coradoc 1.1.5 → 1.1.7

data/lib/coradoc/input/html/cleaner.rb

@@ -1,138 +1,142 @@
-module Coradoc::Input::HTML
-  class Cleaner
-    def tidy(string)
-      if string.is_a? Hash
-        return string.transform_values { |i| tidy(i) }
-      end
-
-      result = HtmlConverter.track_time "Removing inner whitespace" do
-        remove_inner_whitespaces(String.new(string))
-      end
-      result = HtmlConverter.track_time "Removing newlines" do
-        remove_newlines(result)
-      end
-      result = HtmlConverter.track_time "Removing leading newlines" do
-        remove_leading_newlines(result)
-      end
-      result = HtmlConverter.track_time "Cleaning tag borders" do
-        clean_tag_borders(result)
-      end
-      result = HtmlConverter.track_time "Cleaning punctuation characters" do
-        clean_punctuation_characters(result)
-      end
-      result = remove_block_leading_newlines(result)
-      result = remove_section_attribute_newlines(result)
-    end
-
-    def remove_block_leading_newlines(string)
-      string.gsub("]\n****\n\n", "]\n****\n")
-    end
+module Coradoc
+  module Input
+    module Html
+      class Cleaner
+        def tidy(string)
+          if string.is_a? Hash
+            return string.transform_values { |i| tidy(i) }
+          end
+
+          result = HtmlConverter.track_time "Removing inner whitespace" do
+            remove_inner_whitespaces(String.new(string))
+          end
+          result = HtmlConverter.track_time "Removing newlines" do
+            remove_newlines(result)
+          end
+          result = HtmlConverter.track_time "Removing leading newlines" do
+            remove_leading_newlines(result)
+          end
+          result = HtmlConverter.track_time "Cleaning tag borders" do
+            clean_tag_borders(result)
+          end
+          result = HtmlConverter.track_time "Cleaning punctuation characters" do
+            clean_punctuation_characters(result)
+          end
+          result = remove_block_leading_newlines(result)
+          result = remove_section_attribute_newlines(result)
+        end
 
-    def remove_section_attribute_newlines(string)
-      string.gsub("]\n\n==", "]\n==")
-    end
+        def remove_block_leading_newlines(string)
+          string.gsub("]\n****\n\n", "]\n****\n")
+        end
 
-    def remove_newlines(string)
-      string.gsub(/\n{3,}/, "\n\n")
-    end
+        def remove_section_attribute_newlines(string)
+          string.gsub("]\n\n==", "]\n==")
+        end
 
-    def remove_leading_newlines(string)
-      string.gsub(/\A\n+/, "")
-    end
+        def remove_newlines(string)
+          string.gsub(/\n{3,}/, "\n\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
-      result = +""
-      string.each_line do |line|
-        result << preserve_border_whitespaces(line) do
-          line.strip.gsub(/[ \t]{2,}/, " ")
+        def remove_leading_newlines(string)
+          string.gsub(/\A\n+/, "")
         end
-      end
-      result
-    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: Coradoc::Input::HTML.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: Coradoc::Input::HTML.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: Coradoc::Input::HTML.config.tag_border) do
-          match.strip.sub("~~ ", "~~").sub(" ~~", "~~")
+        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
+          result = +""
+          string.each_line do |line|
+            result << preserve_border_whitespaces(line) do
+              line.strip.gsub(/[ \t]{2,}/, " ")
+            end
+          end
+          result
         end
-      end
 
-      result.gsub(/\s?\[.*?\]\s?/) do |match|
-        preserve_border_whitespaces(match) do
-          match.strip.sub("[ ", "[").sub(" ]", "]")
+        # 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: Coradoc::Input::HTML.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: Coradoc::Input::HTML.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: Coradoc::Input::Html.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
-      end
-    end
 
-    def clean_punctuation_characters(string)
-      string.gsub(/(\*\*|~~|__)\s([.!?'"])/, "\\1\\2")
-    end
+        def clean_punctuation_characters(string)
+          string.gsub(/(\*\*|~~|__)\s([.!?'"])/, "\\1\\2")
+        end
 
-    # preprocesses HTML, rather than postprocessing it
-    def preprocess_word_html(string)
-      clean_headings(scrub_whitespace(string.dup))
-    end
+        # preprocesses HTML, rather than postprocessing it
+        def preprocess_word_html(string)
+          clean_headings(scrub_whitespace(string.dup))
+        end
 
-    def scrub_whitespace(string)
-      string.gsub!(/&nbsp;|&#xA0;|\u00a0/i, "&#xA0;") # HTML encoded spaces
-      string = Coradoc.strip_unicode(string) # Strip document-level leading and trailing whitespace
-      string.gsub!(/( +)$/, " ") # line trailing whitespace
-      string.gsub!(/\n\n\n\n/, "\n\n") # Quadruple line breaks
-      # string.delete!('?| ')               # Unicode non-breaking spaces, injected as tabs
-      string
-    end
+        def scrub_whitespace(string)
+          string.gsub!(/&nbsp;|&#xA0;|\u00a0/i, "&#xA0;") # HTML encoded spaces
+          string = Coradoc.strip_unicode(string) # Strip document-level leading and trailing whitespace
+          string.gsub!(/( +)$/, " ") # line trailing whitespace
+          string.gsub!(/\n\n\n\n/, "\n\n") # Quadruple line breaks
+          # string.delete!('?| ')               # Unicode non-breaking spaces, injected as tabs
+          string
+        end
 
