Merge pull request #52 from krzysiek1507/feature/performance-fixes

Performance and memory fixes

lib/hashdiff/diff.rb

@@ -122,13 +122,15 @@ def self.diff(obj1, obj2, options = {}, &block)
     elsif obj1.is_a?(Array) && !opts[:use_lcs]
       result.concat(LinearCompareArray.call(obj1, obj2, opts))
     elsif obj1.is_a?(Hash)
+      obj1_keys = obj1.keys
+      obj2_keys = obj2.keys
 
-      deleted_keys = obj1.keys - obj2.keys
-      common_keys = obj1.keys & obj2.keys
-      added_keys = obj2.keys - obj1.keys
+      deleted_keys = (obj1_keys - obj2_keys).sort_by(&:to_s)
+      common_keys = (obj1_keys & obj2_keys).sort_by(&:to_s)
+      added_keys = (obj2_keys - obj1_keys).sort_by(&:to_s)
 
       # add deleted properties
-      deleted_keys.sort_by { |k, _v| k.to_s }.each do |k|
+      deleted_keys.each do |k|
         change_key = prefix_append_key(opts[:prefix], k, opts)
         custom_result = custom_compare(opts[:comparison], change_key, obj1[k], nil)
 
@@ -140,13 +142,13 @@ def self.diff(obj1, obj2, options = {}, &block)
       end
 
       # recursive comparison for common keys
-      common_keys.sort_by { |k, _v| k.to_s }.each do |k|
+      common_keys.each do |k|
         prefix = prefix_append_key(opts[:prefix], k, opts)
         result.concat(diff(obj1[k], obj2[k], opts.merge(prefix: prefix)))
       end
 
       # added properties
-      added_keys.sort_by { |k, _v| k.to_s }.each do |k|
+      added_keys.each do |k|
         change_key = prefix_append_key(opts[:prefix], k, opts)
         next if obj1.key?(k)
 
@@ -171,28 +173,33 @@ def self.diff(obj1, obj2, options = {}, &block)
   #
   # diff array using LCS algorithm
   def self.diff_array_lcs(arraya, arrayb, options = {})
-    opts = {
-      prefix: '',
-      similarity: 0.8,
-      delimiter: '.'
-    }.merge!(options)
+    return [] if arraya.empty? && arrayb.empty?
 
     change_set = []
-    return [] if arraya.empty? && arrayb.empty?
 
     if arraya.empty?
       arrayb.each_index do |index|
         change_set << ['+', index, arrayb[index]]
       end
+
       return change_set
-    elsif arrayb.empty?
+    end
+
+    if arrayb.empty?
       arraya.each_index do |index|
         i = arraya.size - index - 1
         change_set << ['-', i, arraya[i]]
       end
+
       return change_set
     end
 
+    opts = {
+      prefix: '',
+      similarity: 0.8,
+      delimiter: '.'
+    }.merge!(options)
+
     links = lcs(arraya, arrayb, opts)
 
     # yield common

lib/hashdiff/lcs.rb

@@ -6,12 +6,12 @@ module HashDiff
   # caculate array difference using LCS algorithm
   # http://en.wikipedia.org/wiki/Longest_common_subsequence_problem
   def self.lcs(arraya, arrayb, options = {})
+    return [] if arraya.empty? || arrayb.empty?
+
     opts = { similarity: 0.8 }.merge!(options)
 
     opts[:prefix] = prefix_append_array_index(opts[:prefix], '*', opts)
 
-    return [] if arraya.empty? || arrayb.empty?
-
     a_start = b_start = 0
     a_finish = arraya.size - 1
     b_finish = arrayb.size - 1

lib/hashdiff/util.rb

@@ -7,14 +7,15 @@ module HashDiff
   def self.similar?(obja, objb, options = {})
     return compare_values(obja, objb, options) unless obja.is_a?(Array) || obja.is_a?(Hash) || objb.is_a?(Array) || objb.is_a?(Hash)
 
-    opts = { similarity: 0.8 }.merge(options)
-
     count_a = count_nodes(obja)
     count_b = count_nodes(objb)
-    diffs = count_diff diff(obja, objb, opts)
 
     return true if (count_a + count_b).zero?
 
+    opts = { similarity: 0.8 }.merge!(options)
+
+    diffs = count_diff diff(obja, objb, opts)
+
     (1 - diffs.to_f / (count_a + count_b).to_f) >= opts[:similarity]
   end
 
@@ -83,8 +84,8 @@ def self.node(hash, parts)
   #
   # check for equality or "closeness" within given tolerance
   def self.compare_values(obj1, obj2, options = {})
-    if (options[:numeric_tolerance].is_a? Numeric) &&
-       [obj1, obj2].all? { |v| v.is_a? Numeric }
+    if options[:numeric_tolerance].is_a?(Numeric) &&
+       obj1.is_a?(Numeric) && obj2.is_a?(Numeric)
       return (obj1 - obj2).abs <= options[:numeric_tolerance]
     end
 
@@ -105,9 +106,7 @@ def self.compare_values(obj1, obj2, options = {})
   #
   # check if objects are comparable
   def self.comparable?(obj1, obj2, strict = true)
-    [Array, Hash].each do |type|
-      return true if obj1.is_a?(type) && obj2.is_a?(type)
-    end
+    return true if (obj1.is_a?(Array) || obj1.is_a?(Hash)) && obj2.is_a?(obj1.class)
     return true if !strict && obj1.is_a?(Numeric) && obj2.is_a?(Numeric)
 
     obj1.is_a?(obj2.class) && obj2.is_a?(obj1.class)

spec/hash_diff/util_spec.rb

@@ -20,6 +20,14 @@
     described_class.similar?(a, b, similarity: 1).should be false
   end
 
+  it 'is able to tell similiar empty hash' do
+    described_class.similar?({}, {}, 1).should be true
+  end
+
+  it 'is able to tell similiar empty array' do
+    described_class.similar?([], [], 1).should be true
+  end
+
   it 'is able to tell similiar hash with values within tolerance' do
     a = { 'a' => 1.5, 'b' => 2.25, 'c' => 3, 'd' => 4, 'e' => 5 }
     b = { 'a' => 1.503, 'b' => 2.22, 'c' => 3, 'e' => 5 }