ttf.rb in prawn-1.0.0.rc1

- old
+ new

@@ -1,37 +1,35 @@
 # encoding: utf-8
 
 # prawn/font/ttf.rb : Implements AFM font support for Prawn
 #
-# Copyright May 2008, Gregory Brown / James Healy / Jamis Buck
+# Copyright May 2008, Gregory Brown / James Healy / Jamis Buck  
 # All Rights Reserved.
 #
 # This is free software. Please see the LICENSE and COPYING files for details.
 
 require 'ttfunk'
 require 'ttfunk/subset_collection'
 
 module Prawn
   class Font
-
-    # @private
     class TTF < Font
       attr_reader :ttf, :subsets
 
       def unicode?
         true
       end
-
+      
       def initialize(document, name, options={})
         super
 
         @ttf              = read_ttf_file
         @subsets          = TTFunk::SubsetCollection.new(@ttf)
 
         @attributes       = {}
-        @bounding_boxes   = {}
-        @char_widths      = {}
+        @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)
@@ -42,30 +40,30 @@
         scale = (options[:size] || size) / 1000.0
         if options[:kerning]
           kern(string).inject(0) do |s,r|
             if r.is_a?(Numeric)
               s - r
-            else
+            else 
               r.inject(s) { |s2, u| s2 + character_width_by_code(u) }
             end
           end * scale
         else
-          string.codepoints.inject(0) do |s,r|
+          string.unpack("U*").inject(0) do |s,r|
             s + character_width_by_code(r)
           end * scale
         end
-      end
-
+      end   
+     
       # The font bbox, as an array of integers
-      #
+      # 
       def bbox
         @bbox ||= @ttf.bbox.map { |i| Integer(i * scale_factor) }
       end
 
       # Returns true if the font has kerning data, false otherwise
       def has_kerning_data?
-        @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
@@ -77,11 +75,11 @@
       def encode_text(text,options={})
         text = text.chomp
 
         if options[:kerning]
           last_subset = nil
-          kern(text).inject([]) do |result, element|
+          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
@@ -102,11 +100,11 @@
           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...
@@ -160,47 +158,65 @@
           flags |= 0x0004 # assume the font contains at least some non-latin characters
         end
       end
 
       def normalize_encoding(text)
-        begin
-          text.encode(::Encoding::UTF_8)
-        rescue => e
-          puts e
-          raise Prawn::Errors::IncompatibleStringEncoding, "Encoding " +
+        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*")
+            return text.dup
+          rescue
+            raise Prawn::Errors::IncompatibleStringEncoding, "The string you " +
+              "are attempting to render is not encoded in valid UTF-8."
+          end
         end
       end
 
       def glyph_present?(char)
-        code = char.codepoints.first
+        code = char.unpack("U*").first
         cmap[code] > 0
       end
 
       # Returns the number of characters in +str+ (a UTF-8-encoded string).
       #
       def character_count(str)
-        str.length
+        if str.respond_to?(:encode)
+          str.length
+        else
+          str.unpack("U*").length
+        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.each_codepoint do |r|
+        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]
@@ -223,29 +239,29 @@
 
       def hmtx
         @hmtx ||= @ttf.horizontal_metrics
       end
 
-      def character_width_by_code(code)
+      def character_width_by_code(code)    
         return 0 unless cmap[code]
 
         # Some TTF fonts have nonzero widths for \n (UTF-8 / ASCII code: 10).
         # Patch around this as we'll never be drawing a newline with a width.
         return 0.0 if code == 10
 
         @char_widths[code] ||= Integer(hmtx.widths[cmap[code]] * scale_factor)
-      end
+      end                   
 
       def scale_factor
         @scale ||= 1000.0 / @ttf.header.units_per_em
       end
 
       def register(subset)
         temp_name = @ttf.name.postscript_name.gsub("\0","").to_sym
         ref = @document.ref!(:Type => :Font, :BaseFont => temp_name)
 
-        # Embed the font metrics in the document after everything has been
+        # Embed the font metrics in the document after everything has been 
         # drawn, just before the document is emitted.
         @document.before_render { |doc| embed(ref, subset) }
 
         ref
       end
@@ -262,14 +278,17 @@
         # if their font name is more than 33 bytes long. Strange. But true.
         basename = font.name.postscript_name[0, 33].gsub("\0","")
 
         raise "Can't detect a postscript name for #{file}" if basename.nil?
 
-        fontfile = @document.ref!(:Length1 => font_content.size)
-        fontfile.stream << font_content
-        fontfile.stream.compress!
+        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.to_sym,
                                    :FontFile2   => fontfile,
                                    :FontBBox    => bbox,
                                    :Flags       => pdf_flags,
@@ -295,11 +314,11 @@
         # It offends my inner purist, but it'll do.
 
         map = @subsets[subset].to_unicode_map
 
         ranges = [[]]
-        map.keys.sort.inject("") do |s, code|
+        lines = map.keys.sort.inject("") do |s, code|
           ranges << [] if ranges.last.length >= 100
           unicode = map[code]
           ranges.last << "<%02x><%04x>" % [code, unicode]
         end
 
@@ -309,10 +328,10 @@
 
         to_unicode_cmap = UNICODE_CMAP_TEMPLATE % range_blocks.strip
 
         cmap = @document.ref!({})
         cmap << to_unicode_cmap
-        cmap.stream.compress!
+        cmap.compress_stream
 
         reference.data.update(:Subtype => :TrueType,
                               :BaseFont => basename.to_sym,
                               :FontDescriptor => descriptor,
                               :FirstChar => 32,