Rust

Ooof. Part 1 was easy enough, but for part two I initially went with the naive solution of trying every single seed which took more than a minute (I never really measured). Although that got me the right answer, to me that was just unacceptable.

I proceeded to try and combine all mappings into one but gave up after spending way too much time on it.

Then I had the idea that the lowest number in the end must lie at the beginning of a range somewhere. Either the start of a seed range in the beginning or the start of a range in one of the mappings. Any in-between numbers must end up with a higher result. So I considered the start points of all ranges, went through the mappings in reverse to find out if that point is actually within a seed range, and only tested those starting points.

Finally I had only 183 points to test which ran much faster (0.9ms).

[JavaScript] Well that was by far the hardest out of all of the days, part 1 was relatively fine but part 2 took me awhile of trying different things

Ended up solving it by working backwards by trying different location values and seeing if that can become a valid seed. Takes around 3 secs to compute the answer.

Link to code

// Part 1
// ======

function part1(input) {
  const split = input.split("\r\n\r\n");

  let pastValues = split[0].match(/\d+/g).map((x) => parseInt(x));
  let currentValues = [];

  for (const section of split.slice(1)) {
    for (const line of section.split("\r\n")) {
      const values = line.match(/\d+/g)?.map((x) => parseInt(x));

      if (!values) {
        continue;
      }

      const sourceStart = values[1];
      const destinationStart = values[0];
      const length = values[2];

      for (let i = 0; i < pastValues.length; i++) {
        if (
          pastValues[i] >= sourceStart &&
          pastValues[i] < sourceStart + length
        ) {
          currentValues.push(destinationStart + pastValues[i] - sourceStart);
          pastValues.splice(i, 1);
          i--;
        }
      }
    }

    for (let i = 0; i < pastValues.length; i++) {
      currentValues.push(pastValues[i]);
    }

    pastValues = [...currentValues];
    currentValues = [];
  }

  return Math.min(...pastValues);
}

// Part 2
// ======

function part2(input) {
  const split = input.split("\r\n\r\n");

  let seeds = split[0].match(/\d+/g).map((x) => parseInt(x));
  seeds = seeds
    .filter((x, i) => i % 2 == 0)
    .map((x, i) => [x, seeds[i * 2 + 1]]);

  const maps = split
    .slice(1)
    .map((x) => {
      const lines = x.split("\r\n");
      return lines
        .map((x) => x.match(/\d+/g)?.map((x) => parseInt(x)))
        .filter((x) => x);
    })
    .reverse();

  for (let i = 0; true; i++) {
    let curValue = i;

    for (const map of maps) {
      for (const line of map) {
        const sourceStart = line[1];
        const destinationStart = line[0];
        const length = line[2];

        if (
          curValue >= destinationStart &&
          curValue < destinationStart + length
        ) {
          curValue = sourceStart + curValue - destinationStart;
          break;
        }
      }
    }

    for (const [seedRangeStart, seedRangeLength] of seeds) {
      if (
        curValue >= seedRangeStart &&
        curValue < seedRangeStart + seedRangeLength
      ) {
        return i;
      }
    }
  }
}

Finally. :cries:

Part 1 was fine, I was figuring I might be able to practice classes.

Part 2 told me that nope, memory management required for you. In the end instead of calculating seeds, I resolved the whole thing down to a single mapping of seeds to locations. Then I could just sort by location ranges and try to see if they were a seed. Code is full of old parts from failed solutions but I’ve had enough of day 5, so I no longer care to clean it up.

Haskell

Not hugely proud of this one; part one would have been easier if I’d spend more time reading the question and not started on an overly-general solution, and I lost a lot of time on part two to a missing a +. More haste, less speed, eh?

