Refactor `Site#cleanup`

Move the cleanup logic out into a separate class,
`Jekyll::Cleaner`.

lib/jekyll.rb

@@ -42,6 +42,7 @@ require 'jekyll/filters'
 require 'jekyll/static_file'
 require 'jekyll/errors'
 require 'jekyll/related_posts'
+require 'jekyll/cleaner'
 
 # extensions
 require 'jekyll/plugin'

lib/jekyll/cleaner.rb

@@ -0,0 +1,69 @@
+require 'set'
+
+module Jekyll
+  # Handles the cleanup of a site's destination before the site is built.
+  class Cleaner
+    def initialize(site)
+      @site = site
+    end
+
+    # Cleans up the site's destination directory
+    def cleanup!
+      obsolete_files = existing_files - new_files - new_dirs + replaced_files
+      FileUtils.rm_rf(obsolete_files.to_a)
+    end
+
+    private
+
+    # Private: The list of existing files, except those included in keep_files and hidden files.
+    #
+    # Returns a Set with the file paths
+    def existing_files
+      files = Set.new
+      Dir.glob(File.join(@site.dest, "**", "*"), File::FNM_DOTMATCH) do |file|
+        if @site.keep_files.length > 0
+          files << file unless file =~ /\/\.{1,2}$/ || file =~ keep_file_regex
+        else
+          files << file unless file =~ /\/\.{1,2}$/
+        end
+      end
+      files
+    end
+
+    # Private: The list of files to be created when the site is built.
+    #
+    # Returns a Set with the file paths
+    def new_files
+      files = Set.new
+      @site.each_site_file { |item| files << item.destination(@site.dest) }
+      files
+    end
+
+    # Private: The list of directories to be created when the site is built.
+    # These are the parent directories of the files in #new_files.
+    #
+    # Returns a Set with the directory paths
+    def new_dirs
+      new_files.map { |file| File.dirname(file) }.to_set
+    end
+
+    # Private: The list of existing files that will be replaced by a directory during build
+    #
+    # Returns a Set with the file paths
+    def replaced_files
+      new_dirs.select { |dir| File.file?(dir) }.to_set
+    end
+
+    # Private: creates a regular expression from the config's keep_files array
+    #
+    # Examples
+    #   ['.git','.svn'] creates the following regex: /\/(\.git|\/.svn)/
+    #
+    # Returns the regular expression
+    def keep_file_regex
+      or_list = @site.keep_files.join("|")
+      pattern = "\/(#{or_list.gsub(".", "\.")})"
+      Regexp.new pattern
+    end
+  end
+end

lib/jekyll/site.rb

@@ -1,5 +1,3 @@
-require 'set'
-
 module Jekyll
   class Site
     attr_accessor :config, :layouts, :posts, :pages, :static_files,
@@ -89,6 +87,8 @@ module Jekyll
 
       self.converters = instantiate_subclasses(Jekyll::Converter)
       self.generators = instantiate_subclasses(Jekyll::Generator)
+
+      @site_cleaner = Cleaner.new(self)
     end
 
     # Internal: Setup the plugin search path
@@ -228,43 +228,7 @@ module Jekyll
     #
     # Returns nothing.
     def cleanup
-      # all files and directories in destination, including hidden ones
+      @site_cleaner.cleanup!
-      dest_files = Set.new
-      Dir.glob(File.join(self.dest, "**", "*"), File::FNM_DOTMATCH) do |file|
-        if self.keep_files.length > 0
-          dest_files << file unless file =~ /\/\.{1,2}$/ || file =~ keep_file_regex
-        else
-          dest_files << file unless file =~ /\/\.{1,2}$/
-        end
-      end
-
-      # files to be written
-      files = Set.new
-      each_site_file { |item| files << item.destination(self.dest) }
-
-      # adding files' parent directories
-      dirs = Set.new
-      files.each { |file| dirs << File.dirname(file) }
-      files.merge(dirs)
-
-      # files that are replaced by dirs should be deleted
-      files_to_delete = Set.new
-      dirs.each { |dir| files_to_delete << dir if File.file?(dir) }
-
-      obsolete_files = dest_files - files + files_to_delete
-      FileUtils.rm_rf(obsolete_files.to_a)
-    end
-
-    # Private: creates a regular expression from the keep_files array
-    #
-    # Examples
-    #   ['.git','.svn'] creates the following regex: /\/(\.git|\/.svn)/
-    #
-    # Returns the regular expression
-    def keep_file_regex
-      or_list = self.keep_files.join("|")
-      pattern = "\/(#{or_list.gsub(".", "\.")})"
-      Regexp.new pattern
     end
 
     # Write static files, pages, and posts.