upmark 0.0.1 → 0.1.0

data/lib/upmark/parser/xml.rb

@@ -8,9 +8,9 @@ module Upmark
     #   http://www.w3.org/TR/2000/REC-xml-20001006
     #
     class XML < Parslet::Parser
-      root(:content)
+      root(:node)
 
-      rule(:content) {
+      rule(:node) {
         (
           element.as(:element) |
           text.as(:text)
@@ -20,7 +20,7 @@ module Upmark
       rule(:element) {
         (
           start_tag.as(:start_tag) >>
-          content.as(:content) >>
+          node.as(:children) >>
           end_tag.as(:end_tag)
         ) |
         empty_tag.as(:empty_tag)

data/lib/upmark/transform/ignore.rb

@@ -0,0 +1,19 @@
+module Upmark
+  module Transform
+    # A transform class which marks all elements in a subtree as ignored.
+    class Ignore < Parslet::Transform
+      include TransformHelpers
+
+      element(:*) do |element|
+        {
+          element: {
+            name:       element[:name],
+            attributes: element[:attributes],
+            children:   element[:children],
+            ignore:     true
+          }
+        }
+      end
+    end
+  end
+end

data/lib/upmark/transform/markdown.rb

@@ -5,55 +5,37 @@ module Upmark
     # Transforms an abstract syntax tree (AST) into a Markdown document.
     #
     class Markdown < Parslet::Transform
-      def self.tag(tag_name)
-        tag_name = tag_name.to_s.downcase
-        {
-          tag:     {name: tag_name, attributes: subtree(:attributes)},
-          content: sequence(:values)
-        }
-      end
-
-      def self.map_attributes_subtree(ast)
-        ast.inject({}) do |hash, attribute|
-          hash[attribute[:name].to_sym] = attribute[:value]
-          hash
-        end
-      end
-
-      rule(element: subtree(:values)) { values }
+      include TransformHelpers
 
       rule(text: simple(:value)) { value.to_s }
 
-      rule(tag(:p))      { "#{values.join}\n\n" }
-      rule(tag(:strong)) { "**#{values.join}**" }
-      rule(tag(:em))     { "*#{values.join}*" }
-      rule(tag(:li))     { "#{values.join}" }
-      rule(tag(:h1))     { "# #{values.join}" }
-      rule(tag(:h2))     { "## #{values.join}" }
-      rule(tag(:h3))     { "### #{values.join}" }
-      rule(tag(:br))     { "\n" }
+      element(:p)  {|element| "#{element[:children].join}\n\n" }
+      element(:h1) {|element| "# #{element[:children].join}" }
+      element(:h2) {|element| "## #{element[:children].join}" }
+      element(:h3) {|element| "### #{element[:children].join}" }
+      element(:li) {|element| "#{element[:children].join}" }
 
-      rule(tag(:ul)) do |dictionary|
-        values = dictionary[:values].map {|value| value.strip != "" ? value : nil }.compact
-        values.map {|value| "* #{value}\n" }
+      element(:ul) do |element|
+        children = element[:children].map {|value| value.strip != "" ? value : nil }.compact
+        children.map {|value| "* #{value}\n" }
       end
 
-      rule(tag(:ol)) do |dictionary|
-        values = dictionary[:values].map {|value| value.strip != "" ? value : nil }.compact
-        values.map_with_index {|value, i| "#{i + 1}. #{value}\n" }
+      element(:ol) do |element|
+        children = element[:children].map {|value| value.strip != "" ? value : nil }.compact
+        children.map_with_index {|value, i| "#{i + 1}. #{value}\n" }
       end
 
-      rule(tag(:a)) do |dictionary|
-        attributes = map_attributes_subtree(dictionary[:attributes])
+      element(:a) do |element|
+        attributes = map_attributes_subtree(element[:attributes])
         href       = attributes[:href]
         title      = attributes[:title]
-        values     = dictionary[:values].join
+        children   = element[:children].join
 
-        %Q{[#{values}](#{href} "#{title}")}
+        %Q{[#{children}](#{href} "#{title}")}
       end
 
