reverse_adoc 0.3.6 → 1.0.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 5ada0c5000d417b1fa73962833f53144ae8ef29d4089ec49cd5aa8cfe87f0a99
-  data.tar.gz: d56d80b96a0c52b7ad1d1c0e0133821ba9711f277388f59727fc6c904a2cad12
+  metadata.gz: 69ca642594cadc268f47a51d4d9fc10852d7f0e51f2987ecb7839702187f345a
+  data.tar.gz: 7e0ca6f3c222616d8fc1d0d4b94853a97abb9b310418aa531e2c4abde3ce1a39
 SHA512:
-  metadata.gz: 2cde0734e47f95bb24664dade9a3a11266bf4d109b15211b45e3385d71b6cc5748d7faa238aca2861d31f01045c09975f5e516a251b7c555ad31d48199fc8a90
-  data.tar.gz: 3f318a46c99954e5d69b12b03aaaca6e75b0238740fbdded85ae8dc2c1d147bdb1b5c2e4e9207f34406664150359a92bb6237cfe38622996cd24f89899af8323
+  metadata.gz: a625dfb2610efafebcb38b8427a2f414b8358c0261a463627f60190e3cb45d18f643b1e29c62135c9a76d402db380189bea234e82af3d658871f3ce2f56c6655
+  data.tar.gz: 9076811a8e823ab0b52c4565b1bd8996931721b59d84b603494ab0fe475bc72a9f35f6e585722f6afbb535b945d6abe1c8fdf097c7ef97ac181335b8a1157137

data/.github/workflows/release.yml

@@ -7,11 +7,12 @@ on:
     inputs:
       next_version:
         description: |
-          Next release version. Possible values: x.y.z, major, minor, patch or pre|rc|etc
+          Next release version. Possible values: x.y.z, major, minor, patch (or pre|rc|etc).
+          Also, you can pass 'skip' to skip 'git tag' and do 'gem push' for the current version
         required: true
         default: 'skip'
-  push:
-    tags: [ v* ]
+  repository_dispatch:
+    types: [ do-release ]
 
 jobs:
   release:

data/Gemfile

@@ -1,6 +1,6 @@
 source "https://rubygems.org"
 
-git_source(:github) {|repo_name| "https://github.com/#{repo_name}" }
+git_source(:github) { |repo_name| "https://github.com/#{repo_name}" }
 
 # Specify your gem's dependencies in reverse_adoc.gemspec
 gemspec

data/Rakefile

@@ -1,14 +1,14 @@
-require 'bundler/gem_tasks'
+require "bundler/gem_tasks"
 
-if File.exist?('.codeclimate')
-  ENV["CODECLIMATE_REPO_TOKEN"] = File.read('.codeclimate').strip
+if File.exist?(".codeclimate")
+  ENV["CODECLIMATE_REPO_TOKEN"] = File.read(".codeclimate").strip
 end
 
-require 'rspec/core/rake_task'
+require "rspec/core/rake_task"
 RSpec::Core::RakeTask.new(:spec)
-task :default => :spec
+task default: :spec
 
-desc 'Open an irb session preloaded with this library'
+desc "Open an irb session preloaded with this library"
 task :console do
-  sh 'irb -rubygems -I lib -r reverse_adoc.rb'
+  sh "irb -rubygems -I lib -r reverse_adoc.rb"
 end

data/exe/reverse_adoc

@@ -1,33 +1,34 @@
 #!/usr/bin/env ruby
 # Usage: reverse_adoc [FILE]...
 # Usage: cat FILE | reverse_adoc
-require 'rubygems'
-require 'bundler/setup'
+require "rubygems"
+require "bundler/setup"
 
-require 'reverse_adoc'
-require 'optparse'
-require 'fileutils'
+require "reverse_adoc"
+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|
+  opts.on("-m", "--mathml2asciimath", "Convert MathML to AsciiMath") do |_v|
     ReverseAdoc.config.mathml2asciimath = true
   end
 
