Improve file output in general (#67)

README.md

@@ -21,7 +21,7 @@ starting at \$15/mo.
 - Write using Markdown
 - Book layout support
 - Syntax highlight
-- Generate HTML, PDF, e-Pub, and Mobi
+- Generate HTML, PDF, e-Pub (version 3.3), and Mobi
 - Table of Contents automatically generated from chapter titles
 
 ## Installation
@@ -262,6 +262,15 @@ Kitabu::Markdown.processor = Redcarpet::Markdown.new(
 
 The above options are Kitabu's defaults.
 
+### Exporting PDFs using DocRaptor
+
+If you're not planning to buy PrinceXML, consider using
+[DocRaptor](http://docraptor.com). Here's how you can easily do it:
+
+```bash
+curl -H "Content-Type:application/json" -d'{"user_credentials":"YOUR_CREDENTIALS_HERE", "doc":{"name":"kitabu.pdf", "document_type":"pdf", "test":"false", "document_url":"https://example.com/output/kitabu.pdf.html"}}' http://docraptor.com/docs > kitabu.pdf
+```
+
 ## References
 
 - Markdown: <http://daringfireball.net/projects/markdown/syntax>
@@ -271,15 +280,6 @@ The above options are Kitabu's defaults.
 
 - PrinceXML: [license](http://www.princexml.com/license/)
 
-Alternatives:
-
-- If you're not planning to buy PrinceXML, consider using
-  [DocRaptor](http://docraptor.com). Here's how you can easily do it:
-
-```bash
-curl -H "Content-Type:application/json" -d'{"user_credentials":"YOUR_CREDENTIALS_HERE", "doc":{"name":"kitabu.pdf", "document_type":"pdf", "test":"false", "document_url":"https://dl.dropboxusercontent.com/u/123456789/output/kitabu.pdf.html"}}' http://docraptor.com/docs > kitabu.pdf
-```
-
 ## Maintainer
 
 - [Nando Vieira](https://github.com/fnando)

kitabu.gemspec

@@ -26,14 +26,13 @@ Gem::Specification.new do |s|
   s.require_paths = ["lib"]
 
   s.add_dependency "activesupport"
-  s.add_dependency "eeepub-with-cover-support"
+  s.add_dependency "epub-rb"
   s.add_dependency "i18n"
   s.add_dependency "nokogiri"
   s.add_dependency "redcarpet"
   s.add_dependency "rouge"
   s.add_dependency "rubyzip"
   s.add_dependency "thor"
-  s.add_dependency "zip-zip"
 
   s.add_development_dependency "pry-meta"
   s.add_development_dependency "rake"

lib/kitabu.rb

@@ -2,9 +2,8 @@
 
 require "active_support/all"
 require "digest/md5"
-require "eeepub"
+require "epub"
 require "erb"
-require "logger"
 require "nokogiri"
 require "open3"
 require "optparse"
@@ -30,7 +29,6 @@ module Kitabu
 
   require "kitabu/extensions/string"
   require "kitabu/extensions/rouge"
-  require "kitabu/extensions/eeepub"
   require "kitabu/errors"
   require "kitabu/version"
   require "kitabu/generator"
@@ -49,7 +47,6 @@ module Kitabu
   require "kitabu/footnotes/pdf"
   require "kitabu/hook"
   require "kitabu/toc/html"
-  require "kitabu/toc/html/stream"
   require "kitabu/toc/epub"
   require "kitabu/dependency"
   require "kitabu/stats"
@@ -90,8 +87,4 @@ def self.config(root_dir = nil)
     erb = ERB.new(content).result
     YAML.safe_load(erb).with_indifferent_access
   end
-
-  def self.logger
-    @logger ||= Logger.new(File.open("/tmp/kitabu.log", "a"))
-  end
 end

lib/kitabu/cli.rb

@@ -93,18 +93,34 @@ def permalinks
       inside_ebook!
 
       html = Kitabu::Exporter::HTML.new(root_dir).content
-      toc = Kitabu::TOC::HTML.generate(html)
+      toc = Kitabu::TOC::HTML.generate(Nokogiri::HTML(html))
 