-      rule(tag(:img)) do |dictionary|
-        attributes = map_attributes_subtree(dictionary[:attributes])
+      element(:img) do |element|
+        attributes = map_attributes_subtree(element[:attributes])
         href       = attributes[:src]
         title      = attributes[:title]
         alt_text   = attributes[:alt]
@@ -61,14 +43,23 @@ module Upmark
         %Q{![#{alt_text}](#{href} "#{title}")}
       end
 
-      # Catch-all rule to pass all tags through.
+      element(:b, :strong) {|element| "**#{element[:children].join}**" }
+      element(:i, :em)     {|element| "*#{element[:children].join}*" }
+
+      element(:br) { "\n" }
+
+      # Pass all unmatched elements through.
       rule(
-        tag:     {name: simple(:tag_name), attributes: subtree(:attributes)},
-        content: sequence(:values)
-      ) do |dictionary|
-        attributes = map_attributes_subtree(dictionary[:attributes])
-        values     = dictionary[:values].join
-        tag_name   = dictionary[:tag_name]
+        element: {
+          name:       simple(:name),
+          attributes: subtree(:attributes),
+          children:   sequence(:children),
+          ignore:     simple(:ignore)
+        }
+      ) do |element|
+        attributes = map_attributes_subtree(element[:attributes])
+        children    = element[:children].join
+        name       = element[:name]
 
         attributes_list =
           if attributes.any?
@@ -77,7 +68,11 @@ module Upmark
             ""
           end
 
-        %Q{<#{tag_name}#{attributes_list}>#{values}</#{tag_name}>}
+        if children.empty?
+          %Q{<#{name}#{attributes_list} />}
+        else
+          %Q{<#{name}#{attributes_list}>#{children}</#{name}>}
+        end
       end
     end
   end

data/lib/upmark/transform/pass_through.rb

@@ -0,0 +1,20 @@
+require "upmark/transform/ignore"
+
+module Upmark
+  module Transform
+    class PassThrough < Parslet::Transform
+      include TransformHelpers
+
+      element(:div, :table, :pre) do |element|
+        {
+          element: {
+            name:       element[:name],
+            attributes: element[:attributes],
+            children:   Ignore.new.apply(element[:children]),
+            ignore:     true
+          }
+        }
+      end
+    end
+  end
+end

data/lib/upmark/transform/preprocess.rb

@@ -1,23 +1,36 @@
 module Upmark
   module Transform
+    # A preprocess transform which normalises start/end/empty tags into the same structure.
     class Preprocess < Parslet::Transform
       rule(
-        start_tag: {name: simple(:tag_name), attributes: subtree(:attributes)},
-        end_tag:   {name: simple(:tag_name)},
-        content:   subtree(:values)
+        element: {
+          start_tag: {name: simple(:name), attributes: subtree(:attributes)},
+          end_tag:   {name: simple(:name)},
+          children:  subtree(:children)
+        }
       ) do
         {
-          tag:     {name: tag_name, attributes: attributes},
-          content: values
+          element: {
+            name:       name,
+            attributes: attributes,
+            children:   children,
+            ignore:     false
+          }
         }
       end
 
       rule(
-        empty_tag: {name: simple(:tag_name), attributes: subtree(:attributes)}
+        element: {
+          empty_tag: {name: simple(:name), attributes: subtree(:attributes)}
+        }
       ) do
         {
-          tag:     {name: tag_name, attributes: attributes},
-          content: []
+          element: {
+            name:       name,
+            attributes: attributes,
+            children:   [],
+            ignore:     false
+          }
         }
       end
     end

data/lib/upmark/transform_helpers.rb

@@ -0,0 +1,35 @@
+module Upmark
+  module TransformHelpers
+    def self.included(base)
+      base.send :extend, ClassMethods
+    end
+
+    module ClassMethods
+      def element(*names, &block)
+        names.each do |name|
+          name = name.to_s.downcase
+          rule(
+            {
+              element: {
+                name:       (name != "*" ? name : simple(:name)),
+                attributes: subtree(:attributes),
+                children:   subtree(:children),
+                ignore:     false
+              }
+            }
+          ) do |element|
+            element[:name] ||= name
+            block.call(element)
+          end
+        end
+      end
+
+      def map_attributes_subtree(ast)
+        ast.inject({}) do |hash, attribute|
+          hash[attribute[:name].to_sym] = attribute[:value]
+          hash
+        end
+      end
+    end
+  end
+end

