fast_html_diff 0.8 → 0.8.1

checksums.yaml

@@ -1,15 +1,15 @@
 ---
 !binary "U0hBMQ==":
   metadata.gz: !binary |-
-    MGI5NjhiMjI5ODFlNjJhM2NhODJhNjAxMDRlMWM2Y2ExNGJmYzlkNg==
+    MmI5NmQwZjQyNDNjMDVmODhmZjk0YzgxZDA1MzViNzE5NWRiOTQ4Yw==
   data.tar.gz: !binary |-
-    OTgzNjEwMWQ5NTJiNGMxNzljODljYzJmYmY5ZGEzMjg1NDA5ZjA0OA==
+    YzAzZWM0NmYyYWY2NmFhOGIyYWE2MmJkNjg2ZTM3OWNmMTdkYmQyOA==
 !binary "U0hBNTEy":
   metadata.gz: !binary |-
-    ZTAxYmU5MzEwMGY4ZTYwYmYwYjVhNTIxM2UzM2RjMTg2MDZkZWZjYzYwNjQ1
-    NjMzOGY1YThjMzY1M2E4MjU4ZTZhODBhMmM3MmRlYzA0ZmQ2MDdkZWI5ZjE0
-    NzI0NGVmNjMwMjYxMGM5ZmE0MDdiMWU3ZTAwNmRjZWMyNGIxMWQ=
+    YWQwMjc5NjM4YTU4NzliYWMzNmJkMDk0OGZhNjYzZDhkNDE0Y2VhZWFjN2I2
+    OTM3Zjc4ZjcxODdjYTJhZmZhYmI3MmVmNmExNzBiMzM2ZTYyYzU1ZGMyZjQ2
+    YTQ5ZGUwMDM3MjE1OTI4ZGRhY2RmZDI3OTYxMWMyMjBhMjQ2N2I=
   data.tar.gz: !binary |-
-    ZWYyOWMwMDZkN2JiMjZjMWExNDA3OWNlMGI4ZDZiNDg3OTMxN2YwZjc0NDFk
-    OGE0OGZkYTU0ZTk4MjkwOGJlYmZlZmRiYjZiMmE2MzhjODU3ZjZlOWZkMmIy
-    ODU1NGQ1ZGY3MjU1ZDZhNDMzOTNmNGI5YzNkZjVjOTg0ZDIyMTc=
+    MTAwMjNkMDM1ZjE4NTQ3ZGJjMmNiMmExZDFmZTllZDM3YTY0NjMyMzIxZTZm
+    N2VhNTVjOThkOTNhY2Q1N2NlM2ZmODRlYzYwYWU2ZmUxY2NhMzQ0ZGRmMWVm
+    YTBjNjkxMzliNmQ4N2RjODFkOGM4MjJhN2IwM2E5OThiNTkwMzQ=

data/lib/fast_html_diff.rb

@@ -4,9 +4,7 @@ require 'nokogiri'
 module FastHtmlDiff
   class DiffBuilder
     def initialize(html_str_a,html_str_b,config={})
-      @a = html_str_a
-      @b = html_str_b
-
+      # merge specified config with defaults
       @config = default_config.merge(config)
       if config[:tokenizer_regexp].nil?
         if @config[:ignore_punctuation]
@@ -19,14 +17,13 @@ module FastHtmlDiff
       @word_list = {}
       @insertions = []
       @deletions = []
+      @matches = []
       @split_nodes = Hash.new
       @insertion_nodes = Hash.new
-    end
 
-    def build
-      # parse, tokenize and index
-      @a = Nokogiri::HTML(@a)
-      @b = Nokogiri::HTML(@b)
+      # parse, tokenize and index the input documents
+      @a = Nokogiri::HTML(html_str_a)
+      @b = Nokogiri::HTML(html_str_b)
       if @config[:simplify_html]
         simplify_html(@a)
         simplify_html(@b)
@@ -36,12 +33,40 @@ module FastHtmlDiff
 
       # find the insertions and deletions
       diff_words
+    end
 
-      # update doc a with tags for the insertions and deletions
+    # build output HTML
+    def build
+      # update doc_a with tags for the insertions and deletions
       update_dom
       @a.to_html
     end
 
+    # output statistics on insertions and deletions
+    def statistics
+      result = {
+          insertions: { segments: 0, words: 0, chars: 0 },
+          deletions: { segments: 0, words: 0, chars: 0 },
+          matches: { segments: 0, words: 0, chars: 0}
+      }
+      @insertions.each do |i|
+        result[:insertions][:segments] += 1
+        result[:insertions][:words] += i[:b_end] - i[:b_start] + 1
+        result[:insertions][:chars] += @word_list[:b][i[:b_end]][:end_pos] - @word_list[:b][i[:b_start]][:start_pos]
+      end
+      @deletions.each do |i|
+        result[:deletions][:segments] += 1
+        result[:deletions][:words] += i[:a_end] - i[:a_start] + 1
+        result[:deletions][:chars] += @word_list[:a][i[:a_end]][:end_pos] - @word_list[:a][i[:a_start]][:start_pos]
+      end
+      @matches.each do |i|
+        result[:matches][:segments] += 1
+        result[:matches][:words] += i[:a_end] - i[:a_start] + 1
+        result[:matches][:chars] += @word_list[:a][i[:a_end]][:end_pos] - @word_list[:a][i[:a_start]][:start_pos]
+      end
+      result
+    end
+
     private
 
     # index the words in the document
@@ -140,6 +165,9 @@ module FastHtmlDiff
             prev_operation = :deletion
           end
           doca_i += 1
+        else
+          if prev_operation == :match
+            @matches.last[:a_end] = doca_i
           else
             if prev_operation == :insertion
               @insertions.last[:next_operation] = :match
@@ -147,6 +175,12 @@ module FastHtmlDiff
               @deletions.last[:next_operation] = :match
             end
 
+            @matches << {
+                a_start: doca_i,
+                a_end: doca_i
+            }
+          end
+
           prev_operation = :match
           doca_i += 1
           docb_i += 1

data/lib/fast_html_diff/version.rb

@@ -1,3 +1,3 @@
 module FastHtmlDiff
-  VERSION = "0.8"
+  VERSION = "0.8.1"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: fast_html_diff
 version: !ruby/object:Gem::Version
-  version: '0.8'
+  version: 0.8.1
 platform: ruby
 authors:
 - Kent Mewhort
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2013-06-30 00:00:00.000000000 Z
+date: 2013-07-12 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler