apply-patch.rb

require "English"
require "json"
require "epub/ocf/physical_container/zipruby"
require "epub/parser"
require "epub/parser/cfi"
require "epub/maker"
require "rouge"
require "rouge/lexers/fluentd"

def main(argv)
  epub_file = argv.shift
  patch_file = argv.shift
  if patch_file.nil? || epub_file.nil?
    abort "USAGE: ruby #{$PROGRAM_NAME} EPUB PATCH"
  end

  patch = JSON.parse(File.read(patch_file))
  epub = EPUB::Parser.parse(epub_file, container_adapter: :Zipruby)
  # epub = EPUB::Parser.parse(epub_file)
  unless patch["target"] == epub.release_identifier
    abort "target release identifier doesn't match"
  end

  patch["patch"].each do |operation|
    case operation["op"]
    when "add"
      cfi = EPUB::CFI(operation["cfi"])
      cfi.find_and_edit epub do |elem|
        case elem
        when EPUB::Publication::Package::Manifest::Item
          value = Nokogiri::XML::DocumentFragment.parse(operation["value"]).first_element_child
          # TODO: insert to proper position: just before elem
          elem.manifest.make_item do |item|
            ["id", "href", "media-type", "properties", "media-overlay"].each do |attr|
              item.__send__ "#{attr.gsub('-', '_')}=", value[attr]
            end
            theme = operation["rouge-theme"]
            scope = operation["rouge-scope"]
            theme_class = Rouge::Theme.find(theme)
            if theme_class
              item.content = theme_class.new(scope: scope).render
            else
              item.content = `pygmentize -f html -S #{theme} -a #{scope}`
            end
            item.save # FIXME: bothering
          end
        when Nokogiri::XML::Element
          value = Nokogiri::XML::DocumentFragment.parse(operation["value"]).first_element_child
          elem.add_previous_sibling value
        else
          raise NotImplementedError
        end
      end
    when "replace"
      if operation["lang"]
        highlight epub, operation, patch["rouge-scope"]
      else
        value = Nokogiri::XML::DocumentFragment.parse(operation["value"]).first_element_child
        # TODO: keep and consider order of metadata child elements
        if value["property"] == "dcterms:modified"
          epub.package.edit do |package|
            package.metadata.modified.content = value.content
          end
        else
          raise NotImplementedError
        end
      end
    end
  end
end

def highlight(epub, operation, default_scope)
  formatter = Rouge::Formatters::HTML.new(wrap: false)
  cfi = EPUB::CFI(operation["cfi"])
  cfi.find_and_edit epub do |elem|
    lexer_name = operation["lang"]
    case operation["lang"]
    when "HiveQL", "PigLatin"
      lexer_name = "SQL"
    end
    lexer = Rouge::Lexers.const_get(lexer_name).new
    lexed = lexer.lex(elem.content)
    begin
      elem.inner_html = formatter.format(lexed)
      scope = operation["rouge-scope"] || default_scope
      additional_class = scope[1..-1]
      class_value = elem["class"] ?
                      elem["class"] + " #{additional_class}" :
                      additional_class
      elem["class"] = class_value
    rescue NoMethodError => error
      warn "Can't format #{operation['cfi']} as #{lexer_name}"
    end
  end
end

class EPUB::CFI::Location
  def find_and_edit(epub)
    path_in_package = paths.first.dup
    first_step = path_in_package.steps.shift
    modules = [:metadata, :manifest, :spine]
    mod_name = modules[first_step.value / 2 - 1]
    second_step = path_in_package.steps.shift
    case mod_name
    when :metadata
      if second_step
        raise "Can't resolve CFI" unless path_in_package.steps.empty?
        index = second_step.value / 2 - 1
        # TODO: keep and consider order of metadata child elements
        raise NotImplementedError
      else
        if paths.length == 1
          epub.package.edit do |package|
            yield package.metadata
          end
        else
          raise "Can't resolve indirect instruction"
        end
      end
    when :manifest
      if second_step
        raise "Can't resolve CFI" unless path_in_package.steps.empty?
        index = second_step.value / 2 - 1
        epub.package.edit do |package|
          yield package.manifest.items[index]
        end
      else
        if paths.length == 1
          epub.package.edit do |package|
            yield package.manifest
          end
        else
          raise "Can't find indirect instruction"
        end
      end
    when :spine
      unless second_step
        if paths.length == 1
          # edit spine
        else
          raise "Can't resolve indirect instruction"
        end
      end
      index = second_step.value / 2 - 1
      itemref = epub.package.spine.itemrefs[index]
      raise "Unresolvable step in first path" unless path_in_package.steps.empty?
      if paths.length == 1
        # edit itemref
      else
        itemref.item.edit_with_nokogiri do |doc|
          elem = doc.root
          paths[1].steps.each do |step|
            if step.element?
              index = step.value / 2 - 1
              elem = elem.element_children[index]
            else
              