data/lib/upmark/version.rb

@@ -1,3 +1,3 @@
 module Upmark
-  VERSION = "0.0.1"
+  VERSION = "0.1.0"
 end

data/lib/upmark.rb

@@ -3,18 +3,22 @@ require "parslet"
 require "core_ext/array"
 
 require "upmark/parser/xml"
+require 'upmark/transform_helpers'
 require "upmark/transform/markdown"
+require "upmark/transform/pass_through"
 require "upmark/transform/preprocess"
 require "upmark/version"
 
 module Upmark
   def self.convert(html)
-    xml        = Parser::XML.new
-    preprocess = Transform::Preprocess.new
-    markdown   = Transform::Markdown.new
+    xml          = Parser::XML.new
+    preprocess   = Transform::Preprocess.new
+    pass_through = Transform::PassThrough.new
+    markdown     = Transform::Markdown.new
 
     ast = xml.parse(html.strip)
     ast = preprocess.apply(ast)
+    ast = pass_through.apply(ast)
     ast = markdown.apply(ast)
 
     # The result is either a String or an Array.

data/spec/acceptance/upmark_spec.rb

@@ -3,17 +3,44 @@ require "spec_helper"
 describe Upmark, ".convert" do
   subject { Upmark.convert(html) }
 
-  let(:html) { <<-HTML.strip }
-<h1>messenger bag skateboard</h1>
+  context "<a>" do
+    let(:html) { <<-HTML.strip }
+<p><a href="http://helvetica.com/" title="art party organic">messenger <strong>bag</strong> skateboard</a></p>
+    HTML
 
-<p>messenger bag skateboard</p>
+    it { should == <<-MD.strip }
+[messenger **bag** skateboard](http://helvetica.com/ "art party organic")
+    MD
+  end
 
-<p>messenger <em>bag</em> <strong>skateboard</strong></p>
+  context "<img>" do
+    let(:html) { <<-HTML.strip }
+<img src="http://helvetica.com/image.gif" title="art party organic" alt="messenger bag skateboard" />
 
-<p><a href="http://helvetica.com/" title="art party organic">messenger <strong>bag</strong> skateboard</a></p>
+    HTML
 
-<img src="http://helvetica.com/image.gif" title="art party organic" alt="messenger bag skateboard" />
+    it { should == <<-MD.strip }
+![messenger bag skateboard](http://helvetica.com/image.gif "art party organic")
+    MD
+  end
 
+  context "<p>" do
+    let(:html) { <<-HTML.strip }
+<p>messenger <strong>bag</strong> skateboard</p>
+
+<p>art party<br />organic</p>
+    HTML
+
+    it { should == <<-MD.strip }
+messenger **bag** skateboard
+
+art party
+organic
+    MD
+  end
+
+  context "<ul>" do
+    let(:html) { <<-HTML.strip }
 <ul>
   <li>messenger</li>
   <li><strong>bag</strong></li>
@@ -25,7 +52,23 @@ describe Upmark, ".convert" do
   <li><p><strong>bag</strong></p></li>
   <li><p>skateboard</p></li>
 </ul>
+    HTML
+
+    it { should == <<-MD.strip }
+* messenger
+* **bag**
+* skateboard
+
+* messenger
 
+* **bag**
+
+* skateboard
+    MD
+  end
+
+  context "<ol>" do
+    let(:html) { <<-HTML.strip }
 <ol>
   <li>messenger</li>
   <li><strong>bag</strong></li>
@@ -37,32 +80,9 @@ describe Upmark, ".convert" do
   <li><p><strong>bag</strong></p></li>
   <li><p>skateboard</p></li>
 </ol>
+    HTML
 
