coradoc-html 1.1.18 → 1.1.19

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

@@ -1,134 +0,0 @@
-# frozen_string_literal: true
-
-module Coradoc
-  module Input
-    module Html
-      class Cleaner
-        # Pre-compiled regexes for performance
-        INNER_WHITESPACE_REGEX_1 = /\n stem:\[/
-        INNER_WHITESPACE_REGEX_2 = /(stem:\[([^\]]|\\\])*\])\n(?=\S)/
-        NEWLINES_REGEX = /\n{3,}/
-        LEADING_NEWLINE_REGEX = /\A\n+/
-        WHITESPACE_REGEX = /[ \t\r\n]+/
-        TRAILING_WHITESPACE_REGEX = /[ \t\r\n]+\z/
-
-        def tidy(string)
-          return string.transform_values { |i| tidy(i) } if string.is_a? Hash
-
-          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
-
-        def remove_section_attribute_newlines(string)
-          string.gsub("]\n\n==", "]\n==")
-        end
-
-        def remove_newlines(string)
-          string.gsub(NEWLINES_REGEX, "\n\n")
-        end
-
-        def remove_leading_newlines(string)
-          string.gsub(LEADING_NEWLINE_REGEX, '')
-        end
-
-        def remove_inner_whitespaces(string)
-          unless string.nil?
-            string.gsub!("\n stem:[", "\nstem:[")
-            string.gsub!(INNER_WHITESPACE_REGEX_1, '\\1 ')
-            string.gsub!(INNER_WHITESPACE_REGEX_2, '\\1')
-          end
-          result = +''
-          string.each_line do |line|
-            result << preserve_border_whitespaces(line) do
-              # Use ASCII-only strip to preserve CJK fullwidth spaces
-              line.gsub(/\A[ \t\r\n]+/, '').gsub(/[ \t\r\n]+\z/, '').gsub(/[ \t]{2,}/, ' ')
-            end
-          end
-          result
-        end
-
-        def clean_tag_borders(string)
-          result = string.gsub(/\s?~{2,}.*?~{2,}\s?/) do |match|
-            preserve_border_whitespaces(
-              match,
-              default_border: Coradoc::Html::Input.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
-
-        def clean_punctuation_characters(string)
-          string.gsub(/(\*\*|~~|__)\s([.!?'"])/, '\\1\\2')
-        end
-
-        def preprocess_word_html(string)
-          clean_headings(scrub_whitespace(string.dup))
-        end
-
-        def scrub_whitespace(string)
-          string.gsub!(/&nbsp;|&#xA0;| /i, '&#xA0;')
-          string = Coradoc.strip_unicode(string)
-          string.gsub!(/( +)$/, ' ')
-          string.gsub!("\n\n\n\n", "\n\n")
-          string
-        end
-
-        def clean_headings(string)
-          string.gsub!(%r{<h([1-9])[^>]*></h\1>}, ' ')
-          string.gsub!(
-            %r{<h([1-9])[^>]* style="vertical-align: super;[^>]*>(.+?)</h\1>},
-            '<sup>\\2</sup>'
-          )
-          string
-        end
-
-        private
-
-        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
-
-        def present_or_default(string, default)
-          return default if string.nil? || string.empty?
-
-          string
-        end
-      end
-    end
-  end
-end

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

@@ -1,80 +0,0 @@
-# frozen_string_literal: true
-
-require 'tmpdir'
-
-module Coradoc
-  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
-
-          # Source of HTML
-          # @sourcedir        = nil
-
-          # Image counter, assuming there are max 999 images
-          @image_counter = 1
-          # pad with 0s
-          @image_counter_pattern = '%03d'
-
-          @em_delimiter     = '_'
-          @strong_delimiter = '*'
-          @inline_options   = {}
-          @tag_border       = ' '
-
-          @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 for better portability
-          # across output formats.
-          @doc_width        = 1000
-
-          # Plugin system
-          @plugins          = []
-
-          # 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 self.declare_option(option)
-          attr_accessor option
-
-          original_reader = instance_method(option)
-          define_method(option) do
-            @inline_options[option] || original_reader.bind_call(self)
-          end
-        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
-end

