Post: Advent of code day 7

This commit is contained in:
Jez Cope 2017-12-10 19:05:02 +00:00
parent a934838a83
commit c33d9ec60f
2 changed files with 133 additions and 0 deletions

View File

@ -13,6 +13,8 @@ series: aoc2017
[Today's challenge](http://adventofcode.com/2017/day/6) asks us to follow a recipe for redistributing objects in memory that bears a striking resemblance to the rules of the African game [Mancala](http://en.wikipedia.org/wiki/Mancala).
[→ Full code on GitHub](https://github.com/jezcope/aoc2017/blob/master/06-memory-reallocation.py)
!!! commentary
When I was doing my MSci, one of our programming exercises was to write (in Haskell, IIRC) a program to play a Mancala variant called [Oware](http://en.wikipedia.org/wiki/Oware), so this had a nice ring of nostalgia.

View File

@ -0,0 +1,131 @@
---
title: "Recursive Circus — Ruby — #adventofcode Day 7"
description: "In which I go full object-oriented in Ruby."
slug: day-07
date: 2017-12-10T19:04:36+00:00
tags:
- Technology
- Learning
- Advent of Code
- Ruby
series: aoc2017
---
[Today's challenge](http://adventofcode.com/2017/day/7) introduces a set of processes balancing precariously on top of each other. We find them stuck and unable to get down because one of the processes is the wrong size, unbalancing the whole circus. Our job is to figure out the root from the input and then find the correct weight for the single incorrect process.
[→ Full code on GitHub](https://github.com/jezcope/aoc2017/blob/master/07-recursive-circus.rb)
!!! commentary
So I didn't really intend to take a full polyglot approach to Advent of Code, but it turns out to have been quite fun, so I made a shortlist of languages to try. Building a tree is a classic application for object-orientation using a class to represent tree nodes, and I've always liked the feel of [Ruby](https://www.ruby-lang.org/en/)'s class syntax, so I gave it a go.
First make sure we have access to `Set`, which we'll use later.
```ruby
require 'set'
```
Now to define the `CircusNode` class, which represents nodes in the tree. `attr :s` automatically creates a function `s` that returns the value of the instance attribute `@s`
```ruby
class CircusNode
attr :name, :weight
def initialize(name, weight, children=nil)
@name = name
@weight = weight
@children = children || []
end
```
Add a `<<` operator (the same syntax for adding items to a list) that adds a child to this node.
```ruby
def <<(c)
@children << c
@total_weight = nil
end
```
`total_weight` recursively calculates the weight of this node and everything above it. The `@total_weight ||= blah` idiom caches the value so we only calculate it once.
```ruby
def total_weight
@total_weight ||= @weight + @children.map {|c| c.total_weight}.sum
end
```
`balance_weight` does the hard work of figuring out the proper weight for the incorrect node by recursively searching through the tree.
```ruby
def balance_weight(target=nil)
by_weight = Hash.new{|h, k| h[k] = []}
@children.each{|c| by_weight[c.total_weight] << c}
if by_weight.size == 1 then
if target
return @weight - (total_weight - target)
else
raise ArgumentError, 'This tree seems balanced!'
end
else
odd_one_out = by_weight.select {|k, v| v.length == 1}.first[1][0]
child_target = by_weight.select {|k, v| v.length > 1}.first[0]
return odd_one_out.balance_weight child_target
end
end
```
A couple of utility functions for displaying trees finish off the class.
```ruby
def to_s
"#{@name} (#{@weight})"
end
def print_tree(n=0)
puts "#{' '*n}#{self} -> #{self.total_weight}"
@children.each do |child|
child.print_tree n+1
end
end
end
```
`build_circus` takes input as a list of lists `[name, weight, children]`. We make two passes over this list, first creating all the nodes, then building the tree by adding children to parents.
```ruby
def build_circus(data)
all_nodes = {}
all_children = Set.new
data.each do |name, weight, children|
all_nodes[name] = CircusNode.new name, weight
end
data.each do |name, weight, children|
children.each {|child| all_nodes[name] << all_nodes[child]}
all_children.merge children
end
root_name = (all_nodes.keys.to_set - all_children).first
return all_nodes[root_name]
end
```
Finally, build the tree and solve the problem! Note that we use `String.to_sym` to convert the node names to symbols (written in Ruby as `:symbol`), because they're faster to work with in `Hash`es and `Set`s as we do above.
```ruby
data = readlines.map do |line|
match = /(?<parent>\w+) \((?<weight>\d+)\)(?: -> (?<children>.*))?/.match line
[match['parent'].to_sym,
match['weight'].to_i,
match['children'] ? match['children'].split(', ').map {|x| x.to_sym} : []]
end
root = build_circus data
puts "Root node: #{root}"
puts root.balance_weight
```