Day 8

Advent of Code: Worked Solutions

Setup

Import libraries:

library(tidyverse)

Read input from file:

input <- read_delim(
  file = "../input/day08.txt", 
  delim = " | ",
  col_names = c("signal", "output"),
  show_col_types = FALSE
) |> 
  mutate(across(everything(), ~ str_split(.x, " ")))

Part 1

Count the total number of digits 1, 4, 7, and 8 in each output by counting the number of output values with length 2, 3, 4, or 7:

input |> 
  pull(output) |> 
  list_c() |> 
  str_length() |> 
  keep(~ .x %in% c(2, 3, 4, 7)) |> 
  length()

Part 2

By visual examination, we have the following rules:

• Digits 1, 4, 7, and 8 are identifiable by length alone.
• Digits 0, 6, 9 all have the same length:
  • 9 is the only superset of 4
  • Once 9 is identified, 0 is the only remaining superset of 1
  • 6 is the last remaining value.
• Digits 2, 3, 5 all have the same length.
  • 3 is the only superset of 1
  • 5 is the only subset of 6
  • 2 is the remainder

Using these rules, we can quickly identify which signals correspond to which digits. First, we define a helper function needed to fill null values in a single-value list:

lst_fill <- function(x) {
  x |> 
    discard(is.null) |> 
    unique() |> 
    rep(length(x))
}

Next, prepare our input by splitting and sorting the characters in each code:

df <- input |> 
  mutate(entry_id = row_number()) |> 
  pivot_longer(c(signal, output), names_to = "type", values_to = "code") |> 
  unnest_longer(code) |> 
  mutate(
    code = map(code, ~ sort(str_split_1(.x, ""))),
    length = map_int(code, length)
  )

Identify the digit corresponding to each code:

result <- df |> 
  mutate(
    
    # Identify digits 1, 4, 7, 8 by length only
    digit = case_match(length, 2 ~ 1, 4 ~ 4, 3 ~ 7, 7 ~ 8),
    
    # Identify digits 0, 6, 9 by whether they have 1/4 as a subset
    four = lst_fill(case_when(digit == 4 ~ code)),
    one  = lst_fill(case_when(digit == 1 ~ code)),
    digit = case_when(
      length == 6 & map2_lgl(code, four, ~ all(.y %in% .x)) ~ 9,
      length == 6 & map2_lgl(code, one,  ~ all(.y %in% .x)) ~ 0,
      length == 6 ~ 6,
      .default = digit
    ),
    
    # Identify digits 2, 3, 5 by whether they have 1/6 as a subset/superset
    six = lst_fill(case_when(digit == 6 ~ code)),
    digit = case_when(
      length == 5 & map2_lgl(code, one, ~ all(.y %in% .x)) ~ 3,
      length == 5 & map2_lgl(code, six, ~ all(.x %in% .y)) ~ 5,
      length == 5 ~ 2,
      .default = digit
    ),
    
    .by = entry_id
  ) |> 
  select(entry_id, type, digit, code)

Determine the output value for each entry and take the sum for the final result:

result |> 
  filter(type == "output") |> 
  summarize(
    value = parse_number(str_c(digit, collapse = "")), 
    .by = entry_id
  ) |> 
  pull(value) |> 
  sum()