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

@@ -1,79 +0,0 @@
-# frozen_string_literal: true
-
-require 'coradoc'
-
-module Coradoc
-  module Input
-    module Html
-      module Converters
-        class A < Base
-          INSTANCE = new
-
-          def to_coradoc(node, state = {})
-            # Use treat_children_coradoc to get CoreModel elements
-            content = treat_children_coradoc(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 nil if /^_Toc\d+$|^_GoBack$/.match?(id)
-
-            # For inline anchors - return CoreModel InlineElement with format_type "anchor"
-            if id
-              return Coradoc::CoreModel::InlineElement.new(
-                format_type: 'anchor',
-                target: id
-              )
-            end
-
-            # For cross-references
-            if href.to_s.start_with?('#')
-              ref_id = href.sub(/^#/, '').gsub(/\s/, '').gsub(/__+/, '_')
-              content_str = extract_text_from_content(content)
-              return Coradoc::CoreModel::CrossReferenceElement.new(
-                target: ref_id,
-                content: content_str.strip.empty? ? nil : content_str.strip
-              )
-            end
-
-            return nil if href.to_s.empty?
-
-            # For links
-            ambigous_characters = /[\w.?&#=%;\[\u{ff}-\u{10ffff}]/
-            right_constrain = textnode_after_start_with?(node, ambigous_characters)
-
-            content_str = extract_text_from_content(content)
-
-            out = []
-            # Add leading space if needed
-            if textnode_before_end_with?(node, ambigous_characters)
-              out << Coradoc::CoreModel::TextElement.new(
-                content: ' '
-              )
-            end
-
-            # Create link element
-            link = Coradoc::CoreModel::LinkElement.new(
-              target: href,
-              content: content_str.strip,
-              metadata: {
-                title: (title.strip unless title.to_s.strip.empty?),
-                right_constrain: right_constrain
-              }.compact
-            )
-            out << link
-
-            # Return single element or array
-            out.length == 1 ? out.first : out
-          end
-        end
-
-        register :a, A::INSTANCE
-      end
-    end
-  end
-end

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

@@ -1,22 +0,0 @@
-# frozen_string_literal: true
-
-module Coradoc
-  module Input
-    module Html
-      module Converters
-        class Aside < Base
-          INSTANCE = new
-
-          def to_coradoc(node, state = {})
-            content = treat_children_coradoc(node, state)
-            Coradoc::CoreModel::SidebarBlock.new(
-              children: content
-            )
-          end
-        end
-
-        register :aside, Aside::INSTANCE
-      end
-    end
-  end
-end

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

@@ -1,21 +0,0 @@
-# frozen_string_literal: true
-
-module Coradoc
-  module Input
-    module Html
-      module Converters
-        class Audio < MediaBase
-          INSTANCE = new
-
-          private
-
-          def semantic_type
-            :audio
-          end
-        end
-
-        register :audio, Audio::INSTANCE
-      end
-    end
-  end
-end