import Data.List
import Data.List.Split

readInput :: String -> ([Int], [(String, [(Int, Int, Int)])])
readInput s =
  let (seedsChunk : mapChunks) = splitOn [""] $ lines s
      seeds = map read $ tail $ words $ head seedsChunk
      maps = map readMapChunk mapChunks
   in (seeds, maps)
  where
    readMapChunk (title : rows) =
      let name = head $ words title
          entries = map ((\[a, b, c] -> (a, b, c)) . map read . words) rows
       in (name, entries)

part1 (seeds, maps) =
  let f = foldl1' (flip (.)) $ map (ref . snd) maps
   in minimum $ map f seeds
  where
    ref [] x = x
    ref ((a, b, c) : rest) x =
      let i = x - b
       in if i >= 0 && i < c
            then a + i
            else ref rest x

mapRange :: [(Int, Int, Int)] -> (Int, Int) -> [(Int, Int)]
mapRange entries (start, end) =
  go start $ sortOn (\(_, b, _) -> b) entries
  where
    go i [] = [(i, end)]
    go i es@((a, b, c) : rest)
      | i > end = []
      | b > end = go i []
      | b + c <= i = go i rest
      | i < b = (i, b - 1) : go b es
      | otherwise =
          let d = min (b + c - 1) end
           in (a + i - b, a + d - b) : go (d + 1) rest

part2 (seeds, maps) =
  let seedRanges = map (\[a, b] -> (a, a + b - 1)) $ chunksOf 2 seeds
   in minimum $ map fst $ foldl' (flip mapRanges) seedRanges $ map snd maps
  where
    mapRanges m = concatMap (mapRange m)

main = do
  input <- readInput <$> readFile "input05"
  print $ part1 input
  print $ part2 input

Odin

When I read the problem description I expected the input to also be 2 digit numbers. When I looked at it I just had to say “huh.”

Second part I think you definitely have to do in reverse (edit: if you are doing a linear search for the answer), as that allows you to nope out as soon as you find a match, whereas with doing it forward you have to keep checking just in case.

Formatted code

package day5

import "core:fmt"
import "core:strings"
import "core:slice"
import "core:strconv"

Range :: struct {
    dest: int,
    src: int,
    range: int,
}

Mapper :: struct {
    ranges: []Range,
}

parse_range :: proc(s: string) -> (ret: Range) {
    rest := s

    parseLen := -1

    destOk: bool
    ret.dest, destOk = strconv.parse_int(rest, 10, &parseLen)
    rest = strings.trim_left_space(rest[parseLen:])

    srcOk: bool
    ret.src, srcOk = strconv.parse_int(rest, 10, &parseLen)
    rest = strings.trim_left_space(rest[parseLen:])

    rangeOk: bool
    ret.range, rangeOk = strconv.parse_int(rest, 10, &parseLen)

    return
}

parse_mapper :: proc(ss: []string) -> (ret: Mapper) {
    ret.ranges = make([]Range, len(ss)-1)
    for s, i in ss[1:] {
        ret.ranges[i] = parse_range(s)
    }

    return
}

parse_mappers :: proc(ss: []string) -> []Mapper {
    mapsStr := make([dynamic][]string)
    defer delete(mapsStr)

    restOfLines := ss
    isLineEmpty :: proc(s: string)->bool {return len(s)==0}

    for i, found := slice.linear_search_proc(restOfLines, isLineEmpty); 
        found; 
        i, found  = slice.linear_search_proc(restOfLines, isLineEmpty) {
        
        append(&mapsStr, restOfLines[:i])
        restOfLines = restOfLines[i+1:]
    }
    append(&mapsStr, restOfLines[:])

    return slice.mapper(mapsStr[1:], parse_mapper)
}

apply_mapper :: proc(mapper: Mapper, num: int) -> int {
    for r in mapper.ranges {
        if num >= r.src && num - r.src < r.range do return num - r.src + r.dest
    }

    return num
}

p1 :: proc(input: []string) {
    maps := parse_mappers(input)
    defer {
        for m in maps do delete(m.ranges)
        delete(maps)
    }

    restSeeds := input[0][len("seeds: "):]
    min := 0x7fffffff

    for len(restSeeds) > 0 {
        seedLen := -1
        seed, seedOk := strconv.parse_int(restSeeds, 10, &seedLen)
        restSeeds = strings.trim_left_space(restSeeds[seedLen:])

        fmt.print(seed)
        for m in maps {
            seed = apply_mapper(m, seed)
            fmt.print(" ->", seed)
        }
        fmt.println()

        if seed < min do min = seed
    }

    fmt.println(min)
}

apply_mapper_reverse :: proc(mapper: Mapper, num: int) -> int {
    for r in mapper.ranges {
        if num >= r.dest && num - r.dest < r.range do return num - r.dest + r.src
    }

