Add YamlFrontMatterFilter to the PreProcessPipeliners-frontmatter

This filter will detect YAML Front Matter and convert it to an HTML
table for prettier formatting.

3 files changed, 208 insertions, 0 deletions

diff --git a/lib/banzai/filter/yaml_front_matter_filter.rb b/lib/banzai/filter/yaml_front_matter_filter.rb
new file mode 100644
index 00000000000..ed1371f17c4
--- /dev/null
+++ b/lib/banzai/filter/yaml_front_matter_filter.rb
@@ -0,0 +1,70 @@
+require 'html/pipeline/filter'
+require 'yaml'
+
+module Banzai
+  module Filter
+    class YamlFrontMatterFilter < HTML::Pipeline::Filter
+      DELIM = '---'.freeze
+
+      # Hat-tip to Middleman: https://git.io/v2e0z
+      PATTERN = %r{
+        \A(?:[^\r\n]*coding:[^\r\n]*\r?\n)?
+        (?<start>#{DELIM})[ ]*\r?\n
+        (?<frontmatter>.*?)[ ]*\r?\n?
+        ^(?<stop>#{DELIM})[ ]*\r?\n?
+        \r?\n?
+        (?<additional_content>.*)
+      }mx.freeze
+
+      def call
+        match = PATTERN.match(html)
+
+        return html unless match
+
+        frontmatter = load_yaml(match['frontmatter'])
+
+        return html unless frontmatter
+
+        tableize(frontmatter) << match['additional_content']
+      end
+
+      private
+
+      def load_yaml(yaml_string)
+        YAML.safe_load(yaml_string, [Date, DateTime, Symbol, Time])
+      end
+
+      def escape(html)
+        ERB::Util.html_escape(html)
+      end
+
+      # Converts a YAML object into an HTML table
+      #
+      # Each row is a key-value pair
+      def tableize(yaml)
+        table = "<table>\n"
+
+        if yaml.respond_to?(:each_pair)
+          yaml.each_pair do |key, value|
+            table << "<tr><th>#{escape(key)}</th><td>"
+
+            case value
+            when Array
+              value.inject(table) { |memo, v| memo << tableize(v) }
+            when Hash
+              table << tableize(value)
+            else
+              table << escape(value)
+            end
+
+            table << "</td></tr>\n"
+          end
+        else
+          table << "<tr><td>#{escape(yaml)}</td></tr>"
+        end
+
+        table << "</table>\n"
+      end
+    end
+  end
+end
diff --git a/lib/banzai/pipeline/pre_process_pipeline.rb b/lib/banzai/pipeline/pre_process_pipeline.rb
index c174f0b862d..b3d1e9b661e 100644
--- a/lib/banzai/pipeline/pre_process_pipeline.rb
+++ b/lib/banzai/pipeline/pre_process_pipeline.rb
@@ -3,6 +3,7 @@ module Banzai
     class PreProcessPipeline < BasePipeline
       def self.filters
         [
+          Filter::YamlFrontMatterFilter
         ]
       end
 
diff --git a/spec/lib/banzai/filter/yaml_front_matter_filter_spec.rb b/spec/lib/banzai/filter/yaml_front_matter_filter_spec.rb
new file mode 100644
index 00000000000..451b392c17a
--- /dev/null
+++ b/spec/lib/banzai/filter/yaml_front_matter_filter_spec.rb
@@ -0,0 +1,137 @@
+require 'rails_helper'
+
+describe Banzai::Filter::YamlFrontMatterFilter, lib: true do
+  include FilterSpecHelper
+
+  def parse(html)
+    Nokogiri::HTML.parse(html)
+  end
+
+  it 'allows for `encoding:` before the frontmatter' do
+    content = <<-MD.strip_heredoc
+      # encoding: UTF-8
+      ---
+      foo: foo
+      ---
+
+      # Header
+
+      Content
+    MD
+
+    output = filter(content)
+
+    expect(output).not_to match 'encoding'
+  end
+
+  it 'converts YAML frontmatter to a table' do
+    content = <<-MD.strip_heredoc
+      ---
+      foo: foo string
+      bar: :bar_symbol
+      ---
+
+      # Header
+
+      Content
+    MD
+
+    output = parse(filter(content))
+
+    aggregate_failures do
+      expect(output.xpath('.//tr[1]/th').text).to eq 'foo'
+      expect(output.xpath('.//tr[1]/td').text).to eq 'foo string'
+      expect(output.xpath('.//tr[2]/th').text).to eq 'bar'
+      expect(output.xpath('.//tr[2]/td').text).to eq 'bar_symbol'
+    end
+  end
+
+  it 'loads YAML safely' do
+    pending "Unsure what we can actually put here to verify this."
+
+    content = <<-MD.strip_heredoc
+      ---
+      ---
+    MD
+
+    expect { filter(content) }.to raise_error(Psych::DisallowedClass)
+  end
+
+  it 'HTML-escapes all YAML keys' do
+    content = <<-MD.strip_heredoc
+      ---
+      <b>foo</b>: 'foo'
+      ---
+    MD
+
+    output = filter(content)
+
+    expect(output).to match '&lt;b&gt;foo&lt;/b&gt;'
+  end
+
+  it 'HTML-escapes all YAML values' do
+    content = <<-MD.strip_heredoc
+      ---
+      foo: <th>foo</th>
+      ---
+    MD
+
+    output = filter(content)
+
+    expect(output).to match '&lt;th&gt;foo&lt;/th&gt;'
+  end
+
+  it 'supports Arrays' do
+    content = <<-MD.strip_heredoc
+      ---
+      foo:
+        - bar
+        - baz
+      ---
+    MD
+
+    output = parse(filter(content))
+
+    aggregate_failures do
+      expect(output.css('table').length).to eq 3
+      expect(output.xpath('.//table/tr[1]/td/table[1]/tr[1]/td[1]').text).to eq 'bar'
+    end
+  end
+
+  it 'supports Arrays of Hashes' do
+    content = <<-MD.strip_heredoc
+      ---
+      mvps:
+        - version: 8.4
+          name: Kyungchul Shin
+          date: January 22nd, 2016
+        - version: 8.3
+          name: Greg Smethells
+          date: December 22nd, 2015
+        - version: 8.2
+          name: Cristian Bica
+          date: November 22nd, 2015
+      ---
+    MD
+
+    output = parse(filter(content))
+
+    aggregate_failures do
+      expect(output.css('table').length).to eq 4
+      expect(output.xpath('.//table/tr[1]/td/table[2]/tr[2]/th[1]').text).to eq 'name'
+      expect(output.xpath('.//table/tr[1]/td/table[2]/tr[2]/td[1]').text).to eq 'Greg Smethells'
+    end
+  end
+
+  context 'on content without frontmatter' do
+    it 'returns the content unmodified' do
+      content = <<-MD.strip_heredoc
+        # This is some Markdown
+
+        It has no YAML frontmatter to parse.
+      MD
+
+      expect(filter(content)).to eq content
+    end
+  end
+end