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

@@ -1,118 +0,0 @@
-# frozen_string_literal: true
-
-module Coradoc
-  module Input
-    module Html
-      module Converters
-        class Base
-          def treat_children_coradoc(node, state)
-            results = node.children.map do |child|
-              treat_coradoc(child, state)
-            end.flatten
-
-            results.reject do |x|
-              x.nil? || (x.is_a?(String) && x.strip.empty?)
-            end
-          end
-
-          def treat_coradoc(node, state)
-            Converters.process_coradoc(node, state)
-          end
-
-          def extract_title(node)
-            node['title'].to_s
-          end
-
-          def node_has_ancestor?(node, name)
-            case name
-            when String
-              node.ancestors(name).any?
-            when Array
-              name.any? { |n| node.ancestors(n).any? }
-            end
-          end
-
-          def textnode_before_end_with?(node, str)
-            return false unless [String, Regexp].include?(str.class)
-            return false if str.is_a?(String) && str.empty?
-
-            str = /#{Regexp.escape(str)}/ if str.is_a?(String)
-            str = /(?:#{str})\z/
-
-            node2 = node.at_xpath('preceding-sibling::node()[1]')
-            node2.is_a?(Nokogiri::XML::Text) && node2.text.match?(str)
-          end
-
-          def textnode_after_start_with?(node, str)
-            return false unless [String, Regexp].include?(str.class)
-            return false if str.is_a?(String) && str.empty?
-
-            str = /#{Regexp.escape(str)}/ if str.is_a?(String)
-            str = /\A(?:#{str})/
-
-            node2 = node.at_xpath('following-sibling::node()[1]')
-            node2.is_a?(Nokogiri::XML::Text) && node2.text.match?(str)
-          end
-
-          def extract_leading_trailing_whitespace(node)
-            node.text =~ /^(\s+)/
-            leading_whitespace = ::Regexp.last_match(1)
-            unless leading_whitespace.nil?
-              first_text = node.at_xpath('./text()[1]')
-              first_text&.replace(first_text.text.lstrip)
-              leading_whitespace = ' '
-            end
-            node.text =~ /(\s+)$/
-            trailing_whitespace = ::Regexp.last_match(1)
-            unless trailing_whitespace.nil?
-              last_text = node.at_xpath('./text()[last()]')
-              last_text&.replace(last_text.text.rstrip)
-              trailing_whitespace = ' '
-            end
-            [leading_whitespace, trailing_whitespace]
-          end
-
-          def unconstrained_before?(node)
-            before = node.at_xpath('preceding::node()[1]')
-
-            before &&
-              !before.text.strip.empty? &&
-              before.text[-1]&.match?(/\w/)
-          end
-
-          def unconstrained_after?(node)
-            after = node.at_xpath('following::node()[1]')
-
-            after && !after.text.strip.empty? &&
-              after.text[0]&.match?(/\w|,|;|"|\.\?!/)
-          end
-
-          # Extract plain text from a mixed content array.
-          # Handles String, InlineElement (via .content), and other
-          # CoreModel::Base (via .content or .title).
-          def extract_text_from_content(content)
-            return content if content.is_a?(String)
-            return '' if content.nil?
-
-            content.map do |item|
-              case item
-              when String
-                item
-              when Coradoc::CoreModel::InlineElement
-                item.content.to_s
-              when Coradoc::CoreModel::Base
-                if item.content
-                  item.content.to_s
-                else
-                  ''
-                end
-              else
-                item.to_s
-              end
-            end.join
-          end
-        end
-      end
-    end
-  end
-end

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

@@ -1,27 +0,0 @@
-# frozen_string_literal: true
-
-module Coradoc
-  module Input
-    module Html
-      module Converters
-        class Blockquote < Base
-          INSTANCE = new
-
-          def to_coradoc(node, state = {})
-            id = node['id']
-            cite = node['cite']
-            content = treat_children_coradoc(node, state)
-
-            Coradoc::CoreModel::QuoteBlock.new(
-              children: content,
-              id: id,
-              attribution: cite
-            )
-          end
-        end
-
-        register :blockquote, Blockquote::INSTANCE
-      end
-    end
-  end
-end

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

@@ -1,19 +0,0 @@
-# frozen_string_literal: true
-
-module Coradoc
-  module Input
-    module Html
-      module Converters
-        class Br < Base
-          INSTANCE = new
-
-          def to_coradoc(_node, _state = {})
-            Coradoc::CoreModel::LineBreakElement.new
-          end
-        end
-
-        register :br, Br::INSTANCE
-      end
-    end
-  end
-end

