, 1):
          for dj in (-1, 0, 1):
            adjacent_to_symbols.add((i + di, j + dj))
    numbers = []
    for i, line in enumerate(self. lines):
      for number_match in re.finditer(r'\d+', line):
        is_adjacent_to_symbol = False
        for j in range(number_match.start(), number_match.end()):
          if (i, j) in adjacent_to_symbols:
            is_adjacent_to_symbol = True
        if is_adjacent_to_symbol:
          numbers.append(int(number_match.group()))
    return sum(numbers)

  def solve_second_star(self):
    gear_numbers = collections.defaultdict(list)
    adjacent_to_gears = {}
    for i, line in enumerate(self.lines):
      for j, sym in enumerate(line):
        if sym == '*':
          for di in (-1, 0, 1):
            for dj in (-1, 0, 1):
              adjacent_to_gears[(i + di, j + dj)] = (i, j)
    for i, line in enumerate(self. lines):
      for number_match in re.finditer(r'\d+', line):
        adjacent_to_gear = None
        for j in range(number_match.start(), number_match.end()):
          if (i, j) in adjacent_to_gears:
            adjacent_to_gear = adjacent_to_gears[(i, j)]
        if adjacent_to_gear:
          gear_numbers[adjacent_to_gear].append(int(number_match.group()))
    ratios = []
    for gear_numbers in gear_numbers.values():
      match gear_numbers:
        case [a, b]:
          ratios.append(a * b)
    return sum(ratios)

Python solution

just more parsing problems

Factor on github (with comments and imports):

: symbol-indices ( line -- seq )
  [ ".0123456789" member? not ] find-all [ first ] map
;

: num-spans ( line -- seq )
  >array [ over digit? [ nip ] [ 2drop f ] if ] map-index
  { f } split harvest
  [ [ first ] [ last ] bi 2array ] map
;

: adjacent? ( num-span symbol-indices -- ? )
  swap [ first 1 - ] [ last 1 + ] bi [a,b]
  '[ _ interval-contains? ] any?
;

: part-numbers ( line nearby-symbol-indices -- seq )
  [ dup num-spans ] dip
  '[ _ adjacent? ] filter
  swap '[ first2 1 + _ subseq string>number ] map
;

: part1 ( -- )
  "vocab:aoc-2023/day03/input.txt" utf8 file-lines
  [ [ symbol-indices ] map ] keep
  [
    pick swap [ 1 - ?nth-of ] [ nth-of ] [ 1 + ?nth-of ] 2tri
    3append part-numbers sum
  ] map-index sum nip .
;

: star-indices ( line -- seq )
  [ CHAR: * = ] find-all [ first ] map
;

: gears ( line prev-line next-line -- seq-of-pairs )
  pick star-indices
  [ 1array '[ _ part-numbers ] [ 3dup ] dip tri@ 3append ]
  [ length 2 = ] map-filter [ 3drop ] dip
;

: part2 ( -- )
  "vocab:aoc-2023/day03/input.txt" utf8 file-lines
  dup [
    pick swap [ 1 - ?nth-of ] [ 1 + ?nth-of ] 2bi
    gears [ product ] map-sum
  ] map-index sum nip .
;

F# solution

Had an off-by-one error that took me a while to find

Ruby

My goodness I struggled today. https://github.com/snowe2010/advent-of-code/blob/master/ruby_aoc/2023/day03/day03.rb

I get the feeling that I should include some default types for handling 2D maps in my boilerplate, it’s a very recurring problem in AoC after all.

My solution is reasonably simplistic - and therefore also a bit slow, but the design meant I could do part 2 with just a few extra lines of code on the already processed data, here’s the functional part of it; (I push the previous days solution as part of my workflow for starting with the current day so the full code won’t be up until tomorrow)

Point = Struct.new('Point', :x, :y)
PartNumber = Struct.new('PartNumber', :number, :adjacent) do
  def adjacent?(to); adjacent.include?(to); end
  def irrelevant?; adjacent.empty?; end
  def to_i; number; end
end

class Implementation
  def initialize
    @map = []; @dim = { width: 0, height: 0 }; @symbols = []; @numbers = []
  end

  def input(line)
    @dim[:width] = line.size; @dim[:height] += 1
    @map += line.chars
  end

  def calc
    for y in (0..@dim[:height]-1) do
      for x in (0..@dim[:width]-1) do
        chr = get(x, y); next if chr =~ /\d/ || chr == '.'
        @symbols << Point.new(x, y)
      end
    end

