rouge-alda 3.3.0

data/lib/rouge/formatters/html_legacy.rb

@@ -0,0 +1,45 @@
+# -*- coding: utf-8 -*- #
+# frozen_string_literal: true
+
+# stdlib
+require 'cgi'
+
+module Rouge
+  module Formatters
+    # Transforms a token stream into HTML output.
+    class HTMLLegacy < Formatter
+      tag 'html_legacy'
+
+      # @option opts [String] :css_class ('highlight')
+      # @option opts [true/false] :line_numbers (false)
+      # @option opts [Rouge::CSSTheme] :inline_theme (nil)
+      # @option opts [true/false] :wrap (true)
+      #
+      # Initialize with options.
+      #
+      # If `:inline_theme` is given, then instead of rendering the
+      # tokens as <span> tags with CSS classes, the styles according to
+      # the given theme will be inlined in "style" attributes.  This is
+      # useful for formats in which stylesheets are not available.
+      #
+      # Content will be wrapped in a tag (`div` if tableized, `pre` if
+      # not) with the given `:css_class` unless `:wrap` is set to `false`.
+      def initialize(opts={})
+        @formatter = opts[:inline_theme] ? HTMLInline.new(opts[:inline_theme])
+                   : HTML.new
+
+
+        @formatter = HTMLTable.new(@formatter, opts) if opts[:line_numbers]
+
+        if opts.fetch(:wrap, true)
+          @formatter = HTMLPygments.new(@formatter, opts.fetch(:css_class, 'codehilite'))
+        end
+      end
+
+      # @yield the html output.
+      def stream(tokens, &b)
+        @formatter.stream(tokens, &b)
+      end
+    end
+  end
+end

data/lib/rouge/formatters/html_linewise.rb

@@ -0,0 +1,28 @@
+# -*- coding: utf-8 -*- #
+# frozen_string_literal: true
+
+module Rouge
+  module Formatters
+    class HTMLLinewise < Formatter
+      def initialize(formatter, opts={})
+        @formatter = formatter
+        @class_format = opts.fetch(:class, 'line-%i')
+      end
+
+      def stream(tokens, &b)
+        token_lines(tokens) do |line|
+          yield "<div class=#{next_line_class}>"
+          line.each do |tok, val|
+            yield @formatter.span(tok, val)
+          end
+          yield '</div>'
+        end
+      end
+
+      def next_line_class
+        @lineno ||= 0
+        sprintf(@class_format, @lineno += 1).inspect
+      end
+    end
+  end
+end

data/lib/rouge/formatters/html_pygments.rb

@@ -0,0 +1,18 @@
+# frozen_string_literal: true
+
+module Rouge
+  module Formatters
+    class HTMLPygments < Formatter
+      def initialize(inner, css_class='codehilite')
+        @inner = inner
+        @css_class = css_class
+      end
+
+      def stream(tokens, &b)
+        yield %(<div class="highlight"><pre class="#{@css_class}"><code>)
+        @inner.stream(tokens, &b)
+        yield "</code></pre></div>"
+      end
+    end
+  end
+end

data/lib/rouge/formatters/html_table.rb

@@ -0,0 +1,62 @@
+# -*- coding: utf-8 -*- #
+# frozen_string_literal: true
+
+module Rouge
+  module Formatters
+    class HTMLTable < Formatter
+      tag 'html_table'
+
+      def initialize(inner, opts={})
+        @inner = inner
+        @start_line = opts.fetch(:start_line, 1)
+        @line_format = opts.fetch(:line_format, '%i')
+        @table_class = opts.fetch(:table_class, 'rouge-table')
+        @gutter_class = opts.fetch(:gutter_class, 'rouge-gutter')
+        @code_class = opts.fetch(:code_class, 'rouge-code')
+      end
+
+      def style(scope)
+        yield "#{scope} .rouge-table { border-spacing: 0 }"
+        yield "#{scope} .rouge-gutter { text-align: right }"
+      end
+
+      def stream(tokens, &b)
+        num_lines = 0
+        last_val = ''
+        formatted = String.new('')
+
+        tokens.each do |tok, val|
+          last_val = val
+          num_lines += val.scan(/\n/).size
+          formatted << @inner.span(tok, val)
+        end
+
+        # add an extra line for non-newline-terminated strings
+        if last_val[-1] != "\n"
+          num_lines += 1
+          @inner.span(Token::Tokens::Text::Whitespace, "\n") { |str| formatted << str }
+        end
+
+        # generate a string of newline-separated line numbers for the gutter>
+        formatted_line_numbers = (@start_line..num_lines+@start_line-1).map do |i|
+          sprintf("#{@line_format}", i) << "\n"
+        end.join('')
+
+        numbers = %(<pre class="lineno">#{formatted_line_numbers}</pre>)
+
+        yield %(<table class="#@table_class"><tbody><tr>)
+
+        # the "gl" class applies the style for Generic.Lineno
+        yield %(<td class="#@gutter_class gl">)
+        yield numbers
+        yield '</td>'
+
+        yield %(<td class="#@code_class"><pre>)
+        yield formatted
+        yield '</pre></td>'
+
+        yield "</tr></tbody></table>"
+      end
+    end
+  end
+end

