myers_diff 1.0.0 → 1.1.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 30beabf5813fa2a01a7b2d599b5041afb61c8ec6a1d91c0298cfb9bde2c9a9eb
-  data.tar.gz: a5eaeef03046a444bfa566629b9cf38f4f4f97604423dc7cbb509ad35c27f616
+  metadata.gz: 38ecad0ebd0a04de796e852e6cfc77065daf0b0da1e6384beed39520b4279230
+  data.tar.gz: 285e2e66e2e25facf31c54b2c32d7d45d8bd2f695d7f789f04b84c579ef2abc4
 SHA512:
-  metadata.gz: 534e00e7e71ffa1a878c4f885af92fed85758ad927dde76f8ac1e89b4e4a2b5534bab2cff073df6ec23ced8f16a6ebe777a6e3efe8664e3e282c151aeefabb0b
-  data.tar.gz: 5808a27d91717e5b8fb280f35d23d92afbd1d6be8ac854feb308ac0e33a818119c2bd60e8cc540bd475cf02c80f317a848f0b14e1bdbaa682fde73eeeeeae212
+  metadata.gz: 1fa47b2e26e99fc72dbda6f16155848de64da9b59da6aa68cc8ecdfca127395b0887c8e4ecd3ecaa2a3acab2e937c7175aa50f35c4a6b8d39bef0c906286b2e3
+  data.tar.gz: eeb8d1d8db1bebda97e164932b4e878374945d1940f8515d2f0b308bd922361fff2e1ae035f5044fb43ea405a283109a407c4a9fc32c145b200526ff3afbe213

data/lib/myers_diff.rb

@@ -1 +1,2 @@
 require 'myers_diff/char_diff'
+require 'myers_diff/word_diff'

data/lib/myers_diff/version.rb

@@ -1,3 +1,3 @@
 module MyersDiff
-  VERSION = "1.0.0".freeze
+  VERSION = "1.1.0".freeze
 end

