coradoc 0.3.0 → 1.0.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: c721a05ddb985a8bde74747c3c39e47818beca7525b13115794673394d46b2ab
-  data.tar.gz: f45600c3b4b85877a97904f38a03d5161b0ff83e7259ad6f54da5e1711ac9e46
+  metadata.gz: 71bfd2ac723d58ca75b591bceabcdf834abd9cbbf32e54b96352da21b8f15095
+  data.tar.gz: 1a7ea6bf80d64ab4c1349c948aa2aee9ed704dc36ddd07fb7cce78293dca86cd
 SHA512:
-  metadata.gz: 0063e9145b1a8e46046c7397b4278a480535ee119624842fad72a8e9280def9de4af7c8335dc5f64b5a118040a8e178af800b27879d8c335a00d37d574b34bbb
-  data.tar.gz: 8f02e79025a941ab47c8f359a7638129ce2e7a7aa80352a38bb9004540efdbadff8cd3c0427641004ab834a346f7d39676b9ace64e47f62dcbb9c0b0a357d2ce
+  metadata.gz: 555800621a06ffafc07e03a5c2cdca1210de6369d524a606de51df78b1ee9b19aaba30d32d8a448347237e488362a59677781a02502a1d8891fb89aa1cc267f5
+  data.tar.gz: 22707b18c6ed99fb4c6127885d871875ed1fed2906d2d10dcdd5f4ed9bdd1b1bfe3f19fd092768ccc4d488bdb7e705aba9ba7ab948ceace9cc679f1ee733eddc

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/exe/reverse_adoc

@@ -28,6 +28,18 @@ OptionParser.new do |opts|
     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
@@ -54,6 +66,10 @@ if Coradoc::ReverseAdoc.config.external_images && Coradoc::ReverseAdoc.config.de
   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)
 
@@ -64,7 +80,12 @@ unless Coradoc::ReverseAdoc.config.destination
 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)
+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/lib/coradoc/document.rb

@@ -1,3 +1,4 @@
+require_relative "element/base"
 require_relative "element/title"
 require_relative "element/block"
 require_relative "element/section"

data/lib/coradoc/element/admonition.rb

@@ -1,7 +1,7 @@
 module Coradoc
   module Element
-    class Admonition
-      attr_reader :type, :content, :line_break
+    class Admonition < Base
+      attr_accessor :type, :content, :line_break
 
       def initialize(content, type, options = {})
         @content = content

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

@@ -1,7 +1,9 @@
 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
@@ -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

@@ -1,7 +1,9 @@
 module Coradoc
   module Element
-    class Author
-      attr_reader :email, :last_name, :first_name
+    class Author < Base
+      attr_accessor :email, :last_name, :first_name
+
+      declare_children :email, :last_name, :first_name
 
       def initialize(first_name, last_name, email, middle_name = nil)
         @first_name = first_name

data/lib/coradoc/element/base.rb

@@ -1,15 +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)
 
-      # def extract_document_attributes(asciidoc)
-      #   @document_attributes ||= DocumentAttributes.new(asciidoc.attributes)
-      # end
+            if simplified && !simplified.empty?
+              collected_content << simplified
+            end
+          else
+            collected_content << i
+          end
+        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

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

@@ -3,15 +3,17 @@ require_relative "../inline/anchor"
 module Coradoc
   module Element
     module Block
-      class Core
-        attr_reader :title, :lines, :attributes, :lang, :id
+      class Core < Base
+        attr_accessor :title, :lines, :attributes, :lang, :id
+
+        declare_children :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, {})
+          @attributes = options.fetch(:attributes, AttributeList.new)
           @lang = options.fetch(:lang, nil)
           @id = options.fetch(:id, nil)
           @anchor = @id.nil? ? nil : Inline::Anchor.new(@id)
@@ -33,7 +35,9 @@ module Coradoc
         end
 
         def gen_attributes
-          @attributes.nil? ? "" : "#{@attributes.to_adoc}\n"
+          attrs = @attributes.to_adoc(false)
+          return "#{attrs}\n" if !attrs.empty?
+          ""
         end
 
         def gen_delimiter

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

@@ -4,7 +4,7 @@ module Coradoc
       class Quote < Core
         def initialize(title, options = {})
           @title = title
-          @attributes = options.fetch(:attributes, {})
+          @attributes = options.fetch(:attributes, AttributeList.new)
           @lines = options.fetch(:lines, [])
           @delimiter_char = "_"
           @delimiter_len = options.fetch(:delimiter_len, 4)

data/lib/coradoc/element/break.rb

@@ -1,7 +1,7 @@
 module Coradoc
   module Element
     module Break
-      class ThematicBreak
+      class ThematicBreak < Base
         def to_adoc
           "\n* * *\n"
         end

data/lib/coradoc/element/document_attributes.rb

@@ -1,7 +1,9 @@
 module Coradoc
   module Element
-    class DocumentAttributes
-      attr_reader :data
+    class DocumentAttributes < Base
+      attr_accessor :data
+
+      declare_children :data
 
       def initialize(data = {}, options = {})
         @data = data
@@ -9,10 +11,8 @@ module Coradoc
       end
 
       def to_hash
-        Hash.new.tap do |hash|
-          data.each do |attribute|
-            hash[attribute.key.to_s] = attribute.value.to_s.gsub("'", "")
-          end
+        data.to_h do |attribute|
+          [attribute.key.to_s, attribute.value.to_s.gsub("'", "")]
         end
       end
     end

data/lib/coradoc/element/header.rb

@@ -1,7 +1,9 @@
 module Coradoc
   module Element
-    class Header
-      attr_reader :title, :author, :revision
+    class Header < Base
+      attr_accessor :title, :author, :revision
+
+      declare_children :title
 
       def initialize(title, options = {})
         @title = title

data/lib/coradoc/element/image/block_image.rb

@@ -2,10 +2,21 @@ module Coradoc
   module Element
     module Image
       class BlockImage < Core
-        def initialize(title, id, src, options = ())
-          super(title, id, src, options)
+        def initialize(title, id, src, options = {})
+          super
           @colons = "::"
         end
+
+        def validate_named
+          @attributes.validate_named(VALIDATORS_NAMED, VALIDATORS_NAMED_BLOCK)
+        end
+
+        extend AttributeList::Matchers
+        VALIDATORS_NAMED_BLOCK = {
+          caption: String,
+          align: one("left", "center", "right"),
+          float: one("left", "right"),
+        }
       end
     end
   end

