Moved the migration logic out of Post and into MiddlemanPost

lib/wp2middleman.rb

@@ -1,5 +1,6 @@
 require 'wp2middleman/version'
 require 'wp2middleman/post'
+require 'wp2middleman/middleman_post'
 require 'wp2middleman/post_collection'
 require 'wp2middleman/frontmatter'
 require 'wp2middleman/migrator'

lib/wp2middleman/middleman_post.rb

@@ -0,0 +1,71 @@
+module WP2Middleman
+  class MiddlemanPost
+    def initialize(wp_post, body_to_markdown: false, include_fields: [])
+      @wp_post = wp_post
+      @body_to_markdown = body_to_markdown
+      @include_fields = include_fields
+    end
+
+    def title
+      wp_post.title
+    end
+
+    def title_for_filename
+      title.gsub(/[^\w\s_-]+/, '')
+        .gsub(/(^|\b\s)\s+($|\s?\b)/, '\\1\\2')
+        .gsub(/\s+/, '-')
+    end
+
+    def filename
+      "#{date_published}-#{title_for_filename}"
+    end
+
+    def date_published
+      wp_post.date_published
+    end
+
+    def full_filename output_path
+      "#{output_path}#{filename}.html.markdown"
+    end
+
+    def file_content
+      <<-EOS.gsub(/^ {8}/, '')
+        #{frontmatter.to_yaml}
+        ---
+
+        #{formatted_post_content}
+      EOS
+    end
+
+    def formatted_post_content
+      if body_to_markdown
+        markdown_content
+      else
+        content
+      end
+    end
+
+    def content
+      wp_post.content
+    end
+
+    def markdown_content
+      html = HTMLPage.new :contents => content
+      html.comment do |node,_|
+        "#{node}"
+      end
+      html.iframe do |node,_|
+        "#{node}"
+      end
+      html.markdown
+    end
+
+    private 
+
+    attr_reader :wp_post, :body_to_markdown, :include_fields
+
+    def frontmatter 
+      @frontmatter ||= Frontmatter.new(wp_post, include_fields: include_fields)
+    end
+  end
+end

lib/wp2middleman/migrator.rb

@@ -5,11 +5,10 @@ class Migrator
     attr_reader :posts
 
     def initialize(wp_xml_export_file, body_to_markdown: false, include_fields: [])
-      @posts = WP2Middleman::PostCollection.from_file(
-        wp_xml_export_file, 
-        body_to_markdown: body_to_markdown, 
-        include_fields: include_fields
-      ).without_attachments.only_valid
+      @posts = WP2Middleman::PostCollection.from_file(wp_xml_export_file)
+        .without_attachments
+        .only_valid
+        .to_middleman(body_to_markdown: body_to_markdown, include_fields: include_fields)
     end
 
     def migrate

lib/wp2middleman/post.rb

@@ -4,22 +4,14 @@ module WP2Middleman
   class Post
     attr_accessor :post
 
-    def initialize(nokogiri_post_doc, body_to_markdown: false, include_fields: [])
+    def initialize(nokogiri_post_doc)
       @post = nokogiri_post_doc
-      @body_to_markdown = body_to_markdown
-      @include_fields = include_fields
     end
 
     def title
       post.css('title').text
     end
 
-    def title_for_filename
-      title.gsub(/[^\w\s_-]+/, '')
-      .gsub(/(^|\b\s)\s+($|\s?\b)/, '\\1\\2')
-      .gsub(/\s+/, '-')
-    end
-
     def valid?
       !(post_date.nil? || title.nil? || date_published.nil? || content.nil?)
     end
@@ -28,35 +20,10 @@ def attachment?
       type == 'attachment'
     end
 
-    def filename
-      "#{date_published}-#{title_for_filename}"
-    end
-
-    def full_filename output_path
-      "#{output_path}#{filename}.html.markdown"
-    end
-
     def field(field)
       post.xpath(field).first.inner_text
     end
 