-  opts.on('-oFILENAME', '--output=FILENAME', 'Output file to write to') do |v|
+  opts.on("-oFILENAME", "--output=FILENAME", "Output file to write to") do |v|
     ReverseAdoc.config.destination = File.expand_path(v)
     # puts "output goes to #{ReverseAdoc.config.destination}"
   end
 
-  opts.on('-e', '--external-images', 'Export images if data URI') do |v|
+  opts.on("-e", "--external-images", "Export images if data URI") do |_v|
     ReverseAdoc.config.external_images = true
   end
 
-  opts.on('-u', '--unknown_tags [pass_through, drop, bypass, raise]', 'Unknown tag handling (default: pass_through)') do |v|
+  opts.on("-u", "--unknown_tags [pass_through, drop, bypass, raise]",
+          "Unknown tag handling (default: pass_through)") do |v|
     ReverseAdoc.config.unknown_tags = v
   end
 
-  opts.on('-v', '--version', 'Version information') do |v|
+  opts.on("-v", "--version", "Version information") do |_v|
     puts "reverse_adoc: v#{ReverseAdoc::VERSION}"
     exit
   end
@@ -36,7 +37,6 @@ OptionParser.new do |opts|
     puts opts
     exit
   end
-
 end.parse!
 
 if filename = ARGV.pop

data/exe/w2a

@@ -1,37 +1,37 @@
 #!/usr/bin/env ruby
 # frozen_string_literal: true
 
-require 'rubygems'
-require 'bundler/setup'
+require "rubygems"
+require "bundler/setup"
 
-require 'word-to-markdown'
-require 'optparse'
-require 'reverse_adoc'
+require "word-to-markdown"
+require "optparse"
+require "reverse_adoc"
 
-ARGV.push('-h') if ARGV.empty?
+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|
+  opts.on("-m", "--mathml2asciimath", "Convert MathML to AsciiMath") do |_v|
     ReverseAdoc.config.mathml2asciimath = true
   end
 
-  opts.on('-oFILENAME', '--output=FILENAME', 'Output file to write to') do |v|
+  opts.on("-oFILENAME", "--output=FILENAME", "Output file to write to") do |v|
     ReverseAdoc.config.destination = File.expand_path(v)
     # puts "output goes to #{ReverseAdoc.config.destination}"
   end
 
-  opts.on('-e', '--external-images', 'Export images if data URI') do |v|
+  opts.on("-e", "--external-images", "Export images if data URI") do |_v|
     ReverseAdoc.config.external_images = true
   end
 
-  opts.on('-v', '--version', 'Version information') do |v|
+  opts.on("-v", "--version", "Version information") do |_v|
     puts "reverse_adoc: v#{ReverseAdoc::VERSION}"
     puts "[dependency] WordToMarkdown: v#{WordToMarkdown::VERSION}"
-    unless Gem.win_platform?
-      puts "[dependency] LibreOffice: v#{WordToMarkdown.soffice.version}"
-    else
+    if Gem.win_platform?
       puts "[dependency] LibreOffice: version not available on Windows"
+    else
+      puts "[dependency] LibreOffice: v#{WordToMarkdown.soffice.version}"
     end
     exit
   end
@@ -40,7 +40,6 @@ OptionParser.new do |opts|
     puts opts
     exit
   end
-
 end.parse!
 