data/lib/rouge/formatters/null.rb

@@ -0,0 +1,20 @@
+# -*- coding: utf-8 -*- #
+# frozen_string_literal: true
+
+module Rouge
+  module Formatters
+    # A formatter which renders nothing.
+    class Null < Formatter
+      tag 'null'
+
+      def initialize(*)
+      end
+
+      def stream(tokens, &b)
+        tokens.each do |tok, val|
+          yield "#{tok.qualname} #{val.inspect}\n"
+        end
+      end
+    end
+  end
+end

data/lib/rouge/formatters/terminal256.rb

@@ -0,0 +1,181 @@
+# -*- coding: utf-8 -*- #
+# frozen_string_literal: true
+
+module Rouge
+  module Formatters
+    # A formatter for 256-color terminals
+    class Terminal256 < Formatter
+      tag 'terminal256'
+
+      # @private
+      attr_reader :theme
+
+      # @param [Hash,Rouge::Theme] theme
+      #   the theme to render with.
+      def initialize(theme = Themes::ThankfulEyes.new)
+        if theme.is_a?(Rouge::Theme)
+          @theme = theme
+        elsif theme.is_a?(Hash)
+          @theme = theme[:theme] || Themes::ThankfulEyes.new
+        else
+          raise ArgumentError, "invalid theme: #{theme.inspect}"
+        end
+      end
+
+      def stream(tokens, &b)
+        tokens.each do |tok, val|
+          escape = escape_sequence(tok)
+          yield escape.style_string
+          yield val.gsub("\n", "#{escape.reset_string}\n#{escape.style_string}")
+          yield escape.reset_string
+        end
+      end
+
+      class EscapeSequence
+        attr_reader :style
+        def initialize(style)
+          @style = style
+        end
+
+        def self.xterm_colors
+          @xterm_colors ||= [].tap do |out|
+            # colors 0..15: 16 basic colors
+            out << [0x00, 0x00, 0x00] # 0
+            out << [0xcd, 0x00, 0x00] # 1
+            out << [0x00, 0xcd, 0x00] # 2
+            out << [0xcd, 0xcd, 0x00] # 3
+            out << [0x00, 0x00, 0xee] # 4
+            out << [0xcd, 0x00, 0xcd] # 5
+            out << [0x00, 0xcd, 0xcd] # 6
+            out << [0xe5, 0xe5, 0xe5] # 7
+            out << [0x7f, 0x7f, 0x7f] # 8
+            out << [0xff, 0x00, 0x00] # 9
+            out << [0x00, 0xff, 0x00] # 10
+            out << [0xff, 0xff, 0x00] # 11
+            out << [0x5c, 0x5c, 0xff] # 12
+            out << [0xff, 0x00, 0xff] # 13
+            out << [0x00, 0xff, 0xff] # 14
+            out << [0xff, 0xff, 0xff] # 15
+
+            # colors 16..232: the 6x6x6 color cube
+            valuerange = [0x00, 0x5f, 0x87, 0xaf, 0xd7, 0xff]
+
+            217.times do |i|
+              r = valuerange[(i / 36) % 6]
+              g = valuerange[(i / 6) % 6]
+              b = valuerange[i % 6]
+              out << [r, g, b]
+            end
+
+            # colors 233..253: grayscale
+            1.upto 22 do |i|
+              v = 8 + i * 10
+              out << [v, v, v]
+            end
+          end
+        end
+
+        def fg
+          return @fg if instance_variable_defined? :@fg
+          @fg = style.fg && self.class.color_index(style.fg)
+        end
+
+        def bg
+          return @bg if instance_variable_defined? :@bg
+          @bg = style.bg && self.class.color_index(style.bg)
+        end
+
+        def style_string
+          @style_string ||= begin
+            attrs = []
+
+            attrs << ['38', '5', fg.to_s] if fg
+            attrs << ['48', '5', bg.to_s] if bg
+            attrs << '01' if style[:bold]
+            attrs << '04' if style[:italic] # underline, but hey, whatevs
+            escape(attrs)
+          end
+        end
+