    return num
}

p2 :: proc(input: []string) {
    SeedRange :: struct {
        start: int,
        len: int,
    }

    seeds := make([dynamic]SeedRange)
    restSeeds := input[0][len("seeds: "):]

    for len(restSeeds) > 0 {
        seedLen := -1
        seedS, seedSOk := strconv.parse_int(restSeeds, 10, &seedLen)
        restSeeds = strings.trim_left_space(restSeeds[seedLen:])

        seedL, seedLOk := strconv.parse_int(restSeeds, 10, &seedLen)
        restSeeds = strings.trim_left_space(restSeeds[seedLen:])

        append(&seeds, SeedRange{seedS, seedL})
    }

    maps := parse_mappers(input)
    defer {
        for m in maps do delete(m.ranges)
        delete(maps)
    }

    for i := 0; true; i += 1 {
        rseed := i
        #reverse for m in maps {
            rseed = apply_mapper_reverse(m, rseed)
        }

        found := false
        for sr in seeds {
            if rseed >= sr.start && rseed < sr.start + sr.len {
                found = true
                break
            }
        }
        if found {
            fmt.println(i)
            break
        }
    }
}

Scala3

kind of convoluted, but purely functional

import scala.collection.immutable.NumericRange.Exclusive
import math.max
import math.min

extension [A] (l: List[A])
  def chunk(pred: A => Boolean): List[List[A]] =
    def go(l: List[A], partial_acc: List[A], acc: List[List[A]]): List[List[A]] =
      l match
        case (h :: t) if pred(h) => go(t, List(), partial_acc.reverse :: acc)
        case (h :: t) => go(t, h :: partial_acc, acc)
        case _ => partial_acc.reverse :: acc
    
    go(l, List(), List()).reverse

type R = Exclusive[Long]

def intersectTranslate(r: R, c: R, t: Long): R =
    (t + max(r.start, c.start) - c.start) until (t + min(r.end, c.end) - c.start)

case class MappingEntry(from: R, to: Long)
case class Mapping(entries: List[MappingEntry], produces: String):
    def resolveRange(in: R): List[R] =
        entries.map(e => intersectTranslate(in, e.from, e.to)).filter(!_.isEmpty)

def completeEntries(a: List[MappingEntry]): List[MappingEntry] = 
    a ++ ((0L until 0L) +: a.map(_.from).sorted :+ (Long.MaxValue until Long.MaxValue)).sliding(2).flatMap{ case List(a, b) => Some(MappingEntry(a.end until b.start, a.end)); case _ => None}.toList

def parse(a: List[String], init: List[Long] => List[R]): (List[R], Map[String, Mapping]) =
    def parseEntry(s: String): MappingEntry =
        s match
            case s"$end $start $range" => MappingEntry(start.toLong until start.toLong + range.toLong, end.toLong)

    a.chunk(_ == "") match
        case List(s"seeds: $seeds") :: maps => 
            init(seeds.split(raw"\s+").map(_.toLong).toList) -> (maps.flatMap{ case s"$from-to-$to map:" :: entries => Some(from -> Mapping(completeEntries(entries.map(parseEntry)), to)); case _ => None }).toMap
        case _ => (List(), Map()).ensuring(false)

def singletons(a: List[Long]): List[R] = a.map(s => s until s + 1)
def paired(a: List[Long]): List[R] = a.grouped(2).flatMap{ case List(x, y) => Some(x until x+y); case _ => None }.toList

def chase(d: (List[R], Map[String, Mapping]), initial: String, target: String) =
    val (init, m) = d
    def go(a: List[R], s: String): List[R] =
        if trace(s) == target then a else
            val x = m(s)
            go(a.flatMap(x.resolveRange), x.produces)
    go(trace(init), initial)

def task1(a: List[String]): Long = 
    chase(parse(a, singletons), "seed", "location").min.start

def task2(a: List[String]): Long =
    chase(parse(a, paired), "seed", "location").min.start

Nim

Woof. Part 1 was simple enough. I thought I could adapt my solution to part 2 pretty easily, just add all the values in the ranges to the starting set. Worked fine for the example, but the ranges for the actual input are too large. Ended up taking 16gb of RAM and crunching forever.

