auto replace rules in titles

lib/text_cleaner.rb

@@ -0,0 +1,45 @@
+#
+# Clean up a text
+#
+class TextCleaner
+
+  def self.title_options
+    # cf. http://meta.discourse.org/t/should-we-have-auto-replace-rules-in-titles/5687
+    {
+      deduplicate_exclamation_marks: SiteSetting.title_prettify,
+      deduplicate_question_marks: SiteSetting.title_prettify,
+      replace_all_upper_case: SiteSetting.title_prettify,
+      capitalize_first_letter: SiteSetting.title_prettify,
+      remove_unnecessary_period: SiteSetting.title_prettify,
+      remove_extraneous_space: SiteSetting.title_prettify && SiteSetting.default_locale == "en",
+      fixes_interior_spaces: true,
+      strip_whitespaces: true
+    }
+  end
+
+  def self.clean_title(title)
+    TextCleaner.clean(title, TextCleaner.title_options)
+  end
+
+  def self.clean(text, opts = {})
+    # Replace !!!!! with a single !
+    text.gsub!(/!+/, '!') if opts[:deduplicate_exclamation_marks]
+    # Replace ????? with a single ?
+    text.gsub!(/\?+/, '?') if opts[:deduplicate_question_marks]
+    # Replace all-caps text with regular case letters
+    text.tr!('A-Z', 'a-z') if opts[:replace_all_upper_case] && (text =~ /[A-Z]+/) && (text == text.upcase)
+    # Capitalize first letter
+    text.sub!(/\A([a-z])/) { |first| first.capitalize } if opts[:capitalize_first_letter]
+    # Remove unnecessary period at the end
+    text.sub!(/([^.])\.(\s*)\z/, '\1\2') if opts[:remove_unnecessary_period]
+    # Remove extraneous space before the end punctuation
+    text.sub!(/\s+([!?]\s*)\z/, '\1') if opts[:remove_extraneous_space]
+    # Fixes interior spaces
+    text.gsub!(/ +/, ' ') if opts[:fixes_interior_spaces]
+    # Strip whitespaces
+    text.strip! if opts[:strip_whitespaces]
+
+    text
+  end
+
+end

lib/text_sentinel.rb

@@ -1,31 +1,27 @@
 #
-# Given a string, tell us whether or not is acceptable. Also, remove stuff we don't like
-# such as leading / trailing space.
+# Given a string, tell us whether or not is acceptable.
 #
 class TextSentinel
 
   attr_accessor :text
 
+  def initialize(text, opts=nil)
+    @opts = opts || {}
+    @text = text.encode('UTF-8', invalid: :replace, undef: :replace, replace: '') if text.present?
+  end
+
   def self.non_symbols_regexp
     /[\ -\/\[-\`\:-\@\{-\~]/m
   end
 
-  def initialize(text, opts=nil)
-    @opts = opts || {}
-
-    if text.present?
-      @text = text.encode('UTF-8', invalid: :replace, undef: :replace, replace: '')
-      @text.gsub!(/ +/m, ' ') if @opts[:remove_interior_spaces]
-      @text.strip! if @opts[:strip]
-    end
+  def self.body_sentinel(text)
+    TextSentinel.new(text, min_entropy: SiteSetting.body_min_entropy)
   end
 
   def self.title_sentinel(text)
     TextSentinel.new(text,
                      min_entropy: SiteSetting.title_min_entropy,
-                     max_word_length: SiteSetting.max_word_length,
-                     remove_interior_spaces: true,
-                     strip: true)
+                     max_word_length: SiteSetting.max_word_length)
   end
 
   # Entropy is a number of how many unique characters the string needs.
@@ -35,7 +31,6 @@ class TextSentinel
   end
 
   def valid?
-
     # Blank strings are not valid
     return false if @text.blank? || @text.strip.blank?
 
@@ -47,12 +42,12 @@ class TextSentinel
     return false if non_symbols == 0
 
     # Don't allow super long strings without spaces
-
     return false if @opts[:max_word_length] && @text =~ /\w{#{@opts[:max_word_length]},}(\s|$)/
 
     # We don't allow all upper case content in english
     return false if (@text =~ /[A-Z]+/) && (@text == @text.upcase)
 
+    # It is valid
     true
   end