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

@@ -1,84 +0,0 @@
-# frozen_string_literal: true
-
-module Coradoc
-  module Input
-    module Html
-      module Converters
-        class Bypass < Base
-          INSTANCE = new
-
-          def to_coradoc(node, state = {})
-            treat_children_coradoc(node, state)
-          end
-        end
-
-        register :document, Bypass::INSTANCE
-        register :html,     Bypass::INSTANCE
-        register :body,     Bypass::INSTANCE
-        register :span,     Bypass::INSTANCE
-        register :thead,    Bypass::INSTANCE
-        register :tbody,    Bypass::INSTANCE
-        register :tfoot,    Bypass::INSTANCE
-        register :abbr,     Bypass::INSTANCE
-        register :acronym,  Bypass::INSTANCE
-        register :address,  Bypass::INSTANCE
-        register :applet,   Bypass::INSTANCE
-        register :map,      Bypass::INSTANCE
-        register :area,     Bypass::INSTANCE
-        register :bdi,      Bypass::INSTANCE
-        register :bdo,      Bypass::INSTANCE
-        register :big,      Bypass::INSTANCE
-        register :button,   Bypass::INSTANCE
-        register :canvas,   Bypass::INSTANCE
-        register :data,     Bypass::INSTANCE
-        register :datalist, Bypass::INSTANCE
-        register :del,      Bypass::INSTANCE
-        register :ins,      Bypass::INSTANCE
-        register :dfn,      Bypass::INSTANCE
-        register :dialog,   Bypass::INSTANCE
-        register :embed,    Bypass::INSTANCE
-        register :fieldset, Bypass::INSTANCE
-        register :font,     Bypass::INSTANCE
-        register :footer,   Bypass::INSTANCE
-        register :form,     Bypass::INSTANCE
-        register :frame,    Bypass::INSTANCE
-        register :frameset, Bypass::INSTANCE
-        register :header,   Bypass::INSTANCE
-        register :iframe,   Bypass::INSTANCE
-        register :input,    Bypass::INSTANCE
-        register :label,    Bypass::INSTANCE
-        register :legend,   Bypass::INSTANCE
-        register :main,     Bypass::INSTANCE
-        register :menu,     Bypass::INSTANCE
-        register :menulist, Bypass::INSTANCE
-        register :meter,    Bypass::INSTANCE
-        register :nav,      Bypass::INSTANCE
-        register :noframes, Bypass::INSTANCE
-        register :noscript, Bypass::INSTANCE
-        register :object,   Bypass::INSTANCE
-        register :optgroup, Bypass::INSTANCE
-        register :option,   Bypass::INSTANCE
-        register :output,   Bypass::INSTANCE
-        register :param,    Bypass::INSTANCE
-        register :picture,  Bypass::INSTANCE
-        register :progress, Bypass::INSTANCE
-        register :ruby,     Bypass::INSTANCE
-        register :rt,       Bypass::INSTANCE
-        register :rp,       Bypass::INSTANCE
-        register :s,        Bypass::INSTANCE
-        register :select,   Bypass::INSTANCE
-        register :small,    Bypass::INSTANCE
-        register :strike,   Bypass::INSTANCE
-        register :details,  Bypass::INSTANCE
-        register :section,  Bypass::INSTANCE
-        register :summary,  Bypass::INSTANCE
-        register :svg,      Bypass::INSTANCE
-        register :template, Bypass::INSTANCE
-        register :textarea, Bypass::INSTANCE
-        register :track,    Bypass::INSTANCE
-        register :u,        Bypass::INSTANCE
-        register :wbr,      Bypass::INSTANCE
-      end
-    end
-  end
-end

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

@@ -1,27 +0,0 @@
-# frozen_string_literal: true
-
-module Coradoc
-  module Input
-    module Html
-      module Converters
-        class Code < Markup
-          INSTANCE = new
-
-          def coradoc_format_type
-            'monospace'
-          end
-
-          def markup_ancestor_tag_names
-            %w[code tt kbd samp var]
-          end
-        end
-
-        register :code, Code::INSTANCE
-        register :tt,   Code::INSTANCE
-        register :kbd,  Code::INSTANCE
-        register :samp, Code::INSTANCE
-        register :var,  Code::INSTANCE
-      end
-    end
-  end
-end

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

@@ -1,25 +0,0 @@
-# frozen_string_literal: true
-
-module Coradoc
-  module Input
-    module Html
-      module Converters
-        class Div < Base
-          INSTANCE = new
-
-          def to_coradoc(node, state = {})
-            id = node['id']
-            contents = treat_children_coradoc(node, state)
-
-            Coradoc::CoreModel::OpenBlock.new(
-              children: contents,
-              id: id
-            )
-          end
-        end
-
-        register :div, Div::INSTANCE
-      end
-    end
-  end
-end