reverse_adoc 0.2.3

data/bin/reverse_adoc

@@ -0,0 +1,67 @@
+#!/usr/bin/env ruby
+# Usage: reverse_asciidoctor [FILE]...
+# Usage: cat FILE | reverse_asciidoctor
+require 'reverse_asciidoctor'
+require 'optparse'
+require 'fileutils'
+
+OptionParser.new do |opts|
+  opts.banner = "Usage: reverse_adoc [options] <file>"
+  opts.on('-m', '--mathml2asciimath', 'Convert MathML to AsciiMath') do |v|
+    ReverseAsciidoctor.config.mathml2asciimath = true
+  end
+
+  opts.on('-oFILENAME', '--output=FILENAME', 'Output file to write to') do |v|
+    ReverseAsciidoctor.config.destination = File.expand_path(v)
+    # puts "output goes to #{ReverseAsciidoctor.config.destination}"
+  end
+
+  opts.on('-e', '--external-images', 'Export images if data URI') do |v|
+    ReverseAsciidoctor.config.external_images = true
+  end
+
+  opts.on('-u', '--unknown_tags [pass_through, drop, bypass, raise]', 'Unknown tag handling (default: pass_through)') do |v|
+    ReverseAsciidoctor.config.unknown_tags = v
+  end
+
+  opts.on('-v', '--version', 'Version information') do |v|
+    puts "reverse_adoc: v#{ReverseAsciidoctor::VERSION}"
+    exit
+  end
+
+  opts.on("-h", "--help", "Prints this help") do
+    puts opts
+    exit
+  end
+
+end.parse!
+
+if filename = ARGV.pop
+  input_content = IO.read(filename)
+  ReverseAsciidoctor.config.sourcedir = File.dirname(File.expand_path(filename))
+else
+  if ReverseAsciidoctor.config.external_images
+    raise "The -e | --external-images feature cannot be used with STDIN input. Exiting."
+  end
+
+  input_content = ARGF.read
+end
+
+if ReverseAsciidoctor.config.external_images && ReverseAsciidoctor.config.destination.nil?
+  raise "The -e | --external-images feature must be used with -o | --output. Exiting."
+end
+
+# Read from STDIN
+adoc_content = ReverseAsciidoctor.convert(input_content)
+
+# Print to STDOUT
+unless ReverseAsciidoctor.config.destination
+  puts adoc_content
+  exit
+end
+
+# Write output to ReverseAsciidoctor.config.destination
+FileUtils.mkdir_p(File.dirname(ReverseAsciidoctor.config.destination))
+File.open(ReverseAsciidoctor.config.destination, "w") do |file|
+  file.write(adoc_content)
+end

