Verso - Advent Time Again

It’s Advent of Code time again!

Once again, thousands of programmers around the world are rearranging their lives to solve a series of Christmas-themed puzzles. As I write this, only one puzzle has been released, but this already promises to be an interesting year.

This year, I’ll write a brief post explaining how I solved each puzzle. We start with day 1.

This year’s first puzzle was unexpectedly hard. Part 1 is a simple string manipulation problem. The input consists of a file where each line has lowercase letters and single digits. You just have to read the first and last integers on each line of a file, combine them into a string, read the resulting integer, and sum them.

```{python}
import re

from utils.utils import split_lines

raw = split_lines("inputs/day1.txt")
# No zeroes
naturals = r"[1-9]"
stripped = [re.sub("[a-z]+", "", line) for line in raw]
part1 = sum(int(line[0] + line[-1]) for line in stripped)
print(part1)
```

Part 2 reveals a nasty surprise. You no longer have to just search for digits, but written numbers. one and 1 now both mean 1. I created an awkward regular expression to handle this, then confidently submitted my answer. Wrong. I checked my code on the provided test case, and it was correct.

This happens to me several times each year, but rarely on day 1! I racked my mind for a pathological edge case that would break my code. Immediately, I thought of overlapping matches. By the rules, a string like eightwo should be parsed as 82, but a conventional pattern would match eight, move on to w, and fail to match any more substrings from the string. I needed an overlapping pattern to get the correct answer.

As with most problems, the answer to this one turned out to lie in a purple StackOverflow link. I tweaked my pattern, reran my code, and submitted a new, slightly higher answer. Correct!

```{python}
nums = ["one", "two", "three", "four", "five", "six", "seven", "eight", "nine"]
n = len(nums)
parts = "|".join(nums + [naturals])
pattern = f"(?=({parts}))"
chars = list(map(str, range(1, 1 + n)))
keys = dict(zip(nums, chars))

matches = [re.findall(pattern, line) for line in raw]
part2 = sum(
    int(keys.get(r[0], r[0]) + keys.get(r[-1], r[-1])) if r else 0 for r in matches
)
print(part2)
```

All in all, it was a typical Day 7 or 8 puzzle, which made it a nasty surprise as a Day 1 puzzle. So Day 2 should be easy by comparison - right?

--- title: "Advent Time Again" author: "Ryan Heslin" date: "2023-12-1" categories: ["Advent of Code"] urlcolor: "blue" --- It's [Advent of Code](https://adventofcode.com/) time again! Once again, thousands of programmers around the world are rearranging their lives to solve a series of Christmas-themed puzzles. As I write this, only one puzzle has been released, but this already promises to be an interesting year. This year, I'll write a brief post explaining how I solved each puzzle. We start with day 1. This year's first puzzle was unexpectedly hard. Part 1 is a simple string manipulation problem. The input consists of a file where each line has lowercase letters and single digits. You just have to read the first and last integers on each line of a file, combine them into a string, read the resulting integer, and sum them. ```{python} import re from utils.utils import split_lines raw = split_lines("inputs/day1.txt") # No zeroes naturals = r"[1-9]" stripped = [re.sub("[a-z]+", "", line) for line in raw] part1 = sum(int(line[0] + line[-1]) for line in stripped) print(part1) ``` Part 2 reveals a nasty surprise. You no longer have to just search for digits, but _written numbers_. `one` and `1` now both mean `1`. I created an awkward regular expression to handle this, then confidently submitted my answer. Wrong. I checked my code on the provided test case, and it was correct. This happens to me several times each year, but rarely on day 1! I racked my mind for a pathological edge case that would break my code. Immediately, I thought of overlapping matches. By the rules, a string like `eightwo` should be parsed as `82`, but a conventional pattern would match `eight`, move on to `w`, and fail to match any more substrings from the string. I needed an overlapping pattern to get the correct answer. As with most problems, the answer to this one turned out to lie in [a purple StackOverflow link](https://stackoverflow.com/questions/5616822/how-to-use-regex-to-find-all-overlapping-matches). I tweaked my pattern, reran my code, and submitted a new, slightly higher answer. Correct! ```{python} nums = ["one", "two", "three", "four", "five", "six", "seven", "eight", "nine"] n = len(nums) parts = "|".join(nums + [naturals]) pattern = f"(?=({parts}))" chars = list(map(str, range(1, 1 + n))) keys = dict(zip(nums, chars)) matches = [re.findall(pattern, line) for line in raw] part2 = sum( int(keys.get(r[0], r[0]) + keys.get(r[-1], r[-1])) if r else 0 for r in matches ) print(part2) ``` All in all, it was a typical Day 7 or 8 puzzle, which made it a nasty surprise as a Day 1 puzzle. So Day 2 should be easy by comparison - right?