-    # following added by me
-    def clean_headings(string)
-      string.gsub!(%r{<h([1-9])[^>]*></h\1>}, " ")
-      # I don't know why Libre Office is inserting them, but they need to go
-      string.gsub!(%r{<h([1-9])[^>]* style="vertical-align: super;[^>]*>(.+?)</h\1>},
-                   "<sup>\\2</sup>")
-      # I absolutely don't know why Libre Office is rendering superscripts as h1
-      string
-    end
+        # following added by me
+        def clean_headings(string)
+          string.gsub!(%r{<h([1-9])[^>]*></h\1>}, " ")
+          # I don't know why Libre Office is inserting them, but they need to go
+          string.gsub!(%r{<h([1-9])[^>]* style="vertical-align: super;[^>]*>(.+?)</h\1>},
+                       "<sup>\\2</sup>")
+          # I absolutely don't know why Libre Office is rendering superscripts as h1
+          string
+        end
 
-    private
+        private
 
-    def preserve_border_whitespaces(string, options = {})
-      return string if /\A\s*\Z/.match?(string)
+        def preserve_border_whitespaces(string, options = {})
+          return string if /\A\s*\Z/.match?(string)
 
-      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 /[\[(\])]/.match?(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
+          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 /[\[(\])]/.match?(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)
-      return default if string.nil? || string.empty?
+        def present_or_default(string, default)
+          return default if string.nil? || string.empty?
 
-      string
+          string
+        end
+      end
     end
   end
 end

data/lib/coradoc/input/html/config.rb

@@ -1,75 +1,77 @@
 require "tmpdir"
 
 module Coradoc
-  module Input::HTML
-    class Config
-      def initialize
-        @unknown_tags     = :pass_through
-        @input_format     = :html
-        @mathml2asciimath = false
-        @external_images  = false
+  module Input
+    module Html
+      class Config
+        def initialize
+          @unknown_tags     = :pass_through
+          @input_format     = :html
+          @mathml2asciimath = false
+          @external_images  = false
 
-        # Destination to save file and images
-        @destination      = nil
+          # Destination to save file and images
+          @destination      = nil
 
-        # Source of HTML
-        # @sourcedir        = nil
+          # Source of HTML
+          # @sourcedir        = nil
 
-        # Image counter, assuming there are max 999 images
-        @image_counter = 1
-        # pad with 0s
-        @image_counter_pattern = "%03d"
+          # 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
+          @em_delimiter     = "_".freeze
+          @strong_delimiter = "*".freeze
+          @inline_options   = {}
+          @tag_border       = " ".freeze
 
-        @split_sections   = nil
+          @split_sections   = nil
 
-        # Document width - used to compute table sizes.
-        # This is an assumption for screen size in input document.
-        # If column widths are specified in absolute values, then we
-        # have to convert them to relative values, as AsciiDoc only
-        # supports those.
-        @doc_width        = 1000
+          # Document width - used to compute table sizes.
+          # This is an assumption for screen size in input document.
+          # If column widths are specified in absolute values, then we
+          # have to convert them to relative values, as AsciiDoc only
+          # supports those.
+          @doc_width        = 1000
 
-        # Plugin system
-        @plugins          = []
+          # Plugin system
+          @plugins          = []
 
-        # Debugging options
-        @track_time       = false
-      end
+          # Debugging options
+          @track_time       = false
+        end
 
-      def with(options = {})
-        old_options = @inline_options
-        @inline_options = options
-        result = yield
-        @inline_options = old_options
-        result
-      end
+        def with(options = {})
+          old_options = @inline_options
+          @inline_options = options
+          result = yield
+          @inline_options = old_options
+          result
+        end
+
+        def self.declare_option(option)
+          define_method(option) do
+            @inline_options[option] || instance_variable_get(:"@#{option}")
+          end
 
-      def self.declare_option(option)
-        define_method(option) do
-          @inline_options[option] || instance_variable_get(:"@#{option}")
+          attr_writer option
         end
 
-        attr_writer option
+        declare_option :unknown_tags
+        declare_option :tag_border
+        declare_option :mathml2asciimath
+        declare_option :external_images
+        declare_option :destination
+        declare_option :sourcedir
+        declare_option :image_counter
+        declare_option :image_counter_pattern
+        declare_option :input_format
+        declare_option :split_sections
+        declare_option :doc_width
+        declare_option :plugins
+        declare_option :track_time
       end
-
-      declare_option :unknown_tags
-      declare_option :tag_border
-      declare_option :mathml2asciimath
-      declare_option :external_images
-      declare_option :destination
-      declare_option :sourcedir
-      declare_option :image_counter
-      declare_option :image_counter_pattern
-      declare_option :input_format
-      declare_option :split_sections
-      declare_option :doc_width
-      declare_option :plugins
-      declare_option :track_time
     end
   end
 end

data/lib/coradoc/input/html/converters/a.rb

@@ -1,47 +1,52 @@
 require "coradoc"
 
