The Humble Hash

Hashes seem simple. Set a key to a corresponding value, retrieve the value by key. What else is there to know?

A whole lot, it turns out! Ruby makes several surprising choices about how hashes work, which turn hashes into dynamic powerhouses of functionality.

We'll dive into how hashes work, understand how they form the groundwork for some extremely useful standard library utilities, and learn patterns to leverage the unparalleled versatility of the humble hash to write concise, performant, beautiful code.

Transcript

1. RUBYCONF 2020 Ariel Caplan • @amcaplan • amcaplan.ninja The Humble Hash
3. LIFE
6. cell.next_state = if cell.alive? [2,3].include?(cell.live_neighbors_count) ? 1 : 0 else cell.live_neighbors_count == 3 ? 1 : 0 end
16. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN •
17. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • • Ruby programmer since 2014 • Backend Developer at • Find me online: @amcaplan • https://amcaplan.ninja Hi! I’m Ariel Caplan.
18. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • talk[:basics]
19. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • • Performant key-value pair store • Keys and values can be any Ruby object • Literal syntax: • Explicit syntax: • Strict fetching: • Fetching with default value: • Setting a different default value: What is a Hash? { key: "value" } hash = Hash.new hash[:key] = "value" hash[:key] #=> "value" hash[:non_key] #=> nil hash.fetch(:key) hash = Hash.new(4) hash[:non_key] #=> 4
20. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • talk[:obj_key][:set]
21. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Set require 'set' set = Set.new([1,2,3]) #=> #<Set: {1, 2, 3}> set.include?(3) #=> true set.include?(4) #=> false set.add(4) #=> #<Set: {1, 2, 3, 4}> set.add(3) #=> #<Set: {1, 2, 3, 4}>
22. def add(o) @hash[o] = true self end def merge # ... do_with_enum(enum) { |o| add(o) } # ... end class Set def initialize(enum = nil, &block) @hash ||= Hash.new(false) # some other stuff merge(enum) end RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Set def include?(o) @hash[o] end end
23. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • talk[:obj_key][:cache]
24. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Disclaimers # disclaimer { key: "network", value: "tier_a", text: "To receive the lowest out-of-pocket costs, please choose a provider in the Tier A network." } # provider metadata { specialty: "periodontist", network: "tier_a" }
25. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Disclaimers # 500 x 300 = 150,000 operations # once per disclaimer - 300 disclaimers disclaimers.select { |disclaimer| # once per provider - 500 providers providers.any? { |provider| provider.metadata[disclaimer[:key]] == disclaimer[:value] } }.map { |disclaimer| disclaimer[:text] }
26. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Disclaimers # 500 + 300 = 800 operations # once per provider - 500 providers provider_data = Set.new providers.each do |provider| provider.metadata.each do |key, value| provider_data << { key: key, value: value } end end # once per disclaimer - 300 disclaimers disclaimers.select { |disclaimer| data = { key: disclaimer[:key], value: disclaimer[:value] } provider_data.include?(data) }.map { |disclaimer| disclaimer[:text] }
27. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • ! • String • Symbol • Numeric • !1.0.eql?(1) • Array • Hash " • Anything for which equality isn’t obvious Hash Key Warning! Containing valid types {
28. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • talk[:defaults][:wizardry]
29. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • • 1 object that will always be returned • Doesn’t set a key-value pair • Set at hash initialization: • Set after hash initialization Defaults Deep Dive: Default Value Hash.new(4) hash = {} hash.default = 4
30. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • • Gotcha: It will always be the same object! Defaults Deep Dive: Default Value hash = Hash.new("abc") str = hash[:key] str << "d" hash[:another_key] #=> "abcd"
31. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • • Proc that will be run when a key is not found • Can set a key-value pair, if you want • Set at hash initialization: • Set after hash initialization Defaults Deep Dive: Default Proc hash = Hash.new { |h, k| h[k] = 4 } hash = {} hash.default_proc = ->(h, k) { h[k] = 4 }
32. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Default Proc Can Reference Itself! default_proc = ->(h, k) { h[k] = Hash.new(&default_proc) } hash = Hash.new(&default_proc) hash[:a][:b][:c] = 4 hash #=> {:a=>{:b=>{:c=>4}}}
34. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Fibonacci Sequence Fn = Fn-2 + Fn-1 0, 1, 1, 2, 3, 5, 8, 13, 21, 34, 55…
35. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Fibonacci Sequence: Iterative def fib(n) i, j = 0, 1 counter = 0 while counter < n i, j = j, i + j counter += 1 end i end
36. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Fibonacci Sequence: Recursive def fib(n) case n when 0, 1 n else fib(n - 2) + fib(n - 1) end end
37. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Fibonacci Sequence: Default Proc hash = Hash.new { |h, k| h[k] = h[k-2] + h[k-1] } hash[0] = 0 hash[1] = 1
38. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Fibonacci Performance fib(4) fib(2) fib(1) fib(0) fib(3) fib(2) fib(1) fib(1) fib(0)
39. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Fibonacci Performance hash[4] hash[2] hash[1] hash[0] hash[1] hash[3] hash[2]
40. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Fibonacci Performance hash[4] hash[2] hash[1] hash[0] hash[1] hash[3] hash[2] fib(4) fib(2) fib(1) fib(0) fib(3) fib(2) fib(1) fib(1) fib(0) 9 calls 7 calls
41. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Fibonacci Performance hash[4] hash[0] hash[1] hash[3] hash[2] fib(4) fib(2) fib(1) fib(0) fib(3) fib(2) fib(1) fib(1) fib(0) fib(5) fib(1) fib(3) fib(2) fib(1) fib(0) hash[5] hash[1] hash[3] hash[2] 15 calls 9 calls
42. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Fibonacci Performance hash[4] hash[0] hash[1] hash[3] hash[2] fib(4) fib(2) fib(1) fib(0) fib(3) fib(2) fib(1) fib(1) fib(0) fib(5) fib(1) fib(3) fib(2) fib(1) fib(0) hash[5] hash[1] hash[3] hash[2] 25 calls 11 calls fib(6) fib(4) fib(2) fib(1) fib(0) fib(3) fib(2) fib(1) fib(1) fib(0) hash[6] hash[4]
43. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Fibonacci Performance 29,860,703 calls 69 calls fib(35) hash[35]
44. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Conway’s Game of Life def life_func(x, y, step) if step == 0 INITIAL_STATES[[x, y]] || 0 else neighbor_diffs = [[-1, -1], [-1, 0], [-1, 1], [ 0, -1], [ 0, 1], [ 1, -1], [ 1, 0], [ 1, 1]] live_neighbors = neighbor_diffs.sum { |x_diff, y_diff| life_func((x + x_diff) % WIDTH, (y + y_diff) % HEIGHT, step - 1) } if life_func(x, y, step - 1) == 1 [2, 3].include?(live_neighbors) ? 1 : 0 else live_neighbors == 3 ? 1 : 0 end end end
45. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Conway’s Game of Life life_hash = Hash.new do |h, (x, y, step)| h[[x, y, step]] = if step == 0 INITIAL_STATES[[x, y]] || 0 else neighbor_diffs = [[-1, -1], [-1, 0], [-1, 1], [ 0, -1], [ 0, 1], [ 1, -1], [ 1, 0], [ 1, 1]] live_neighbors = neighbor_diffs.sum { |x_diff, y_diff| h[[(x + x_diff) % WIDTH, (y + y_diff) % HEIGHT, step - 1]] } if h[[x, y, step - 1]] == 1 [2, 3].include?(live_neighbors) ? 1 : 0 else live_neighbors == 3 ? 1 : 0 end end end
46. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Conway’s Game of Life life_hash = Hash.new do |h, (x, y, step)| h[[x, y, step]] = if step == 0 INITIAL_STATES[[x, y]] || 0 else neighbor_diffs = [[-1, -1], [-1, 0], [-1, 1], [ 0, -1], [ 0, 1], [ 1, -1], [ 1, 0], [ 1, 1]] live_neighbors = neighbor_diffs.sum { |x_diff, y_diff| h[[(x + x_diff) % WIDTH, (y + y_diff) % HEIGHT, step - 1]] } if h[[x, y, step - 1]] == 1 [2, 3].include?(live_neighbors) ? 1 : 0 else live_neighbors == 3 ? 1 : 0 end end end def life_func(x, y, step) if step == 0 INITIAL_STATES[[x, y]] || 0 else neighbor_diffs = [[-1, -1], [-1, 0], [-1, 1], [ 0, -1], [ 0, 1], [ 1, -1], [ 1, 0], [ 1, 1]] live_neighbors = neighbor_diffs.sum { |x_diff, y_diff| life_func((x + x_diff) % WIDTH, (y + y_diff) % HEIGHT, step - 1) } if life_func(x, y, step - 1) == 1 [2, 3].include?(live_neighbors) ? 1 : 0 else live_neighbors == 3 ? 1 : 0 end end end
47. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Conway’s Game of Life
48. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • talk[:defaults][:abbrev]
49. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Abbrev require 'abbrev' Abbrev.abbrev(["conf", "code"]) #=> {"conf"=>"conf", "con"=>"conf", "code"=>"code", "cod"=>"code"}
50. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Abbrev def abbrev(words, pattern = nil) table = {} seen = Hash.new(0) if pattern.is_a?(String) pattern = /\A#{Regexp.quote(pattern)}/ # regard as a prefix end words.each do |word| next if word.empty? word.size.downto(1) { |len| abbrev = word[0...len] next if pattern && pattern !~ abbrev case seen[abbrev] += 1 when 1 table[abbrev] = word when 2 table.delete(abbrev) else break end } end words.each do |word| next if pattern && pattern !~ word table[word] = word end table end
51. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Abbrev Seen Table conf => conf
52. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Abbrev Seen Table conf => conf conf 1 => conf
53. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Abbrev Seen Table conf => conf conf 1 con => conf
54. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Abbrev Seen Table conf => conf conf 1 con con 1 => conf => conf
55. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Abbrev Seen Table conf conf 1 con con 1 co co 1 c c 1 => conf => conf => conf => conf code => code
56. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Abbrev Seen Table conf conf 1 con con 1 co co 1 c c 1 => conf => conf => conf => conf code => code code 1 => code cod cod 1 => code co
57. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Abbrev Seen Table conf conf 1 con con 1 co co 1 c c 1 => conf => conf => conf => conf code => code code 1 => code cod cod 1 => code co 2
58. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Abbrev Seen Table conf conf 1 con con 1 co 1 c c 1 => conf => conf => conf code => code code 1 => code cod cod 1 => code 2
59. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Abbrev Seen Table conf conf 1 con con 1 co 1 c c 1 => conf => conf => conf code => code code 1 => code cod cod 1 => code 2 c
60. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Abbrev Seen Table conf conf 1 con con 1 co 1 c 1 => conf => conf code => code code 1 => code cod cod 1 => code 2 2
61. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Abbrev def abbrev(words) table = {} seen = Hash.new(0) words.each do |word| word.size.downto(1) { |len| abbrev = word[0...len] case seen[abbrev] += 1 when 1 table[abbrev] = word when 2 table.delete(abbrev) else break end } end table end Default Value Return Value Add Abbreviations Seen Once Remove Ambiguous Abbreviations Never set seen[abbrev] to 0! Iterate over each word portion
62. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • talk[:defaults][:app_code]
63. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Abbrev-like Default Value totals = Hash.new(0) items = ["a", "a", "b", "c", "c", "c", "d", "b"] items.each { |item| totals[item] += 1 } totals #=> {"a"=>2, "b"=>2, "c"=>3, "d"=>1} totals["e"] #=> 0
64. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Real-Life Defaults! totals = Hash.new(0) # { "sum" => 1000 } by_company = Hash.new { |h,k| h[k] = Hash.new(0) } # { "a_company" => { "sum" => 100 }} totals["unset_key"] #=> 0 by_company["some_company"]["sum"] #=> 0
65. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Real-Life Defaults! def some_expensive_calculation @some_expensive_calculation ||= _do_the_expensive_calculation end
66. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Real-Life Defaults! def some_expensive_calculation_a @some_expensive_calculation_a ||= _do_the_expensive_calculation(:a) end def some_expensive_calculation_b @some_expensive_calculation_b ||= _do_the_expensive_calculation(:b) end def some_expensive_calculation_c @some_expensive_calculation_c ||= _do_the_expensive_calculation(:c) end
67. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Real-Life Defaults! def some_expensive_calculation(argument) @some_expensive_calculation ||= Hash.new do |h, k| h[k] = _do_the_expensive_calculation(k) end @some_expensive_calculation[argument] end
68. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • •Flexibility: any object can be a key or value •Dynamicity and Resilience: default value/proc •Recursion-ish •Caching strategies What have we covered?
69. RUBYCONF 2020 • ARIEL CAPLAN • @AMCAPLAN • Please say hi in the Slack chat!