Chris@210: require 'set'
Chris@210: 
Chris@210: module CodeRay
Chris@210: module Encoders
Chris@210: 
Chris@210:   # = HTML Encoder
Chris@210:   #
Chris@210:   # This is CodeRay's most important highlighter:
Chris@210:   # It provides save, fast XHTML generation and CSS support.
Chris@210:   #
Chris@210:   # == Usage
Chris@210:   #
Chris@210:   #  require 'coderay'
Chris@210:   #  puts CodeRay.scan('Some /code/', :ruby).html  #-> a HTML page
Chris@210:   #  puts CodeRay.scan('Some /code/', :ruby).html(:wrap => :span)
Chris@210:   #  #-> <span class="CodeRay"><span class="co">Some</span> /code/</span>
Chris@210:   #  puts CodeRay.scan('Some /code/', :ruby).span  #-> the same
Chris@210:   #  
Chris@210:   #  puts CodeRay.scan('Some code', :ruby).html(
Chris@210:   #    :wrap => nil,
Chris@210:   #    :line_numbers => :inline,
Chris@210:   #    :css => :style
Chris@210:   #  )
Chris@210:   #  #-> <span class="no">1</span>  <span style="color:#036; font-weight:bold;">Some</span> code
Chris@210:   #
Chris@210:   # == Options
Chris@210:   #
Chris@210:   # === :tab_width
Chris@210:   # Convert \t characters to +n+ spaces (a number.)
Chris@210:   # Default: 8
Chris@210:   #
Chris@210:   # === :css
Chris@210:   # How to include the styles; can be :class or :style.
Chris@210:   #
Chris@210:   # Default: :class
Chris@210:   #
Chris@210:   # === :wrap
Chris@210:   # Wrap in :page, :div, :span or nil.
Chris@210:   #
Chris@210:   # You can also use Encoders::Div and Encoders::Span.
Chris@210:   #
Chris@210:   # Default: nil
Chris@210:   #
Chris@210:   # === :title
Chris@210:   # 
Chris@210:   # The title of the HTML page (works only when :wrap is set to :page.)
Chris@210:   #
Chris@210:   # Default: 'CodeRay output'
Chris@210:   #
Chris@210:   # === :line_numbers
Chris@210:   # Include line numbers in :table, :inline, :list or nil (no line numbers)
Chris@210:   #
Chris@210:   # Default: nil
Chris@210:   #
Chris@210:   # === :line_number_start
Chris@210:   # Where to start with line number counting.
Chris@210:   #
Chris@210:   # Default: 1
Chris@210:   #
Chris@210:   # === :bold_every
Chris@210:   # Make every +n+-th number appear bold.
Chris@210:   #
Chris@210:   # Default: 10
Chris@210:   #
Chris@210:   # === :highlight_lines
Chris@210:   # 
Chris@210:   # Highlights certain line numbers.
Chris@210:   # Can be any Enumerable, typically just an Array or Range, of numbers.
Chris@210:   # 
Chris@210:   # Bolding is deactivated when :highlight_lines is set. It only makes sense
Chris@210:   # in combination with :line_numbers.
Chris@210:   #
Chris@210:   # Default: nil
Chris@210:   #
Chris@210:   # === :hint
Chris@210:   # Include some information into the output using the title attribute.
Chris@210:   # Can be :info (show token type on mouse-over), :info_long (with full path)
Chris@210:   # or :debug (via inspect).
Chris@210:   #
Chris@210:   # Default: false
Chris@210:   class HTML < Encoder
Chris@210: 
Chris@210:     include Streamable
Chris@210:     register_for :html
Chris@210: 
Chris@210:     FILE_EXTENSION = 'html'
Chris@210: 
Chris@210:     DEFAULT_OPTIONS = {
Chris@210:       :tab_width => 8,
Chris@210: 
Chris@210:       :css => :class,
Chris@210: 
Chris@210:       :style => :cycnus,
Chris@210:       :wrap => nil,
Chris@210:       :title => 'CodeRay output',
Chris@210: 
Chris@210:       :line_numbers => nil,
Chris@210:       :line_number_start => 1,
Chris@210:       :bold_every => 10,
Chris@210:       :highlight_lines => nil,
Chris@210: 
Chris@210:       :hint => false,
Chris@210:     }
Chris@210: 
Chris@210:     helper :output, :css
Chris@210: 
Chris@210:     attr_reader :css
Chris@210: 
Chris@210:   protected
Chris@210: 
Chris@210:     HTML_ESCAPE = {  #:nodoc:
Chris@210:       '&' => '&amp;',
Chris@210:       '"' => '&quot;',
Chris@210:       '>' => '&gt;',
Chris@210:       '<' => '&lt;',
Chris@210:     }
Chris@210: 
Chris@210:     # This was to prevent illegal HTML.
