Source: Day 1: Trebuchet?!
Full solution for today (spoilers!). Note: I did slightly change my solutions template after writing this blog post, so the final solution is structured slightly differently than the code in this post. The functionality itself hasn’t changed.
Part 1
Given a list of alphanumeric strings, find the first and last digit on each line (they may be the same). Concatenate each pair and sum them.
We’re starting out fairly light. To find the digits, we can filter out anything that is not a digit. Then concatenate, numberize, and sum.
fn part1(filename: &Path) -> Result<String> {
Ok(iter_lines(filename)
.filter_map(|l| {
let mut first = None;
let mut last = None;
for c in l.chars() {
if c.is_numeric() {
if first.is_none() {
first = Some(c);
}
last = Some(c);
}
}
Some(10 * first?.to_digit(10)? + last?.to_digit(10)?)
})
.sum::<u32>()
.to_string())
}
Ironically, we’re not really using the filter_map here. I’m only using it because if anything fails to parse, the ? will return None, which we can filter out.
Also, I’m not that thrilled with this answer. I’d really like to be able to filter inline, extract the first and last elements, and product them. Something like:
l.chars()
.filter(|c| c.is_numeric())
.first_and_last()
.collect::<String>()
.parse::<u32>()
But that would require a first_and_last function…
Okay, let’s try it.
mod first_and_last {
pub(crate) trait IteratorExt: Iterator {
fn first_and_last(mut self) -> [Self::Item; 2]
where
Self: Sized,
Self::Item: Clone,
{
let first = self.next().unwrap();
let last = self.last().or_else(|| Some(first.clone())).unwrap();
[first, last]
}
}
impl<T: ?Sized> IteratorExt for T where T: Iterator {}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn test_first_and_last() {
assert_eq!(vec![1, 2, 3, 4, 5].into_iter().first_and_last(), [1, 5]);
assert_eq!(vec![1].into_iter().first_and_last(), [1, 1]);
}
}
}
I’ve… no idea if this is a good idea or not. But it does simplify part 1 a bit (IMO):
fn part1(filename: &Path) -> Result<String> {
Ok(iter_lines(filename)
.map(|l| {
l.chars()
.filter(|c| c.is_numeric())
.first_and_last()
.iter()
.collect::<String>()
.parse::<u32>()
.unwrap()
})
.sum::<u32>()
.to_string())
}
I suppose it’s up to you which you prefer!
(And this is only part 1 of day 1…)
Part 2
This time, treat
one,two, etc as valid digits (that can be first or last).
One thing to note here (minor spoilers): there are a few cases where numbers can overlap, such as eightwo being both 8 and 2. A simple regex won’t handle that. Here’s how I did it:
fn part2(filename: &Path) -> Result<String> {
let digit_words = [
"zero", "one", "two", "three", "four", "five", "six", "seven", "eight", "nine",
];
Ok(iter_lines(filename)
.filter_map(|l| {
let mut first = None;
let mut last = None;
for (i, c) in l.chars().enumerate() {
// Match literal digits
if c.is_numeric() {
let c = c.to_digit(10)? as usize;
if first.is_none() {
first = Some(c);
}
last = Some(c);
continue;
}
// Match digit words
for (digit, word) in digit_words.iter().enumerate() {
if l[i..].starts_with(word) {
if first.is_none() {
first = Some(digit);
}
last = Some(digit);
break;
}
}
}
Some(10 * first? + last?)
})
.sum::<usize>()
.to_string())
}
Performance
$ cargo run --release --bin 01-calibrationinator 1 data/01.txt
Compiling aoc2023 v0.1.0 (/Users/jp/Projects/advent-of-code/2023)
Finished release [optimized] target(s) in 0.19s
Running `target/release/01-calibrationinator 1 data/01.txt`
53651
took 500.417µs
$ cargo run --release --bin 01-calibrationinator 2 data/01.txt
Finished release [optimized] target(s) in 0.00s
Running `target/release/01-calibrationinator 2 data/01.txt`
53894
took 671.791µs
Yeah… µs. Nothing much to bother here.