data/lib/myers_diff/word_diff.rb

@@ -0,0 +1,187 @@
+module MyersDiff
+  class WordDiff
+    def diff(s1, s2, **options)
+      old_string = cast_input(s1)
+      new_string = cast_input(s2)
+
+      old_string = remove_empty(tokenize(old_string))
+      new_string = remove_empty(tokenize(new_string))
+
+      new_len = new_string.size
+      old_len = old_string.size
+      edit_length = 1
+      max_edit_length = new_len + old_len
+      best_path = { }
+      best_path[0] = { new_pos: -1, components: [] }
+
+      old_pos = extract_common(best_path[0], new_string, old_string, 0)
+      if best_path[0][:new_pos] + 1 >= new_len && old_pos + 1 >= old_len
+        return [ { value: join(new_string), count: new_string.size } ]
+      end
+
+      exec_edit_length = lambda do
+        diagonal_path = -1 * edit_length
+        while diagonal_path <= edit_length
+          add_path = best_path[diagonal_path - 1]
+          remove_path = best_path[diagonal_path + 1]
+          old_pos = (remove_path ? remove_path[:new_pos] : 0) - diagonal_path
+          best_path[diagonal_path - 1] = nil if add_path
+
+          can_add = add_path && add_path[:new_pos] + 1 < new_len
+          can_remove = remove_path && 0 <= old_pos && old_pos < old_len
+          if !can_add && !can_remove
+            best_path[diagonal_path] = nil
+            diagonal_path += 2
+            next
+          end
+
+          base_path = if !can_add || (can_remove && add_path[:new_pos] < remove_path[:new_pos])
+            p = clone_path(remove_path)
+            push_component(p[:components], nil, true)
+            p
+          else
+            p = add_path
+            p[:new_pos] += 1
+            push_component(p[:components], true, nil)
+            p
+          end
+
+          old_pos = extract_common(base_path, new_string, old_string, diagonal_path)
+
+          if base_path[:new_pos] + 1 >= new_len && old_pos + 1 >= old_len
+            return build_values(base_path[:components], new_string, old_string)
+          else
+            best_path[diagonal_path] = base_path
+          end
+
+          diagonal_path += 2
+        end
+
+        edit_length += 1
+        nil
+      end
+
+      while edit_length <= max_edit_length
+        if res = exec_edit_length.call
+          return res
+        end
+      end
+
+      'death'
+    end
+
+    def push_component(components, added, removed)
+      last = components.last
+      if last && last[:added] == added && last[:removed] == removed
+        components[-1] = { added: last[:added], removed: last[:removed], count: last[:count] + 1 }
+      else
+        components.push(count: 1, added: added, removed: removed)
+      end
+    end
+
+    # base_path : { new_pos: int, components: [] }
+    # diagonal_path : int
+    def extract_common(base_path, new_string, old_string, diagonal_path)
+      new_len = new_string.size
+      old_len = old_string.size
+      new_pos = base_path[:new_pos]
+      old_pos = new_pos - diagonal_path
+      common_count = 0
+
+      while new_pos + 1 < new_len && old_pos + 1 < old_len && equals(new_string[new_pos + 1], old_string[old_pos + 1])
+        new_pos += 1
+        old_pos += 1
+        common_count += 1
+      end
+
+      if common_count > 0
+        base_path[:components].push(count: common_count)
+      end
+
+      base_path[:new_pos] = new_pos
+      old_pos
+    end
+
+    def equals(l, r)
+      l == r
+      # TODO: support custom comparator
+      # TODO: support case-insensitive
+    end
+
+    def remove_empty(array)
+      array.compact
+    end
+
+    def cast_input(str)
+      str
+    end
+
+    def tokenize(str)
+      str.split(' ')
+    end
+
+    def join(chars)
+      chars.join(' ')
+    end
+
+    # new_string - tokenized string i.e. array of strings
+    def build_values(components, new_string, old_string, use_longest_token = true)
+      component_pos = 0
+      component_len = components.size
+      new_pos = 0
+      old_pos = 0
+
+      while component_pos < component_len
+        component = components[component_pos]
+        if !component[:removed]
+          if !component[:added] && use_longest_token
+            value = new_string[new_pos, component[:count]]
+            value = value.map.with_index do |val, i|
+              old_val = old_string[old_pos + i]
+              old_val.size > val.size ? old_val : val
+            end
+
+            component[:value] = join(value)
+          else
+            component[:value] = join(new_string[new_pos, component[:count]])
+          end
+
+          new_pos += component[:count]
+          old_pos += component[:count] unless component[:added]
+        else
+          component[:value] = join(old_string[old_pos, component[:count]])
+          old_pos += component[:count]
+
+          if component_pos && 0 <= component_pos - 1 && components[component_pos - 1][:added]
+            tmp = components[component_pos - 1]
+            components[component_pos - 1] = components[component_pos]
+            components[component_pos] = tmp
+          end
+        end
+
+        component_pos += 1
+      end
+
+      last_component = components[component_len - 1]
+      if component_len > 1 &&
+          last_component[:value].is_a?(String) &&
+          (last_component[:added] || last_component[:removed]) &&
+          equals('', last_component[:value])
+        components[component_len - 2][:value] += last_component[:value]
+        components.pop
+      end
+
+      recount(components)
+
+      components
+    end
+
+    def clone_path(path_hash)
+      { new_pos: path_hash[:new_pos], components: path_hash[:components].dup }
+    end
+
+    def recount(components)
+      components.each { |component| component[:count] = component[:value].size }
+    end
+  end
+end

metadata

@@ -1,17 +1,17 @@
 --- !ruby/object:Gem::Specification
 name: myers_diff
 version: !ruby/object:Gem::Version
-  version: 1.0.0
+  version: 1.1.0
 platform: ruby
 authors:
 - Alex Tsui
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2020-05-22 00:00:00.000000000 Z
+date: 2020-05-27 00:00:00.000000000 Z
 dependencies: []
-description: Implementation of Myers 1986 text diff algorithmthat started as a port
-  of the jsdiff project withplans to branch out to human-friendly diffs
+description: Implementation of Myers 1986 text diff algorithm that started as a port
+  of the jsdiff project with plans to branch out to human-friendly diffs
 email: alextsui@pm.me
 executables: []
 extensions: []
@@ -20,6 +20,7 @@ files:
 - lib/myers_diff.rb
 - lib/myers_diff/char_diff.rb
 - lib/myers_diff/version.rb
+- lib/myers_diff/word_diff.rb
 homepage: https://github.com/alextsui05/myers_diff
 licenses:
 - MIT