It’s that time of year again! Just like last year, I’ll be posting my solutions to the Advent of Code puzzles. This year, I’ll be solving the puzzles in Kotlin. I’ll post my solutions and code to GitHub as well. If you haven’t given AoC a try, I encourage you to do so along with me!
Day 10 - Pipe Maze
Find the problem description HERE.
The Input - Loop the Loop
Today’s puzzle had a serious amount of parsing, which is really more along the lines of what I’d expect from a weekend at this point in the calendar. On the plus side, it made sense to me to use a builder pattern for building up our input one step at a time today.
/**
* This class represents a location in a grid of PipeTiles
*
* @param row The row of this location in a grid of PipeTiles
* @param col The column of this location in a grid of PipeTiles
*/
data class Location(val row: Int, val col: Int) {
fun toTheSouth(): Location = Location(row + 1, col)
fun toTheNorth(): Location = Location(row - 1, col)
fun toTheWest(): Location = Location(row, col - 1)
fun toTheEast(): Location = Location(row, col + 1)
}
/**
* This class represents a tile in the grid of pipes.
*
* Each tile is accessible via two different directions: north, south,
* east, or west. The exceptions are tiles with no pipe, which aren't
* accessible via any direction, and the default value for the start,
* which is accessible in any direction. This will need to be corrected
* before tracing the loop of pipes.
*
* @property north Is this tile accessible via the north?
* @property south Is this tile accessible via the south?
* @property east Is this tile accessible via the east?
* @property west Is this tile accessible via the west?
*/
data class PipeTile(
var north: Boolean,
var south: Boolean,
var east: Boolean,
var west: Boolean
) {
companion object {
fun default(): PipeTile = PipeTile(false, false, false, false)
fun start(): PipeTile = PipeTile(true, true, true, true)
/**
* Produce a [PipeTile] from a character.
*
* Maps the characters from the input file to the corresponding [PipeTile].
*
* @param char The character to map.
* @return The corresponding [PipeTile].
*/
fun fromChar(char: Char): PipeTile {
return when (char) {
'|' -> PipeTile(north = true, south = true, east = false, west = false)
'-' -> PipeTile(north = false, south = false, east = true, west = true)
'L' -> PipeTile(north = true, south = false, east = true, west = false)
'J' -> PipeTile(north = true, south = false, east = false, west = true)
'7' -> PipeTile(north = false, south = true, east = false, west = true)
'F' -> PipeTile(north = false, south = true, east = true, west = false)
'.' -> default()
'S' -> start()
else -> throw IllegalArgumentException(
"$char does not represent a valid pipe section!"
)
}
}
}
}
/**
* Represents a two-dimensional grid of [PipeTile]s
*
* @property grid A two-dimensional array containing [PipeTile]s.
*/
data class PipeGrid(val grid: List<List<PipeTile>>) {
val rows: Int get() = grid.size
val cols: Int get() = grid.first().size
/**
* Override indexing operations so that the values in the grid can be
* accessed directly via [Location]. This operation includes bounds-checking,
* and will return a representation of an empty tile space for an attempt to
* access a location outside the grid.
*
* @param location The location of the tile to access.
*/
operator fun get(location: Location): PipeTile {
val (row, col) = location
val indexOutOfBounds = (row < 0 || row >= rows || col < 0 || col >= cols)
return if (indexOutOfBounds) PipeTile.default() else grid[row][col]
}
}
/**
* Provides a builder interface for producing a [PipeMap].
*
* Starting with the input file data, this class provides an API for
* building up a [PipeMap] one step at a time.
*/
data class PipeMapBuilder(
var grid: PipeGrid? = null,
var start: Location? = null,
var loopTiles: HashMap<Location, Int>? = null
) {
/**
* Read in the input file and fill in the [PipeGrid]
*
* @param input The list of lines from the input file.
* @return The partially completed [PipeMapBuilder] with grid added.
*/
fun readInputToGrid(input: List<String>): PipeMapBuilder {
grid = PipeGrid(input.map { line ->
line.map { char -> PipeTile.fromChar(char) }
})
return this
}
/**
* Identify the starting location of the [PipeGrid].
*
* Iterate through the grid tiles and pick out the starting Location. The
* starting [PipeTile] will need to be corrected from the default starting
* configuration based on its surrounding tiles.
*
* @return The partially completed [PipeMapBuilder] with starting location added.
