Do not double escape text when processing markdown.

lib/kitabu.rb

@@ -54,6 +54,8 @@ module Kitabu
   require "kitabu/front_matter"
   require "kitabu/context"
 
+  I18n.load_path += Dir[File.join(__dir__, "kitabu/locales/*.yml")]
+
   def self.config(root_dir = nil)
     root_dir ||= Pathname.new(Dir.pwd)
     path = root_dir.join("config/kitabu.yml")

lib/kitabu/exporter/base.rb

@@ -7,11 +7,9 @@ def self.default_i18n_load_path
     end
 
     def self.load_translations(root_dir:)
-      paths = default_i18n_load_path.dup
-      paths += Dir[root_dir.join("config/locales/**/*.{yml,rb}").to_s]
-      paths << File.expand_path(File.join(__dir__, "../../../templates/en.yml"))
-
-      I18n.load_path = paths
+      I18n.load_path += Dir[root_dir.join("config/locales/**/*.{yml,rb}").to_s]
+      I18n.load_path << File.expand_path(File.join(__dir__,
+                                                   "../../../templates/en.yml"))
 
       I18n.backend.reload!
       I18n.backend.eager_load!

lib/kitabu/extensions/string.rb

@@ -1,13 +1,17 @@
 # frozen_string_literal: true
 
-class String
-  def to_permalink
-    str = dup.unicode_normalize(:nfkd)
-    str = str.gsub(/[^\x00-\x7F]/, "").to_s
-    str.gsub!(/[^-\w]+/xim, "-")
-    str.gsub!(/-+/xm, "-")
-    str.gsub!(/^-?(.*?)-?$/, '\1')
-    str.downcase!
-    str
+module Kitabu
+  module Extensions
+    refine String do
+      def to_permalink
+        str = dup.unicode_normalize(:nfkd)
+        str = str.gsub(/[^\x00-\x7F]/, "").to_s
+        str.gsub!(/[^-\w]+/xim, "-")
+        str.gsub!(/-+/xm, "-")
+        str.gsub!(/^-?(.*?)-?$/, '\1')
+        str.downcase!
+        str
+      end
+    end
   end
 end

lib/kitabu/locales/en.yml

@@ -0,0 +1,12 @@
+---
+en:
+  contents: Contents
+  left_blank: This page intentionally left blank
+  chapter: Chapter
+
+  alerts:
+    warning: Warning!
+    note: Just so you know…
+    tip: Here’s a tip!
+    caution: Caution!
+    important: Important!

lib/kitabu/markdown.rb

@@ -2,13 +2,46 @@
 
 module Kitabu
   module Markdown
+    # N.B.: Redcarpet pass down escaped HTML. There's no need to escape the
+    # content and we can just interpolate it.
     class Renderer < Redcarpet::Render::HTML
       include Redcarpet::Render::SmartyPants
       include Rouge::Plugins::Redcarpet
+      using Extensions
 
       # Be more flexible than github and support any arbitrary name.
       ALERT_MARK = /^\[!(?<type>[A-Z]+)\](?<title>.*?)?$/
 
+      HEADING_ID = /^(?<text>.*?)(?: {#(?<id>.*?)})?$/
+
+      def escape(text)
+        Nokogiri::HTML.fragment(text).text
+      end
+
+      def heading_counter
+        @heading_counter ||= Hash.new {|h, k| h[k] = 0 }
+      end
+
+      def header(text, level)
+        matches = text.strip.match(HEADING_ID)
+        title = matches[:text].strip
+        html = Nokogiri::HTML.fragment("<h#{level}>#{title}</h#{level}>")
+        heading = html.first_element_child
+        title = heading.text
+
+        id = matches[:id]
+        id ||= title.to_permalink
+
+        heading_counter[id] += 1
+        id = "#{id}-#{heading_counter[id]}" if heading_counter[id] > 1
+
+        heading.add_child %[<a class="anchor" href="##{id}" aria-hidden="true" tabindex="-1"></a>] # rubocop:disable Style/LineLength
+        heading.set_attribute :tabindex, "-1"
+        heading.set_attribute(:id, id)
+
+        heading.to_s
+      end
+
       # Support alert boxes just like github.
       # https://github.com/orgs/community/discussions/16925
       def block_quote(quote)
@@ -26,7 +59,7 @@ def block_quote(quote)
         title = matches[:title].strip
 
         if title == ""
-          title = I18n.t(type, scope: :notes, default: type.titleize)
+          title = I18n.t(type, scope: :alerts, default: type.titleize)
         end
 
         html = Nokogiri::HTML.fragment <<~HTML.strip_heredoc
@@ -44,10 +77,9 @@ def block_quote(quote)
       def table_cell(content, alignment, header)
         tag = header ? "th" : "td"
 