-<div>messenger <strong>bag</strong> skateboard</div>
-<div id="tofu" class="art party">messenger <strong>bag</strong> skateboard</div>
-  HTML
-
-  it { should == <<-MD.strip }
-# messenger bag skateboard
-
-messenger bag skateboard
-
-messenger *bag* **skateboard**
-
-[messenger **bag** skateboard](http://helvetica.com/ "art party organic")
-
-![messenger bag skateboard](http://helvetica.com/image.gif "art party organic")
-
-* messenger
-* **bag**
-* skateboard
-
-* messenger
-
-* **bag**
-
-* skateboard
-
+    it { should == <<-MD.strip }
 1. messenger
 2. **bag**
 3. skateboard
@@ -72,8 +92,90 @@ messenger *bag* **skateboard**
 2. **bag**
 
 3. skateboard
+    MD
+  end
+
+  context "<h1>" do
+    let(:html) { <<-HTML.strip }
+<h1>messenger bag skateboard</h1>
+<h2>messenger bag skateboard</h2>
+<h3>messenger bag skateboard</h3>
+    HTML
+
+    it { should == <<-MD.strip }
+# messenger bag skateboard
+## messenger bag skateboard
+### messenger bag skateboard
+    MD
+  end
+
+  context "<div>" do
+    let(:html) { <<-HTML.strip }
+<div>messenger <strong>bag</strong> skateboard</div>
+<div id="tofu" class="art party">messenger <strong>bag</strong> skateboard</div>
+    HTML
 
-<div>messenger **bag** skateboard</div>
-<div id="tofu" class="art party">messenger **bag** skateboard</div>
-  MD
+    it { should == <<-MD.strip }
+<div>messenger <strong>bag</strong> skateboard</div>
+<div id="tofu" class="art party">messenger <strong>bag</strong> skateboard</div>
+    MD
+  end
+
+  context "<table>" do
+    let(:html) { <<-HTML.strip }
+<table>
+  <tr>
+    <td>messenger</td>
+  </tr>
+  <tr>
+    <td><strong>bag</strong></td>
+  </tr>
+  <tr>
+    <td>skateboard</td>
+  </tr>
+</table>
+    HTML
+
+    it { should == <<-MD.strip }
+<table>
+  <tr>
+    <td>messenger</td>
+  </tr>
+  <tr>
+    <td><strong>bag</strong></td>
+  </tr>
+  <tr>
+    <td>skateboard</td>
+  </tr>
+</table>
+    MD
+  end
+
+  context "<pre>" do
+    let(:html) { <<-HTML.strip }
+<pre>
+  <code>
+    messenger bag skateboard
+  </code>
+</pre>
+    HTML
+
+    it { should == <<-MD.strip }
+<pre>
+  <code>
+    messenger bag skateboard
+  </code>
+</pre>
+    MD
+  end
+
+  context "<span>" do
+    let(:html) { <<-HTML.strip }
+<span>messenger <strong>bag</strong> skateboard</span>
+    HTML
+
+    it { should == <<-MD.strip }
+<span>messenger **bag** skateboard</span>
+    MD
+  end
 end

data/spec/unit/lib/upmark/parser/xml_spec.rb

@@ -3,8 +3,8 @@ require "spec_helper"
 describe Upmark::Parser::XML do
   let(:parser) { Upmark::Parser::XML.new }
 
-  context "#content" do
-    subject { parser.content }
+  context "#node" do
+    subject { parser.node }
 
     it { should parse "" }
     it { should parse "messenger bag skateboard" }
@@ -58,6 +58,7 @@ describe Upmark::Parser::XML do
   context "#empty_tag" do
     subject { parser.empty_tag }
 
+    it { should     parse %q{<tofu />} }
     it { should     parse %q{<tofu art="party" />} }
     it { should     parse %q{<tofu art="party" synth="letterpress" />} }
     it { should_not parse "<tofu>" }
@@ -91,7 +92,7 @@ describe Upmark::Parser::XML do
   context "#parse" do
     subject { parser.parse(html) }
 
-    context "single element" do
+    context "single tag" do
       let(:html) { "<p>messenger</p>" }
 
       it do
@@ -100,14 +101,28 @@ describe Upmark::Parser::XML do
             element: {
               start_tag: {name: "p", attributes: []},
               end_tag:   {name: "p"},
-              content:   [{text: "messenger"}]
+              children:  [{text: "messenger"}]
             }
           }
         ]
       end
     end
 
