sevk 0.1.0 → 1.1.0

data/lib/sevk/markup/style_helpers.rb

@@ -0,0 +1,131 @@
+# frozen_string_literal: true
+
+module Sevk
+  module Markup
+    module StyleHelpers
+      STYLE_ORDER = %w[
+        color background-color background-image background-size background-position background-repeat
+        font-size font-family font-weight line-height text-align text-decoration
+        width height max-width max-height min-width min-height
+        padding padding-top padding-right padding-bottom padding-left
+        margin margin-top margin-right margin-bottom margin-left
+        border border-color border-width border-style
+        border-top border-right border-bottom border-left
+        border-radius border-top-left-radius border-top-right-radius
+        border-bottom-left-radius border-bottom-right-radius
+        border-collapse list-style-type vertical-align
+      ].freeze
+
+      private
+
+      def parse_attributes(attrs_str)
+        attrs = {}
+        attrs_str.scan(/([\w-]+)=(?:"([^"]*)"|'([^']*)')/) do |key, dq_val, sq_val|
+          attrs[key] = dq_val.nil? || dq_val.empty? ? (sq_val || '') : dq_val
+        end
+        attrs
+      end
+
+      def extract_all_style_attributes(attrs)
+        attrs = attrs.reject { |_, value| value.to_s.empty? }
+        style = {}
+
+        if attrs['text-color']
+          style['color'] = attrs['text-color']
+        elsif attrs['color']
+          style['color'] = attrs['color']
+        end
+        style['background-color'] = attrs['background-color'] if attrs['background-color']
+        # Background image group lives between background-color and font-size to
+        # match the reference extractStyleAttributes insertion order.
+        bg_image = attrs['background-image']
+        style['background-image'] = "url('#{bg_image}')" if bg_image
+        if attrs['background-size']
+          style['background-size'] = attrs['background-size']
+        elsif bg_image
+          style['background-size'] = 'cover'
+        end
+        if attrs['background-position']
+          style['background-position'] = attrs['background-position']
+        elsif bg_image
+          style['background-position'] = 'center'
+        end
+        if attrs['background-repeat']
+          style['background-repeat'] = attrs['background-repeat']
+        elsif bg_image
+          style['background-repeat'] = 'no-repeat'
+        end
+        style['font-size'] = attrs['font-size'] if attrs['font-size']
+        style['font-family'] = attrs['font-family'] if attrs['font-family']
+        style['font-weight'] = attrs['font-weight'] if attrs['font-weight']
+        style['line-height'] = attrs['line-height'] if attrs['line-height']
+        style['text-align'] = attrs['text-align'] if attrs['text-align']
+        style['text-decoration'] = attrs['text-decoration'] if attrs['text-decoration']
+
+        style['width'] = attrs['width'] if attrs['width']
+        style['height'] = attrs['height'] if attrs['height']
+        style['max-width'] = attrs['max-width'] if attrs['max-width']
+        style['max-height'] = attrs['max-height'] if attrs['max-height']
+        style['min-width'] = attrs['min-width'] if attrs['min-width']
+        style['min-height'] = attrs['min-height'] if attrs['min-height']
+
+        # Spacing - Padding (shorthand + per-side both emit; per-side wins per CSS)
+        style['padding'] = attrs['padding'] if attrs['padding']
+        style['padding-top'] = attrs['padding-top'] if attrs['padding-top']
+        style['padding-right'] = attrs['padding-right'] if attrs['padding-right']
+        style['padding-bottom'] = attrs['padding-bottom'] if attrs['padding-bottom']
+        style['padding-left'] = attrs['padding-left'] if attrs['padding-left']
+
+        # Spacing - Margin (shorthand + per-side both emit; per-side wins per CSS)