*/
fun identifyStartLocation(): PipeMapBuilder {
val grid = grid ?:
throw Exception("Must establish grid before identify start location.")
// Find the starting location.
var startRow = 0
var startCol = 0
loop@ for ((rowIdx, row) in grid.grid.withIndex()) {
for ((colIdx, pipe) in row.withIndex()) {
if (pipe == PipeTile.start()) {
startRow = rowIdx
startCol = colIdx
break@loop
}
}
}
val startLocation = Location(startRow, startCol)
// Correct the start location to identify which directions are actually valid
val startPipe = PipeTile(true, true, true, true)
if (startPipe.north && !grid[startLocation.toTheNorth()].south) {
grid[startLocation].north = false
}
if (startPipe.south && !grid[startLocation.toTheSouth()].north) {
grid[startLocation].south = false
}
if (startPipe.east && !grid[startLocation.toTheEast()].west) {
grid[startLocation].east = false
}
if (startPipe.west && !grid[startLocation.toTheWest()].east) {
grid[startLocation].west = false
}
start = startLocation
return this
}
/**
* Trace the loop of pipes attached to the start location.
*
* Perform a breadth-first search through the pipes, producing a mapping
* of location to the number of steps that location is from the start.
*
* @return The partially completed [PipeMapBuilder] with completed `loopTiles`.
*/
fun tracePipeLoop(): PipeMapBuilder {
val grid = grid ?:
throw Exception("Must establish grid before identify start location.")
val start = start ?:
throw Exception("Must identify starting location before tracing the pipe loop!")
// We'll walk through using a Breadth-first search, so that we can walk away
// from the starting location in both directions at the same time.
val queue = ArrayDeque(listOf(0 to start))
val seenLocations: HashMap<Location, Int> = HashMap<Location, Int>()
// So long as there are tiles left to visit...
while (queue.isNotEmpty()) {
// Get the last location from the queue and add it to the
// mapping of seen locations to number of steps taken, unless
// we've already been there. That's probably not a concern with
// this puzzle, but it's good practice.
val (steps, location) = queue.removeLast()
if (seenLocations.contains(location)) continue
seenLocations[location] = steps
// Add all the neighbors that haven't been visited yet to the queue.
for (neighbor in grid.getNeighborsOf(location)) {
if (seenLocations.contains(neighbor)) continue
queue.addFirst(steps + 1 to neighbor)
}
}
loopTiles = seenLocations
return this
}
/**
* Finalize the [PipeMapBuilder] into a [PipeMap]
*
* @return A [PipeMap] with all the non-null properties of the [PipeMapBuilder].
*/
fun toPipeMap(): PipeMap {
val grid = grid ?:
throw Exception("Must establish grid before finalizing.")
val start = start ?:
throw Exception("Must identify starting location before finalizing.")
val loopTiles = loopTiles ?:
throw Exception("Must identify loop tiles before finalizing.")
return PipeMap(grid, start, loopTiles)
}
}
/**
* Represents a finalized [PipeMap].
*/
data class PipeMap(
val grid: PipeGrid,
var start: Location,
var loopTiles: HashMap<Location, Int>
)
class Day10(input: List<String>) {
// It's my first builder pattern in Kotlin!
private val parsed = PipeMapBuilder()
.readInputToGrid(input)
.identifyStartLocation()
.tracePipeLoop()
.toPipeMap()
}
Okay, to be fair, there’s a good amount of the code that will eventually be used to solve part one in there. Thing is, it didn’t make sense to walk over all the pipe tiles in order to identify which ones were in the loop without counting the steps while we did it. So, with this, we should pretty much be done with part one, right?
Part One - Critter Chaser
In a strange and alien environment, a tiny robot flees into a tangle of pipes. Obviously, we need to go after it! But, we need to be smart about it. First, we’ll figure out how far away from us it could possibly be by determining the maximum number of steps any piece of pipe could be from the starting location.
// data class Location(val row: Int, val col: Int) { ... }
// data class PipeTile(
// var north: Boolean,
// var south: Boolean,
// var east: Boolean,
// var west: Boolean
// ) { ... }
// data class PipeGrid(val grid: List<List<PipeTile>>) { ... }
// data class PipeMapBuilder(
// var grid: PipeGrid? = null,
// var start: Location? = null,
// var loopTiles: HashMap<Location, Int>? = null
// ) { ... }
/**
* Represents a finalized [PipeMap].