 filename = ARGV.pop
@@ -53,10 +52,10 @@ end
 ReverseAdoc.config.sourcedir = Dir.mktmpdir
 
 doc = WordToMarkdown.new(filename, ReverseAdoc.config.sourcedir)
-#File.open("test.html", "w:UTF-8") { |f| f.write doc.document.html }
+# File.open("test.html", "w:UTF-8") { |f| f.write doc.document.html }
 adoc_content = ReverseAdoc.convert(
   ReverseAdoc.cleaner.preprocess_word_html(doc.document.html),
-  WordToMarkdown::REVERSE_MARKDOWN_OPTIONS
+  WordToMarkdown::REVERSE_MARKDOWN_OPTIONS,
 )
 # puts scrub_whitespace(doc.document.html)
 

data/lib/reverse_adoc/cleaner.rb

@@ -34,20 +34,21 @@ module ReverseAdoc
     # 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: ReverseAdoc.config.tag_border) do
-          match.strip.sub("** ", "**").sub(" **", "**")
-        end
-      end
+      # result = string.gsub(/\s?\*{2,}.*?\*{2,}\s?/) do |match|
+      # preserve_border_whitespaces(match, default_border: ReverseAdoc.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: ReverseAdoc.config.tag_border) do
-          match.strip.sub("__ ", "__").sub(" __", "__")
-        end
-      end
+      # result = string.gsub(/\s?_{2,}.*?_{2,}\s?/) do |match|
+      #   preserve_border_whitespaces(match, default_border: ReverseAdoc.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: ReverseAdoc.config.tag_border) do
+      result = string.gsub(/\s?~{2,}.*?~{2,}\s?/) do |match|
+        preserve_border_whitespaces(match,
+                                    default_border: ReverseAdoc.config.tag_border) do
           match.strip.sub("~~ ", "~~").sub(" ~~", "~~")
         end
       end
@@ -60,7 +61,7 @@ module ReverseAdoc
     end
 
     def clean_punctuation_characters(string)
-      string.gsub(/(\*\*|~~|__)\s([.!?'"])/, "\\1".strip + "\\2")
+      string.gsub(/(\*\*|~~|__)\s([.!?'"])/, "#{'\\1'.strip}\\2")
     end
 
     # preprocesses HTML, rather than postprocessing it
@@ -104,11 +105,9 @@ module ReverseAdoc
     end
 
     def present_or_default(string, default)
-      if string.nil? || string.empty?
-        default
-      else
-        string
-      end
+      return default if string.nil? || string.empty?
+
+      string
     end
   end
 end

data/lib/reverse_adoc/config.rb

@@ -1,9 +1,9 @@
-require 'tmpdir'
+require "tmpdir"
 
 module ReverseAdoc
   class Config
     attr_accessor :unknown_tags, :tag_border, :mathml2asciimath, :external_images,
-      :destination, :sourcedir, :image_counter, :image_counter_pattern, :input_format
+                  :destination, :sourcedir, :image_counter, :image_counter_pattern, :input_format
 
     def initialize
       @unknown_tags     = :pass_through
@@ -18,14 +18,14 @@ module ReverseAdoc
       # @sourcedir        = nil
 
       # Image counter, assuming there are max 999 images
-      @image_counter    = 1
+      @image_counter = 1
       # pad with 0s
-      @image_counter_pattern = '%03d'
+      @image_counter_pattern = "%03d"
 
-      @em_delimiter     = '_'.freeze
-      @strong_delimiter = '*'.freeze
+      @em_delimiter     = "_".freeze
+      @strong_delimiter = "*".freeze
       @inline_options   = {}
-      @tag_border       = ' '.freeze
+      @tag_border       = " ".freeze
     end
 
     def with(options = {})

data/lib/reverse_adoc/converters/a.rb

@@ -1,40 +1,40 @@
-require "uri"
+require "coradoc"
 
 module ReverseAdoc
   module Converters
     class A < Base
-      def convert(node, state = {})
+      def to_coradoc(node, state = {})
         name  = treat_children(node, state)
-        href  = node['href']
+
+        href  = node["href"]
         title = extract_title(node)
-        id = node['id'] || node['name']
+        id = node["id"] || node["name"]
 
         id = id&.gsub(/\s/, "")&.gsub(/__+/, "_")
 
