redcloth.rb

来自「用ruby on rails写的一个博客程序,还不错..ruby on rail」· RB 代码 · 共 1,131 行 · 第 1/3 页

                /(\W)
                (#{rcq})
                (#{C})
                (?::(\S+?))?
                (\S.*?\S|\S)
                #{rcq}
                (?=\W)/x
            else
                /(#{rcq})
                (#{C})
                (?::(\S+))?
                (\S.*?\S|\S)
                #{rcq}/xm 
            end
        [rc, ht, re, rtype]
    end

    # Elements to handle
    GLYPHS = [
    #   [ /([^\s\[{(>])?\'([dmst]\b|ll\b|ve\b|\s|:|$)/, '\1’\2' ], # single closing
        [ /([^\s\[{(>#{PUNCT_Q}][#{PUNCT_Q}]*)\'/, '\1’' ], # single closing
        [ /\'(?=[#{PUNCT_Q}]*(s\b|[\s#{PUNCT_NOQ}]))/, '’' ], # single closing
        [ /\'/, '‘' ], # single opening
        [ /</, '&lt;' ], # less-than
        [ />/, '&gt;' ], # greater-than
    #   [ /([^\s\[{(])?"(\s|:|$)/, '\1&#8221;\2' ], # double closing
        [ /([^\s\[{(>#{PUNCT_Q}][#{PUNCT_Q}]*)"/, '\1&#8221;' ], # double closing
        [ /"(?=[#{PUNCT_Q}]*[\s#{PUNCT_NOQ}])/, '&#8221;' ], # double closing
        [ /"/, '&#8220;' ], # double opening
        [ /\b( )?\.{3}/, '\1&#8230;' ], # ellipsis
        [ /\b([A-Z][A-Z0-9]{2,})\b(?:[(]([^)]*)[)])/, '<acronym title="\2">\1</acronym>' ], # 3+ uppercase acronym
        [ /(^|[^"][>\s])([A-Z][A-Z0-9 ]+[A-Z0-9])([^<A-Za-z0-9]|$)/, '\1<span class="caps">\2</span>\3', :no_span_caps ], # 3+ uppercase caps
        [ /(\.\s)?\s?--\s?/, '\1&#8212;' ], # em dash
        [ /\s->\s/, ' &rarr; ' ], # right arrow
        [ /\s-\s/, ' &#8211; ' ], # en dash
        [ /(\d+) ?x ?(\d+)/, '\1&#215;\2' ], # dimension sign
        [ /\b ?[(\[]TM[\])]/i, '&#8482;' ], # trademark
        [ /\b ?[(\[]R[\])]/i, '&#174;' ], # registered
        [ /\b ?[(\[]C[\])]/i, '&#169;' ] # copyright
    ]

    H_ALGN_VALS = {
        '<' => 'left',
        '=' => 'center',
        '>' => 'right',
        '<>' => 'justify'
    }

    V_ALGN_VALS = {
        '^' => 'top',
        '-' => 'middle',
        '~' => 'bottom'
    }

    #
    # Flexible HTML escaping
    #
    def htmlesc( str, mode )
        str.gsub!( '&', '&amp;' )
        str.gsub!( '"', '&quot;' ) if mode != :NoQuotes
        str.gsub!( "'", '&#039;' ) if mode == :Quotes
        str.gsub!( '<', '&lt;')
        str.gsub!( '>', '&gt;')
    end

    # Search and replace for Textile glyphs (quotes, dashes, other symbols)
    def pgl( text )
        GLYPHS.each do |re, resub, tog|
            next if tog and method( tog ).call
            text.gsub! re, resub
        end
    end

    # Parses Textile attribute lists and builds an HTML attribute string
    def pba( text_in, element = "" )
        
        return '' unless text_in

        style = []
        text = text_in.dup
        if element == 'td'
            colspan = $1 if text =~ /\\(\d+)/
            rowspan = $1 if text =~ /\/(\d+)/
            style << "vertical-align:#{ v_align( $& ) };" if text =~ A_VLGN
        end

        style << "#{ $1 };" if not filter_styles and
            text.sub!( /\{([^}]*)\}/, '' )

        lang = $1 if
            text.sub!( /\[([^)]+?)\]/, '' )

        cls = $1 if
            text.sub!( /\(([^()]+?)\)/, '' )
                        
        style << "padding-left:#{ $1.length }em;" if
            text.sub!( /([(]+)/, '' )

        style << "padding-right:#{ $1.length }em;" if text.sub!( /([)]+)/, '' )