*/
data class PipeMap(
val grid: PipeGrid,
var start: Location,
var loopTiles: HashMap<Location, Int>
) {
/**
* Return the maximum number of steps to any tile from the starting location.
*
* @return The maximum number of steps.
*/
fun stepsToFurthestPointFromStart(): Int {
return loopTiles.values.max()
}
}
class Day10(input: List<String>) {
// private val parsed = ...
// In part one, we chase a furry critter through a loop of pipes.
fun solvePart1(): Int = parsed.stepsToFurthestPointFromStart()
}
Yeah, I’ll admit that most of the work was done by the parsing on this one. Unfortunately, that doesn’t really hold true for part two…
Part Two - Enhance!
Having chased the “animal” into the tangled nest of pipes, we find it has eluded us. It probably has a nest somewhere! Clearly, disturbing this thing in its nest is likely to lead to us getting the machine parts we came for and not multiple lacerations from mechanical critter teeth. The thing is, our original map representation is poorly suited for determining which tiles are completely enclosed in the loop of pipes because it’s too compact. By expanding the map’s resolution, we can reliably determine which tiles can be accessed by walking between pipes (and thus excluded from our final count.)
// data class Location(val row: Int, val col: Int) { ... }
data class PipeTile(
var north: Boolean,
var south: Boolean,
var east: Boolean,
var west: Boolean
) {
// companion object { ... }
/**
* Expands a [PipeTile] from a 1x1 to a 3x3 representation.
*
* Expands this [PipeTile] so that it can occupy a 3x3 grid area for filling
* the tiles outside the loop in part two. Each direction that was accessible
* from the [PipeTile] now contains a corresponding PIPE variant, while the
* other tiles are PASSABLE. For example, if '#' is PIPE and '-' is PASSABLE,
* then the following transformations would apply:
*
* --- -#- -#- ---
* '7' -> ##- 'J' -> ##- '|' -> -#- '-' -> ###
* -#- --- -#- ---
*
* @return An [ExpandedMapTile]
*/
fun expand(): List<List<ExpandedMapTile>> {
val pipes = MutableList(3) { MutableList(3) { ExpandedMapTile.PASSABLE } }
pipes[1][1] = ExpandedMapTile.PIPE
if (north) pipes[0][1] = ExpandedMapTile.PIPE
if (south) pipes[2][1] = ExpandedMapTile.PIPE
if (east) pipes[1][2] = ExpandedMapTile.PIPE
if (west) pipes[1][0] = ExpandedMapTile.PIPE
return pipes
}
}
// data class PipeGrid(val grid: List<List<PipeTile>>) { ... }
// data class PipeMapBuilder(
// var grid: PipeGrid? = null,
// var start: Location? = null,
// var loopTiles: HashMap<Location, Int>? = null
// ) { ... }
// data class PipeMap(
// val grid: PipeGrid,
// var start: Location,
// var loopTiles: HashMap<Location, Int>
// ) { ... }
/**
* Represents a tile type in the [ExpandedPipeMap]
*/
enum class ExpandedMapTile {
EMPTY, // Nothing there
PASSABLE, // A passable tile next to a pipe
PIPE, // An impassable pipe tile
OUTER, // An empty tile outside the pipe loop
}
/**
* Represents an expansion of the [PipeMap] for counting inner loop tiles.
*
* In Part two, we need to identify tiles _inside_ the loop of the pipes, which
* means all tiles completely enclosed by the loop. The original grid makes
* it difficult to determine which tiles are able to be traveled to from outside
* the loop. By exapanding each original tile from a 1x1 to a 3x3 configuration,
* we can walk the grid between the pipes.
*
* @param grid A two-dimensional grid of [ExpandedMapTile]s.
*/
data class ExpandedPipeMap(var grid: MutableList<MutableList<ExpandedMapTile>>) {
companion object {
/**
* Expands a [PipeMap] into an [ExpandedPipeMap].
*
* @param pipeMap The [PipeMap] to expand.
* @return The expanded pipe map.