data/bin/w2a

@@ -0,0 +1,85 @@
+#!/usr/bin/env ruby
+# frozen_string_literal: true
+
+require 'word-to-markdown'
+require 'optparse'
+require 'reverse_asciidoctor'
+
+def scrub_whitespace(string)
+  string = string.dup
+  string.gsub!(' ', ' ')       # HTML encoded spaces
+  string.sub!(/\A[[:space:]]+/, '') # document leading whitespace
+  string.sub!(/[[:space:]]+\z/, '') # document trailing whitespace
+  string.gsub!(/([ ]+)$/, '')       # line trailing whitespace
+  string.gsub!(/\n\n\n\n/, "\n\n")  # Quadruple line breaks
+  string.delete!(' ')               # Unicode non-breaking spaces, injected as tabs
+  # following added by me
+  string.gsub!(%r{<h[1-9][^>]*></h1>}, " ") # I don't know why Libre Office is inserting them, but they need to go
+  string.gsub!(%r{<h1[^>]* style="vertical-align: super;[^>]*>([^<]+)</h1>},
+              "<sup>\\1</sup>")         # I absolutely don't know why Libre Office is rendering superscripts as h1
+  string
+end
+
+ARGV.push('-h') if ARGV.empty?
+
+OptionParser.new do |opts|
+  opts.banner = "Usage: w2a [options] <file>"
+  opts.on('-m', '--mathml2asciimath', 'Convert MathML to AsciiMath') do |v|
+    ReverseAsciidoctor.config.mathml2asciimath = true
+  end
+
+  opts.on('-oFILENAME', '--output=FILENAME', 'Output file to write to') do |v|
+    ReverseAsciidoctor.config.destination = File.expand_path(v)
+    # puts "output goes to #{ReverseAsciidoctor.config.destination}"
+  end
+
+  opts.on('-e', '--external-images', 'Export images if data URI') do |v|
+    ReverseAsciidoctor.config.external_images = true
+  end
+
+  opts.on('-v', '--version', 'Version information') do |v|
+    puts "reverse_adoc: v#{ReverseAsciidoctor::VERSION}"
+    puts "[dependency] WordToMarkdown: v#{WordToMarkdown::VERSION}"
+    unless Gem.win_platform?
+      puts "[dependency] LibreOffice: v#{WordToMarkdown.soffice.version}"
+    else
+      puts "[dependency] LibreOffice: version not available on Windows"
+    end
+    exit
+  end
+
+  opts.on("-h", "--help", "Prints this help") do
+    puts opts
+    exit
+  end
+
+end.parse!
+
+filename = ARGV.pop
+raise "Please provide an input file to process. Exiting." unless filename
+
+if ReverseAsciidoctor.config.external_images && ReverseAsciidoctor.config.destination.nil?
+  raise "The -e | --external-images feature must be used with -o | --output. Exiting."
+end
+
+ReverseAsciidoctor.config.sourcedir = Dir.mktmpdir
+# puts "ReverseAsciidoctor.config.sourcedir #{ReverseAsciidoctor.config.sourcedir}"
+
+doc = WordToMarkdown.new(filename, ReverseAsciidoctor.config.sourcedir)
+adoc_content = ReverseAsciidoctor.convert(
+  scrub_whitespace(doc.document.html),
+  WordToMarkdown::REVERSE_MARKDOWN_OPTIONS
+)
+# puts scrub_whitespace(doc.document.html)
+
+# Print to STDOUT
+unless ReverseAsciidoctor.config.destination
+  puts adoc_content
+  exit
+end
+
+# Write output to ReverseAsciidoctor.config.destination
+FileUtils.mkdir_p(File.dirname(ReverseAsciidoctor.config.destination))
+File.open(ReverseAsciidoctor.config.destination, "w") do |file|
+  file.write(adoc_content)
+end

data/lib/reverse_asciidoctor.rb

@@ -0,0 +1,70 @@
+require 'digest'
+require 'nokogiri'
+require 'reverse_asciidoctor/version'
+require 'reverse_asciidoctor/errors'
+require 'reverse_asciidoctor/cleaner'
+require 'reverse_asciidoctor/config'
+require 'reverse_asciidoctor/converters'
+require 'reverse_asciidoctor/converters/base'
+require 'reverse_asciidoctor/converters/a'
+require 'reverse_asciidoctor/converters/aside'
+require 'reverse_asciidoctor/converters/audio'
+require 'reverse_asciidoctor/converters/blockquote'
+require 'reverse_asciidoctor/converters/br'
+require 'reverse_asciidoctor/converters/bypass'
+require 'reverse_asciidoctor/converters/code'
+require 'reverse_asciidoctor/converters/div'
+require 'reverse_asciidoctor/converters/drop'
+require 'reverse_asciidoctor/converters/em'
+require 'reverse_asciidoctor/converters/figure'
+require 'reverse_asciidoctor/converters/h'
+require 'reverse_asciidoctor/converters/head'
+require 'reverse_asciidoctor/converters/hr'
+require 'reverse_asciidoctor/converters/ignore'
+require 'reverse_asciidoctor/converters/img'
+require 'reverse_asciidoctor/converters/mark'
+require 'reverse_asciidoctor/converters/li'
+require 'reverse_asciidoctor/converters/ol'
+require 'reverse_asciidoctor/converters/p'
+require 'reverse_asciidoctor/converters/pass_through'
+require 'reverse_asciidoctor/converters/pre'
+require 'reverse_asciidoctor/converters/q'
+require 'reverse_asciidoctor/converters/strong'
+require 'reverse_asciidoctor/converters/sup'
+require 'reverse_asciidoctor/converters/sub'
+require 'reverse_asciidoctor/converters/table'
+require 'reverse_asciidoctor/converters/td'
+require 'reverse_asciidoctor/converters/th'
+require 'reverse_asciidoctor/converters/text'
+require 'reverse_asciidoctor/converters/tr'
+require 'reverse_asciidoctor/converters/video'
+require 'reverse_asciidoctor/converters/math'
+
+module ReverseAsciidoctor
+
+  def self.convert(input, options = {})
+    root = case input
+      when String                  then Nokogiri::HTML(input).root
+      when Nokogiri::XML::Document then input.root
+      when Nokogiri::XML::Node     then input
+    end
+
+    root or return ''
+
+    config.with(options) do
+      result = ReverseAsciidoctor::Converters.lookup(root.name).convert(root)
+      cleaner.tidy(result)
+    end
+  end
+
+  def self.config
+    @config ||= Config.new
+    yield @config if block_given?
+    @config
+  end
+
+  def self.cleaner
+    @cleaner ||= Cleaner.new
+  end
+
+end