Chris@210:     # Strange chars should still be avoided in codes.
Chris@210:     evil_chars = Array(0x00...0x20) - [?\n, ?\t, ?\s]
Chris@210:     evil_chars.each { |i| HTML_ESCAPE[i.chr] = ' ' }
Chris@210:     #ansi_chars = Array(0x7f..0xff)
Chris@210:     #ansi_chars.each { |i| HTML_ESCAPE[i.chr] = '&#%d;' % i }
Chris@210:     # \x9 (\t) and \xA (\n) not included
Chris@210:     #HTML_ESCAPE_PATTERN = /[\t&"><\0-\x8\xB-\x1f\x7f-\xff]/
Chris@210:     HTML_ESCAPE_PATTERN = /[\t"&><\0-\x8\xB-\x1f]/
Chris@210: 
Chris@210:     TOKEN_KIND_TO_INFO = Hash.new { |h, kind|
Chris@210:       h[kind] =
Chris@210:         case kind
Chris@210:         when :pre_constant
Chris@210:           'Predefined constant'
Chris@210:         else
Chris@210:           kind.to_s.gsub(/_/, ' ').gsub(/\b\w/) { $&.capitalize }
Chris@210:         end
Chris@210:     }
Chris@210: 
Chris@210:     TRANSPARENT_TOKEN_KINDS = [
Chris@210:       :delimiter, :modifier, :content, :escape, :inline_delimiter,
Chris@210:     ].to_set
Chris@210: 
Chris@210:     # Generate a hint about the given +classes+ in a +hint+ style.
Chris@210:     #
Chris@210:     # +hint+ may be :info, :info_long or :debug.
Chris@210:     def self.token_path_to_hint hint, classes
Chris@210:       title =
Chris@210:         case hint
Chris@210:         when :info
Chris@210:           TOKEN_KIND_TO_INFO[classes.first]
Chris@210:         when :info_long
Chris@210:           classes.reverse.map { |kind| TOKEN_KIND_TO_INFO[kind] }.join('/')
Chris@210:         when :debug
Chris@210:           classes.inspect
Chris@210:         end
Chris@210:       title ? " title=\"#{title}\"" : ''
Chris@210:     end
Chris@210: 
Chris@210:     def setup options
Chris@210:       super
Chris@210: 
Chris@210:       @HTML_ESCAPE = HTML_ESCAPE.dup
Chris@210:       @HTML_ESCAPE["\t"] = ' ' * options[:tab_width]
Chris@210: 
Chris@210:       @opened = [nil]
Chris@210:       @css = CSS.new options[:style]
Chris@210: 
Chris@210:       hint = options[:hint]
Chris@210:       if hint and not [:debug, :info, :info_long].include? hint
Chris@210:         raise ArgumentError, "Unknown value %p for :hint; \
Chris@210:           expected :info, :debug, false, or nil." % hint
Chris@210:       end
Chris@210: 
Chris@210:       case options[:css]
Chris@210: 
Chris@210:       when :class
Chris@210:         @css_style = Hash.new do |h, k|
Chris@210:           c = CodeRay::Tokens::ClassOfKind[k.first]
Chris@210:           if c == :NO_HIGHLIGHT and not hint
Chris@210:             h[k.dup] = false
Chris@210:           else
Chris@210:             title = if hint
Chris@210:               HTML.token_path_to_hint(hint, k[1..-1] << k.first)
Chris@210:             else
Chris@210:               ''
Chris@210:             end
Chris@210:             if c == :NO_HIGHLIGHT
Chris@210:               h[k.dup] = '<span%s>' % [title]
Chris@210:             else
Chris@210:               h[k.dup] = '<span%s class="%s">' % [title, c]
Chris@210:             end
Chris@210:           end
Chris@210:         end
Chris@210: 
Chris@210:       when :style
Chris@210:         @css_style = Hash.new do |h, k|
Chris@210:           if k.is_a? ::Array
Chris@210:             styles = k.dup
Chris@210:           else
Chris@210:             styles = [k]
Chris@210:           end
Chris@210:           type = styles.first
Chris@210:           classes = styles.map { |c| Tokens::ClassOfKind[c] }
Chris@210:           if classes.first == :NO_HIGHLIGHT and not hint
Chris@210:             h[k] = false
Chris@210:           else
Chris@210:             styles.shift if TRANSPARENT_TOKEN_KINDS.include? styles.first
Chris@210:             title = HTML.token_path_to_hint hint, styles
Chris@210:             style = @css[*classes]
Chris@210:             h[k] =
Chris@210:               if style
Chris@210:                 '<span%s style="%s">' % [title, style]
Chris@210:               else
Chris@210:                 false
Chris@210:               end
Chris@210:           end
Chris@210:         end
Chris@210: 
Chris@210:       else
Chris@210:         raise ArgumentError, "Unknown value %p for :css." % options[:css]
Chris@210: 
Chris@210:       end
Chris@210:     end
Chris@210: 
Chris@210:     def finish options
Chris@210:       not_needed = @opened.shift
Chris@210:       @out << '</span>' * @opened.size
Chris@210:       unless @opened.empty?
