coradoc 0.2.0 → 1.0.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: '09284ec8117d3052ad45275f82328e45e600fabc4f86379093af0dbe4ee9ebc4'
-  data.tar.gz: 8c9e2cef678a473459b091d05c303110a0caca9179d5ea89e1f4a07608d7bb9c
+  metadata.gz: 71bfd2ac723d58ca75b591bceabcdf834abd9cbbf32e54b96352da21b8f15095
+  data.tar.gz: 1a7ea6bf80d64ab4c1349c948aa2aee9ed704dc36ddd07fb7cce78293dca86cd
 SHA512:
-  metadata.gz: 35510139cf9cd1cf0eedaaf12b0bc80709628660d88e0178f3651d7f213c870abf68eb04de6f23a8056af4681875f7ba59804228937bc65bfb6550753a4fb769
-  data.tar.gz: 714c0f79bdd9bb05d58de71c6ab5bcdedddcd48a784fad4b019e168078c221ee6e477c573822a6a8f2db328fa02da5a0b9ef60542f239adb0b109c40ff309a66
+  metadata.gz: 555800621a06ffafc07e03a5c2cdca1210de6369d524a606de51df78b1ee9b19aaba30d32d8a448347237e488362a59677781a02502a1d8891fb89aa1cc267f5
+  data.tar.gz: 22707b18c6ed99fb4c6127885d871875ed1fed2906d2d10dcdd5f4ed9bdd1b1bfe3f19fd092768ccc4d488bdb7e705aba9ba7ab948ceace9cc679f1ee733eddc

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/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/README.md

@@ -66,4 +66,8 @@ Coradoc::Parser.parse(sample_asciidoc)
 
 This interface will return the abstract syntax tree.
 
+### Converting from HTML to AsciiDoc (reverse_adoc)
+
+See: [reverse_adoc README](https://github.com/metanorma/coradoc/blob/main/lib/coradoc/reverse_adoc/README.adoc)
+
 [sandi-metz]: http://robots.thoughtbot.com/post/50655960596/sandi-metz-rules-for-developers

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 +1,91 @@
+#!/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("-r", "--require RUBYMODULE", "Require additional Ruby file") do |v|
+    require v
+  end
+
+  opts.on("--track-time", "Track time spent on each step") do
+    Coradoc::ReverseAdoc.config.track_time = true
+  end
+
+  opts.on("--split-sections LEVEL", "Split sections up to LEVEL") do |i|
+    Coradoc::ReverseAdoc.config.split_sections = i.to_i
+  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
+
+if Coradoc::ReverseAdoc.config.split_sections && Coradoc::ReverseAdoc.config.destination.nil?
+  raise "The --split_sections 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
+adoc_content = {nil => adoc_content} unless adoc_content.is_a? Hash
+
+adoc_content.each do |file, content|
+  destination = Coradoc::ReverseAdoc.config.destination
+  destdir = File.dirname(destination)
+  filename = file ? "#{destdir}/#{file}" : destination
+  FileUtils.mkdir_p(File.dirname(filename))
+  File.write(filename, 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

@@ -1,3 +1,4 @@
+require_relative "element/base"
 require_relative "element/title"
 require_relative "element/block"
 require_relative "element/section"
@@ -20,7 +21,6 @@ require_relative "element/break"
 
 module Coradoc
   class Document
-
     class << self
       def from_adoc(filename)
         ast = Coradoc::Parser.parse(filename)
@@ -43,22 +43,22 @@ module Coradoc
           end
         end
 
-        self.new(
+        new(
           document_attributes: @document_attributes,
           header: @header,
-          sections: @sections
+          sections: @sections,
         )
       end
     end
 
     attr_accessor :header, :document_attributes, :sections
 
-    def initialize(options={})
-      @document_attributes = options.fetch(:document_attributes, Coradoc::Element::DocumentAttributes.new)
+    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