+        style['margin'] = attrs['margin'] if attrs['margin']
+        style['margin-top'] = attrs['margin-top'] if attrs['margin-top']
+        style['margin-right'] = attrs['margin-right'] if attrs['margin-right']
+        style['margin-bottom'] = attrs['margin-bottom'] if attrs['margin-bottom']
+        style['margin-left'] = attrs['margin-left'] if attrs['margin-left']
+
+        # Borders cascade: shorthand -> per-property -> per-side (most specific last)
+        style['border'] = attrs['border'] if attrs['border']
+        style['border-color'] = attrs['border-color'] if attrs['border-color']
+        style['border-width'] = attrs['border-width'] if attrs['border-width']
+        style['border-style'] = attrs['border-style'] if attrs['border-style']
+        style['border-top'] = attrs['border-top'] if attrs['border-top']
+        style['border-right'] = attrs['border-right'] if attrs['border-right']
+        style['border-bottom'] = attrs['border-bottom'] if attrs['border-bottom']
+        style['border-left'] = attrs['border-left'] if attrs['border-left']
+
+        # Border Radius (shorthand + per-corner both emit; per-corner wins per CSS)
+        style['border-radius'] = attrs['border-radius'] if attrs['border-radius']
+        style['border-top-left-radius'] = attrs['border-top-left-radius'] if attrs['border-top-left-radius']
+        style['border-top-right-radius'] = attrs['border-top-right-radius'] if attrs['border-top-right-radius']
+        style['border-bottom-left-radius'] = attrs['border-bottom-left-radius'] if attrs['border-bottom-left-radius']
+        style['border-bottom-right-radius'] = attrs['border-bottom-right-radius'] if attrs['border-bottom-right-radius']
+
+        style
+      end
+
+      def class_attribute(attrs)
+        value = attrs['class'].to_s.empty? ? attrs['className'] : attrs['class']
+        value.to_s.empty? ? '' : %( class="#{value}")
+      end
+
+      def style_to_string(style)
+        seen = {}
+        parts = []
+        STYLE_ORDER.each do |key|
+          value = style[key]
+          next if value.nil? || value == ''
+
+          parts << "#{key}:#{value}"
+          seen[key] = true
+        end
+        style.each do |key, value|
+          next if seen[key]
+          next if value.nil? || value == ''
+
+          parts << "#{key}:#{value}"
+        end
+        parts.join(';')
+      end
+    end
+  end
+end

data/lib/sevk/markup/syntax_highlighting.rb

@@ -0,0 +1,227 @@
+# frozen_string_literal: true
+
+module Sevk
+  module Markup
+    module SyntaxHighlighting
+      MAX_CODEBLOCK_LENGTH = 2500
+      MAX_TOTAL_CODEBLOCK_LENGTH = 2500
+
+      LANGUAGES = %w[
+        plain extend plaintext text txt markup html mathml svg xml ssml atom rss css clike javascript js typescript ts
+        jsx tsx python py java c cpp csharp cs dotnet php ruby rb go rust sql bash sh shell json webmanifest
+        yaml yml markdown md
+      ].freeze
+
+      LANGUAGE_ALIASES = { 'jsx' => 'javascript', 'tsx' => 'typescript' }.freeze
+
+      ROUGE_LEXERS = {
+        'markup' => 'html', 'mathml' => 'xml', 'svg' => 'xml', 'ssml' => 'xml', 'atom' => 'xml', 'rss' => 'xml',
+        'dotnet' => 'csharp', 'webmanifest' => 'json', 'plain' => nil, 'plaintext' => nil, 'text' => nil,
+        'txt' => nil, 'clike' => nil
+      }.freeze
+
+      THEME_BASES = {
+        'vscDarkPlus' => {
+          'color' => '#d4d4d4', 'fontSize' => '13px', 'textShadow' => 'none',
+          'fontFamily' => "Menlo, Monaco, Consolas, 'Andale Mono', 'Ubuntu Mono', 'Courier New', monospace",
+          'direction' => 'ltr', 'textAlign' => 'left', 'whiteSpace' => 'pre', 'wordSpacing' => 'normal',
+          'wordBreak' => 'normal', 'lineHeight' => '1.5', 'MozTabSize' => '4', 'OTabSize' => '4', 'tabSize' => '4',
+          'WebkitHyphens' => 'none', 'MozHyphens' => 'none', 'hyphens' => 'none', 'padding' => '16px',