I finally abandoned my quick and dirty approach when rewriting part 2, and made some proper types and functions. Treated each range as an object, and used set operations on them. The difference operation tends to fragment the range that it’s used on, so I meant to write some code to defragment the ranges after each round of mappings. Forgot to do so, but the code ran quick enough this time anyway.

• Day 5, part 1
• Day 5, part 2 first attempt
• Day 5, part 2 rewrite

This was interesting! So iterating through the solution space would be infeasible here and it seems we need to look for boundaries between regions and follow them to find places where a solution could occur.

Python: https://pastebin.com/8Ckx36fu

• Make a list of places where location mappings are discontinuous (0, the end of each mapping, and the number before)
• Repeat this for discontinuities in each intermediate layer
• Trace each such location to its seed, and filter by seed ranges
• Run the very minimal set of interesting seed numbers (around 2000 seeds) through the existing part1 algorithm

Like many others, I really didn’t enjoy this one. I particularly struggled with part 02, which ended up with me just brute forcing it and checking each seed. On my system it took over 15 minutes to run, which is truly awful. I’m open to pointers on how I could better have solved part two.

Solution in Rust 🦀

Formatted Solution on GitLab

use std::{
    env, fs,
    io::{self, 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 mut reader1 = BufReader::new(file1);
    let mut reader2 = BufReader::new(file2);

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

#[derive(Debug)]
struct Map {
    lines: Vec,
}

impl Map {
    fn map_to_lines(&self, key: u32) -> u32 {
        for line in &self.lines {
            if line.in_range(key) {
                return line.map(key);
            }
        }
        key
    }
}

#[derive(Debug)]
struct MapLine {
    dest_range: u32,
    source_range: u32,
    range_length: u32,
}

impl MapLine {
    fn map(&self, key: u32) -> u32 {
        let diff = key - self.source_range;
        if self.dest_range as i64 + diff as i64 > 0 {
            return (self.dest_range as i64 + diff as i64) as u32;
        }
        key
    }

    fn in_range(&self, key: u32) -> bool {
        self.source_range <= key
            && (key as i64) < self.source_range as i64 + self.range_length as i64
    }
}

fn parse_input(reader: &amp;mut BufReader) -> (Vec, Vec<map>) {
    let mut almanac = String::new();
    reader
        .read_to_string(&amp;mut almanac)
        .expect("read successful");
    let parts: Vec&lt;&amp;str> = almanac.split("\n\n").collect();
    let (seeds, others) = parts.split_first().expect("at least one part");
    let seeds: Vec&lt;_> = seeds
        .split(": ")
        .last()
        .expect("at least one")
        .split_whitespace()
        .map(|s| s.to_string())
        .collect();
    let maps: Vec&lt;_> = others
        .iter()
        .map(|item| {
            let lines_iter = item
                .split(':')
                .last()
                .expect("exists")
                .trim()
                .split('\n')
                .map(|nums| {
                    let nums_split = nums.split_whitespace().collect::>();
                    MapLine {
                        dest_range: nums_split[0].parse().expect("is digit"),
                        source_range: nums_split[1].parse().expect("is digit"),
                        range_length: nums_split[2].parse().expect("is digit"),
                    }
                });
            Map {
                lines: lines_iter.collect(),
            }
        })
        .collect();
    (seeds, maps)
}

fn process_part_one(reader: &amp;mut BufReader) -> u32 {
    let (seeds, maps) = parse_input(reader);
    let mut res = u32::MAX;
    for seed in &amp;seeds {
        let mut val = seed.parse::().expect("is digits");
        for map in &amp;maps {
            val = map.map_to_lines(val);
        }
        res = u32::min(res, val);
    }
    res
}

fn process_part_two(reader: &amp;mut BufReader) -> u32 {
    let (seeds, maps) = parse_input(reader);
    let seed_chunks: Vec&lt;_> = seeds.chunks(2).collect();
    let mut res = u32::MAX;
    for chunk in seed_chunks {
        let range_start: u32 = chunk[0].parse().expect("is digits");
        let range_length: u32 = chunk[1].parse().expect("is digits");
        let range_end: u32 = range_start + range_length;
        for seed in range_start..range_end {
            let mut val = seed;
            for map in &amp;maps {
                val = map.map_to_lines(val);
            }
            res = u32::min(res, val);
        }
    }
    res
}