            end
          end
          yield elem
        end
      end
    end
  end
end

main(ARGV)

server-infra-log.epubpatch.json

{
  "schema-version": "0.5.0",
  "modified": "2016-06-03T00:00:00Z",
  "target": "urn:uuid:5847863b-7653-4319-8c0b-ccee8a1dcac3@2014-08-08T09:00:00Z",
  "rouge-scope": ".etupirika-monokai-sublime",
  "patch": [
    {"op": "replace", "cfi": "/6/20!/4/2/42/2", "value": "", "lang": "Ruby"},
    {"op": "replace", "cfi": "/6/20!/4/2/80/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/20!/4/2/94/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/30!/4/2/70/2", "value": "", "lang": "JSON"},
    {"op": "replace", "cfi": "/6/30!/4/2/126/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/30!/4/2/236/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/30!/4/2/248/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/30!/4/2/252/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/30!/4/2/270/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/30!/4/2/284/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/30!/4/2/296/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/30!/4/2/302/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/30!/4/2/320/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/32!/4/2/92/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/32!/4/2/96/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/32!/4/2/122/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/32!/4/2/126/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/32!/4/2/130/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/34!/4/2/46/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/34!/4/2/50/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/34!/4/2/54/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/34!/4/2/136/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/34!/4/2/140/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/34!/4/2/166/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/34!/4/2/178/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/34!/4/2/188/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/34!/4/2/224/2", "value": "", "lang": "PHP"},
    {"op": "replace", "cfi": "/6/34!/4/2/228/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/34!/4/2/280/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/34!/4/2/284/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/34!/4/2/310/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/34!/4/2/314/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/34!/4/2/326/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/34!/4/2/378/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/34!/4/2/396/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/34!/4/2/404/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/34!/4/2/466/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/44!/4/2/62/2", "value": "", "lang": "JSON"},
    {"op": "replace", "cfi": "/6/44!/4/2/88/2", "value": "", "lang": "JSON"},
    {"op": "replace", "cfi": "/6/44!/4/2/124/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/44!/4/2/134/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/44!/4/2/142/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/44!/4/2/166/2", "value": "", "lang": "JSON"},
    {"op": "replace", "cfi": "/6/44!/4/2/206/2", "value": "", "lang": "JSON"},
    {"op": "replace", "cfi": "/6/44!/4/2/238/2", "value": "", "lang": "JSON"},
    {"op": "replace", "cfi": "/6/44!/4/2/286/2", "value": "", "lang": "JSON"},
    {"op": "replace", "cfi": "/6/44!/4/2/348/2", "value": "", "lang": "JSON"},
    {"op": "replace", "cfi": "/6/44!/4/2/354/2", "value": "", "lang": "JSON"},
    {"op": "replace", "cfi": "/6/46!/4/2/20/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/46!/4/2/40/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/48!/4/2/228/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/48!/4/2/236/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/50!/4/2/50/2", "value": "", "lang": "Java"},
    {"op": "replace", "cfi": "/6/50!/4/2/58/2", "value": "", "lang": "Java"},