    for y in (0..@dim[:height]-1) do
      buf = ""; adj = []
      for x in (0..@dim[:width]) do # Going one over, to fake a non-number as an end char on all lines
        chr = get(x, y)
        if chr =~ /\d/
          buf += chr
          (-1..1).each do |adj_x|
            (-1..1).each do |adj_y|
              next if adj_x == 0 && adj_y == 0 ||
                (x + adj_x < 0) || (x + adj_x >= @dim[:width]) ||
                (y + adj_y < 0) || (y + adj_y >= @dim[:height])
              sym = Point.new(x + adj_x, y + adj_y)
              adj << sym if @symbols.any? sym
            end
          end
        elsif !buf.empty?
          @numbers << PartNumber.new(buf.to_i, adj)
          buf = ""; adj = []
        end
      end
    end
  end

  def output
    part1 = @numbers.reject(&:irrelevant?).map(&:to_i).sum
    puts "Part 1:", part1

    gears = @symbols.select do |sym|
      next unless get(sym) == '*'
      next unless @numbers.select { |num| num.adjacent? sym }.size == 2
      true
    end
    part2 = gears.sum { |gear| @numbers.select { |num| num.adjacent? gear }.map(&:to_i).inject(:*) }

    puts "Part 2:", part2
  end

  private

  def get(x, y = -1)
    y = x.y if x.is_a?(Point)
    x = x.x if x.is_a?(Point)
    return unless (0..@dim[:width]-1).include?(x) && (0..@dim[:height]-1).include?(y)

    @map[y * @dim[:width] + x % @dim[:width]]
  end
end

Input parsing AGAIN?

-- SPDX-FileCopyrightText: 2023 Jummit
--
-- SPDX-License-Identifier: GPL-3.0-or-later

local lines = {}
for line in io.open("3.input"):lines() do
	table.insert(lines, "."..line..".")
end
local width = #lines[1]
local height = #lines
local function at(x, y, w)
	if y < 1 or y > height then return nil end
	return lines[y]:sub(x, x + w - 1)
end
local sum = 0
local gears = {}
for y, line in ipairs(lines) do
	local start = 1
	local outLine = line
	while true do
		local newStart, numEnd = line:find("%d+", start)
		if not newStart then break end
		local symbol = false
		local num = tonumber(line:sub(newStart, numEnd))
		for y = y - 1, y + 1 do
			local surrounding = at(newStart - 1, y, numEnd - newStart + 3)
			if surrounding then
				if surrounding and surrounding:match("[^.%d]") then
					symbol = true
				end
				for i = 1, #surrounding do
					local gear = surrounding:sub(i, i) == "*"
					if gear then
						if not gears[y] then
							gears[y] = {}
						end
						local x = i + newStart - 2
						if not gears[y][x] then
							gears[y][i + newStart - 2] = {}
						end
						table.insert(gears[y][x], num)
					end
				end
			end
		end
		if symbol then
			sum = sum + num
		end
		start = numEnd + 1
	end
end
print(sum)
local ratio = 0
for _, line in pairs(gears) do
	for _, gears in pairs(line) do
		if #gears == 2 then
			ratio = ratio + gears[1] * gears[2]
		end
	end
end
print(ratio)

// SPDX-FileCopyrightText: 2023 Jummit
//
// SPDX-License-Identifier: GPL-3.0-or-later

use strings;
use regex;
use fmt;
use os;
use bufio;
use io;
use strconv;
use types;

fn star_in(lines: []str, x: uint, y: uint, w: uint) bool = {
	let start = y;
	if (start > 0) start -= 1;
	let end = y + 1;
	if (end >= len(lines)) end -= 1;
	const re = regex::compile(`[^.0-9]`)!;
	for (let h = start; h <= end; h += 1) {
		fmt::println(strings::sub(lines[h], x, x + w))!;
		if (regex::test(&re, strings::sub(lines[h], x, x + w))) {
			fmt::println("")!;
			return true;
		};
	};
	fmt::println("")!;
	return false;
};