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

@@ -1,24 +1,53 @@
 module Coradoc
   module Element
     module Image
-      class Core
-        attr_reader :title, :id, :src, :attributes
+      class Core < Base
+        attr_accessor :title, :id, :src, :attributes
 
-        def initialize(title, id, src, options = ())
+        declare_children :id, :src, :title, :attributes
+
+        def initialize(title, id, src, options = {})
           @title = title
           @id = id
           @anchor = @id.nil? ? nil : Coradoc::Element::Inline::Anchor.new(@id)
           @src = src
-          @attributes = options.fetch(:attributes, [])
+          @attributes = options.fetch(:attributes, AttributeList.new)
           @title = options.fetch(:title, nil)
+          if @attributes.any?
+            @attributes.validate_positional(VALIDATORS_POSITIONAL)
+            @attributes.validate_named(VALIDATORS_NAMED)
+          end
         end
 
         def to_adoc
           anchor = @anchor.nil? ? "" : "#{@anchor.to_adoc}\n"
           title = ".#{@title}\n" unless @title.to_s.empty?
-          attrs = @attributes.empty? ? "\[\]" : @attributes.to_adoc
+          attrs = @attributes.to_adoc
           [anchor, title, "image", @colons, @src, attrs].join("")
         end
+
+        extend AttributeList::Matchers
+        VALIDATORS_POSITIONAL = [
+          [:alt, String],
+          [:width, Integer],
+          [:height, Integer],
+        ]
+
+        VALIDATORS_NAMED = {
+          id: String,
+          alt: String,
+          fallback: String,
+          title: String,
+          width: Integer,
+          height: Integer,
+          link: String, # change to that URI regexp
+          window: String,
+          scale: Integer,
+          scaledwidth: /\A[0-9]{1,2}%\z/,
+          pdfwidth: /\A[0-9]+vw\z/,
+          role: many(/.*/, "left", "right", "th", "thumb", "related", "rel"),
+          opts: many("nofollow", "noopener", "inline", "interactive"),
+        }
       end
     end
   end

data/lib/coradoc/element/image/inline_image.rb

@@ -2,8 +2,8 @@ module Coradoc
   module Element
     module Image
       class InlineImage < Core
-        def initialize(title, id, src, options = ())
-          super(title, id, src, options)
+        def initialize(title, id, src, options = {})
+          super
           @colons = ":"
         end
       end

data/lib/coradoc/element/inline/anchor.rb

@@ -1,8 +1,10 @@
 module Coradoc
   module Element
     module Inline
-      class Anchor
-        attr_reader :id
+      class Anchor < Base
+        attr_accessor :id
+
+        declare_children :id
 
         def initialize(id)
           @id = id

data/lib/coradoc/element/inline/bold.rb

@@ -1,18 +1,23 @@
 module Coradoc
   module Element
     module Inline
-      class Bold
+      class Bold < Base
         attr_accessor :content, :unconstrained
 
-        def initialize(content, unconstrained = true)
+        declare_children :content
+
+        def initialize(content, unconstrained: true)
           @content = content
           @unconstrained = unconstrained
         end
 
         def to_adoc
           content = Coradoc::Generator.gen_adoc(@content)
-          doubled = @unconstrained ? "*" : ""
-          "#{doubled}*#{content.strip}*#{doubled}#{content[/\s+$/]}"
+          if @unconstrained
+            "**#{content}**"
+          else
+            "*#{content}*"
+          end
         end
       end
     end

data/lib/coradoc/element/inline/cross_reference.rb

@@ -1,8 +1,10 @@
 module Coradoc
   module Element
     module Inline
-      class CrossReference
-        attr_reader :href, :name
+      class CrossReference < Base
+        attr_accessor :href, :name
+
+        declare_children :href, :name
 
         def initialize(href, name = nil)
           @href = href

data/lib/coradoc/element/inline/hard_line_break.rb

@@ -1,7 +1,7 @@
 module Coradoc
   module Element
     module Inline
-      class HardLineBreak
+      class HardLineBreak < Base
         def to_adoc
           " +\n"
         end

data/lib/coradoc/element/inline/highlight.rb

@@ -1,18 +1,23 @@
 module Coradoc
   module Element
     module Inline
-      class Highlight
-        attr_accessor :content
+      class Highlight < Base
+        attr_accessor :content, :unconstrained
 
-        def initialize(content, constrained = true)
+        declare_children :content
+
+        def initialize(content, unconstrained: true)
           @content = content
-          @constrained = constrained
+          @unconstrained = unconstrained
         end
 
         def to_adoc
           content = Coradoc::Generator.gen_adoc(@content)
-          doubled = @constrained ? "" : "#"
-          "#{content[/^\s*/]}#{doubled}##{content.strip}##{doubled}#{content[/\s*$/]}"
+          if @unconstrained
+            "###{content}##"
+          else
+            "##{content}#"
+          end
         end
       end
     end