coradoc 0.1.0 → 0.3.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 1808574d50d22e93126d258859b0a7156015dae5e1830332b70a705643c3ab50
-  data.tar.gz: e1593143645bd24b419ad84f5961335de1351599fc9e9363d40c49ddb5a1d1b9
+  metadata.gz: c721a05ddb985a8bde74747c3c39e47818beca7525b13115794673394d46b2ab
+  data.tar.gz: f45600c3b4b85877a97904f38a03d5161b0ff83e7259ad6f54da5e1711ac9e46
 SHA512:
-  metadata.gz: a40e182cf07edd62e2af114fdbc752e759aef7b44a693cab17a6fba93869f999a5b2532f25e80146f30ed19b997414061b333836e4165045826e5c3853474403
-  data.tar.gz: 51c085c1a3437ea76897568b318a10ed7cc42f3998448903c0337500455fc4712ca869cc50ce1355bcb707461fef16c825f8c5a206fd86050f8b0295e82d198c
+  metadata.gz: 0063e9145b1a8e46046c7397b4278a480535ee119624842fad72a8e9280def9de4af7c8335dc5f64b5a118040a8e178af800b27879d8c335a00d37d574b34bbb
+  data.tar.gz: 8f02e79025a941ab47c8f359a7638129ce2e7a7aa80352a38bb9004540efdbadff8cd3c0427641004ab834a346f7d39676b9ace64e47f62dcbb9c0b0a357d2ce

data/.docker/Dockerfile

@@ -3,7 +3,7 @@ ARG RUBY_IMAGE=ruby:3.1.2-slim
 FROM ${RUBY_IMAGE}
 
 RUN apt-get update \
-    && apt-get install -y build-essential git \
+    && apt-get install -y build-essential git libreoffice \
     && apt-get clean && rm -rf /var/lib/apt/lists/*
 
 # install latest bundler

data/.docker/docker-compose.yml

@@ -7,8 +7,8 @@ services:
       dockerfile: ./.docker/Dockerfile
 
     volumes:
-      - .:/workspace
-      - bundle:/bundle
+      - .:/workspace:z
+      - bundle:/bundle:z
 
 volumes:
   bundle:

data/.editorconfig

@@ -0,0 +1,15 @@
+# EditorConfig is awesome: http://EditorConfig.org
+
+# top-most EditorConfig file
+root = true
+
+# Unix-style newlines with a newline ending every file
+[*]
+charset = utf-8
+end_of_line = lf
+
+[{*.adoc,*.html,*.js,*.json,*.rake,*.rb,*.rf,*.yaml,*.yml,Rakefile,rakefile}]
+indent_style = space
+indent_size = 2
+insert_final_newline = true
+trim_trailing_whitespace = true

data/.pryrc.sample

@@ -0,0 +1 @@
+doc = Coradoc::Parser.parse("spec/fixtures/sample.adoc")

data/CHANGELOG.md

@@ -1,5 +1,9 @@
 ## [Unreleased]
 
+## [0.3.0] - 2024-05-21
+
+- Merge reverse_adoc into coradoc
+
 ## [0.1.0] - 2023-01-08
 
 - Initial release

data/Rakefile

@@ -1,7 +1,17 @@
 # frozen_string_literal: true
 
 require "bundler/gem_tasks"
+
+if File.exist?(".codeclimate")
+  ENV["CODECLIMATE_REPO_TOKEN"] = File.read(".codeclimate").strip
+end
+
 require "rspec/core/rake_task"
 
 RSpec::Core::RakeTask.new(:spec)
 task default: :spec
+
+desc "Open an irb session preloaded with this library"
+task :console do
+  sh "irb -Ilib -rcoradoc -rcoradoc/reverse_adoc"
+end

data/coradoc.gemspec

@@ -28,11 +28,20 @@ Gem::Specification.new do |spec|
   spec.require_paths = ["lib"]
   spec.required_ruby_version = ">= 2.7.0"
 
