require 'ttfunk'
require 'ttfunk/subset_collection'

module Prawn
  class Font
    class TTF < Font
      attr_reader :ttf, :subsets
      
      def initialize(document, name, options={})
        super

        @ttf              = read_ttf_file
        @subsets          = TTFunk::SubsetCollection.new(@ttf)

        @attributes       = {}
        @bounding_boxes   = {} 
        @char_widths      = {}   
        @has_kerning_data = @ttf.kerning.exists? && @ttf.kerning.tables.any?

        @ascender         = Integer(@ttf.ascent * scale_factor)
        @descender        = Integer(@ttf.descent * scale_factor)
        @line_gap         = Integer(@ttf.line_gap * scale_factor)
      end

      # +string+ must be UTF8-encoded.
      def width_of(string, options={})
        scale = (options[:size] || size) / 1000.0
        if options[:kerning]
          kern(string).inject(0) do |s,r|
            if r.is_a?(Numeric)
              s - r
            else 
              r.inject(s) { |s2, u| s2 + character_width_by_code(u) }
            end
          end * scale
        else
          string.unpack("U*").inject(0) do |s,r|
            s + character_width_by_code(r)
          end * scale
        end
      end   
      
      def bbox
        @bbox ||= @ttf.bbox.map { |i| Integer(i * scale_factor) }
      end

      def has_kerning_data?
        @has_kerning_data 
      end

      # Perform any changes to the string that need to happen
      # before it is rendered to the canvas. Returns an array of
      # subset "chunks", where the even-numbered indices are the
      # font subset number, and the following entry element is
      # either a string or an array (for kerned text).
      #
      # The +text+ parameter must be UTF8-encoded.
      def encode_text(text,options={})
        text = text.chomp

        if options[:kerning]
          last_subset = nil
          kern(text).inject([]) do |result, element| 
            if element.is_a?(Numeric)
              result.last[1] = [result.last[1]] unless result.last[1].is_a?(Array)
              result.last[1] << element
              result
            else
              encoded = @subsets.encode(element)

              if encoded.first[0] == last_subset
                result.last[1] << encoded.first[1]
                encoded.shift
              end

              if encoded.any?
                last_subset = encoded.last[0]
                result + encoded
              else
                result
              end
            end
          end
        else
          @subsets.encode(text.unpack("U*"))
        end
      end
      
      def basename
        @basename ||= @ttf.name.postscript_name
      end

      # not sure how to compute this for true-type fonts...
      def stemV
        0
      end

      def italic_angle
        @italic_angle ||= if @ttf.postscript.exists?
          raw = @ttf.postscript.italic_angle
          hi, low = raw >> 16, raw & 0xFF
          hi = -((hi ^ 0xFFFF) + 1) if hi & 0x8000 != 0
          "#{hi}.#{low}".to_f
        else
          0
        end
      end

      def cap_height
        @cap_height ||= begin
          height = @ttf.os2.exists? && @ttf.os2.cap_height || 0
          height == 0 ? ascender : height
        end
      end

      def x_height
        # FIXME: seems like if os2 table doesn't exist, we could
        # just find the height of the lower-case 'x' glyph?
        @ttf.os2.exists? && @ttf.os2.x_height || 0
      end

      def family_class
        @family_class ||= (@ttf.os2.exists? && @ttf.os2.family_class || 0) >> 8
      end

      def serif?
        @serif ||= [1,2,3,4,5,7].include?(family_class)
      end

      def script?
        @script ||= family_class == 10
      end

      def pdf_flags
        @flags ||= begin
          flags = 0
          flags |= 0x0001 if @ttf.postscript.fixed_pitch?
          flags |= 0x0002 if serif?
          flags |= 0x0008 if script?
          flags |= 0x0040 if italic_angle != 0
          flags |= 0x0004 # assume the font contains at least some non-latin characters
        end
      end

      def normalize_encoding(text)
        if text.respond_to?(:encode!)
          # if we're running under a M17n aware VM, ensure the string provided is
          # UTF-8 (by converting it if necessary)
          begin
            text.encode!("UTF-8")
          rescue
            raise Prawn::Errors::IncompatibleStringEncoding, "Encoding " +
            "#{text.encoding} can not be transparently converted to UTF-8. " +
            "Please ensure the encoding of the string you are attempting " +
            "to use is set correctly"
          end
        else
          # on a non M17N aware VM, use unpack as a hackish way to verify the
          # string is valid utf-8. I thought it was better than loading iconv
          # though.
          begin
            text.unpack("U*")
          rescue
            raise Prawn::Errors::IncompatibleStringEncoding, "The string you " +
              "are attempting to render is not encoded in valid UTF-8."
          end
        end
      end

      private

      def cmap
        @cmap ||= @ttf.cmap.unicode.first or raise("no unicode cmap for font")
      end
      