@@ -1,11 +1,13 @@
 module Coradoc
-  class Element::Admonition
-    attr_reader :type, :content, :line_break
+  module Element
+    class Admonition < Base
+      attr_accessor :type, :content, :line_break
 
-    def initialize(content, type, options = {})
-      @content = content
-      @type = type.downcase.to_sym
-      @line_break = options.fetch(: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/element/attribute.rb

@@ -1,7 +1,7 @@
 module Coradoc
   module Element
-    class Attribute
-      attr_reader :key, :value
+    class Attribute < Base
+      attr_accessor :key, :value
 
       def initialize(key, value, _options = {})
         @key = key.to_s

data/lib/coradoc/element/attribute_list.rb

@@ -1,45 +1,124 @@
 module Coradoc
   module Element
-    class AttributeList
-      attr_reader :positional, :named
+    class AttributeList < Base
+      attr_accessor :positional, :named, :rejected_positional, :rejected_named
+
+      declare_children :positional, :named
 
       def initialize(*positional, **named)
         @positional = positional || []
         @named = named || {}
+        @rejected_positional = []
+        @rejected_named = []
       end
 
-      def add_positional(attr)
-        @positional << attr
+      def add_positional(*attr)
+        @positional += attr
       end
 
       def add_named(name, value)
         @named[name] = value
       end
 
+      def any?
+        !empty?
+      end
+
       def empty?
         @positional.empty? && @named.empty?
       end
 
-      def to_adoc
+      def validate_attr(attr, matcher)
+        matcher === attr
+      end
+
+      def validate_positional(validators)
+        @positional.each_with_index do |value, i|
+          # TODO: Decide what to do with this value
+          _positional_name = validators[i][0]
+
+          validator = validators[i][1]
+
+          unless validator && validate_attr(value, validator)
+            @positional[i] = nil
+            @rejected_positional << [i, value]
+          end
+        end
+
+        @positional.pop while !@positional.empty? && @positional.last.nil?
+      end
+
+      def validate_named(validators)
+        @named.each_with_index do |(name, value), i|
+          name = name.to_sym
+          validator = validators[name]
+
+          unless validator && validate_attr(value, validator)
+            @named.delete(name)
+            @rejected_named << [name, value]
+          end
+        end
+      end
+
+      def to_adoc(show_empty = true)
         return "[]" if [@positional, @named].all?(:empty?)
 
-        adoc = ""
-        adoc << @positional.join(",") if @positional.any?
+        adoc = +""
+        if !@positional.empty?
+          adoc << @positional.map { |p| [nil, ""].include?(p) ? '""' : p }.join(",")
+        end
         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}\""
+            v = v.gsub("\"", "\\\"")
+            if v.include?(" ") || v.include?(",") || v.include?('"')
+              v = "\"#{v}\""
             end
           elsif v.is_a?(Array)
-            v2 = "\"#{v.join(",")}\""
+            v = "\"#{v.join(',')}\""
           end
-          [k.to_s, "=", v2].join
+          [k.to_s, "=", v].join
         end.join(",")
-        adoc = "[#{adoc}]" if @positional.any? || @named.any?
-        adoc
+
+        if !empty? || (empty? && show_empty)
+          "[#{adoc}]"
+        elsif empty? && !show_empty
+          adoc
+        end
+      end
+
+      module Matchers
+        def one(*args)
+          One.new(*args)
+        end
+
+        class One
+          def initialize(*possibilities)
+            @possibilities = possibilities
+          end
+
+          def ===(other)
+            @possibilities.any? { |i| i === other }
+          end
+        end
+
+        def many(*args)
+          Many.new(*args)
+        end
+
+        # TODO: Find a way to only reject some values but not all?
+        class Many
+          def initialize(*possibilities)
+            @possibilities = possibilities
+          end
+
+          def ===(other)
+            other = other.split(",") if other.is_a?(String)
+
+            other.is_a?(Array) &&
+              other.all? { |i| @possibilities.any? { |p| p === i } }
+          end
+        end
       end
     end
   end

