Add functionality for importing Blogger comments

lib/jekyll-import/importers/blogger.rb

@@ -5,6 +5,7 @@ def self.specify_options(c)
         c.option 'source', '--source NAME', 'The XML file (blog-MM-DD-YYYY.xml) path to import'
         c.option 'no-blogger-info', '--no-blogger-info', 'not to leave blogger-URL info (id and old URL) in the front matter (default: false)'
         c.option 'replace-internal-link', '--replace-internal-link', 'replace internal links using the post_url liquid tag. (default: false)'
+        c.option 'comments', '--comments', 'import comments to _comments collection'
       end
 
       def self.validate(options)
@@ -41,6 +42,7 @@ def self.process(options)
         listener = BloggerAtomStreamListener.new
 
         listener.leave_blogger_info = ! options.fetch('no-blogger-info', false),
+        listener.comments = options.fetch('comments', false),
 
         File.open(source, 'r') do |f|
           f.flock(File::LOCK_SH)
@@ -95,11 +97,12 @@ def initialize
           extend BloggerAtomStreamListenerMethods
 
           @leave_blogger_info = true
+          @comments = false
         end
       end
 
       module BloggerAtomStreamListenerMethods
-        attr_accessor :leave_blogger_info
+        attr_accessor :leave_blogger_info, :comments
         attr_reader :original_url_base
 
         def tag_start(tag, attrs)
@@ -143,6 +146,10 @@ def tag_start(tag, attrs)
             if @in_entry_elem
               @in_entry_elem[:meta][:thumbnail] = attrs['url']
             end
+          when 'thr:in-reply-to'
+            if @in_entry_elem
+              @in_entry_elem[:meta][:post_id] = attrs['ref']
+            end
           end
         end
 
@@ -185,6 +192,23 @@ def tag_end(tag)
 
                 FileUtils.mkdir_p(target_dir)
 
+                file_name = URI::decode("#{post_data[:filename]}.html")
+                File.open(File.join(target_dir, file_name), 'w') do |f|
+                  f.flock(File::LOCK_EX)
+
+                  f << post_data[:header].to_yaml
+                  f << "---\n\n"
+                  f << post_data[:body]
+                end
+              end
+            elsif @in_entry_elem[:meta][:kind] == 'comment' and @comments
+              post_data = get_post_data_from_in_entry_elem_info
+
+              if post_data
+                target_dir = '_comments'
+
+                FileUtils.mkdir_p(target_dir)
+
                 file_name = URI::decode("#{post_data[:filename]}.html")
                 File.open(File.join(target_dir, file_name), 'w') do |f|
                   f.flock(File::LOCK_EX)
@@ -251,6 +275,48 @@ def get_post_data_from_in_entry_elem_info
               body.gsub!(/{%/, '{{ "{%" }}')
             end
 
+            { :filename => filename, :header => header, :body => body }
+          elsif @in_entry_elem[:meta][:kind] == 'comment'
+            timestamp = Time.parse(@in_entry_elem[:meta][:published]).strftime('%Y-%m-%d')
+            if @in_entry_elem[:meta][:original_url]
+              if not @comment_seq
+                @comment_seq = 1
+              end
+
+              original_uri = URI.parse(@in_entry_elem[:meta][:original_url])
+              original_path = original_uri.path.to_s
+              filename = "%s-%s-%s" %
+                [timestamp,
+                 File.basename(original_path, File.extname(original_path)),
+                 @comment_seq]
+
+              @comment_seq = @comment_seq + 1
+
+              @original_url_base = "#{original_uri.scheme}://#{original_uri.host}"
+            else
+              raise 'Original URL is missing'
+            end
+
+            header = {
+              'date' => @in_entry_elem[:meta][:published],
+              'author' => @in_entry_elem[:meta][:author],
+              'blogger_post_id' => @in_entry_elem[:meta][:post_id],
+            }
+            header['modified_time'] = @in_entry_elem[:meta][:updated] if @in_entry_elem[:meta][:updated] && @in_entry_elem[:meta][:updated] != @in_entry_elem[:meta][:published]
+            header['thumbnail'] = @in_entry_elem[:meta][:thumbnail] if @in_entry_elem[:meta][:thumbnail]
+            header['blogger_id'] = @in_entry_elem[:meta][:id] if @leave_blogger_info
+            header['blogger_orig_url'] = @in_entry_elem[:meta][:original_url] if @leave_blogger_info && @in_entry_elem[:meta][:original_url]
+
+            body = @in_entry_elem[:body]
+
+            # body escaping associated with liquid
+            if body =~ /{{/
+              body.gsub!(/{{/, '{{ "{{" }}')
+            end
+            if body =~ /{%/
+              body.gsub!(/{%/, '{{ "{%" }}')
+            end
+
             { :filename => filename, :header => header, :body => body }
           else
             nil