-module Coradoc::Input::HTML
-  module Converters
-    class A < Base
-      def to_coradoc(node, state = {})
-        name  = treat_children(node, state)
-
-        href  = node["href"]
-        title = extract_title(node)
-        id = node["id"] || node["name"]
-
-        id = id&.gsub(/\s/, "")&.gsub(/__+/, "_")
-        id = nil if id&.empty?
-
-        return "" if /^_Toc\d+$|^_GoBack$/.match?(id)
-
-        return Coradoc::Element::Inline::Anchor.new(id) if id
-
-        if href.to_s.start_with?("#")
-          href = href.sub(/^#/, "").gsub(/\s/, "").gsub(/__+/, "_")
-          return Coradoc::Element::Inline::CrossReference.new(href, name)
+module Coradoc
+  module Input
+    module Html
+      module Converters
+        class A < Base
+          def to_coradoc(node, state = {})
+            name  = treat_children(node, state)
+
+            href  = node["href"]
+            title = extract_title(node)
+            id = node["id"] || node["name"]
+
+            id = id&.gsub(/\s/, "")&.gsub(/__+/, "_")
+            id = nil if id&.empty?
+
+            return "" if /^_Toc\d+$|^_GoBack$/.match?(id)
+
+            return Coradoc::Element::Inline::Anchor.new(id) if id
+
+            if href.to_s.start_with?("#")
+              href = href.sub(/^#/, "").gsub(/\s/, "").gsub(/__+/, "_")
+              return Coradoc::Element::Inline::CrossReference.new(href, name)
+            end
+
+            return name if href.to_s.empty?
+
+            ambigous_characters = /[\w.?&#=%;\[\u{ff}-\u{10ffff}]/
+            if name&.strip == href
+              name = ""
+              right_constrain = textnode_after_start_with?(node,
+                                                           ambigous_characters)
+            end
+
+            out = []
+            out << " " if textnode_before_end_with?(node, ambigous_characters)
+            out << Coradoc::Element::Inline::Link.new(
+              path: href,
+              name: name.strip,
+              title: title.strip,
+              right_constrain: right_constrain,
+            )
+            out
+          end
         end
 
-        return name if href.to_s.empty?
-
-        ambigous_characters = /[\w.?&#=%;\[\u{ff}-\u{10ffff}]/
-        if name&.strip == href
-          name = ""
-          right_constrain = textnode_after_start_with?(node, ambigous_characters)
-        end
-
-        out = []
-        out << " " if textnode_before_end_with?(node, ambigous_characters)
-        out << Coradoc::Element::Inline::Link.new(
-          path: href,
-          name: name.strip,
-          title: title.strip,
-          right_constrain: right_constrain,
-        )
-        out
+        register :a, A.new
       end
     end
-
-    register :a, A.new
   end
 end

data/lib/coradoc/input/html/converters/aside.rb

@@ -1,12 +1,16 @@
-module Coradoc::Input::HTML
-  module Converters
-    class Aside < Base
-      def to_coradoc(node, state = {})
-        content = treat_children(node, state)
-        Coradoc::Element::Block::Side.new(lines: content.lines)
+module Coradoc
+  module Input
+    module Html
+      module Converters
+        class Aside < Base
+          def to_coradoc(node, state = {})
+            content = treat_children(node, state)
+            Coradoc::Element::Block::Side.new(lines: content.lines)
+          end
+        end
+
+        register :aside, Aside.new
       end
     end
-
-    register :aside, Aside.new
   end
 end

data/lib/coradoc/input/html/converters/audio.rb

@@ -1,25 +1,29 @@
-module Coradoc::Input::HTML
-  module Converters
-    class Audio < Base
-      def to_coradoc(node, _state = {})
-        src = node["src"]
-        id = node["id"]
-        title = extract_title(node)
-        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
+module Coradoc
+  module Input
+    module Html
+      module Converters
+        class Audio < Base
+          def to_coradoc(node, _state = {})
+            src = node["src"]
+            id = node["id"]
+            title = extract_title(node)
+            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 options(node)
+            autoplay = node["autoplay"]
+            loop_attr = node["loop"]
+            controls = node["controls"]
+            [autoplay, loop_attr, controls].compact
+          end
+        end
 
-      def options(node)
-        autoplay = node["autoplay"]
-        loop_attr = node["loop"]
-        controls = node["controls"]
-        [autoplay, loop_attr, controls].compact
+        register :audio, Audio.new
       end
     end
-
-    register :audio, Audio.new
   end
 end