+        def reset_string
+          @reset_string ||= begin
+            attrs = []
+            attrs << '39' if fg # fg reset
+            attrs << '49' if bg # bg reset
+            attrs << '00' if style[:bold] || style[:italic]
+
+            escape(attrs)
+          end
+        end
+
+      private
+        def escape(attrs)
+          return '' if attrs.empty?
+          "\e[#{attrs.join(';')}m"
+        end
+
+        def self.color_index(color)
+          @color_index_cache ||= {}
+          @color_index_cache[color] ||= closest_color(*get_rgb(color))
+        end
+
+        def self.get_rgb(color)
+          color = $1 if color =~ /#([0-9a-f]+)/i
+          hexes = case color.size
+          when 3
+            color.chars.map { |c| "#{c}#{c}" }
+          when 6
+            color.scan(/../)
+          else
+            raise "invalid color: #{color}"
+          end
+
+          hexes.map { |h| h.to_i(16) }
+        end
+
+        # max distance between two colors, #000000 to #ffffff
+        MAX_DISTANCE = 257 * 257 * 3
+
+        def self.closest_color(r, g, b)
+          @@colors_cache ||= {}
+          key = (r << 16) + (g << 8) + b
+          @@colors_cache.fetch(key) do
+            distance = MAX_DISTANCE
+
+            match = 0
+
+            xterm_colors.each_with_index do |(cr, cg, cb), i|
+              d = (r - cr)**2 + (g - cg)**2 + (b - cb)**2
+              next if d >= distance
+
+              match = i
+              distance = d
+            end
+
+            match
+          end
+        end
+      end
+
+    # private
+      def escape_sequence(token)
+        @escape_sequences ||= {}
+        @escape_sequences[token.qualname] ||=
+          EscapeSequence.new(get_style(token))
+      end
+
+      def get_style(token)
+        return text_style if token.ancestors.include? Token::Tokens::Text
+
+        theme.get_own_style(token) || text_style
+      end
+
+      def text_style
+        style = theme.get_style(Token['Text'])
+        # don't highlight text backgrounds
+        style.delete :bg
+        style
+      end
+    end
+  end
+end

data/lib/rouge/guesser.rb

@@ -0,0 +1,57 @@
+# frozen_string_literal: true
+
+module Rouge
+  class Guesser
+    class Ambiguous < StandardError
+      attr_reader :alternatives
+      def initialize(alternatives); @alternatives = alternatives; end
+
+      def message
+        "Ambiguous guess: can't decide between #{alternatives.map(&:tag).inspect}"
+      end
+    end
+
+    def self.guess(guessers, lexers)
+      original_size = lexers.size
+
+      guessers.each do |g|
+        new_lexers = case g
+        when Guesser then g.filter(lexers)
+        when proc { |x| x.respond_to? :call } then g.call(lexers)
+        else raise "bad guesser: #{g}"
+        end
+
+        lexers = new_lexers && new_lexers.any? ? new_lexers : lexers
+      end
+
+      # if we haven't filtered the input at *all*,
+      # then we have no idea what language it is,
+      # so we bail and return [].
+      lexers.size < original_size ? lexers : []
+    end
+
+    def collect_best(lexers, opts={}, &scorer)
+      best = []
+      best_score = opts[:threshold]
+
+      lexers.each do |lexer|
+        score = scorer.call(lexer)
+
+        next if score.nil?
+
+        if best_score.nil? || score > best_score
+          best_score = score
+          best = [lexer]
+        elsif score == best_score
+          best << lexer
+        end
+      end
+
+      best
+    end
+
+    def filter(lexers)
+      raise 'abstract'
+    end
+  end
+end