        style << "text-align:#{ h_align( $& ) };" if text =~ A_HLGN

        cls, id = $1, $2 if cls =~ /^(.*?)#(.*)$/
        
        atts = ''
        atts << " style=\"#{ style.join }\"" unless style.empty?
        atts << " class=\"#{ cls }\"" unless cls.to_s.empty?
        atts << " lang=\"#{ lang }\"" if lang
        atts << " id=\"#{ id }\"" if id
        atts << " colspan=\"#{ colspan }\"" if colspan
        atts << " rowspan=\"#{ rowspan }\"" if rowspan
        
        atts
    end

    TABLE_RE = /^(?:table(_?#{S}#{A}#{C})\. ?\n)?^(#{A}#{C}\.? ?\|.*?\|)(\n\n|\Z)/m
    
    # Parses a Textile table block, building HTML from the result.
    def block_textile_table( text ) 
        text.gsub!( TABLE_RE ) do |matches|

            tatts, fullrow = $~[1..2]
            tatts = pba( tatts, 'table' )
            tatts = shelve( tatts ) if tatts
            rows = []

            fullrow.
            split( /\|$/m ).
            delete_if { |x| x.empty? }.
            each do |row|

                ratts, row = pba( $1, 'tr' ), $2 if row =~ /^(#{A}#{C}\. )(.*)/m
                
                cells = []
                row.split( '|' ).each do |cell|
                    ctyp = 'd'
                    ctyp = 'h' if cell =~ /^_/

                    catts = ''
                    catts, cell = pba( $1, 'td' ), $2 if cell =~ /^(_?#{S}#{A}#{C}\. ?)(.*)/

                    unless cell.strip.empty?
                        catts = shelve( catts ) if catts
                        cells << "\t\t\t<t#{ ctyp }#{ catts }>#{ cell }</t#{ ctyp }>" 
                    end
                end
                ratts = shelve( ratts ) if ratts
                rows << "\t\t<tr#{ ratts }>\n#{ cells.join( "\n" ) }\n\t\t</tr>"
            end
            "\t<table#{ tatts }>\n#{ rows.join( "\n" ) }\n\t</table>\n\n"
        end
    end

    LISTS_RE = /^([#*]+?#{C} .*?)$(?![^#*])/m
    LISTS_CONTENT_RE = /^([#*]+)(#{A}#{C}) (.*)$/m

    # Parses Textile lists and generates HTML
    def block_textile_lists( text ) 
        text.gsub!( LISTS_RE ) do |match|
            lines = match.split( /\n/ )
            last_line = -1
            depth = []
            lines.each_with_index do |line, line_id|
                if line =~ LISTS_CONTENT_RE 
                    tl,atts,content = $~[1..3]
                    if depth.last
                        if depth.last.length > tl.length
                            (depth.length - 1).downto(0) do |i|
                                break if depth[i].length == tl.length
                                lines[line_id - 1] << "</li>\n\t</#{ lT( depth[i] ) }l>\n\t"
                                depth.pop
                            end
                        end
                        if depth.last and depth.last.length == tl.length
                            lines[line_id - 1] << '</li>'
                        end
                    end
                    unless depth.last == tl
                        depth << tl
                        atts = pba( atts )
                        atts = shelve( atts ) if atts
                        lines[line_id] = "\t<#{ lT(tl) }l#{ atts }>\n\t<li>#{ content }"
                    else
                        lines[line_id] = "\t\t<li>#{ content }"
                    end
                    last_line = line_id

                else
                    last_line = line_id
                end
                if line_id - last_line > 1 or line_id == lines.length - 1
                    depth.delete_if do |v|
                        lines[last_line] << "</li>\n\t</#{ lT( v ) }l>"
                    end
                end
            end
            lines.join( "\n" )
        end
    end

    CODE_RE = /(\W)
        @
        (?:\|(\w+?)\|)?
        (.+?)
        @
        (?=\W)/x

    def inline_textile_code( text ) 
        text.gsub!( CODE_RE ) do |m|
            before,lang,code,after = $~[1..4]
            lang = " lang=\"#{ lang }\"" if lang
            rip_offtags( "#{ before }<code#{ lang }>#{ code }</code>#{ after }" )
        end
    end

    def lT( text ) 
        text =~ /\#$/ ? 'o' : 'u'
    end

    def hard_break( text )
        text.gsub!( /(.)\n(?!\Z| *([#*=]+(\s|$)|[{|]))/, "\\1<br />" ) if hard_breaks
    end

    BLOCKS_GROUP_RE = /\n{2,}(?! )/m

    def blocks( text, deep_code = false )
        text.replace( text.split( BLOCKS_GROUP_RE ).collect do |blk|
            plain = blk !~ /\A[#*> ]/