export fn main() void = {
	const file = os::open("3.input")!;
	defer io::close(file)!;
	const buf = bufio::newscanner(file, types::SIZE_MAX);
	let lines: []str = [];
	defer strings::freeall(lines);
	for (true) {
		match (bufio::scan_line(&buf)!) {
		case io::EOF =>
			break;
		case let line: const str =>
			append(lines, strings::dup(line));
		};
	};
	const height = len(lines);
	const width = len(lines[0]);
	let sum: uint = 0;
	let gears: [](uint, uint) = [];
	const num_re = regex::compile(`[0-9]+`)!;
	for (let y = 0u; y < len(lines); y += 1) {
		let nums = regex::findall(&num_re, lines[y]);
		defer regex::result_freeall(nums);
		for (let i = 0z; i < len(nums); i += 1) {
			for (let j = 0z; j < len(nums[i]); j += 1) {
				const find = nums[i][j];
				const num = strconv::stou(find.content)!;
				let start = find.start: uint;
				let w = len(find.content): uint + 2;
				if (start > 0) {
					start -= 1;
				} else {
					w -= 1;
				};
				if (star_in(lines, start, y, w)) {
					sum += num;
				};
			};
		};
	};
	fmt::printfln("{}", sum)!;
};

Language: C

Part 2 stumped me for a little bit, it wasn’t an obvious extension of part 1. Part 1 was about numbers (with one or more …) while part 2 worked from the symbols (with exactly two …). Going the other way would require more bookkeeping to avoid double counting.

And for the implementation: if you loop over the grid and check surrounding cells for digits you’d have to account for a bunch of cases, e.g. NW/N or N/NE being part of the same number or NW and NE being part of separate numbers. And you’d have to parse the numbers again. But building a graph or reference list of some sort is both unergonomic with C and not necessarily any simpler.

I ended up just writing out the cases, and honestly it didn’t turn out too bad.

GitHub link

int main(int argc, char **argv)
{
	static char G[GSZ][GSZ];
	static int N[GSZ][GSZ];
	int p1=0,p2=0, h=0, x,y, dx,dy, n=0,sym=0,r;
	
	for (h=0; fgets(&G[h+1][1], GSZ-1, stdin); h++)
		assert(h < GSZ);

	/*
	 * Pass 1: parse numbers and solve part 1. For every digit in
	 * G, the full number it is part of is stored in N.
	 */
	for (y=1; y<=h; y++)
	for (x=1; G[y][x]; x++)
		if (isdigit(G[y][x])) {
			n = n*10 + G[y][x]-'0';

			for (dy=-1; dy<2; dy++)
			for (dx=-1; dx<2; dx++)
				sym = sym || (x && y &&
				    G[y+dy][x+dx] != '.' &&
				    ispunct(G[y+dy][x+dx]));
		} else {
			for (dx=-1; isdigit(G[y][x+dx]); dx--)
				N[y][x+dx] = n;
			if (sym)
				p1 += n;
			n = sym = 0;
		}

	/*
	 * Pass 2: solve part 2 by finding all '*', then counting and
	 * multiplying adjecent numbers.
	 *
	 * Horizontal adjecency is trivial but vertical/diagonal has
	 * two situations: if there's a digit directly North of the '+',
	 * it must be a single number: NW and NE would connect to it.
	 * If N isn't a digit, digits in NW and NE belong to separate
	 * numbers.
	 */
	for (y=1; y<=h; y++)
	for (x=1; G[y][x]; x++) {
		if (G[y][x] != '*')
			continue;

		n = 0; r = 1;

		if (N[y][x-1]) { n++; r *= N[y][x-1]; }
		if (N[y][x+1]) { n++; r *= N[y][x+1]; }

		if (N[y-1][x]) { n++; r *= N[y-1][x]; } else {
			if (N[y-1][x-1]) { n++; r *= N[y-1][x-1]; }
			if (N[y-1][x+1]) { n++; r *= N[y-1][x+1]; }
		}

		if (N[y+1][x]) { n++; r *= N[y+1][x]; } else {
			if (N[y+1][x-1]) { n++; r *= N[y+1][x-1]; }
			if (N[y+1][x+1]) { n++; r *= N[y+1][x+1]; }
		}

		if (n == 2)
			p2 += r;
	}

	printf("%d %d\n", p1, p2);
	return 0;
}

Language: C++

Efficiency? Elegant Code? Nope but It works. Luckily I did part 1 by looking at the symbols first anyway, so extending to part two was trivial. Also originally had a bug where I treated all symbols as cogs, not only ‘*’. Interestingly it worked anyway as only '*'s had two adjacent numbers in my data. It is fixed in this version. Hacked together combined code (originally I did each part as separate programs but they shared so much that I ended up combining then so the post is shorter): https://pastebin.com/Dij2XSYe

Edit: anything in angle brackets is not displaying even with backslashes, idk why but i have moved the code to a pastebin.