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

    const INPUT: &amp;str = "seeds: 79 14 55 13

seed-to-soil map:
50 98 2
52 50 48

soil-to-fertilizer map:
0 15 37
37 52 2
39 0 15

fertilizer-to-water map:
49 53 8
0 11 42
42 0 7
57 7 4

water-to-light map:
88 18 7
18 25 70

light-to-temperature map:
45 77 23
81 45 19
68 64 13

temperature-to-humidity map:
0 69 1
1 0 69

humidity-to-location map:
60 56 37
56 93 4";

    #[test]
    fn test_process_part_one() {
        let input_bytes = INPUT.as_bytes();
        assert_eq!(35, process_part_one(&amp;mut BufReader::new(input_bytes)));
    }

    #[test]
    fn test_process_part_two() {
        let input_bytes = INPUT.as_bytes();
        assert_eq!(46, process_part_two(&amp;mut BufReader::new(input_bytes)));
    }
}

Nim

Part 1 was really easy.
Part 2, I struggled to solve efficiently, so I just ran naive bruteforce for 5 minutes until I got the answer.
Later, I’ve rewritten my solution for Part 2. The idea is to handle ranges as ranges, check seed ranges against map ranges, transform overlaps, but keep not-overlapping parts.

Total runtime after rewrite: ~ 0.4 ms.
Today’s puzzle was nice - 8.5/10.

Code: day_05/solution.nim

Honestly this one was quite easy for me. Not so much for my computer though…

https://git.sr.ht/~aidenisik/aoc23/tree/master/item/day5

C solutions. Disclaimer, part 2 has not finished running. But it’s mostly the same code as part 1 and it works on the small sample data so it’ll be fine.

Language: Python

solution

deleted by creator

Python

Questions and feedback welcome!

import portion as P

from .solver import Solver

_maps = [
  'seed-to-soil',
  'soil-to-fertilizer',
  'fertilizer-to-water',
  'water-to-light',
  'light-to-temperature',
  'temperature-to-humidity',
  'humidity-to-location',
]

def group_lines_in_maps(lines):
  group = []
  for line in lines:
    if not line:
      yield group
      group = []
      continue
    group.append(line)
  yield group

class Day05(Solver):
  def __init__(self):
    super().__init__(5)
    self.seeds = []
    self.mappings = {}

  def presolve(self, input: str):
    lines = input.rstrip().split('\n')
    self.seeds = list(map(int, lines[0].split(' ')[1:]))
    self.mappings = {}
    for mapping in group_lines_in_maps(lines[2:]):
      mapping_name = mapping[0].split(' ')[0]
      mapping_ranges = map(lambda rng: tuple(map(int, rng.split(' '))), mapping[1:])
      self.mappings[mapping_name] = list(mapping_ranges)


  def solve_first_star(self):
    locations = []
    for seed in self.seeds:
      location = seed
      for mapping in map(self.mappings.get, _maps):
        assert mapping
        for dest, source, length in mapping:
          if 0 <= location - source < length:
            location = dest + (location - source)
            break
      locations.append(location)
    return min(locations)


  def solve_second_star(self):
    current_set = P.empty()
    for i in range(0, len(self.seeds), 2):
      current_set = current_set | P.closedopen(self.seeds[i], self.seeds[i] + self.seeds[i + 1])
    for mapping in map(self.mappings.get, _maps):
      assert mapping
      unmapped = current_set
      next_set = P.empty()
      for dest, source, length in mapping:
        delta = dest - source
        source_range = P.closedopen(source, source + length)
        mappable = unmapped & source_range
        mapped_to_destination = mappable.apply(
            lambda x: (x.left, x.lower + delta, x.upper + delta, x.right))  # pylint: disable=cell-var-from-loop
        next_set = next_set | mapped_to_destination
        unmapped = unmapped - source_range
      current_set = next_set | unmapped
    return next(P.iterate(current_set, step=1))

Language: Python

Github

Catching up missed days.