-    def file_content
-      <<-EOS.gsub(/^ {8}/, '')
-        #{frontmatter.to_yaml}
-        ---
-
-        #{formatted_post_content}
-      EOS
-    end
-
-    def formatted_post_content
-      if body_to_markdown
-        markdown_content
-      else
-        content
-      end
-    end
-
     def post_date
       post.xpath("wp:post_date").first.inner_text
     end
@@ -81,17 +48,6 @@ def content
       post.at_xpath(".//content:encoded").inner_text
     end
 
-    def markdown_content
-      html = HTMLPage.new :contents => content
-      html.comment do |node,_|
-        "#{node}"
-      end
-      html.iframe do |node,_|
-        "#{node}"
-      end
-      html.markdown
-    end
-
     def tags
       tags = []
       categories = post.xpath("category")
@@ -104,13 +60,5 @@ def tags
 
       tags
     end
-
-    private 
-
-    attr_reader :body_to_markdown, :include_fields
-
-    def frontmatter 
-      @frontmatter ||= Frontmatter.new(self, include_fields: include_fields)
-    end
   end
 end

lib/wp2middleman/post_collection.rb

@@ -4,9 +4,9 @@ module WP2Middleman
   class PostCollection
     include Enumerable
 
-    def self.from_file(wp_xml_export_file, body_to_markdown: false, include_fields: []) 
+    def self.from_file(wp_xml_export_file) 
       xml = Nokogiri::XML(File.open("#{Dir.pwd}/#{wp_xml_export_file}"))
-      new xml.css('item').collect { |wp_post| WP2Middleman::Post.new(wp_post, body_to_markdown: body_to_markdown, include_fields: include_fields) }
+      new xml.css('item').collect { |raw_wp_post| WP2Middleman::Post.new(raw_wp_post) }
     end
 
     def initialize(posts=[])
@@ -33,6 +33,11 @@ def only_valid
       self.class.new(select(&:valid?))
     end
 
+    def to_middleman(body_to_markdown: false, include_fields: [])
+      middleman_posts = collect { |p| WP2Middleman::MiddlemanPost.new(p, body_to_markdown: body_to_markdown, include_fields: include_fields) }
+      self.class.new(middleman_posts)
+    end
+
     private
 