            # skip blocks that are complex HTML
            if blk =~ /^<\/?(\w+).*>/ and not SIMPLE_HTML_TAGS.include? $1
                blk
            else
                # search for indentation levels
                blk.strip!
                if blk.empty?
                    blk
                else
                    code_blk = nil
                    blk.gsub!( /((?:\n(?:\n^ +[^\n]*)+)+)/m ) do |iblk|
                        flush_left iblk
                        blocks iblk, plain
                        iblk.gsub( /^(\S)/, "\t\\1" )
                        if plain
                            code_blk = iblk; ""
                        else
                            iblk
                        end
                    end

                    block_applied = 0 
                    @rules.each do |rule_name|
                        block_applied += 1 if ( rule_name.to_s.match /^block_/ and method( rule_name ).call( blk ) )
                    end
                    if block_applied.zero?
                        if deep_code
                            blk = "\t<pre><code>#{ blk }</code></pre>"
                        else
                            blk = "\t<p>#{ blk }</p>"
                        end
                    end
                    # hard_break blk
                    blk + "\n#{ code_blk }"
                end
            end

        end.join( "\n\n" ) )
    end

    def textile_bq( tag, atts, cite, content )
        cite, cite_title = check_refs( cite )
        cite = " cite=\"#{ cite }\"" if cite
        atts = shelve( atts ) if atts
        "\t<blockquote#{ cite }>\n\t\t<p#{ atts }>#{ content }</p>\n\t</blockquote>"
    end

    def textile_p( tag, atts, cite, content )
        atts = shelve( atts ) if atts
        "\t<#{ tag }#{ atts }>#{ content }</#{ tag }>"
    end

    alias textile_h1 textile_p
    alias textile_h2 textile_p
    alias textile_h3 textile_p
    alias textile_h4 textile_p
    alias textile_h5 textile_p
    alias textile_h6 textile_p

    def textile_fn_( tag, num, atts, cite, content )
        atts << " id=\"fn#{ num }\""
        content = "<sup>#{ num }</sup> #{ content }"
        atts = shelve( atts ) if atts
        "\t<p#{ atts }>#{ content }</p>"
    end

    BLOCK_RE = /^(([a-z]+)(\d*))(#{A}#{C})\.(?::(\S+))? (.*)$/m

    def block_textile_prefix( text ) 
        if text =~ BLOCK_RE
            tag,tagpre,num,atts,cite,content = $~[1..6]
            atts = pba( atts )

            # pass to prefix handler
            if respond_to? "textile_#{ tag }", true
                text.gsub!( $&, method( "textile_#{ tag }" ).call( tag, atts, cite, content ) )
            elsif respond_to? "textile_#{ tagpre }_", true
                text.gsub!( $&, method( "textile_#{ tagpre }_" ).call( tagpre, num, atts, cite, content ) )
            end
        end
    end
    
    SETEXT_RE = /\A(.+?)\n([=-])[=-]* *$/m
    def block_markdown_setext( text )
        if text =~ SETEXT_RE
            tag = if $2 == "="; "h1"; else; "h2"; end
            blk, cont = "<#{ tag }>#{ $1 }</#{ tag }>", $'
            blocks cont
            text.replace( blk + cont )
        end
    end

    ATX_RE = /\A(\#{1,6})  # $1 = string of #'s
              [ ]*
              (.+?)       # $2 = Header text
              [ ]*
              \#*         # optional closing #'s (not counted)
              $/x
    def block_markdown_atx( text )
        if text =~ ATX_RE
            tag = "h#{ $1.length }"
            blk, cont = "<#{ tag }>#{ $2 }</#{ tag }>\n\n", $'
            blocks cont
            text.replace( blk + cont )
        end
    end

    MARKDOWN_BQ_RE = /\A(^ *> ?.+$(.+\n)*\n*)+/m

    def block_markdown_bq( text )
        text.gsub!( MARKDOWN_BQ_RE ) do |blk|
            blk.gsub!( /^ *> ?/, '' )
            flush_left blk
            blocks blk
            blk.gsub!( /^(\S)/, "\t\\1" )
            "<blockquote>\n#{ blk }\n</blockquote>\n\n"
        end
    end

    MARKDOWN_RULE_RE = /^(#{
        ['*', '-', '_'].collect { |ch| '( ?' + Regexp::quote( ch ) + ' ?){3,}' }.join( '|' )
    })$/

    def block_markdown_rule( text )
        text.gsub!( MARKDOWN_RULE_RE ) do |blk|
            "<hr />"
        end
    end

    # XXX TODO XXX
    def block_markdown_lists( text )
    end

    def inline_textile_span( text ) 
        QTAGS.each do |qtag_rc, ht, qtag_re, rtype|
            text.gsub!( qtag_re ) do |m|
             
                case rtype
                when :limit
                    sta,qtag,atts,cite,content = $~[1..5]
                else
                    qtag,atts,cite,content = $~[1..4]
                    sta = ''
                end
                atts = pba( atts )
                atts << " cite=\"#{ cite }\"" if cite
                atts = shelve( atts ) if atts