#!/usr/bin/env ruby require 'ffi' require 'stringio' require 'cgi' require 'set' require 'uri' module CMark extend FFI::Library ffi_lib ['libcmark', 'cmark'] typedef :pointer, :node enum :node_type, [:document, :blockquote, :list, :list_item, :fenced_code, :indented_code, :html, :paragraph, :atx_header, :setext_header, :hrule, :reference_def, :str, :softbreak, :linebreak, :code, :inline_html, :emph, :strong, :link, :image] enum :list_type, [:no_list, :bullet_list, :ordered_list] attach_function :cmark_free_nodes, [:node], :void attach_function :cmark_node_unlink, [:node], :void attach_function :cmark_markdown_to_html, [:string, :int], :string attach_function :cmark_parse_document, [:string, :int], :node attach_function :cmark_node_first_child, [:node], :node attach_function :cmark_node_parent, [:node], :node attach_function :cmark_node_next, [:node], :node attach_function :cmark_node_previous, [:node], :node attach_function :cmark_node_get_type, [:node], :node_type attach_function :cmark_node_get_string_content, [:node], :string attach_function :cmark_node_get_url, [:node], :string attach_function :cmark_node_get_title, [:node], :string attach_function :cmark_node_get_header_level, [:node], :int attach_function :cmark_node_get_list_type, [:node], :list_type attach_function :cmark_node_get_list_start, [:node], :int attach_function :cmark_node_get_list_tight, [:node], :bool end class Node attr_accessor :type, :children, :string_content, :header_level, :list_type, :list_start, :list_tight, :url, :title def initialize(pointer) if pointer.null? return nil end @pointer = pointer @type = CMark::cmark_node_get_type(pointer) @children = [] first_child = CMark::cmark_node_first_child(pointer) b = first_child while !b.null? @children << Node.new(b) b = CMark::cmark_node_next(b) end @string_content = CMark::cmark_node_get_string_content(pointer) if @type == :atx_header || @type == :setext_header @header_level = CMark::cmark_node_get_header_level(pointer) end if @type == :list @list_type = CMark::cmark_node_get_list_type(pointer) @list_start = CMark::cmark_node_get_list_start(pointer) @list_tight = CMark::cmark_node_get_list_tight(pointer) end if @type == :link || @type == :image @url = CMark::cmark_node_get_url(pointer) if !@url then @url = "" end @title = CMark::cmark_node_get_title(pointer) if !@title then @title = "" end end if @type == :document self.free end end def self.parse_string(s) Node.new(CMark::cmark_parse_document(s, s.bytesize)) end def self.parse_file(f) s = f.read() self.parse_string(s) end def free CMark::cmark_free_nodes(@pointer) end end class Renderer attr_accessor :in_tight, :warnings, :in_plain def initialize(stream = nil) if stream @stream = stream @stringwriter = false else @stringwriter = true @stream = StringIO.new end @need_blocksep = false @warnings = Set.new [] @in_tight = false @in_plain = false end def outf(format, *args) @stream.printf(format, *args) end def out(*args) args.each do |arg| if arg.kind_of?(String) @stream.write(arg) elsif arg.kind_of?(Node) self.render(arg) elsif arg.kind_of?(Array) arg.each { |x| self.out(x) } else @stream.write(arg) end end end def render(node) @node = node if node.type == :document self.document(node) self.out("\n") if @stringwriter return @stream.string end elsif self.in_plain && node.type != :str && node.type != :softbreak # pass through looking for str, softbreak node.children.each do |child| render(child) end else begin self.send(node.type, node) rescue NoMethodError => e @warnings.add("WARNING: " + node.type.to_s + " not implemented.") raise e end end end def document(node) self.out(node.children) end def indented_code(node) self.code_block(node) end def fenced_code(node) self.code_block(node) end def setext_header(node) self.header(node) end def atx_header(node) self.header(node) end def reference_def(node) end def blocksep self.out("\n") end def containersep if !self.in_tight self.out("\n") end end def block(&blk) if @need_blocksep self.blocksep end blk.call @need_blocksep = true end def container(starter, ender, &blk) self.out(starter) self.containersep @need_blocksep = false blk.call self.containersep self.out(ender) end def plain(&blk) old_in_plain = @in_plain @in_plain = true blk.call @in_plain = old_in_plain end end class HtmlRenderer < Renderer def header(node) block do self.out("", node.children, "") end end def paragraph(node) block do if self.in_tight self.out(node.children) else self.out("

", node.children, "

") end end end def list(node) old_in_tight = self.in_tight self.in_tight = node.list_tight block do if node.list_type == :bullet_list container("") do self.out(node.children) end else start = node.list_start == 1 ? '' : (' start="' + node.list_start.to_s + '"') container(start, "") do self.out(node.children) end end end self.in_tight = old_in_tight end def list_item(node) block do container("
  • ", "
  • ") do self.out(node.children) end end end def blockquote(node) block do container("
    ", "
    ") do self.out(node.children) end end end def hrule(node) block do self.out("
    ") end end def code_block(node) block do self.out("
    ")
          self.out(CGI.escapeHTML(node.string_content))
          self.out("
    ") end end def html(node) block do self.out(node.string_content) end end def inline_html(node) self.out(node.string_content) end def emph(node) self.out("", node.children, "") end def strong(node) self.out("", node.children, "") end def link(node) self.out(' 0 self.out(' title="', CGI.escapeHTML(node.title), '"') end self.out('>', node.children, '') end def image(node) self.out(' 0 self.out(' title="', CGI.escapeHTML(node.title), '"') end plain do self.out(' alt="', node.children, '" />') end end def str(node) self.out(CGI.escapeHTML(node.string_content)) end def code(node) self.out("") self.out(CGI.escapeHTML(node.string_content)) self.out("") end def linebreak(node) self.out("
    ") self.softbreak(node) end def softbreak(node) self.out("\n") end end doc = Node.parse_file(ARGF) renderer = HtmlRenderer.new(STDOUT) renderer.render(doc) renderer.warnings.each do |w| STDERR.write(w) STDERR.write("\n") end # def markdown_to_html(s) # len = s.bytes.length # CMark::cmark_markdown_to_html(s, len) # end # print markdown_to_html(STDIN.read())