-  spec.add_dependency "parslet"
+  spec.add_dependency "marcel", "~> 1.0.0"
+  spec.add_dependency "mathml2asciimath"
+  spec.add_dependency "nokogiri", "~> 1.13"
   spec.add_dependency "oscal", "~> 0.1.1"
-
+  spec.add_dependency "parslet"
+  spec.add_dependency "premailer", "~> 1.11.0"
+  spec.add_dependency "word-to-markdown"
+  spec.add_development_dependency "codeclimate-test-reporter"
   spec.add_development_dependency "pry"
   spec.add_development_dependency "rake"
+  spec.add_development_dependency "redcarpet"
   spec.add_development_dependency "rspec"
   spec.add_development_dependency "rubocop"
+  spec.add_development_dependency "rubocop-performance"
+  spec.add_development_dependency "simplecov"
+  # spec.add_runtime_dependency "thor"
 end

data/exe/reverse_adoc

@@ -0,0 +1,70 @@
+#!/usr/bin/env ruby
+# Usage: reverse_adoc [FILE]...
+# Usage: cat FILE | reverse_adoc
+require "rubygems"
+require "bundler/setup"
+
+require "coradoc/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|
+    Coradoc::ReverseAdoc.config.mathml2asciimath = true
+  end
+
+  opts.on("-oFILENAME", "--output=FILENAME", "Output file to write to") do |v|
+    Coradoc::ReverseAdoc.config.destination = File.expand_path(v)
+    # puts "output goes to #{Coradoc::ReverseAdoc.config.destination}"
+  end
+
+  opts.on("-e", "--external-images", "Export images if data URI") do |_v|
+    Coradoc::ReverseAdoc.config.external_images = true
+  end
+
+  opts.on("-u", "--unknown_tags [pass_through, drop, bypass, raise]",
+          "Unknown tag handling (default: pass_through)") do |v|
+    Coradoc::ReverseAdoc.config.unknown_tags = v
+  end
+
+  opts.on("-v", "--version", "Version information") do |_v|
+    puts "reverse_adoc: v#{Coradoc::ReverseAdoc::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)
+  Coradoc::ReverseAdoc.config.sourcedir = File.dirname(File.expand_path(filename))
+else
+  if Coradoc::ReverseAdoc.config.external_images
+    raise "The -e | --external-images feature cannot be used with STDIN input. Exiting."
+  end
+
+  input_content = ARGF.read
+end
+
+if Coradoc::ReverseAdoc.config.external_images && Coradoc::ReverseAdoc.config.destination.nil?
+  raise "The -e | --external-images feature must be used with -o | --output. Exiting."
+end
+
+# Read from STDIN
+adoc_content = Coradoc::ReverseAdoc.convert(input_content)
+
+# Print to STDOUT
+unless Coradoc::ReverseAdoc.config.destination
+  puts adoc_content
+  exit
+end
+
+# Write output to Coradoc::ReverseAdoc.config.destination
+FileUtils.mkdir_p(File.dirname(Coradoc::ReverseAdoc.config.destination))
+File.open(Coradoc::ReverseAdoc.config.destination, "w") do |file|
+  file.write(adoc_content)
+end

data/exe/w2a

@@ -0,0 +1,72 @@
+#!/usr/bin/env ruby
+# frozen_string_literal: true
+
+require "rubygems"
+require "bundler/setup"
+
+require "word-to-markdown"
+require "optparse"
+require "coradoc/reverse_adoc"
+
+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|
+    Coradoc::ReverseAdoc.config.mathml2asciimath = true
+  end
+
+  opts.on("-oFILENAME", "--output=FILENAME", "Output file to write to") do |v|
+    Coradoc::ReverseAdoc.config.destination = File.expand_path(v)
+    # puts "output goes to #{Coradoc::ReverseAdoc.config.destination}"
+  end
+
+  opts.on("-e", "--external-images", "Export images if data URI") do |_v|
+    Coradoc::ReverseAdoc.config.external_images = true
+  end
+
+  opts.on("-v", "--version", "Version information") do |_v|
+    puts "reverse_adoc: v#{Coradoc::ReverseAdoc::VERSION}"
+    puts "[dependency] WordToMarkdown: v#{WordToMarkdown::VERSION}"
+    if Gem.win_platform?
+      puts "[dependency] LibreOffice: version not available on Windows"
+    else
+      puts "[dependency] LibreOffice: v#{WordToMarkdown.soffice.version}"
+    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 Coradoc::ReverseAdoc.config.external_images && Coradoc::ReverseAdoc.config.destination.nil?
+  raise "The -e | --external-images feature must be used with -o | --output. Exiting."
+end
+
+Coradoc::ReverseAdoc.config.sourcedir = Dir.mktmpdir
+
+doc = WordToMarkdown.new(filename, Coradoc::ReverseAdoc.config.sourcedir)
+# File.open("test.html", "w:UTF-8") { |f| f.write doc.document.html }
+adoc_content = Coradoc::ReverseAdoc.convert(
+  Coradoc::ReverseAdoc.cleaner.preprocess_word_html(doc.document.html),
+  WordToMarkdown::REVERSE_MARKDOWN_OPTIONS,
+)
+# puts scrub_whitespace(doc.document.html)
+
+# Print to STDOUT
+unless Coradoc::ReverseAdoc.config.destination
+  puts adoc_content
+  exit
+end
+
+# Write output to Coradoc::ReverseAdoc.config.destination
+FileUtils.mkdir_p(File.dirname(Coradoc::ReverseAdoc.config.destination))
+File.open(Coradoc::ReverseAdoc.config.destination, "w") do |file|
+  file.write(adoc_content)
+end