-    context "single element with attributes" do
+    context "empty tag" do
+      let(:html) { "<br />" }
+
+      it do
+        should == [
+          {
+            element: {
+              empty_tag: {name: "br", attributes: []}
+            }
+          }
+        ]
+      end
+    end
+
+    context "single tag with attributes" do
       let(:html) { %q{<a href="http://helvetica.com/" title="art party organic">messenger bag skateboard</a>} }
 
       it do
@@ -121,15 +136,15 @@ describe Upmark::Parser::XML do
                   {name: "title", value: "art party organic"}
                 ]
               },
-              end_tag: {name: "a"},
-              content: [{text: "messenger bag skateboard"}]
+              end_tag:  {name: "a"},
+              children: [{text: "messenger bag skateboard"}]
             }
           }
         ]
       end
     end
 
-    context "multiple inline elements" do
+    context "multiple inline tags" do
       let(:html) { "<p>messenger</p><p>bag</p><p>skateboard</p>" }
 
       it do
@@ -138,26 +153,26 @@ describe Upmark::Parser::XML do
             element: {
               start_tag: {name: "p", attributes: []},
               end_tag:   {name: "p"},
-              content:   [{text: "messenger"}]
+              children:  [{text: "messenger"}]
             }
           }, {
             element: {
               start_tag: {name: "p", attributes: []},
               end_tag:   {name: "p"},
-              content:   [{text: "bag"}]
+              children:  [{text: "bag"}]
             }
           }, {
             element: {
               start_tag: {name: "p", attributes: []},
               end_tag:   {name: "p"},
-              content:   [{text: "skateboard"}]
+              children:  [{text: "skateboard"}]
             }
           }
         ]
       end
     end
 
-    context "multiple elements" do
+    context "multiple tags" do
       let(:html) { "<p>messenger</p>\n<p>bag</p>\n<p>skateboard</p>" }
 
       it do
@@ -166,7 +181,7 @@ describe Upmark::Parser::XML do
             element: {
               start_tag: {name: "p", attributes: []},
               end_tag:   {name: "p"},
-              content:   [{text: "messenger"}]
+              children:  [{text: "messenger"}]
             }
           }, {
             text: "\n"
@@ -174,7 +189,7 @@ describe Upmark::Parser::XML do
             element: {
               start_tag: {name: "p", attributes: []},
               end_tag:   {name: "p"},
-              content:   [{text: "bag"}]
+              children:  [{text: "bag"}]
             }
           }, {
             text: "\n"
@@ -182,14 +197,14 @@ describe Upmark::Parser::XML do
             element: {
               start_tag: {name: "p", attributes: []},
               end_tag:   {name: "p"},
-              content:   [{text: "skateboard"}]
+              children:  [{text: "skateboard"}]
             }
           }
         ]
       end
     end
 
-    context "nested elements" do
+    context "nested tags" do
       let(:html) { "<p>messenger <strong>bag</strong> skateboard</p>" }
 
       it do
@@ -198,12 +213,13 @@ describe Upmark::Parser::XML do
             element: {
               start_tag: {name: "p", attributes: []},
               end_tag:   {name: "p"},
-              content:   [
-                {text: "messenger "},
+              children:  [
                 {
+                  text: "messenger "
+                }, {
                   element: {
                     start_tag: {name: "strong", attributes: []},
-                    content:   [{text: "bag"}],
+                    children:  [{text: "bag"}],
                     end_tag:   {name: "strong"}
                   }
                 }, {

data/spec/unit/lib/upmark/transform/markdown_spec.rb

@@ -7,13 +7,15 @@ describe Upmark::Transform::Markdown do
     subject { transform.apply(ast) }
 
     context "<p>" do
-      context "single element" do
+      context "single tag" do
         let(:ast) do
           [
             {
               element: {
-                tag:     {name: "p", attributes: []},
-                content: [{text: "messenger bag skateboard"}]
+                name: "p",
+                attributes: [],
+                children: [{text: "messenger bag skateboard"}],
+                ignore: false
               }
             }
           ]
@@ -22,23 +24,29 @@ describe Upmark::Transform::Markdown do
         it { should == ["messenger bag skateboard\n\n"] }
       end
 