-      toc.toc.each do |options|
-        level = options[:level] - 1
-        title = " #{options[:text]}: "
-        permalink = "##{options[:permalink]}"
+      renderer = lambda do |hierarchy, level = 1|
+        hierarchy.each do |entry|
+          title = " #{entry[:label]}: "
+          permalink = "##{entry[:content]}"
 
-        text = "*" * level
-        text << color(title, :blue)
-        text << color(permalink, :yellow)
-        say(text)
+          text = "*" * level
+          text << color(title, :blue)
+          text << color(permalink, :yellow)
+          say(text)
+
+          renderer.call(entry[:hierarchy], level + 1) if entry[:hierarchy].any?
+        end
       end
+
+      renderer.call(toc.hierarchy)
+
+      # toc.hierarchy.each do |options|
+      #   level = options[:level] - 1
+      #   title = " #{options[:text]}: "
+      #   permalink = "##{options[:permalink]}"
+
+      #   text = "*" * level
+      #   text << color(title, :blue)
+      #   text << color(permalink, :yellow)
+      #   say(text)
+      # end
     end
 
     desc "stats", "Display some stats about your e-book"

lib/kitabu/exporter/epub.rb

@@ -3,9 +3,12 @@
 module Kitabu
   class Exporter
     class Epub < Base
+      SECTION_SELECTOR =
+        "div.chapter, div.section, section.chapter, section.section"
+
       def sections
         @sections ||=
-          html.css("div.chapter").each_with_index.map do |chapter, index|
+          html.css(SECTION_SELECTOR).each_with_index.map do |chapter, index|
             html = Nokogiri::HTML(chapter.inner_html)
 
             OpenStruct.new(
@@ -18,7 +21,18 @@ def sections
       end
 
       def epub
-        @epub ||= EeePub::Maker.new
+        @epub ||= ::Epub.new(
+          root_dir: tmp_dir,
+          title: config[:title],
+          subtitle: config[:subtitle],
+          language: config[:language],
+          copyright: config[:copyright],
+          publisher: config[:publisher],
+          contributors: Array(config[:contributors]),
+          creators: Array(config[:creators]),
+          identifiers: Array(config[:identifiers]),
+          date: config[:published_at] || Date.today.to_s
+        )
       end
 
       def html
@@ -30,19 +44,15 @@ def export
 
         copy_styles!
         copy_images!
-        File.open(root_dir.join("output/epub/cover.html"), "w") do |io|
-          io << render_template(
-            root_dir.join("templates/epub/cover.erb"),
-            config
-          )
-        end
-        set_metadata!
+
         write_sections!
+        write_cover!
         write_toc!
 
-        epub.files sections.map(&:filepath) + assets
-        epub.nav      navigation
-        epub.toc_page toc_path
+        epub.files += [tmp_dir.join("cover.html"), tmp_dir.join("toc.html")]
+        epub.files += tmp_dir.glob("**/*").select do |entry|
+          !epub.files.include?(entry) && entry.file?
+        end
 
         epub.save(epub_path)
 
@@ -61,56 +71,31 @@ def copy_images!
         copy_directory("output/images", "output/epub/images")
       end
 
-      def set_metadata!
-        epub.title        config[:title]
-        epub.language     config[:language]
-        epub.creator      config[:authors].to_sentence
-        epub.publisher    config[:publisher]
-        epub.date         config[:published_at]
-        epub.uid          config[:uid]
-        epub.identifier   config[:identifier][:id],
-                          scheme: config[:identifier][:type]
-
-        # epubchecker complains when assigning an image directly,
-        # but if we don't, then Apple Books doesn't render the cover.
-        # Need to investigate some more.
-        epub.cover_page   cover_image if cover_image && File.exist?(cover_image)
-      end
-
       def write_toc!
-        toc = TOC::Epub.new(navigation)
-
-        File.open(toc_path, "w") do |file|
-          file << toc.to_html
+        root_dir.join(toc_path).open("w") do |io|
+          io << render_template(
+            root_dir.join("templates/epub/toc.erb"),
+            config.merge(
+              navigation: ::Epub::Navigation.extract_html(
+                tmp_dir.glob("**/*.{xhtml,html}"),
+                root_dir: tmp_dir
+              )
+            )
+          )
         end
       end
 