data/lib/rouge/guessers/disambiguation.rb

@@ -0,0 +1,106 @@
+# frozen_string_literal: true
+
+module Rouge
+  module Guessers
+    class Disambiguation < Guesser
+      include Util
+      include Lexers
+
+      def initialize(filename, source)
+        @filename = File.basename(filename)
+        @source = source
+      end
+
+      def filter(lexers)
+        return lexers if lexers.size == 1
+        return lexers if lexers.size == Lexer.all.size
+
+        @analyzer = TextAnalyzer.new(get_source(@source))
+
+        self.class.disambiguators.each do |disambiguator|
+          next unless disambiguator.match?(@filename)
+
+          filtered = disambiguator.decide!(self)
+          return filtered if filtered
+        end
+
+        return lexers
+      end
+
+      def contains?(text)
+        return @analyzer.include?(text)
+      end
+
+      def matches?(re)
+        return !!(@analyzer =~ re)
+      end
+
+      @disambiguators = []
+      def self.disambiguate(*patterns, &decider)
+        @disambiguators << Disambiguator.new(patterns, &decider)
+      end
+
+      def self.disambiguators
+        @disambiguators
+      end
+
+      class Disambiguator
+        include Util
+
+        def initialize(patterns, &decider)
+          @patterns = patterns
+          @decider = decider
+        end
+
+        def decide!(guesser)
+          out = guesser.instance_eval(&@decider)
+          case out
+          when Array then out
+          when nil then nil
+          else [out]
+          end
+        end
+
+        def match?(filename)
+          @patterns.any? { |p| test_glob(p, filename) }
+        end
+      end
+
+      disambiguate '*.pl' do
+        next Perl if contains?('my $')
+        next Prolog if contains?(':-')
+        next Prolog if matches?(/\A\w+(\(\w+\,\s*\w+\))*\./)
+      end
+
+      disambiguate '*.h' do
+        next ObjectiveC if matches?(/@(end|implementation|protocol|property)\b/)
+        next ObjectiveC if contains?('@"')
+
+        C
+      end
+
+      disambiguate '*.m' do
+        next ObjectiveC if matches?(/@(end|implementation|protocol|property)\b/)
+        next ObjectiveC if contains?('@"')
+
+        next Mathematica if contains?('(*')
+        next Mathematica if contains?(':=')
+
+        next Matlab if matches?(/^\s*?%/)
+      end
+
+      disambiguate '*.php' do
+        # PHP always takes precedence over Hack
+        PHP
+      end
+
+      disambiguate '*.hh' do
+        next Cpp if matches?(/^\s*#include/)
+        next Hack if matches?(/^<\?hh/)
+        next Hack if matches?(/(\(|, ?)\$\$/)
+
+        Cpp
+      end
+    end
+  end
+end

data/lib/rouge/guessers/filename.rb

@@ -0,0 +1,27 @@
+# frozen_string_literal: true
+
+module Rouge
+  module Guessers
+    class Filename < Guesser
+      attr_reader :fname
+      def initialize(filename)
+        @filename = filename
+      end
+
+      # returns a list of lexers that match the given filename with
+      # equal specificity (i.e. number of wildcards in the pattern).
+      # This helps disambiguate between, e.g. the Nginx lexer, which
+      # matches `nginx.conf`, and the Conf lexer, which matches `*.conf`.
+      # In this case, nginx will win because the pattern has no wildcards,
+      # while `*.conf` has one.
+      def filter(lexers)
+        mapping = {}
+        lexers.each do |lexer|
+          mapping[lexer.name] = lexer.filenames || []
+        end
+
+        GlobMapping.new(mapping, @filename).filter(lexers)
+      end
+    end
+  end
+end