    {"op": "replace", "cfi": "/6/50!/4/2/68/2", "value": "", "lang": "Java"},
    {"op": "replace", "cfi": "/6/50!/4/2/76/2", "value": "", "lang": "Java"},
    {"op": "replace", "cfi": "/6/50!/4/2/102/2", "value": "", "lang": "HiveQL", "rouge-scope": ".etupirika-colorful"},
    {"op": "replace", "cfi": "/6/50!/4/2/112/2", "value": "", "lang": "HiveQL", "rouge-scope": ".etupirika-colorful"},
    {"op": "replace", "cfi": "/6/50!/4/2/132/2", "value": "", "lang": "PigLatin"},
    {"op": "replace", "cfi": "/6/50!/4/2/140/2", "value": "", "lang": "PigLatin"},
    {"op": "replace", "cfi": "/6/50!/4/2/150/2", "value": "", "lang": "PigLatin"},
    {"op": "replace", "cfi": "/6/56!/4/2/48/2", "value": "", "lang": "Javascript", "rouge-scope": ".etupirika-emacs"},
    {"op": "replace", "cfi": "/6/56!/4/2/54/2", "value": "", "lang": "Javascript"},
    {"op": "replace", "cfi": "/6/66!/4/2/28/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/66!/4/2/44/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/66!/4/2/66/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/66!/4/2/76/2", "value": "", "lang": "Fluentd"},
    {"op": "replace", "cfi": "/6/66!/4/2/92/2", "value": "", "lang": "Nginx"},
    {"op": "replace", "cfi": "/6/66!/4/2/108/2", "value": "", "lang": "Nginx"},
    {"op": "replace", "cfi": "/6/66!/4/2/118/2", "value": "", "lang": "Nginx"},
    {"op": "add", "cfi": "/4/2", "value": "<item id=\"etupirika-monoki-sublime-css\" href=\"etupirika/monokai.sublime.css\" media-type=\"text/css\" />", "rouge-theme": "monokai.sublime", "rouge-scope": ".etupirika-monokai-sublime", "comment": "adding item means adding file together"},
    {"op": "add", "cfi": "/4/2", "value": "<item id=\"etupirika-colorful-css\" href=\"etupirika/colorful.css\" media-type=\"text/css\" />", "rouge-theme": "colorful", "rouge-theme": "colorful", "rouge-scope": ".etupirika-colorful"},
    {"op": "add", "cfi": "/4/2", "value": "<item id=\"etupirika-emacs-css\" href=\"etupirika/emacs.css\" media-type=\"text/css\" />", "rouge-theme": "emacs", "rouge-theme": "emacs", "rouge-scope": ".etupirika-emacs"},
    {"op": "add", "cfi": "/6/14!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/monokai.sublime.css\" />", "rouge-scope": ".etupirika-monokai-sublime", "rouge-theme": "monokai.sublime"},
    {"op": "add", "cfi": "/6/20!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/monokai.sublime.css\" />", "rouge-scope": ".etupirika-monokai-sublime", "rouge-theme": "monokai.sublime"},
    {"op": "add", "cfi": "/6/30!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/monokai.sublime.css\" />", "rouge-scope": ".etupirika-monokai-sublime", "rouge-theme": "monokai.sublime"},
    {"op": "add", "cfi": "/6/32!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/monokai.sublime.css\" />", "rouge-scope": ".etupirika-monokai-sublime", "rouge-theme": "monokai.sublime"},
    {"op": "add", "cfi": "/6/34!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/monokai.sublime.css\" />", "rouge-scope": ".etupirika-monokai-sublime", "rouge-theme": "monokai.sublime"},
    {"op": "add", "cfi": "/6/44!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/monokai.sublime.css\" />", "rouge-scope": ".etupirika-monokai-sublime", "rouge-theme": "monokai.sublime"},
    {"op": "add", "cfi": "/6/46!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/monokai.sublime.css\" />", "rouge-scope": ".etupirika-monokai-sublime", "rouge-theme": "monokai.sublime"},
    {"op": "add", "cfi": "/6/48!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/monokai.sublime.css\" />", "rouge-scope": ".etupirika-monokai-sublime", "rouge-theme": "monokai.sublime"},
    {"op": "add", "cfi": "/6/50!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/monokai.sublime.css\" />", "rouge-scope": ".etupirika-monokai-sublime", "rouge-theme": "monokai.sublime"},
    {"op": "add", "cfi": "/6/56!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/monokai.sublime.css\" />", "rouge-scope": ".etupirika-monokai-sublime", "rouge-theme": "monokai.sublime"},