-        if /^_Toc\d+$|^_GoBack$/.match id
-          ""
-        elsif !id.nil? && !id.empty?
-          "[[#{id}]]"
-        elsif href.to_s.start_with?('#')
+        return "" if /^_Toc\d+$|^_GoBack$/.match?(id)
+
+        if !id.nil? && !id.empty?
+          return Coradoc::Element::Inline::Anchor.new(id)
+        end
+
+        if href.to_s.start_with?("#")
           href = href.sub(/^#/, "").gsub(/\s/, "").gsub(/__+/, "_")
-          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
+          return Coradoc::Element::Inline::CrossReference.new(href, name)
         end
-      end
 
-      private
+        if href.to_s.empty?
+          return name
+        end
+
+        Coradoc::Element::Inline::Link.new(path: href,
+                                           name: name,
+                                           title: title)
+      end
 
+      def convert(node, state = {})
+        Coradoc::Generator.gen_adoc(to_coradoc(node, state))
+      end
     end
 
     register :a, A.new

data/lib/reverse_adoc/converters/aside.rb

@@ -1,11 +1,13 @@
 module ReverseAdoc
   module Converters
     class Aside < Base
+      def to_coradoc(node, state = {})
+        content = treat_children(node, state)
+        Coradoc::Element::Block::Side.new(lines: content.lines)
+      end
+
       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"
+        Coradoc::Generator.gen_adoc(to_coradoc(node, state))
       end
     end
 

data/lib/reverse_adoc/converters/audio.rb

@@ -1,31 +1,26 @@
 module ReverseAdoc
   module Converters
     class Audio < Base
-      def convert(node, state = {})
-        autoplay   = node['autoplay']
-        loop_attr   = node['loop']
-        controls   = node['controls']
-        src   = node['src']
-        id = node['id']
-        anchor = id ? "[[#{id}]]\n" : ""
+      def to_coradoc(node, _state = {})
+        src = node["src"]
+        id = node["id"]
         title = extract_title(node)
-        title = ".#{title}\n" unless title.empty?
-        [anchor, title, "audio::", src, "[", options(node), "]"].join("")
+        attributes = Coradoc::Element::AttributeList.new
+        options = options(node)
+        attributes.add_named("options", options) if options.any?
+        Coradoc::Element::Audio.new(title, id: id, src: src,
+                                           attributes: attributes)
+      end
+
+      def convert(node, state = {})
+        Coradoc::Generator.gen_adoc(to_coradoc(node, state))
       end
 
       def options(node)
-        autoplay   = node['autoplay']
-        loop_attr   = node['loop']
-        controls   = node['controls']
-        ret = ""
-        if autoplay || loop_attr || controls
-          out = []
-          out << "autoplay" if autoplay
-          out << "loop" if loop_attr
-          out << "controls" if controls
-          ret = %{options="#{out.join(',')}"}
-        end
-        ret
+        autoplay = node["autoplay"]
+        loop_attr = node["loop"]
+        controls = node["controls"]
+        [autoplay, loop_attr, controls].compact
       end
     end
 

data/lib/reverse_adoc/converters/base.rb

@@ -2,7 +2,7 @@ module ReverseAdoc
   module Converters
     class Base
       def treat_children(node, state)
-        node.children.inject('') do |memo, child|
+        node.children.inject("") do |memo, child|
           memo << treat(child, state)
         end
       end
@@ -11,15 +11,89 @@ module ReverseAdoc
         ReverseAdoc::Converters.lookup(node.name).convert(node, state)
       end
 