-      def write_sections!
-        # First we need to get all ids, which are used as
-        # the anchor target.
-        #
-        links = sections.each_with_object({}) do |section, buffer|
-          section.html.css("[id]").each do |element|
-            anchor = "##{element['id']}"
-            buffer[anchor] = "#{section.filename}#{anchor}"
-          end
+      def write_cover!
+        root_dir.join(cover_path).open("w") do |io|
+          io << render_template(
+            root_dir.join("templates/epub/cover.erb"),
+            config
+          )
         end
+      end
 
-        # Then we can normalize all links and
-        # manipulate other paths.
-        #
+      def write_sections!
         sections.each do |section|
-          section.html.css("a[href^='#']").each do |link|
-            href = link["href"]
-            link.set_attribute("href", links.fetch(href, href))
-          end
-
-          # Normalize <ol>
-          #
-          section.html.css("ol[start]").each do |node|
-            node.remove_attribute("start")
-          end
-
           # Remove anchors (only useful for html exports).
           #
           section.html.css("a.anchor").each(&:remove)
@@ -121,82 +106,23 @@ def write_sections!
             node.remove_attribute("tabindex")
           end
 
-          # Replace all srcs.
-          #
-          section.html.css("[src]").each do |element|
-            src = File.basename(element["src"]).gsub(/\.svg$/, ".png")
-            element.set_attribute("src", src)
-            element.set_attribute("alt", "")
-            element.node_name = "img"
-          end
-
           FileUtils.mkdir_p(tmp_dir)
 
           # Save file to disk.
           #
           File.open(section.filepath, "w") do |file|
-            body = section.html.css("body").to_xhtml.gsub(
+            section.html = Kitabu.run_hooks(:process_epub_section, section.html)
+
+            content = section.html.css("body").to_xhtml.gsub(
               %r{<body>(.*?)</body>}m, "\\1"
             )
-            file << render_chapter(body)
-          end
-        end
-      end
 
-      def render_chapter(content)
-        locals = config.merge(content:)
-        render_template(template_path, locals)
-      end
+            page_title = section.html.css("h2").first.text.strip
+            locals = config.merge(content:, page_title:)
 
-      def assets
-        @assets ||= begin
-          assets = Dir[root_dir.join("output/epub/styles/**/*.css")]
-          assets += Dir[root_dir.join("images/**/*.{jpg,png,gif}")]
-          assets
-        end
-      end
-
-      def cover_image
-        path = Dir[root_dir.join("output/epub/images/cover.{jpg,png,gif}").to_s]
-               .first
-
-        path if path && File.exist?(path)
-      end
-
-      def navigation
-        klass = Struct.new(:level, :data, :parent, keyword_init: true)
-
-        root = klass.new(level: 1, data: {nav: []})
-        current = root
-
-        sections.each do |section|
-          section.html.css("h2, h3, h4, h5, h6").each do |node|
-            label = CGI.escape_html(node.text.strip)
-            level = node.name[1].to_i
-
-            data = {
-              label:,
-              content: "#{section.filename}##{node.attributes['id']}",
-              nav: []
-            }
-
-            if level > current.level
-              current = klass.new(level:, data:, parent: current)
-            elsif level == current.level
-              current = klass.new(level:, data:, parent: current.parent)
-            else
-              while current.parent && current.parent.level >= level
-                current = current.parent
-              end
-
-              current = klass.new(level:, data:, parent: current.parent)
-            end
-
-            current.parent.data[:nav] << data
+            file << render_template(template_path, locals)
           end
         end
-
-        root.data[:nav]
       end
 
       def template_path
@@ -218,6 +144,10 @@ def tmp_dir
       def toc_path
         tmp_dir.join("toc.html")
       end
+
+      def cover_path
+        tmp_dir.join("cover.html")
+      end
     end
   end
 end