Chris@210:         warn '%d tokens still open: %p' % [@opened.size, @opened]
Chris@210:       end
Chris@210: 
Chris@210:       @out.extend Output
Chris@210:       @out.css = @css
Chris@210:       @out.numerize! options[:line_numbers], options
Chris@210:       @out.wrap! options[:wrap]
Chris@210:       @out.apply_title! options[:title]
Chris@210: 
Chris@210:       super
Chris@210:     end
Chris@210: 
Chris@210:     def token text, type = :plain
Chris@210:       case text
Chris@210:       
Chris@210:       when nil
Chris@210:         # raise 'Token with nil as text was given: %p' % [[text, type]] 
Chris@210:       
Chris@210:       when String
Chris@210:         if text =~ /#{HTML_ESCAPE_PATTERN}/o
Chris@210:           text = text.gsub(/#{HTML_ESCAPE_PATTERN}/o) { |m| @HTML_ESCAPE[m] }
Chris@210:         end
Chris@210:         @opened[0] = type
Chris@210:         if text != "\n" && style = @css_style[@opened]
Chris@210:           @out << style << text << '</span>'
Chris@210:         else
Chris@210:           @out << text
Chris@210:         end
Chris@210:         
Chris@210:       
Chris@210:       # token groups, eg. strings
Chris@210:       when :open
Chris@210:         @opened[0] = type
Chris@210:         @out << (@css_style[@opened] || '<span>')
Chris@210:         @opened << type
Chris@210:       when :close
Chris@210:         if @opened.empty?
Chris@210:           # nothing to close
Chris@210:         else
Chris@210:           if $CODERAY_DEBUG and (@opened.size == 1 or @opened.last != type)
Chris@210:             raise 'Malformed token stream: Trying to close a token (%p) \
Chris@210:               that is not open. Open are: %p.' % [type, @opened[1..-1]]
Chris@210:           end
Chris@210:           @out << '</span>'
Chris@210:           @opened.pop
Chris@210:         end
Chris@210:       
Chris@210:       # whole lines to be highlighted, eg. a deleted line in a diff
Chris@210:       when :begin_line
Chris@210:         @opened[0] = type
Chris@210:         if style = @css_style[@opened]
Chris@210:           if style['class="']
Chris@210:             @out << style.sub('class="', 'class="line ')
Chris@210:           else
Chris@210:             @out << style.sub('>', ' class="line">')
Chris@210:           end
Chris@210:         else
Chris@210:           @out << '<span class="line">'
Chris@210:         end
Chris@210:         @opened << type
Chris@210:       when :end_line
Chris@210:         if @opened.empty?
Chris@210:           # nothing to close
Chris@210:         else
Chris@210:           if $CODERAY_DEBUG and (@opened.size == 1 or @opened.last != type)
Chris@210:             raise 'Malformed token stream: Trying to close a line (%p) \
Chris@210:               that is not open. Open are: %p.' % [type, @opened[1..-1]]
Chris@210:           end
Chris@210:           @out << '</span>'
Chris@210:           @opened.pop
Chris@210:         end
Chris@210:       
Chris@210:       else
Chris@210:         raise 'unknown token kind: %p' % [text]
Chris@210:         
Chris@210:       end
Chris@210:     end
Chris@210: 
Chris@210:   end
Chris@210: 
Chris@210: end
Chris@210: end