Sha256: 2230b9435eed9375330bc1308ca6a96873ab7bf870917651b2e3c8e495c66eb8
Contents?: true
Size: 1.79 KB
Versions: 8
Compression:
Stored size: 1.79 KB
Contents
require 'optparse'
require 'epub/parser'
options = {:format => :line}
opt = OptionParser.new do |opt|
opt.banner = <<EOB
Show metadata of an EPUB file
Usage: epubinfo [options] EPUBFILE
EOB
opt.version = EPUB::Parser::VERSION
formats = [:line, :json, :yaml]
opt.on '-f', '--format=FORMAT', formats, "format of output(#{formats[0..-2].join(', ')} or #{formats.last}), defaults to line(for console)" do |format|
options[:format] = format
end
opt.on '--words', 'count words of content documents' do
options[:words] = true
end
opt.on '--chars', 'count charactors of content documents' do
options[:chars] = true
end
end
opt.parse!(ARGV)
file = ARGV.shift
unless file
$stdout.puts "error: you must supply an EPUB file name"
$stdout.puts opt.help
abort
end
book = EPUB::Parser.parse(file)
data = {'Title' => [book.title]}
data.merge!(book.metadata.to_h)
data['Unique identifier'] = [book.metadata.unique_identifier]
data['EPUB Version'] = [book.package.version]
counts = {:chars => 0, :words => 0}
if options[:words] or options[:chars]
book.resources.select(&:xhtml?).each do |xhtml|
begin
doc = xhtml.content_document.nokogiri
body = doc.search('body').first
content = body.content
if body
counts[:words] += content.scan(/\S+/).length
counts[:chars] += content.gsub(/\r|\n/, '').length
end
rescue => error
warn "#{xhtml.href}: #{error}"
end
end
end
data['Words'] = [counts[:words]] if options[:words]
data['Characters'] = [counts[:chars]] if options[:chars]
if options[:format] == :line
key_width = data.keys.map {|k| k.length}.max + 3
data.each_pair do |k, v|
puts (k.to_s.capitalize + ':').ljust(key_width) + v.join(', ')
end
else
require options[:format].to_s
puts data.__send__("to_#{options[:format]}")
end
Version data entries
8 entries across 8 versions & 2 rubygems