Replace load-in of YAML data with Jekyll::Document logic.

COLLECTIONS IS COMING

lib/jekyll.rb

@@ -34,6 +34,7 @@ require 'jekyll/utils'
 require 'jekyll/stevenson'
 require 'jekyll/deprecator'
 require 'jekyll/configuration'
+require 'jekyll/document'
 require 'jekyll/plugin_manager'
 require 'jekyll/site'
 require 'jekyll/convertible'

lib/jekyll/document.rb

@@ -0,0 +1,83 @@
+module Jekyll
+  class Document
+
+    attr_reader   :path
+    attr_accessor :content
+
+    # Create a new Document.
+    #
+    # site - the Jekyll::Site instance to which this Document belongs
+    # path - the path to the file
+    #
+    # Returns nothing.
+    def initialize(site, path)
+      @site = site
+      @path = path
+    end
+
+    # Fetch the Document's data.
+    #
+    # Returns a Hash containing the data. An empty hash is returned if
+    #   no data was read.
+    def data
+      @data ||= Hash.new
+    end
+
+    def extname
+      File.extname(path)
+    end
+
+    def yaml_file?
+      %w[.yaml .yml].include?(extname)
+    end
+
+    # Returns merged option hash for File.read of self.site (if exists)
+    # and a given param
+    #
+    # opts - override options
+    #
+    # Return
+    def merged_file_read_opts(opts)
+      (site ? site.file_read_opts : {}).merge(opts)
+    end
+
+    # Whether the file is published or not, as indicated in YAML front-matter
+    #
+    # Returns true if the 'published' key is specified in the YAML front-matter and not `false`.
+    def published?
+      !(data.has_key?('published') && data['published'] == false)
+    end
+
+    # Read in the file and assign the content and data based on the file contents.
+    #
+    # Returns nothing.
+    def read
+      if yaml_file?
+        @data = SafeYAML.load_file(path)
+      else
+        begin
+          @content = File.read(path, merged_file_read_opts(opts))
+          if content =~ /\A(---\s*\n.*?\n?)^(---\s*$\n?)/m
+            @content = $POSTMATCH
+            @data    = SafeYAML.load($1)
+          end
+        rescue SyntaxError => e
+          puts "YAML Exception reading #{path}: #{e.message}"
+        rescue Exception => e
+          puts "Error reading file #{path}: #{e.message}"
+        end
+      end
+    end
+
+    # Create a Liquid-understandable version of this Document.
+    #
+    # Returns a Hash representing this Document's data.
+    def to_liquid
+      data.merge({
+        "content" => content,
+        "path"    => path
+      })
+    end
+
+  end
+end

lib/jekyll/site.rb

@@ -177,7 +177,7 @@ module Jekyll
         next if File.symlink?(path) && safe
 
         key = sanitize_filename(File.basename(entry, '.*'))
-        self.data[key] = SafeYAML.load_file(path)
+        (self.data[key] = Jekyll::Document.new(self, path)).read
       end
     end
 

test/test_site.rb

@@ -364,8 +364,8 @@ class TestSite < Test::Unit::TestCase
 
         file_content = SafeYAML.load_file(File.join(source_dir, '_data', 'members.yaml'))
 
-        assert_equal site.data['members'], file_content
+        assert_equal site.data['members'].data, file_content
-        assert_equal site.site_payload['site']['data']['members'], file_content
+        assert_equal site.site_payload['site']['data']['members'].data, file_content
       end
 
       should 'auto load yml files' do
@@ -374,8 +374,8 @@ class TestSite < Test::Unit::TestCase
 
         file_content = SafeYAML.load_file(File.join(source_dir, '_data', 'languages.yml'))
 
-        assert_equal site.data['languages'], file_content
+        assert_equal site.data['languages'].data, file_content
-        assert_equal site.site_payload['site']['data']['languages'], file_content
+        assert_equal site.site_payload['site']['data']['languages'].data, file_content
       end
 
       should "load symlink files in unsafe mode" do
@@ -384,8 +384,8 @@ class TestSite < Test::Unit::TestCase
 
         file_content = SafeYAML.load_file(File.join(source_dir, '_data', 'products.yml'))
 
-        assert_equal site.data['products'], file_content
+        assert_equal site.data['products'].data, file_content
-        assert_equal site.site_payload['site']['data']['products'], file_content
+        assert_equal site.site_payload['site']['data']['products'].data, file_content
       end
 
       should "not load symlink files in safe mode" do