data/lib/coradoc/document.rb

@@ -1,22 +1,26 @@
-require "coradoc/document/title"
-require "coradoc/document/block"
-require "coradoc/document/section"
-require "coradoc/document/attribute"
-require "coradoc/document/admonition"
-require "coradoc/document/text_element"
-require "coradoc/document/author"
-require "coradoc/document/revision"
-require "coradoc/document/header"
-require "coradoc/document/bibdata"
-require "coradoc/document/paragraph"
-require "coradoc/document/table"
-require "coradoc/document/list"
+require_relative "element/title"
+require_relative "element/block"
+require_relative "element/section"
+require_relative "element/attribute"
+require_relative "element/attribute_list"
+require_relative "element/admonition"
+require_relative "element/text_element"
+require_relative "element/author"
+require_relative "element/revision"
+require_relative "element/header"
+require_relative "element/document_attributes"
+require_relative "element/paragraph"
+require_relative "element/table"
+require_relative "element/list"
+require_relative "element/inline"
+require_relative "element/image"
+require_relative "element/audio"
+require_relative "element/video"
+require_relative "element/break"
 
 module Coradoc
-  module Document
+  class Document
     class << self
-      attr_reader :header, :bibdata, :sections
-
       def from_adoc(filename)
         ast = Coradoc::Parser.parse(filename)
         Coradoc::Transformer.transform(ast)
@@ -26,21 +30,34 @@ module Coradoc
         @sections = []
 
         elements.each do |element|
-          if element.is_a?(Coradoc::Document::Bibdata)
-            @bibdata = element
-          end
+          case element
+          when Coradoc::Element::DocumentAttributes
+            @document_attributes = element
 
-          if element.is_a?(Coradoc::Document::Header)
+          when Coradoc::Element::Header
             @header = element
-          end
 
-          if element.is_a?(Coradoc::Document::Section)
+          when Coradoc::Element::Section
             @sections << element
           end
         end
 
-        self
+        new(
+          document_attributes: @document_attributes,
+          header: @header,
+          sections: @sections,
+        )
       end
     end
+
+    attr_accessor :header, :document_attributes, :sections
+
+    def initialize(options = {})
+      @document_attributes = options.fetch(:document_attributes,
+                                           Coradoc::Element::DocumentAttributes.new)
+      @header = options.fetch(:header, Coradoc::Element::Header.new(""))
+      @sections = options.fetch(:sections, [])
+      self
+    end
   end
 end

data/lib/coradoc/element/admonition.rb

@@ -0,0 +1,13 @@
+module Coradoc
+  module Element
+    class Admonition
+      attr_reader :type, :content, :line_break
+
+      def initialize(content, type, options = {})
+        @content = content
+        @type = type.downcase.to_sym
+        @line_break = options.fetch(:line_break, "")
+      end
+    end
+  end
+end

data/lib/coradoc/{document → element}/attribute.rb

@@ -1,5 +1,5 @@
 module Coradoc
-  module Document
+  module Element
     class Attribute
       attr_reader :key, :value
 