data/lib/rouge/guessers/glob_mapping.rb

@@ -0,0 +1,45 @@
+# frozen_string_literal: true
+
+module Rouge
+  module Guessers
+    # This class allows for custom behavior
+    # with glob -> lexer name mappings
+    class GlobMapping < Guesser
+      include Util
+
+      def self.by_pairs(mapping, filename)
+        glob_map = {}
+        mapping.each do |(glob, lexer_name)|
+          lexer = Lexer.find(lexer_name)
+
+          # ignore unknown lexers
+          next unless lexer
+
+          glob_map[lexer.name] ||= []
+          glob_map[lexer.name] << glob
+        end
+
+        new(glob_map, filename)
+      end
+
+      attr_reader :glob_map, :filename
+      def initialize(glob_map, filename)
+        @glob_map = glob_map
+        @filename = filename
+      end
+
+      def filter(lexers)
+        basename = File.basename(filename)
+
+        collect_best(lexers) do |lexer|
+          score = (@glob_map[lexer.name] || []).map do |pattern|
+            if test_glob(pattern, basename)
+              # specificity is better the fewer wildcards there are
+              -pattern.scan(/[*?\[]/).size
+            end
+          end.compact.min
+        end
+      end
+    end
+  end
+end

data/lib/rouge/guessers/mimetype.rb

@@ -0,0 +1,16 @@
+# frozen_string_literal: true
+
+module Rouge
+  module Guessers
+    class Mimetype < Guesser
+      attr_reader :mimetype
+      def initialize(mimetype)
+        @mimetype = mimetype
+      end
+
+      def filter(lexers)
+        lexers.select { |lexer| lexer.mimetypes.include? @mimetype }
+      end
+    end
+  end
+end

data/lib/rouge/guessers/modeline.rb

@@ -0,0 +1,46 @@
+# frozen_string_literal: true
+
+module Rouge
+  module Guessers
+    class Modeline < Guesser
+      include Util
+
+      # [jneen] regexen stolen from linguist
+      EMACS_MODELINE = /-\*-\s*(?:(?!mode)[\w-]+\s*:\s*(?:[\w+-]+)\s*;?\s*)*(?:mode\s*:)?\s*([\w+-]+)\s*(?:;\s*(?!mode)[\w-]+\s*:\s*[\w+-]+\s*)*;?\s*-\*-/i
+
+      # First form vim modeline
+      # [text]{white}{vi:|vim:|ex:}[white]{options}
+      # ex: 'vim: syntax=ruby'
+      VIM_MODELINE_1 = /(?:vim|vi|ex):\s*(?:ft|filetype|syntax)=(\w+)\s?/i
+
+      # Second form vim modeline (compatible with some versions of Vi)
+      # [text]{white}{vi:|vim:|Vim:|ex:}[white]se[t] {options}:[text]
+      # ex: 'vim set syntax=ruby:'
+      VIM_MODELINE_2 = /(?:vim|vi|Vim|ex):\s*se(?:t)?.*\s(?:ft|filetype|syntax)=(\w+)\s?.*:/i
+
+      MODELINES = [EMACS_MODELINE, VIM_MODELINE_1, VIM_MODELINE_2]
+
+      def initialize(source, opts={})
+        @source = source
+        @lines = opts[:lines] || 5
+      end
+
+      def filter(lexers)
+        # don't bother reading the stream if we've already decided
+        return lexers if lexers.size == 1
+
+        source_text = get_source(@source)
+
+        lines = source_text.split(/\n/)
+
+        search_space = (lines.first(@lines) + lines.last(@lines)).join("\n")
+
+        matches = MODELINES.map { |re| re.match(search_space) }.compact
+        return lexers unless matches.any?
+        
+        match_set = Set.new(matches.map { |m| m[1] })
+        lexers.select { |l| match_set.include?(l.tag) || l.aliases.any? { |a| match_set.include?(a) } }
+      end
+    end
+  end
+end

data/lib/rouge/guessers/source.rb

@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+module Rouge
+  module Guessers
+    class Source < Guesser
+      include Util
+
+      attr_reader :source
+      def initialize(source)
+        @source = source
+      end
+
+      def filter(lexers)
+        # don't bother reading the input if
+        # we've already filtered to 1
+        return lexers if lexers.size == 1
+
+        source_text = get_source(@source)
+
+        Lexer.assert_utf8!(source_text)
+
+        source_text = TextAnalyzer.new(source_text)
+
+        collect_best(lexers) do |lexer|
+          next unless lexer.methods(false).include? :detect?
+          lexer.detect?(source_text) ? 1 : nil
+        end
+      end
+    end
+  end
+end