data/lib/reverse_asciidoctor/cleaner.rb

@@ -0,0 +1,90 @@
+module ReverseAsciidoctor
+  class Cleaner
+
+    def tidy(string)
+      result = remove_inner_whitespaces(string)
+      result = remove_newlines(result)
+      result = remove_leading_newlines(result)
+      result = clean_tag_borders(result)
+      clean_punctuation_characters(result)
+    end
+
+    def remove_newlines(string)
+      string.gsub(/\n{3,}/, "\n\n")
+    end
+
+    def remove_leading_newlines(string)
+      string.gsub(/\A\n+/, '')
+    end
+
+    def remove_inner_whitespaces(string)
+      unless string.nil?
+        string.gsub!(/\n stem:\[/, "\nstem:[")
+        string.gsub!(/(stem:\[([^\]]|\\\])*\])\n(?=\S)/, "\\1 ")
+        string.gsub!(/(stem:\[([^\]]|\\\])*\])\s+(?=[\^-])/, "\\1")
+      end
+      string.each_line.inject("") do |memo, line|
+        memo + preserve_border_whitespaces(line) do
+          line.strip.gsub(/[ \t]{2,}/, ' ')
+        end
+      end
+    end
+
+    # Find non-asterisk content that is enclosed by two or
+    # more asterisks. Ensure that only one whitespace occurs
+    # in the border area.
+    # Same for underscores and brackets.
+    def clean_tag_borders(string)
+      result = string.gsub(/\s?\*{2,}.*?\*{2,}\s?/) do |match|
+        preserve_border_whitespaces(match, default_border: ReverseAsciidoctor.config.tag_border) do
+          match.strip.sub('** ', '**').sub(' **', '**')
+        end
+      end
+
+      result = result.gsub(/\s?\_{2,}.*?\_{2,}\s?/) do |match|
+        preserve_border_whitespaces(match, default_border: ReverseAsciidoctor.config.tag_border) do
+          match.strip.sub('__ ', '__').sub(' __', '__')
+        end
+      end
+
+      result = result.gsub(/\s?~{2,}.*?~{2,}\s?/) do |match|
+        preserve_border_whitespaces(match, default_border: ReverseAsciidoctor.config.tag_border) do
+          match.strip.sub('~~ ', '~~').sub(' ~~', '~~')
+        end
+      end
+
+      result.gsub(/\s?\[.*?\]\s?/) do |match|
+        preserve_border_whitespaces(match) do
+          match.strip.sub('[ ', '[').sub(' ]', ']')
+        end
+      end
+    end
+
+    def clean_punctuation_characters(string)
+      string.gsub(/(\*\*|~~|__)\s([\.!\?'"])/, "\\1".strip + "\\2")
+    end
+
+    private
+
+    def preserve_border_whitespaces(string, options = {}, &block)
+      return string if string =~ /\A\s*\Z/
+      default_border = options.fetch(:default_border, '')
+      # If the string contains part of a link so the characters [,],(,)
+      # then don't add any extra spaces
+      default_border = '' if string =~ /[\[\(\]\)]/
+      string_start   = present_or_default(string[/\A\s*/], default_border)
+      string_end     = present_or_default(string[/\s*\Z/], default_border)
+      result         = yield
+      string_start + result + string_end
+    end
+
+    def present_or_default(string, default)
+      if string.nil? || string.empty?
+        default
+      else
+        string
+      end
+    end
+
+  end
+end