data/lib/coradoc/element/audio.rb

@@ -1,13 +1,15 @@
 module Coradoc
   module Element
-    class Audio
-      attr_reader :id, :title, :src, :options, :anchor
+    class Audio < Base
+      attr_accessor :id, :title, :src, :options, :anchor, :attributes
+
+      declare_children :id, :title, :anchor, :attributes
 
       def initialize(title, options = {})
         @title = title
         @id = options.fetch(:id, nil)
         @anchor = Inline::Anchor.new(@id) if @id
-        @src = options.fetch(:src, '')
+        @src = options.fetch(:src, "")
         @attributes = options.fetch(:attributes, [])
       end
 
@@ -17,6 +19,15 @@ module Coradoc
         attrs = @attributes.empty? ? "\[\]" : @attributes.to_adoc
         [anchor, title, "audio::", @src, attrs].join("")
       end
+
+      extend AttributeList::Matchers
+      VALIDATORS_NAMED = {
+        title: String,
+        start: Integer,
+        end: Integer,
+        options: many("nofollow", "noopener", "inline", "interactive"),
+        opts: many("nofollow", "noopener", "inline", "interactive"),
+      }
     end
   end
 end

data/lib/coradoc/element/author.rb

@@ -1,20 +1,24 @@
 module Coradoc
-  class Element::Author
-    attr_reader :email, :last_name, :first_name
+  module Element
+    class Author < Base
+      attr_accessor :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
+      declare_children :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}"
-      adoc << " <#{@email}>" if @email
-      adoc
+      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

@@ -1,17 +1,78 @@
 module Coradoc
   module Element
     class Base
-      # attr_reader :document_attributes
+      # The idea here, is that HTML content generators may often introduce
+      # a lot of unnecessary markup, that only makes sense in the HTML+CSS
+      # context. The idea is that certain cases can be simplified, making it
+      # so that the result is equivalent, but much simpler, allowing us to
+      # generate a nicer AsciiDoc syntax for those cases.
+      def simplify_block_content(content)
+        content = Array(content)
+        collected_content = []
+        content.each do |i|
+          case i
+          when Coradoc::Element::Section
+            return content unless i.safe_to_collapse?
 
-      # def initialize(asciidoc)
-      #   @document_attributes = extract_document_attributes(asciidoc)
-      # end
+            simplified = simplify_block_content(i.contents)
 
-      private
+            if simplified && !simplified.empty?
+              collected_content << simplified
+            end
+          else
+            collected_content << i
+          end
+        end
 
-      # def extract_document_attributes(asciidoc)
-      #   @document_attributes ||= DocumentAttributes.new(asciidoc.attributes)
-      # end
+        collected_content = collected_content.compact
+
+        # We can safely do this optimization only if there's just one other
+        # element inside this structure.
+        if collected_content.length <= 1
+          collected_content
+        else
+          content
+        end
+      end
+
+      def self.declare_children(*children)
+        @children = (@children || []).dup + children
+      end
+
+      def self.visit(element, &block)
+        element = yield element, :pre
+        element = if element.respond_to? :visit
+                    element.visit(&block)
+                  elsif element.is_a? Array
+                    element.map { |child| visit(child, &block) }.flatten.compact
+                  elsif element.is_a? Hash
+                    element.to_h do |k, v|
+                      [visit(k, &block), visit(v, &block)]
+                    end
+                  else
+                    element
+                  end
+        yield element, :post
+      end
+
+      def self.children_accessors
+        @children || []
+      end
+
+      def children_accessors
+        self.class.children_accessors
+      end
+
+      def visit(&block)
+        children_accessors.each do |accessor|
+          child = public_send(accessor)
+          result = self.class.visit(child, &block)
+          if result != child
+            public_send(:"#{accessor}=", result)
+          end
+        end
+        self
+      end
     end
   end
 end