-        html = Nokogiri::HTML.fragment("<#{tag}></#{tag}>")
+        html = Nokogiri::HTML.fragment("<#{tag}>#{content}</#{tag}>")
         node = html.children.first
         node.append_class("align-#{alignment}") if alignment
-        node.content = content
 
         html.to_s
       end

lib/kitabu/toc/html.rb

@@ -8,7 +8,6 @@ class HTML
       Result = Struct.new(:toc, :html, :hierarchy, keyword_init: true)
 
       def self.generate(html)
-        html = normalize(html)
         file = Tempfile.new("kitabu")
         file.write(html.to_s)
         file.close
@@ -26,29 +25,6 @@ def self.generate(html)
       ensure
         file.unlink
       end
-
-      def self.normalize(html)
-        counter = {}
-
-        html.search("h1, h2, h3, h4, h5, h6").each do |tag|
-          title = tag.inner_text.strip
-          permalink = title.to_permalink
-
-          counter[permalink] ||= 0
-          counter[permalink] += 1
-
-          if counter[permalink] > 1
-            permalink = "#{permalink}-#{counter[permalink]}"
-          end
-
-          tag.set_attribute("id", permalink)
-          tag["tabindex"] = "-1"
-
-          tag.prepend_child %[<a class="anchor" href="##{permalink}" aria-hidden="true" tabindex="-1"></a>] # rubocop:disable Style/LineLength
-        end
-
-        html
-      end
     end
   end
 end

spec/kitabu/extensions/string_spec.rb

@@ -5,9 +5,12 @@
 require "spec_helper"
 
 describe String do
+  using Kitabu::Extensions
+
   describe "#to_permalink" do
     it "normalizes strings" do
       {
+        "Simple > Complex" => "simple-complex",
         "This IS a Tripped out title!!.!1  (well/ not really)" => "this-is-a-tripped-out-title-1-well-not-really",
         "////// meph1sto r0x ! \\\\\\" => "meph1sto-r0x",
         "āčēģīķļņū" => "acegiklnu",

spec/kitabu/markdown_spec.rb

@@ -157,4 +157,30 @@ class User
     expect(html).not_to have_tag("figure")
     expect(html).not_to have_tag("figcaption")
   end
+
+  it "sets automatic ids when defining headers" do
+    html = Kitabu::Markdown.render <<~TEXT
+      # My header
+
+      ## My header
+
+      ### My header {#my-custom-id}
+    TEXT
+
+    html = Nokogiri::HTML.fragment(html)
+
+    expect(html).to have_tag("h1#my-header", "My header")
+    expect(html).to have_tag("h2#my-header-2", "My header")
+    expect(html).to have_tag("h3#my-custom-id", "My header")
+  end
+
+  it "sets custom ids when defining headers" do
+    html = Kitabu::Markdown.render <<~TEXT
+      # My header {#header}
+    TEXT
+
+    html = Nokogiri::HTML.fragment(html)
+
+    expect(html).to have_tag("h1#header", "My header")
+  end
 end

spec/kitabu/toc/html_spec.rb

@@ -8,28 +8,26 @@ def regexp(text)
   end
 
   def input
-    (+<<-HTML).force_encoding("utf-8")
-      <h2>Item 1</h2>
-      <h3>Item 1.1</h3>
-      <h4>Item 1.1.1</h4>
-      <h5>Item 1.1.1.1</h5>
-      <h6>Item 1.1.1.1.1</h6>
+    html = Kitabu::Markdown.render <<~MARKDOWN
+      ## Item 1
+      ### Item 1.1
+      #### Item 1.1.1
+      ##### Item 1.1.1.1
+      ###### Item 1.1.1.1.1
+      ## Item 2
+      ### Item 2.1
+      #### Item 2.1.1
+      #### Item 2.1.2
+      ## Item 3
+      ## Internacionalização
+      ## Title
+      ## Title
+    MARKDOWN
 
-      <h2>Item 2</h2>
-      <h3>Item 2.1</h3>
-      <h4>Item 2.1.1</h4>
-      <h4>Item 2.1.2</h4>
-
-      <h2>Item 3</h2>
-      <h2>Internacionalização</h2>
-
-      <!-- Duplicated titles -->
-      <h2>Title</h2>
-      <h2>Title</h2>
-    HTML
+    html.to_s
   end
 
-  let(:result) { described_class.generate(Nokogiri::HTML(input)) }
+  let(:result) { described_class.generate(input) }
   let(:toc) { Nokogiri::HTML(result.toc) }
   let(:html) { result.html }
 

spec/spec_helper.rb

@@ -61,7 +61,9 @@ def bundle_install
   config.before { FileUtils.mkdir_p(TMPDIR) }
 
   config.before do
+    I18n.load_path = Dir["#{__dir__}/../lib/kitabu/locales/*.yml"]
     Kitabu::Markdown.hooks.clear
     Kitabu::Markdown.setup_default_hooks
+    Kitabu::Markdown.processor.renderer.heading_counter.clear
   end
 end

spec/support/mybook/config/locales/en.yml

@@ -4,7 +4,7 @@ en:
   left_blank: This page intentionally left blank
   chapter: Chapter
 
-  notes:
+  alerts:
     warning: Warning!
     note: Just so you know…
     tip: Here’s a tip!

templates/en.yml

@@ -4,7 +4,7 @@ en:
   left_blank: This page intentionally left blank
   chapter: Chapter
 
-  notes:
+  alerts:
     warning: Warning!
     note: Just so you know…
     tip: Here’s a tip!