     attr_reader :posts

spec/lib/wp2middleman/middleman_post_spec.rb

@@ -0,0 +1,116 @@
+require 'spec_helper'
+
+describe WP2Middleman::MiddlemanPost do
+  let(:post_one) { double :post, 
+    title: "A Title", 
+    date_published: Date.new(2012,6,8).to_s, 
+    content: "Paragraph one.\n\n      Paragraph two.\n    ", 
+    tags: [], 
+    published?: true
+  }
+
+  let(:post_two) { double :post, 
+    title: "A second title", 
+    date_published: Date.new(2011,7,25).to_s, 
+    content: " <strong>Foo</strong>", 
+    tags: ["some_tag", "another tag", "tag"], 
+    published?: true
+  }
+
+  let(:post_three) { double :post, 
+    title: "A third title: With colon", 
+    date_published: Date.new(2011,7,26).to_s, 
+    content: "Foo", 
+    tags: ["some_tag", "another tag", "tag"], 
+    published?: false
+  }
+
+  let(:post_with_iframe) { double :post, 
+    title: "A fourth item with iframe and comment", 
+    date_published: Date.new(2011,7,26).to_s, 
+    content: "Here's a post with an iframe and a comment.\n\n<!--more-->\n\n<iframe width=\"400\" height=\"100\" style=\"position: relative; display: block; width: 400px; height: 100px;\" src=\"http://bandcamp.com/EmbeddedPlayer/v=2/track=833121761/size=venti/bgcol=FFFFFF/linkcol=4285BB/\" allowtransparency=\"true\" frameborder=\"0\"><a href=\"http://dihannmoore.bandcamp.com/track/you-do-it-for-me\">&quot;YOU DO IT FOR ME&quot; by DIHANN MOORE</a></iframe>", 
+    tags: ["some_tag", "another tag", "tag"], 
+    published?: false
+  }
+
+  it "has a title formatted for a filename" do
+    wp_post = double :post, title: "A Title"
+    post = WP2Middleman::MiddlemanPost.new(wp_post)
+    expect(post.title_for_filename).to eq "A-Title"
+  end
+
+  it "removes odd characters (like colons) from title for filename" do
+    wp_post = double :post, title: "A third title: With colon"
+    post = WP2Middleman::MiddlemanPost.new(wp_post)
+    expect(post.title_for_filename).to eq "A-third-title-With-colon"
+  end
+
+  it "prepends the date published onto the title for the filename" do
+    wp_post = double :post, title: "A Title", date_published: Date.new(2012,6,8)
+    post = WP2Middleman::MiddlemanPost.new(wp_post)
+    expect(post.filename).to eq "2012-06-08-A-Title"
+  end
+
+  it "returns the full filename for a Middleman-style markdown post" do
+    wp_post = double :post, title: "A Title", date_published: Date.new(2012,6,8)
+    post = WP2Middleman::MiddlemanPost.new(wp_post)
+    expect(post.full_filename('/some/path/')).to eq("/some/path/2012-06-08-A-Title.html.markdown")
+  end
+
+  it "converts the content to markdown" do
+    wp_post = double :post, content: "<strong>Foo</strong>"
+    post = WP2Middleman::MiddlemanPost.new(wp_post)
+    expect(post.markdown_content).to eq "**Foo**"
+  end
+
+  it "has no formatting change to post body by default" do
+    post = WP2Middleman::MiddlemanPost.new post_one
+
+    expect(post.file_content).to eq(
+      "---\ntitle: A Title\ndate: '2012-06-08'\ntags: []\n---\n\nParagraph one.\n\n      Paragraph two.\n    \n"
+    )
+  end
+
+  it "formats the post body as markdown" do
+    post = WP2Middleman::MiddlemanPost.new post_two, body_to_markdown: true
+
+    expect( post.file_content ).to eq(
+      "---\ntitle: A second title\ndate: '2011-07-25'\ntags:\n- some_tag\n- another tag\n- tag\n---\n\n**Foo**\n"
+    )
+  end
+
+  it "ignores iframe and comment tags when converting to markdown" do
+    post = WP2Middleman::MiddlemanPost.new post_with_iframe, body_to_markdown: true
+
+    expect(post.file_content).to eq("---\ntitle: A fourth item with iframe and comment\ndate: '2011-07-26'\ntags:\n- some_tag\n- another tag\n- tag\npublished: false\n---\n\nHere's a post with an iframe and a comment.\n\n\n<!--more-->\n\n\n<iframe width=\"400\" height=\"100\" style=\"position: relative; display: block; width: 400px; height: 100px;\" src=\"http://bandcamp.com/EmbeddedPlayer/v=2/track=833121761/size=venti/bgcol=FFFFFF/linkcol=4285BB/\" allowtransparency=\"true\" frameborder=\"0\"><a href=\"http://dihannmoore.bandcamp.com/track/you-do-it-for-me\">\"YOU DO IT FOR ME\" by DIHANN MOORE</a></iframe>\n")
+  end
+
+  it "appends included fields in with frontmatter" do
+    post_two.should_receive(:field).with('wp:post_id').and_return('209')
+    post = WP2Middleman::MiddlemanPost.new post_two, include_fields: ['wp:post_id']
+
+    expect( post.file_content ).to eq(
+      "---\ntitle: A second title\ndate: '2011-07-25'\ntags:\n- some_tag\n- another tag\n- tag\nwp:post_id: '209'\n---\n\n <strong>Foo</strong>\n"
+    )
+  end
+
+  it "reports 'published: false' in the post's frontmatter when post is not published" do
+    post = WP2Middleman::MiddlemanPost.new post_three
+
+    expect( post.file_content ).to eq(
+      "---\ntitle: 'A third title: With colon'\ndate: '2011-07-26'\ntags:\n- some_tag\n- another tag\n- tag\npublished: false\n---\n\nFoo\n"
+    )
+  end
+
+  it "does no formatting by default" do
+    post = WP2Middleman::MiddlemanPost.new post_two
+
+    expect( post.formatted_post_content ).to eq(" <strong>Foo</strong>")
+  end
+
+  it "does markdown formatting" do
+    post = WP2Middleman::MiddlemanPost.new post_two, body_to_markdown: true
+
+    expect( post.formatted_post_content ).to eq("**Foo**")
+  end
+end