data/lib/coradoc/element/attribute_list.rb

@@ -0,0 +1,46 @@
+module Coradoc
+  module Element
+    class AttributeList
+      attr_reader :positional, :named
+
+      def initialize(*positional, **named)
+        @positional = positional || []
+        @named = named || {}
+      end
+
+      def add_positional(attr)
+        @positional << attr
+      end
+
+      def add_named(name, value)
+        @named[name] = value
+      end
+
+      def empty?
+        @positional.empty? && @named.empty?
+      end
+
+      def to_adoc
+        return "[]" if [@positional, @named].all?(:empty?)
+
+        adoc = ""
+        adoc << @positional.join(",") if @positional.any?
+        adoc << "," if @positional.any? && @named.any?
+        adoc << @named.map do |k, v|
+          if v.is_a?(String)
+            v2 = v.to_s
+            v2 = v2.include?("\"") ? v2.gsub("\"", "\\\"") : v2
+            if v2.include?(" ") || v2.include?(",") || v2.include?("\"")
+              v2 = "\"#{v2}\""
+            end
+          elsif v.is_a?(Array)
+            v2 = "\"#{v.join(',')}\""
+          end
+          [k.to_s, "=", v2].join
+        end.join(",")
+        adoc = "[#{adoc}]" if @positional.any? || @named.any?
+        adoc
+      end
+    end
+  end
+end

data/lib/coradoc/element/audio.rb

@@ -0,0 +1,22 @@
+module Coradoc
+  module Element
+    class Audio
+      attr_reader :id, :title, :src, :options, :anchor
+
+      def initialize(title, options = {})
+        @title = title
+        @id = options.fetch(:id, nil)
+        @anchor = Inline::Anchor.new(@id) if @id
+        @src = options.fetch(:src, "")
+        @attributes = options.fetch(:attributes, [])
+      end
+
+      def to_adoc
+        anchor = @anchor.nil? ? "" : "#{@anchor.to_adoc}\n"
+        title = ".#{@title}\n" unless @title.empty?
+        attrs = @attributes.empty? ? "\[\]" : @attributes.to_adoc
+        [anchor, title, "audio::", @src, attrs].join("")
+      end
+    end
+  end
+end

data/lib/coradoc/element/author.rb

@@ -0,0 +1,22 @@
+module Coradoc
+  module Element
+    class Author
+      attr_reader :email, :last_name, :first_name
+
+      def initialize(first_name, last_name, email, middle_name = nil)
+        @first_name = first_name
+        @last_name = last_name
+        @email = email
+        @middle_name = middle_name
+      end
+
+      def to_adoc
+        adoc = "#{@first_name} "
+        adoc << "#{@middle_name} " if @middle_name
+        adoc << @last_name.to_s
+        adoc << " <#{@email}>" if @email
+        adoc
+      end
+    end
+  end
+end

data/lib/coradoc/element/base.rb

@@ -0,0 +1,15 @@
+module Coradoc
+  module Element
+    class Base
+      # attr_reader :document_attributes
+
+      # def initialize(asciidoc)
+      #   @document_attributes = extract_document_attributes(asciidoc)
+      # end
+
+      # def extract_document_attributes(asciidoc)
+      #   @document_attributes ||= DocumentAttributes.new(asciidoc.attributes)
+      # end
+    end
+  end
+end

data/lib/coradoc/element/block/core.rb

@@ -0,0 +1,69 @@
+require_relative "../inline/anchor"
+
+module Coradoc
+  module Element
+    module Block
+      class Core
+        attr_reader :title, :lines, :attributes, :lang, :id
+
+        def initialize(title, options = {})
+          @title = title
+          @lines = options.fetch(:lines, [])
+          @type_str = options.fetch(:type, nil)
+          @delimiter = options.fetch(:delimiter, "")
+          @attributes = options.fetch(:attributes, {})
+          @lang = options.fetch(:lang, nil)
+          @id = options.fetch(:id, nil)
+          @anchor = @id.nil? ? nil : Inline::Anchor.new(@id)
+        end
+
+        def type
+          @type ||= defined_type || type_from_delimiter
+        end
+
+        def gen_anchor
+          @anchor.nil? ? "" : "#{@anchor.to_adoc}\n"
+        end
+
+        def gen_title
+          t = Coradoc::Generator.gen_adoc(@title)
+          return "" if t.empty?
+
+          ".#{t}\n"
+        end
+
+        def gen_attributes
+          @attributes.nil? ? "" : "#{@attributes.to_adoc}\n"
+        end
+
+        def gen_delimiter
+          @delimiter_char * @delimiter_len
+        end
+
+        def gen_lines
+          Coradoc::Generator.gen_adoc(@lines)
+        end
+
+        private
+
+        def defined_type
+          @type_str&.to_s&.to_sym
+        end
+
+        def type_from_delimiter
+          type_hash.fetch(@delimiter, nil)
+        end
+
+        def type_hash
+          @type_hash ||= {
+            "____" => :quote,
+            "****" => :side,
+            "----" => :source,
+            "====" => :example,
+            "...." => :literal,
+          }
+        end
+      end
+    end
+  end
+end