+          'overflow' => 'auto', 'backgroundColor' => '#1e1e1e'
+        },
+        'vs' => {
+          'color' => '#393A34',
+          'fontFamily' => "'Consolas', 'Bitstream Vera Sans Mono', 'Courier New', Courier, monospace",
+          'direction' => 'ltr', 'textAlign' => 'left', 'whiteSpace' => 'pre', 'wordSpacing' => 'normal',
+          'wordBreak' => 'normal', 'fontSize' => '.9em', 'lineHeight' => '1.2em', 'MozTabSize' => '4',
+          'OTabSize' => '4', 'tabSize' => '4', 'WebkitHyphens' => 'none', 'MozHyphens' => 'none',
+          'hyphens' => 'none', 'padding' => '16px', 'overflow' => 'auto', 'border' => '1px solid #dddddd',
+          'backgroundColor' => 'white'
+        },
+        'oneDark' => {
+          'backgroundColor' => '#282c34', 'color' => '#abb2bf', 'textShadow' => '0 1px rgba(0, 0, 0, 0.3)',
+          'fontFamily' => "'Fira Code', 'Fira Mono', Menlo, Consolas, 'DejaVu Sans Mono', monospace",
+          'direction' => 'ltr', 'textAlign' => 'left', 'whiteSpace' => 'pre', 'wordSpacing' => 'normal',
+          'wordBreak' => 'normal', 'lineHeight' => '1.5', 'MozTabSize' => '2', 'OTabSize' => '2', 'tabSize' => '2',
+          'WebkitHyphens' => 'none', 'MozHyphens' => 'none', 'hyphens' => 'none', 'padding' => '16px',
+          'overflow' => 'auto', 'borderRadius' => '6px'
+        },
+        'oneLight' => {
+          'backgroundColor' => '#fafafa', 'color' => '#383a42',
+          'fontFamily' => "'Fira Code', 'Fira Mono', Menlo, Consolas, 'DejaVu Sans Mono', monospace",
+          'direction' => 'ltr', 'textAlign' => 'left', 'whiteSpace' => 'pre', 'wordSpacing' => 'normal',
+          'wordBreak' => 'normal', 'lineHeight' => '1.5', 'MozTabSize' => '2', 'OTabSize' => '2', 'tabSize' => '2',
+          'WebkitHyphens' => 'none', 'MozHyphens' => 'none', 'hyphens' => 'none', 'padding' => '16px',
+          'overflow' => 'auto', 'borderRadius' => '6px'
+        }
+      }.freeze
+
+      THEME_TOKENS = {
+        'vscDarkPlus' => {
+          'comment' => { 'color' => '#6a9955' }, 'prolog' => { 'color' => '#6a9955' },
+          'punctuation' => { 'color' => '#d4d4d4' }, 'property' => { 'color' => '#9cdcfe' },
+          'tag' => { 'color' => '#569cd6' }, 'boolean' => { 'color' => '#569cd6' },
+          'number' => { 'color' => '#b5cea8' }, 'constant' => { 'color' => '#9cdcfe' },
+          'symbol' => { 'color' => '#b5cea8' }, 'inserted' => { 'color' => '#b5cea8' },
+          'selector' => { 'color' => '#d7ba7d' }, 'attr-name' => { 'color' => '#9cdcfe' },
+          'string' => { 'color' => '#ce9178' }, 'char' => { 'color' => '#ce9178' },
+          'builtin' => { 'color' => '#ce9178' }, 'deleted' => { 'color' => '#ce9178' },
+          'operator' => { 'color' => '#d4d4d4' }, 'entity' => { 'color' => '#569cd6' },
+          'atrule' => { 'color' => '#ce9178' }, 'keyword' => { 'color' => '#569CD6' },
+          'function' => { 'color' => '#dcdcaa' }, 'regex' => { 'color' => '#d16969' },
+          'important' => { 'color' => '#569cd6' }, 'italic' => { 'font-style' => 'italic' },
+          'class-name' => { 'color' => '#4ec9b0' }, 'variable' => { 'color' => '#9cdcfe' },
+          'attr-value' => { 'color' => '#ce9178' }, 'cdata' => { 'color' => '#808080' },
+          'namespace' => { 'color' => '#4ec9b0' }
+        },
+        'vs' => {
+          'comment' => { 'color' => '#008000', 'font-style' => 'italic' },
+          'prolog' => { 'color' => '#008000', 'font-style' => 'italic' },
+          'doctype' => { 'color' => '#008000', 'font-style' => 'italic' },
+          'cdata' => { 'color' => '#008000', 'font-style' => 'italic' }, 'namespace' => { 'opacity' => '.7' },
+          'string' => { 'color' => '#A31515' }, 'punctuation' => { 'color' => '#393A34' },
+          'operator' => { 'color' => '#393A34' }, 'url' => { 'color' => '#36acaa' },