      # +string+ must be UTF8-encoded.
      #
      # Returns an array. If an element is a numeric, it represents the
      # kern amount to inject at that position. Otherwise, the element
      # is an array of UTF-16 characters.
      def kern(string)
        a = []

        string.unpack("U*").each do |r|
          if a.empty?
            a << [r]
          elsif (kern = kern_pairs_table[[cmap[a.last.last], cmap[r]]])
            kern *= scale_factor
            a << -kern << [r]
          else
            a.last << r
          end
        end

        a
      end

      def kern_pairs_table
        @kerning_data ||= has_kerning_data? ? @ttf.kerning.tables.first.pairs : {}
      end

      def cid_to_gid_map
        max = cmap.code_map.keys.max
        (0..max).map { |cid| cmap[cid] }.pack("n*")
      end

      def hmtx
        @hmtx ||= @ttf.horizontal_metrics
      end

      def character_width_by_code(code)    
        return 0 unless cmap[code]
        @char_widths[code] ||= Integer(hmtx.widths[cmap[code]] * scale_factor)
      end                   

      def scale_factor
        @scale ||= 1000.0 / @ttf.header.units_per_em
      end

      def register(subset)
        temp_name = @ttf.name.postscript_name
        @document.ref(:Type => :Font, :BaseFont => temp_name) { |ref| embed(ref, subset) }
      end

      def embed(reference, subset)
        font_content = @subsets[subset].encode

        # FIXME: we need postscript_name and glyph widths from the font
        # subset. Perhaps this could be done by querying the subset,
        # rather than by parsing the font that the subset produces?
        font = TTFunk::File.new(font_content)
        basename = font.name.postscript_name

        raise "Can't detect a postscript name for #{file}" if basename.nil?

        compressed_font = Zlib::Deflate.deflate(font_content)

        fontfile = @document.ref(:Length => compressed_font.size,
                                 :Length1 => font_content.size,
                                 :Filter => :FlateDecode )
        fontfile << compressed_font

        descriptor = @document.ref(:Type        => :FontDescriptor,
                                   :FontName    => basename,
                                   :FontFile2   => fontfile,
                                   :FontBBox    => bbox,
                                   :Flags       => pdf_flags,
                                   :StemV       => stemV,
                                   :ItalicAngle => italic_angle,
                                   :Ascent      => ascender,
                                   :Descent     => descender,
                                   :CapHeight   => cap_height,
                                   :XHeight     => x_height)

        hmtx = font.horizontal_metrics
        widths = font.cmap.tables.first.code_map.map { |gid|
          Integer(hmtx.widths[gid] * scale_factor) }

        reference.data.update(:Subtype => :TrueType,
                              :BaseFont => basename,
                              :FontDescriptor => descriptor,
                              :FirstChar => 0,
                              :LastChar => 255,
                              :Widths => @document.ref(widths))

        if @subsets[subset].unicode?
          map = @subsets[subset].to_unicode_map

          ranges = [[]]
          lines = map.keys.sort.inject("") do |s, code|
            ranges << [] if ranges.last.length >= 100
            unicode = map[code]
            ranges.last << "<%02x><%04x>" % [code, unicode]
          end

          range_blocks = ranges.inject("") do |s, list|
            s << "%d beginbfchar\n%s\nendbfchar\n" % [list.length, list.join("\n")]
          end

          to_unicode_cmap = UNICODE_CMAP_TEMPLATE % range_blocks.strip

          cmap = @document.ref({})
          cmap << to_unicode_cmap
          cmap.compress_stream

          @references[subset].data[:ToUnicode] = cmap
        else
          @references[subset].data[:Encoding] = :MacRomanEncoding
        end
      end

      UNICODE_CMAP_TEMPLATE = <<-STR.strip.gsub(/^\s*/, "")
        /CIDInit /ProcSet findresource begin
        12 dict begin
        begincmap
        /CIDSystemInfo <<
          /Registry (Adobe)
          /Ordering (UCS)
          /Supplement 0
        >> def
        /CMapName /Adobe-Identity-UCS def
        /CMapType 2 def
        1 begincodespacerange
        <00><ff>
        endcodespacerange
        %s
        endcmap
        CMapName currentdict /CMap defineresource pop
        end
        end
      STR

      def read_ttf_file
        TTFunk::File.open(@name)
      end
    end
  end
end