data/lib/coradoc/element/block/example.rb

@@ -0,0 +1,22 @@
+require_relative "core"
+
+module Coradoc
+  module Element
+    module Block
+      class Example < Core
+        def initialize(title, options = {})
+          @title = title
+          @id = options.fetch(:id, nil)
+          @anchor = @id.nil? ? nil : Inline::Anchor.new(@id)
+          @lines = options.fetch(:lines, [])
+          @delimiter_char = "="
+          @delimiter_len = options.fetch(:delimiter_len, 4)
+        end
+
+        def to_adoc
+          "\n\n#{gen_anchor}#{gen_title}#{gen_delimiter}\n" << gen_lines << "\n#{gen_delimiter}\n\n"
+        end
+      end
+    end
+  end
+end

data/lib/coradoc/element/block/literal.rb

@@ -0,0 +1,19 @@
+module Coradoc
+  module Element
+    module Block
+      class Literal < Core
+        def initialize(_title, options = {})
+          @id = options.fetch(:id, nil)
+          @anchor = @id.nil? ? nil : Inline::Anchor.new(@id)
+          @lines = options.fetch(:lines, [])
+          @delimiter_char = "."
+          @delimiter_len = options.fetch(:delimiter_len, 4)
+        end
+
+        def to_adoc
+          "\n\n#{gen_anchor}#{gen_delimiter}\n" << gen_lines << "\n#{gen_delimiter}\n\n"
+        end
+      end
+    end
+  end
+end

data/lib/coradoc/element/block/quote.rb

@@ -0,0 +1,19 @@
+module Coradoc
+  module Element
+    module Block
+      class Quote < Core
+        def initialize(title, options = {})
+          @title = title
+          @attributes = options.fetch(:attributes, {})
+          @lines = options.fetch(:lines, [])
+          @delimiter_char = "_"
+          @delimiter_len = options.fetch(:delimiter_len, 4)
+        end
+
+        def to_adoc
+          "\n\n#{gen_attributes}#{gen_delimiter}\n" << gen_lines << "\n#{gen_delimiter}\n\n"
+        end
+      end
+    end
+  end
+end

data/lib/coradoc/element/block/side.rb

@@ -0,0 +1,17 @@
+module Coradoc
+  module Element
+    module Block
+      class Side < Core
+        def initialize(options = {})
+          @lines = options.fetch(:lines, [])
+          @delimiter_char = "*"
+          @delimiter_len = options.fetch(:delimiter_len, 4)
+        end
+
+        def to_adoc
+          "\n\n#{gen_delimiter}" << gen_lines << "\n#{gen_delimiter}\n\n"
+        end
+      end
+    end
+  end
+end

data/lib/coradoc/element/block/sourcecode.rb

@@ -0,0 +1,20 @@
+module Coradoc
+  module Element
+    module Block
+      class SourceCode < Core
+        def initialize(_title, options = {})
+          @id = options.fetch(:id, nil)
+          @anchor = @id.nil? ? nil : Inline::Anchor.new(@id)
+          @lang = options.fetch(:lang, "")
+          @lines = options.fetch(:lines, [])
+          @delimiter_char = "-"
+          @delimiter_len = options.fetch(:delimiter_len, 4)
+        end
+
+        def to_adoc
+          "\n\n#{gen_anchor}[source,#{@lang}]\n#{gen_delimiter}\n" << gen_lines << "\n#{gen_delimiter}\n\n"
+        end
+      end
+    end
+  end
+end

data/lib/coradoc/element/block.rb

@@ -0,0 +1,13 @@
+module Coradoc
+  module Element
+    module Block
+    end
+  end
+end
+
+require_relative "block/core"
+require_relative "block/example"
+require_relative "block/literal"
+require_relative "block/quote"
+require_relative "block/side"
+require_relative "block/sourcecode"

data/lib/coradoc/element/break.rb

@@ -0,0 +1,11 @@
+module Coradoc
+  module Element
+    module Break
+      class ThematicBreak
+        def to_adoc
+          "\n* * *\n"
+        end
+      end
+    end
+  end
+end