+          'symbol' => { 'color' => '#36acaa' }, 'number' => { 'color' => '#36acaa' },
+          'boolean' => { 'color' => '#36acaa' }, 'variable' => { 'color' => '#36acaa' },
+          'constant' => { 'color' => '#36acaa' }, 'inserted' => { 'color' => '#36acaa' },
+          'atrule' => { 'color' => '#0000ff' }, 'keyword' => { 'color' => '#0000ff' },
+          'attr-value' => { 'color' => '#0000ff' }, 'function' => { 'color' => '#393A34' },
+          'deleted' => { 'color' => '#9a050f' }, 'selector' => { 'color' => '#800000' },
+          'important' => { 'color' => '#e90', 'font-weight' => 'bold' }, 'bold' => { 'font-weight' => 'bold' },
+          'italic' => { 'font-style' => 'italic' }, 'class-name' => { 'color' => '#2B91AF' },
+          'tag' => { 'color' => '#800000' }, 'attr-name' => { 'color' => '#ff0000' },
+          'property' => { 'color' => '#ff0000' }, 'regex' => { 'color' => '#ff0000' },
+          'entity' => { 'color' => '#ff0000' }
+        },
+        'oneDark' => {
+          'comment' => { 'color' => '#5c6370', 'font-style' => 'italic' }, 'prolog' => { 'color' => '#5c6370' },
+          'cdata' => { 'color' => '#5c6370' }, 'doctype' => { 'color' => '#abb2bf' },
+          'punctuation' => { 'color' => '#abb2bf' }, 'entity' => { 'color' => '#abb2bf', 'cursor' => 'help' },
+          'attr-name' => { 'color' => '#d19a66' }, 'class-name' => { 'color' => '#d19a66' },
+          'boolean' => { 'color' => '#d19a66' }, 'constant' => { 'color' => '#d19a66' },
+          'number' => { 'color' => '#d19a66' }, 'atrule' => { 'color' => '#d19a66' },
+          'keyword' => { 'color' => '#c678dd' }, 'property' => { 'color' => '#e06c75' },
+          'tag' => { 'color' => '#e06c75' }, 'symbol' => { 'color' => '#e06c75' },
+          'deleted' => { 'color' => '#e06c75' }, 'important' => { 'color' => '#e06c75' },
+          'selector' => { 'color' => '#98c379' }, 'string' => { 'color' => '#98c379' },
+          'char' => { 'color' => '#98c379' }, 'builtin' => { 'color' => '#98c379' },
+          'inserted' => { 'color' => '#98c379' }, 'regex' => { 'color' => '#98c379' },
+          'attr-value' => { 'color' => '#98c379' }, 'variable' => { 'color' => '#61afef' },
+          'operator' => { 'color' => '#61afef' }, 'function' => { 'color' => '#61afef' },
+          'url' => { 'color' => '#56b6c2' }, 'bold' => { 'font-weight' => 'bold' },
+          'italic' => { 'font-style' => 'italic' }, 'namespace' => { 'opacity' => '0.8' }
+        },
+        'oneLight' => {
+          'comment' => { 'color' => '#a0a1a7', 'font-style' => 'italic' }, 'prolog' => { 'color' => '#a0a1a7' },
+          'cdata' => { 'color' => '#a0a1a7' }, 'doctype' => { 'color' => '#383a42' },
+          'punctuation' => { 'color' => '#383a42' }, 'entity' => { 'color' => '#383a42', 'cursor' => 'help' },
+          'attr-name' => { 'color' => '#b76b01' }, 'class-name' => { 'color' => '#b76b01' },
+          'boolean' => { 'color' => '#b76b01' }, 'constant' => { 'color' => '#b76b01' },
+          'number' => { 'color' => '#b76b01' }, 'atrule' => { 'color' => '#b76b01' },
+          'keyword' => { 'color' => '#a626a4' }, 'property' => { 'color' => '#e45649' },
+          'tag' => { 'color' => '#e45649' }, 'symbol' => { 'color' => '#e45649' },
+          'deleted' => { 'color' => '#e45649' }, 'important' => { 'color' => '#e45649' },
+          'selector' => { 'color' => '#50a14f' }, 'string' => { 'color' => '#50a14f' },
+          'char' => { 'color' => '#50a14f' }, 'builtin' => { 'color' => '#50a14f' },
+          'inserted' => { 'color' => '#50a14f' }, 'regex' => { 'color' => '#50a14f' },
+          'attr-value' => { 'color' => '#50a14f' }, 'variable' => { 'color' => '#4078f2' },
+          'operator' => { 'color' => '#4078f2' }, 'function' => { 'color' => '#4078f2' },
+          'url' => { 'color' => '#0184bc' }, 'bold' => { 'font-weight' => 'bold' },
