reflect-errata-api-design.rb

# coding: utf-8
gem 'epub-parser', '>= 0.2.4'
gem 'epub-maker', '>= 0.0.2'
require 'English'
require 'epub/maker'
require 'epub/parser/cfi'
require 'nokogiri/xml/range'

TARGET_RELEASE_IDENTIFIER = 'urn:uuid:14eadcd9-214a-4f45-9623-d8ac54f22af7@2015-12-04T09:00:00Z'

ERRATA = [
  {target: '/6/36!/4/2/16/5,:25,:27', operation: :replace, replace: 'ri'},
  {target: '/6/36!/4/2/18/7,:33,:35', operation: :replace, replace: 'ri'},
  {target: '/6/46!/4/2/70/6/1,:0,:4', operation: :replace, replace: 'send'},
  {target: '/6/52!/4/2/28/1,:61,:62', operation: :remove},
  {target: '/6/68!/4/2/44/1:267', operation: :add, add: 'が'},
  {target: '/6/98!/4/2/12/9:7', operation: :add, add: '*10'},
  {target: '/6/118!/4/2/50/3,:45,:46', operation: :remove}
]

UNOFFICIAL_ERRATA = [
  {target: '/6/84!/4/2/34/4/1,:53,:54', operation: :remove}
]

NEW_MODIFIED = '2015-12-30T14:03:00Z'

def main(argv)
  epub_path = argv.shift
  if epub_path.nil? or !File.file?(epub_path)
    $stderr.puts "EPUBファイルを指定してください。"
    $stderr.puts usage
    abort
  end

  epub = EPUB::Parser.parse(epub_path)

  $stderr.puts "#{$PROGRAM_NAME}: このプログラムは指定された本（#{epub_path}）を上書きします。よろしいですか？(y/N)"
  unless gets.chomp.downcase == "y"
    exit
  end

  unless assert_release_identifier(epub)
    $stderr.puts "指定された本は、エラッタが適用可能なバージョン（Release Identifier）ではありません。"
    $stderr.puts "  エラッタの対象バージョン：#{TARGET_RELEASE_IDENTIFIER}"
    $stderr.puts "  指定された本のバージョン：#{epub.release_identifier}"
    abort
  end
  reflect_errata(epub)
  update_modified(epub)
rescue => error
  $stderr.puts error
  $stderr.puts error.backtrace
  $stderr.puts usage
  abort
end

def usage
  "Usage: ruby #{$PROGRAM_NAME} EPUB_PATH"
end

def assert_release_identifier(epub)
  epub.release_identifier == TARGET_RELEASE_IDENTIFIER
end

def reflect_errata(epub)
  errata = setup_errata
  errata.each do |erratum|
    reflect_erratum erratum, epub
  end
end

def update_modified(epub)
  epub.package.edit do |package|
    package.metadata.modified.content = NEW_MODIFIED
  end
end

def reflect_erratum(erratum, epub)
  if erratum[:target].kind_of? EPUB::CFI::Location
    item, start_node = end_node = get_item_and_element(erratum[:target], epub)
    start_offset = end_offset = erratum[:target].paths.last.offset.value
  else
    # TODO: Search nodes from parent subpath
    item, start_node = get_item_and_element(erratum[:target].first, epub)
    _, end_node = get_item_and_element(erratum[:target].last, epub)
    start_offset = erratum[:target].first.paths.last.offset.value
    end_offset = erratum[:target].last.paths.last.offset.value
  end

  range = Nokogiri::XML::Range.new(start_node, start_offset, end_node, end_offset)

  # page_title = (range.document/'h1').first.content.gsub(/\s+/, ' ')
  # puts '==========================='
  # puts "#{page_title}(#{item.href})"
  # puts '-----------BEFORE----------'
  # puts range.start_node.parent.content

  case erratum[:operation]
  when :add
    range.insert_node Nokogiri::XML::Text.new(erratum[:add], range.document)
  when :remove
    range.delete_contents
  when :replace
    range.delete_contents
    range.insert_node Nokogiri::XML::Text.new(erratum[:replace], range.document)
  end

  # puts '-----------AFTER----------'
  # puts range.start_node.parent.content

  item.content = range.document.to_xml
  item.save
end

def get_item_and_element(cfi, epub)
  path_in_package = cfi.paths.first
  step_to_itemref = path_in_package.steps[1]
  itemref = epub.spine.itemrefs[step_to_itemref.step / 2 - 1]

  doc = itemref.item.content_document.nokogiri
  path_in_doc = cfi.paths[1]
  current_node = doc.root
  path_in_doc.steps.each do |step|
    if step.element?
      current_node = current_node.element_children[step.value / 2 - 1]
    else
      element_index = (step.value - 1) / 2 - 1
      if element_index == -1
        current_node = current_node.children.first
      else
        prev = current_node.element_children[element_index]
        break unless prev
        current_node = prev.next_sibling
        break unless current_node
      end
    end
  end
  [itemref.item, current_node]
end

def setup_errata
  (ERRATA + UNOFFICIAL_ERRATA).map {|erratum|
    erratum[:target] = EPUB::CFI(erratum[:target])
    erratum
  }.sort_by {|erratum|
    cfi = erratum[:target]
    cfi = cfi.first if cfi.kind_of? EPUB::CFI::Range
    cfi
  }.reverse
end

main(ARGV)