    {"op": "add", "cfi": "/6/66!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/monokai.sublime.css\" />", "rouge-scope": ".etupirika-monokai-sublime", "rouge-theme": "monokai.sublime"},
    {"op": "add", "cfi": "/6/14!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/colorful.css\" />", "rouge-scope": ".etupirika-colorful", "rouge-theme": "colorful"},
    {"op": "add", "cfi": "/6/20!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/colorful.css\" />", "rouge-scope": ".etupirika-colorful", "rouge-theme": "colorful"},
    {"op": "add", "cfi": "/6/30!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/colorful.css\" />", "rouge-scope": ".etupirika-colorful", "rouge-theme": "colorful"},
    {"op": "add", "cfi": "/6/32!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/colorful.css\" />", "rouge-scope": ".etupirika-colorful", "rouge-theme": "colorful"},
    {"op": "add", "cfi": "/6/34!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/colorful.css\" />", "rouge-scope": ".etupirika-colorful", "rouge-theme": "colorful"},
    {"op": "add", "cfi": "/6/44!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/colorful.css\" />", "rouge-scope": ".etupirika-colorful", "rouge-theme": "colorful"},
    {"op": "add", "cfi": "/6/46!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/colorful.css\" />", "rouge-scope": ".etupirika-colorful", "rouge-theme": "colorful"},
    {"op": "add", "cfi": "/6/48!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/colorful.css\" />", "rouge-scope": ".etupirika-colorful", "rouge-theme": "colorful"},
    {"op": "add", "cfi": "/6/50!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/colorful.css\" />", "rouge-scope": ".etupirika-colorful", "rouge-theme": "colorful"},
    {"op": "add", "cfi": "/6/56!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/colorful.css\" />", "rouge-scope": ".etupirika-colorful", "rouge-theme": "colorful"},
    {"op": "add", "cfi": "/6/66!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/colorful.css\" />", "rouge-scope": ".etupirika-colorful", "rouge-theme": "colorful"},
    {"op": "add", "cfi": "/6/14!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/emacs.css\" />", "rouge-scope": ".etupirika-emacs", "rouge-theme": "emacs"},
    {"op": "add", "cfi": "/6/20!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/emacs.css\" />", "rouge-scope": ".etupirika-emacs", "rouge-theme": "emacs"},
    {"op": "add", "cfi": "/6/30!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/emacs.css\" />", "rouge-scope": ".etupirika-emacs", "rouge-theme": "emacs"},
    {"op": "add", "cfi": "/6/32!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/emacs.css\" />", "rouge-scope": ".etupirika-emacs", "rouge-theme": "emacs"},
    {"op": "add", "cfi": "/6/34!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/emacs.css\" />", "rouge-scope": ".etupirika-emacs", "rouge-theme": "emacs"},
    {"op": "add", "cfi": "/6/44!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/emacs.css\" />", "rouge-scope": ".etupirika-emacs", "rouge-theme": "emacs"},
    {"op": "add", "cfi": "/6/46!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/emacs.css\" />", "rouge-scope": ".etupirika-emacs", "rouge-theme": "emacs"},
    {"op": "add", "cfi": "/6/48!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/emacs.css\" />", "rouge-scope": ".etupirika-emacs", "rouge-theme": "emacs"},
    {"op": "add", "cfi": "/6/50!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/emacs.css\" />", "rouge-scope": ".etupirika-emacs", "rouge-theme": "emacs"},
    {"op": "add", "cfi": "/6/56!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/emacs.css\" />", "rouge-scope": ".etupirika-emacs", "rouge-theme": "emacs"},
    {"op": "add", "cfi": "/6/66!/2/2", "value": "<link rel=\"stylesheet\" href=\"../etupirika/emacs.css\" />", "rouge-scope": ".etupirika-emacs", "rouge-theme": "emacs"},
    {"op": "replace", "cfi": "/2/16", "value": "<meta property=\"dcterms:modified\">2016-06-03T00:00:00Z</meta>", "comment": "duplicated with top level modified; TODO: replace only content"}
  ]
}