+          'italic' => { 'font-style' => 'italic' }, 'namespace' => { 'opacity' => '0.8' }
+        }
+      }.freeze
+
+      ROUGE_TOKENS = {
+        'Comment' => 'comment', 'Keyword' => 'keyword', 'Keyword.Constant' => 'boolean', 'Keyword.Type' => 'class-name',
+        'Literal.String' => 'string', 'Literal.String.Regex' => 'regex', 'Literal.String.Symbol' => 'symbol',
+        'Literal.String.Escape' => 'string', 'Literal.Number' => 'number', 'Name.Function' => 'function',
+        'Name.Class' => 'class-name', 'Name.Constant' => 'constant', 'Name.Builtin' => 'builtin',
+        'Name.Variable' => 'variable', 'Name.Tag' => 'tag', 'Name.Attribute' => 'attr-name',
+        'Name.Decorator' => 'function', 'Name.Namespace' => 'namespace', 'Name.Entity' => 'entity',
+        'Name.Label' => 'symbol', 'Name.Property' => 'property', 'Operator' => 'operator',
+        'Operator.Word' => 'keyword', 'Punctuation' => 'punctuation', 'Generic.Deleted' => 'deleted',
+        'Generic.Inserted' => 'inserted', 'Generic.Emph' => 'italic', 'Generic.Strong' => 'bold',
+        'Generic.Heading' => 'important', 'Generic.Subheading' => 'important'
+      }.freeze
+
+      TOKEN_SPACE = "\u00A0\u200D\u200B"
+
+      private
+
+      def process_codeblock(attrs, code)
+        requested = (attrs['language'].to_s.empty? ? 'javascript' : attrs['language']).downcase
+        language = LANGUAGE_ALIASES.fetch(requested, requested)
+        theme = THEME_BASES.key?(attrs['theme'].to_s) ? attrs['theme'] : 'oneDark'
+        custom_style = extract_all_style_attributes(attrs)
+
+        return '<pre><code></code></pre>' if code.empty?
+
+        length = code.length + code.count("\u{10000}-\u{10FFFF}")
+
+        if !LANGUAGES.include?(language) || length > MAX_CODEBLOCK_LENGTH ||
+           @codeblock_budget + length > MAX_TOTAL_CODEBLOCK_LENGTH
+          style = pre_style(THEME_BASES[theme], { 'width' => '100%' }, custom_style)
+          escaped = code.gsub('<', '&lt;').gsub('>', '&gt;')
+          return %(<pre style="#{style}"><code>#{escaped}</code></pre>)
+        end
+
+        @codeblock_budget += length
+        inherited = attrs['font-family'].to_s.empty? ? {} : { 'font-family' => attrs['font-family'] }
+        lines = highlight_lines(code, language, theme, inherited)
+        numbered = attrs['line-numbers'] == 'true'
+        html = lines.each_with_index.map do |line, index|
+          number = numbered ? %(<span style="width:2em;display:inline-block">#{index + 1}</span>) : ''
+          %(<p style="margin:0;min-height:1em">#{number}#{line}</p>)
+        end.join
+        style = pre_style(THEME_BASES[theme], { 'width' => '100%', 'box-sizing' => 'border-box' }, custom_style)
+        %(<pre style="#{style}"><code>#{html}</code></pre>)
+      end
+
+      def pre_style(*layers)
+        style = {}
+        layers.each { |layer| layer.each { |key, value| style[key] = value } }
+        style.map do |key, value|
+          property = key.gsub(/[A-Z]/) { |letter| "-#{letter.downcase}" }
+          property = "-#{property}" if property.start_with?('ms-')
+          "#{property}:#{value}"
+        end.join(';')
+      end
+
+      def highlight_lines(code, language, theme, inherited)
+        lexer_name = ROUGE_LEXERS.fetch(language, language)
+        lexer = lexer_name && Rouge::Lexer.find(lexer_name)&.new
+        tokens = [[nil, code]]
+        tokens = lexer.lex(code).map { |type, text| [token_style(theme, type, inherited), text] } if lexer
+        inherited_style = style_to_string(inherited)
+        lines = [+'']
+
+        tokens.each do |style, text|
+          style = inherited_style if style.nil? && !inherited.empty?
+          text.split("\n", -1).each_with_index do |part, index|
+            lines << +'' if index.positive?
+            next if part.empty?