data/lib/reverse_asciidoctor/config.rb

@@ -0,0 +1,53 @@
+require 'tmpdir'
+
+module ReverseAsciidoctor
+  class Config
+    attr_accessor :unknown_tags, :tag_border, :mathml2asciimath, :external_images,
+      :destination, :sourcedir, :image_counter, :image_counter_pattern
+
+    def initialize
+      @unknown_tags     = :pass_through
+      @mathml2asciimath = false
+      @external_images  = false
+
+      # Destination to save file and images
+      @destination      = nil
+
+      # Source of HTML
+      # @sourcedir        = nil
+
+      # Image counter, assuming there are max 999 images
+      @image_counter    = 1
+      # pad with 0s
+      @image_counter_pattern = '%03d'
+
+      @em_delimiter     = '_'.freeze
+      @strong_delimiter = '*'.freeze
+      @inline_options   = {}
+      @tag_border       = ' '.freeze
+    end
+
+    def with(options = {})
+      @inline_options = options
+      result = yield
+      @inline_options = {}
+      result
+    end
+
+    def unknown_tags
+      @inline_options[:unknown_tags] || @unknown_tags
+    end
+
+    def mathml2asciimath
+      @inline_options[:mathml2asciimath] || @mathml2asciimath
+    end
+
+    def external_images
+      @inline_options[:external_images] || @external_images
+    end
+
+    def tag_border
+      @inline_options[:tag_border] || @tag_border
+    end
+  end
+end

data/lib/reverse_asciidoctor/converters.rb

@@ -0,0 +1,33 @@
+module ReverseAsciidoctor
+  module Converters
+    def self.register(tag_name, converter)
+      @@converters ||= {}
+      @@converters[tag_name.to_sym] = converter
+    end
+
+    def self.unregister(tag_name)
+      @@converters.delete(tag_name.to_sym)
+    end
+
+    def self.lookup(tag_name)
+      @@converters[tag_name.to_sym] or default_converter(tag_name)
+    end
+
+    private
+
+    def self.default_converter(tag_name)
+      case ReverseAsciidoctor.config.unknown_tags.to_sym
+      when :pass_through
+        ReverseAsciidoctor::Converters::PassThrough.new
+      when :drop
+        ReverseAsciidoctor::Converters::Drop.new
+      when :bypass
+        ReverseAsciidoctor::Converters::Bypass.new
+      when :raise
+        raise UnknownTagError, "unknown tag: #{tag_name}"
+      else
+        raise InvalidConfigurationError, "unknown value #{ReverseAsciidoctor.config.unknown_tags.inspect} for ReverseAsciidoctor.config.unknown_tags"
+      end
+    end
+  end
+end

data/lib/reverse_asciidoctor/converters/a.rb

@@ -0,0 +1,38 @@
+require "uri"
+
+module ReverseAsciidoctor
+  module Converters
+    class A < Base
+      def convert(node, state = {})
+        name  = treat_children(node, state)
+        href  = node['href']
+        title = extract_title(node)
+        id = node['id'] || node['name']
+
+        if !id.nil? && !id.empty?
+          "[[#{id}]]"
+        elsif href.to_s.start_with?('#')
+            href = href.sub(/^#/, "")
+          if name.empty?
+            "<<#{href}>>"
+          else
+            "<<#{href},#{name}>>"
+          end
+        elsif href.to_s.empty?
+          name
+        else
+          name = title if name.empty?
+          href = "link:#{href}" unless href.to_s =~ URI::DEFAULT_PARSER.make_regexp
+          link = "#{href}[#{name}]"
+          link.prepend(' ')
+          link
+        end
+      end
+
+      private
+
+    end
+
+    register :a, A.new
+  end
+end

data/lib/reverse_asciidoctor/converters/aside.rb

@@ -0,0 +1,14 @@
+module ReverseAsciidoctor
+  module Converters
+    class Aside < Base
+      def convert(node, state = {})
+        id = node['id']
+        anchor = id ? "[[#{id}]]\n" : ""
+        content = treat_children(node, state).strip
+        "\n\n****\n" << treat_children(node, state) << "\n****\n\n"
+      end
+    end
+
+    register :aside, Aside.new
+  end
+end