*/
fun fromPipeMap(pipeMap: PipeMap): ExpandedPipeMap {
// Expand all the original [PipeTile]s.
val expansions =
pipeMap.loopTiles.keys.map {
location -> location to pipeMap.grid[location].expand()
}
// For each expanded tile, transfer the [ExpandedPipeTile] variants to
// correct location in a grid three times the size of the original grid.
val grid = MutableList(pipeMap.grid.rows * 3) {
MutableList(pipeMap.grid.cols * 3) { ExpandedMapTile.EMPTY }
}
// For each [ExpandedPipeTile] in each expansion, transfer it to the
// appropriate location int the expanded grid.
for ((location, expansion) in expansions) {
val (originalRow, originalCol) = location
for ((rowIdx, row) in expansion.withIndex()) {
val newRow = (originalRow * 3) + rowIdx
for ((colIdx, isPipe) in row.withIndex()) {
val newCol = (originalCol * 3) + colIdx
grid[newRow][newCol] = isPipe
}
}
}
return ExpandedPipeMap(grid)
}
}
val rows: Int get() = grid.size
val cols: Int get() = grid.first().size
/**
* Indicates whether the given location is accessible in the grid.
*
* In the expanded grid, a neighbor is accessible if it is in the grid
* and it _isn't_ a PIPE tile.
*
* @param location The [Location] to check.
* @return Whether the location is accessible
*/
private fun isAccessible(location: Location): Boolean {
if (location.row < 0 || location.row >= rows) return false
if (location.col < 0 || location.col >= cols) return false
if (grid[location.row][location.col] == ExpandedMapTile.PIPE) return false
return true
}
/**
* An updated function for finding neighbors in this new expanded grid.
*
* @param location The location to get neighbors for.
* @return A list of accessible neighbor locations.
*/
private fun getNeighborsOf(location: Location): List<Location> {
val (row, col) = location
val neighbors = mutableListOf<Location>()
if (isAccessible(Location(row - 1, col))) neighbors.add(Location(row - 1, col))
if (isAccessible(Location(row + 1, col))) neighbors.add(Location(row + 1, col))
if (isAccessible(Location(row, col - 1))) neighbors.add(Location(row, col - 1))
if (isAccessible(Location(row, col + 1))) neighbors.add(Location(row, col + 1))
return neighbors
}
/**
* Count the tiles enclosed by the loop of pipe.
*
* In part two, we identify the tiles that _aren't_ enclosed in
* the loop by "flood-filling" from an identified outside tile. How
* was this starting tile identified? Visual inspection of the input!
*
* @return The number of tiles enclosed.
*/
fun countInnerTiles(): Int {
// Flood fill to populate the list of seen locations. Perform a
// breadth-first search starting from a known outside tile and
// convert any tile found into an OUTSIDE tile.
val queue = ArrayDeque(listOf(Location(0, 0)))
val seenLocations: HashSet<Location> = HashSet<Location>()
while (queue.isNotEmpty()) {
val location = queue.removeLast()
if (seenLocations.contains(location)) continue
seenLocations.add(location)
// Mark the found tile as an OUTER tile
val (row, col) = location
grid[row][col] = ExpandedMapTile.OUTER
// Check each neighbor that hasn't been visited yet.
for (neighbor in this.getNeighborsOf(location)) {
if (seenLocations.contains(neighbor)) continue
queue.addFirst(neighbor)
}
}
// Return a count of all the EMPTY tiles. The total count is
// divided by 9, because expanding a [PipeTile] to an [ExpandedMapTile]
// increases the number of grid spaces by a factor of 9.
return (grid.sumOf { row -> row.count { it == ExpandedMapTile.EMPTY } }) / 9
}
}
class Day10(input: List<String>) {
// private val parsed = ...
// fun solvePart1(): Int = ...
// In part two we increase the resolution of the map to make it possible
// to flag all the OUTER tiles, leaving the EMPTY tiles as the inner tiles.
fun solvePart2(): Int = ExpandedPipeMap.fromPipeMap(parsed).countInnerTiles()
}
There’s quite a bit of code there, and I really hope the comments are sufficient to explain what’s going on here.
Wrap Up
Today was a doozy! Not so much from the perspective of solving the puzzle, but just the sheer amount of code I wrote to parse, and then re-parse the parsed input into an expanded format. Granted, from the perspective of “learning Kotlin”, it was a good opportunity to write more code. And, I got to test-drive the builder pattern (even though it may not have been completely necessary). A lot of code on a weekend isn’t anything to get upset about either. All told, it was a good day!
Top comments (0)