-      context "multiple elements" do
+      context "multiple tags" do
         let(:ast) do
           [
             {
               element: {
-                tag:     {name: "p", attributes: []},
-                content: [{text: "messenger"}]
+                name: "p",
+                attributes: [],
+                children: [{text: "messenger"}],
+                ignore: false
               }
             }, {
               element: {
-                tag:     {name: "p", attributes: []},
-                content: [{text: "bag"}]
+                name: "p",
+                attributes: [],
+                children: [{text: "bag"}],
+                ignore: false
               }
             }, {
               element: {
-                tag:     {name: "p", attributes: []},
-                content: [{text: "skateboard"}]
+                name: "p",
+                attributes: [],
+                children: [{text: "skateboard"}],
+                ignore: false
               }
             }
           ]
@@ -49,19 +57,18 @@ describe Upmark::Transform::Markdown do
     end
 
     context "<a>" do
-      context "single element" do
+      context "single tag" do
         let(:ast) do
           [
             {
               element: {
-                tag: {
-                  name: "a",
-                  attributes: [
-                    {name: "href",  value: "http://helvetica.com/"},
-                    {name: "title", value: "art party organic"}
-                  ]
-                },
-                content: [{text: "messenger bag skateboard"}]
+                name: "a",
+                attributes: [
+                  {name: "href",  value: "http://helvetica.com/"},
+                  {name: "title", value: "art party organic"}
+                ],
+                children: [{text: "messenger bag skateboard"}],
+                ignore: false
               }
             }
           ]
@@ -72,20 +79,19 @@ describe Upmark::Transform::Markdown do
     end
 
     context "<img>" do
-      context "empty element" do
+      context "empty tag" do
         let(:ast) do
           [
             {
               element: {
-                tag: {
-                  name: "img",
-                  attributes: [
-                    {name: "src",   value: "http://helvetica.com/image.gif"},
-                    {name: "title", value: "art party organic"},
-                    {name: "alt",   value: "messenger bag skateboard"}
-                  ]
-                },
-                content: []
+                name: "img",
+                attributes: [
+                  {name: "src",   value: "http://helvetica.com/image.gif"},
+                  {name: "title", value: "art party organic"},
+                  {name: "alt",   value: "messenger bag skateboard"}
+                ],
+                children: [],
+                ignore: false
               }
             }
           ]

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: upmark
 version: !ruby/object:Gem::Version
-  version: 0.0.1
+  version: 0.1.0
   prerelease: 
 platform: ruby
 authors:
@@ -10,11 +10,11 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2011-09-23 00:00:00.000000000 Z
+date: 2011-09-25 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rspec
-  requirement: &70160857809560 !ruby/object:Gem::Requirement
+  requirement: &70218294771840 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ! '>='
@@ -22,10 +22,10 @@ dependencies:
         version: '0'
   type: :development
   prerelease: false
-  version_requirements: *70160857809560
+  version_requirements: *70218294771840
 - !ruby/object:Gem::Dependency
   name: parslet
-  requirement: &70160857808520 !ruby/object:Gem::Requirement
+  requirement: &70218294771320 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ! '>='
@@ -33,7 +33,7 @@ dependencies:
         version: '0'
   type: :runtime
   prerelease: false
-  version_requirements: *70160857808520
+  version_requirements: *70218294771320
 description: Upmark has the skills to convert your HTML to Markdown.
 email: dev@theconversation.edu.au
 executables:
@@ -51,8 +51,11 @@ files:
 - lib/core_ext/array.rb
 - lib/upmark.rb
 - lib/upmark/parser/xml.rb
+- lib/upmark/transform/ignore.rb
 - lib/upmark/transform/markdown.rb
+- lib/upmark/transform/pass_through.rb
 - lib/upmark/transform/preprocess.rb
+- lib/upmark/transform_helpers.rb
 - lib/upmark/version.rb
 - spec/acceptance/upmark_spec.rb
 - spec/spec_helper.rb