# Advent 2022 - day 3 - sets

Day 3 ended up being a quick one. That is, as long as your language supports set operations. You can do it without set operations but it's easier if you've got them.

## Part 1

Input is a file of text where each line is an input.

You have to read each line, split it in half and then find the common letter between the left and right halves and then apply their scoring rules on that letter.

For scoring, lower case letters are valued a=1, b=2 etc. and for upper case, A=27, B=28 and so on.

For scoring, this is a nice opportunity to talk about ASCII values and the relationship between characters and integers. For lower case letters you can use the formula int(letter) - 97 + 1 where int(x) returns the ASCII code for letter `x`. The 97 is the value for 'a' so that will map the lower case letter to a value of 0 for 'a,' 1 for 'b,' etc.. We add one becuase the question says to start with 1 for 'a.'

For the upper case letters, we can use int(letter)-65 + 27 where 65 is the value for 'A.' In this case, we add an offset of 27 since that's are lowest upper case score.

Here's the clojure code but it would be similar in Python or other languages:

 ``````1 2 3 4 5 `````` ``````(defn score [letter] (if (Character/isUpperCase letter) (+ 27 (- (int letter) (int \A))) (inc (- (int letter) (int \a))) ))``````

The processing of the data is also pretty straight forward. Here's a Clojure solution:

 ``````1 2 3 4 5 6 `````` `````` (defn parse-part1 [line] (let [size (/ (count line) 2) left (take size line) right (drop size line) common (set/intersection (set left) (set right))] (score (first common))))``````

Line 2's Python equivalent would be `size = len(line)/2`.

Line 3 takes the first half and stores it in left and line 4 drops the first half and keeps the right hand side.

Line 5 uses Clojure's set operations to find the letter in common.

To solve part 1 we just split the data, which starts as a long string, split on newlines and then run the parse-part1 function defined above on each line. We then use reduce to add up all the results.

 ``````1 2 3 4 `````` ``````(defn part1 [data] (->> (str/split data #"\n") (map parse-part1) (reduce +)))``````

Full code can be found here.

A Python solution would look pretty similar. Java would be tricker for beginners because they wouldn't know about sets and using them requires more overhead than either Clojure or Python. The problem can still be solved by beginners. An easy, if inefficient way of finding the intersection would be looping over all the characters in one string and looking for it in the other.

## Part two

Part two was very similar. Instead of splitting each line in half, we had to group entire lines by three.

Here's the clojure code:

 ``````1 2 3 4 5 6 7 8 `````` ``````(defn part2 [data] (->> (str/split data #"\n") (map set) (partition 3) (map #(reduce set/intersection %)) (map first) (map score) (reduce +)))``````

Instead of munging the data and then turning strings into sets, we first use line 2 to transform the list of strings into a list of sets.

Then, we use (partition 3) in line 4. This groups the list 3 at a time. Then, going through the rest of the lines

• line 5 - find the intersection within the group fo 3
• line 6 - there will be one letter in the result, but we have to use first to extract the value since the result is a set.
• line 7 - score all the letters
• line 8 - add them up

A nice little problem. With beginners, you get to talk about the ASCII character relationship and if your language supports it, you can play with sets.