+      def treat_children_coradoc(node, state)
+        node.children.inject([]) do |memo, child|
+          memo << treat_coradoc(child, state)
+        end.flatten.reject { |x| x == "" || x.nil? }
+      end
+
+      def treat_coradoc(node, state)
+        ReverseAdoc::Converters.lookup(node.name).to_coradoc(node, state)
+      end
+
       def escape_keychars(string)
-        subs = { '*' => '\*', '_' => '\_' }
+        subs = { "*" => '\*', "_" => '\_' }
         string
-          .gsub(/((?<=\s)[\*_]+)|[\*_]+(?=\s)/) { |n| n.chars.map { |char| subs[char] }.join }
+          .gsub(/((?<=\s)[\*_]+)|[\*_]+(?=\s)/) do |n|
+          n.chars.map do |char|
+            subs[char]
+          end.join
+        end
       end
 
       def extract_title(node)
-        title = escape_keychars(node['title'].to_s)
-        title.empty? ? '' : %[ #{title}]
+        title = escape_keychars(node["title"].to_s)
+        title.empty? ? "" : %[ #{title}]
+      end
+
+      def node_has_ancestor?(node, name)
+        case name
+        when String
+          node.ancestors.map(&:name).include?(name)
+        when Array
+          (node.ancestors.map(&:name) & name).any?
+        end
+      end
+
+      def textnode_before_end_with?(node, str)
+        return nil if !str.is_a?(String) || str.empty?
+
+        node2 = node.at_xpath("preceding-sibling::node()[1]")
+        node2.respond_to?(:text) && node2.text.end_with?(str)
+      end
+
+      def unconstrained_before?(node)
+        before = node.at_xpath("preceding::node()[1]")
+
+        before &&
+          !before.text.strip.empty? &&
+          before.text[-1]&.match?(/\w/)
+      end
+
+      # TODO: This logic ought to be cleaned up.
+      def unconstrained_after?(node)
+        after = node.at_xpath("following::node()[1]")
+
+        after && !after.text.strip.empty? &&
+          after.text[0]&.match?(/\w|,|;|"|\.\?!/)
+      end
+
+      # def trailing_whitespace?(node)
+
+      # TODO: This logic ought to be cleaned up.
+      def constrained?(node)
+        before = node.at_xpath("preceding::node()[1]").to_s[-1]
+        before = if before
+                   before&.match?(/\s/) ? true : false
+                 else
+                   true
+                 end
+
+        if !before && (node.to_s[0] =~ /\s/)
+          before = true
+        end
+
+        after = node.at_xpath("following::node()[1]").to_s[0]
+        after = if after
+                  after&.match?(/\s|,|;|"|\.\?!/) ? true : false
+                else
+                  true
+                end
+        if !after && (node.to_s[-1] =~ /\s/)
+          after = true
+        end
+
+        before && after
       end
     end
   end

data/lib/reverse_adoc/converters/blockquote.rb

@@ -1,15 +1,24 @@
 module ReverseAdoc
   module Converters
     class Blockquote < Base
-      def convert(node, state = {})
-        id = node['id']
-        anchor = id ? "[[#{id}]]\n" : ""
-        cite = node['cite']
-        attrs = cite ? "[quote, #{cite}]\n" : ""
+      def to_coradoc(node, state = {})
+        node["id"]
+        cite = node["cite"]
+        attributes = if cite.nil?
+                       nil
+                     else
+                       Coradoc::Element::AttributeList.new(
+                         "quote", cite
+                       )
+                     end
         content = treat_children(node, state).strip
         content = ReverseAdoc.cleaner.remove_newlines(content)
-        #"\n\n> " << content.lines.to_a.join('> ') << "\n\n"
-        "\n\n#{attrs}____\n" << content.lines.to_a.join('') << "\n____\n\n"
+        Coradoc::Element::Block::Quote.new(nil, lines: content,
+                                                attributes: attributes)
+      end
+
+      def convert(node, state = {})
+        Coradoc::Generator.gen_adoc(to_coradoc(node, state))
       end
     end
 

data/lib/reverse_adoc/converters/br.rb

@@ -1,8 +1,12 @@
 module ReverseAdoc
   module Converters
     class Br < Base
+      def to_coradoc(_node, _state = {})
+        Coradoc::Element::Inline::HardLineBreak.new
+      end
+
       def convert(node, state = {})
-        " +\n"
+        Coradoc::Generator.gen_adoc(to_coradoc(node, state))
       end
     end