+
+            escaped = part.gsub('&', '&amp;').gsub('<', '&lt;').gsub('>', '&gt;').gsub('"', '&quot;')
+                          .gsub("'", '&#39;').gsub(' ', TOKEN_SPACE)
+            lines.last << (style.nil? || style.empty? ? escaped : %(<span style="#{style}">#{escaped}</span>))
+          end
+        end
+
+        lines
+      end
+
+      def token_style(theme, type, inherited)
+        qualname = type.qualname
+        token = nil
+        until token || qualname.nil?
+          token = ROUGE_TOKENS[qualname]
+          dot = qualname.rindex('.')
+          qualname = dot ? qualname[0...dot] : nil
+        end
+        return nil if token.nil?
+
+        style_to_string(inherited.merge(THEME_TOKENS[theme].fetch(token, {})))
+      end
+    end
+  end
+end

data/lib/sevk/markup/tag_scanner.rb

@@ -0,0 +1,112 @@
+# frozen_string_literal: true
+
+module Sevk
+  module Markup
+    module TagScanner
+      TagPatterns = Struct.new(:open, :close, :nested, :close_length)
+
+      def tag_patterns(tag_name)
+        TagPatterns.new(/<#{tag_name}([^>]*)>/i, %r{</#{tag_name}>}i, /<#{tag_name}/i, tag_name.length + 3)
+      end
+
+      def replace_self_closing(source, tag, self_close_only: false)
+        open_pattern = /<#{tag}/i
+        name_length = tag.length + 1
+        output = +''
+        last_index = 0
+        position = 0
+
+        while (opening = source.index(open_pattern, position))
+          closing_bracket = source.index('>', opening + name_length)
+          break unless closing_bracket
+
+          position = closing_bracket + 1
+          next if self_close_only && source[closing_bracket - 1] != '/'
+
+          attributes_end = self_close_only ? closing_bracket - 1 : closing_bracket
+          output << source[last_index...opening] << yield(source[(opening + name_length)...attributes_end])
+          last_index = closing_bracket + 1
+        end
+
+        output << source[last_index..]
+      end
+
+      def get_first_tag_content(source, tag)
+        scanner = StringScanner.new(source)
+        return nil unless scanner.skip_until(/<#{tag}/i) && scanner.skip_until(/>/)
+
+        content_start = scanner.pos
+        return nil unless scanner.skip_until(%r{</#{tag}>}i)
+
+        source.byteslice(content_start, scanner.pos - (tag.length + 3) - content_start)
+      end
+
+      def get_all_tag_contents(source, tag)
+        open_pattern = /<#{tag}/i
+        close_pattern = %r{</#{tag}>}i
+        scanner = StringScanner.new(source)
+        contents = []
+
+        while scanner.skip_until(open_pattern) && scanner.skip_until(/>/)
+          content_start = scanner.pos
+          break unless scanner.skip_until(close_pattern)
+
+          contents << source.byteslice(content_start, scanner.pos - (tag.length + 3) - content_start)
+        end
+
+        contents
+      end
+
+      def strip_tag_pairs(source, tag)
+        open_pattern = /<#{tag}/i
+        close_pattern = %r{</#{tag}>}i
+        scanner = StringScanner.new(source)
+        output = +''
+        last_index = 0
+
+        while scanner.skip_until(open_pattern)
+          opening = scanner.pos - scanner.matched_size
+          break unless scanner.skip_until(/>/) && scanner.skip_until(close_pattern)
+
+          output << source.byteslice(last_index, opening - last_index)
+          last_index = scanner.pos
+        end
+
+        output << source.byteslice(last_index, source.bytesize - last_index)
+      end
+
+      def get_tag_attributes(source, tag)
+        open_pattern = /<#{tag}/i
+        scanner = StringScanner.new(source)
+        attributes = []
+
+        while scanner.skip_until(open_pattern)
+          attributes_start = scanner.pos
+          break unless scanner.skip_until(/>/)
+
+          attributes << source.byteslice(attributes_start, scanner.pos - 1 - attributes_start)
+        end
+
+        attributes
+      end
+
+      def find_hidden_preview_div(html)
+        position = 0
+
+        while (opening = html.index(/<div/i, position))
+          closing_bracket = html.index('>', opening + 4)
+          return nil unless closing_bracket
+
+          if html[opening..closing_bracket].match?(/style=["'][^"']*display:\s*none/i)
+            closing = html.index(%r{</div>}i, closing_bracket + 1)
+            return nil unless closing
+
+            return [html[opening...(closing + 6)], html[(closing_bracket + 1)...closing]]
+          end
+
+          position = closing_bracket + 1
+        end
+      end
+    end
+  end
+end