pdfrb 0.8.3 → 0.8.5

data/lib/pdfrb/font/sfnt.rb

@@ -0,0 +1,73 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Font
+    # sfnt container (OpenType/TTF) table directory: the shared seam
+    # for reading and rebuilding the wrapper around font outlines.
+    # Both subsetters (TrueType glyf and CFF) rebuild the directory
+    # around their subset tables; this module owns that format so
+    # the two implementations cannot drift (they once disagreed even
+    # on whether checksums are computed).
+    module Sfnt
+      Entry = Struct.new(:tag, :offset, :byte_length)
+
+      module_function
+
+      # Parse the table directory of an sfnt byte string.
+      # Returns [sfnt_version, Array<Entry>].
+      def parse_directory(sfnt)
+        num_tables = (sfnt.getbyte(4) * 256) + sfnt.getbyte(5)
+        version = sfnt.byteslice(0, 4)
+        entries = (0...num_tables).map do |i|
+          base = 12 + (i * 16)
+          Entry.new(
+            sfnt.byteslice(base, 4),
+            sfnt.byteslice(base + 8, 4).unpack1("N"),
+            sfnt.byteslice(base + 12, 4).unpack1("N")
+          )
+        end
+        [version, entries]
+      end
+
+      # Raw bytes of +tag+'s table, or nil.
+      def table_bytes(sfnt, tag)
+        entry = parse_directory(sfnt)[1].find { |e| e.tag == tag }
+        entry && sfnt.byteslice(entry.offset, entry.byte_length)
+      end
+
+      # Rebuild an sfnt wrapper: +sfnt_version+ is the 4-byte magic
+      # (e.g. "\x00\x01\x00\x00" or "OTTO"), +tables+ an Array of
+      # [tag, data] pairs. Checksums are computed per the sfnt spec
+      # (sum of big-endian u32s over zero-padded data).
+      def rebuild(sfnt_version, tables)
+        num = tables.length
+        search_range = (2**Math.log2(num).floor) * 16
+        entry_selector = Math.log2(num).floor
+        range_shift = (num * 16) - search_range
+
+        out = +"".b
+        out << sfnt_version.b
+        out << [num, search_range, entry_selector, range_shift].pack("n4")
+
+        sorted = tables.sort_by { |tag, _| tag }
+        offset = 12 + (num * 16)
+        sorted.each do |tag, data|
+          out << tag.b << [checksum(data), offset, data.bytesize].pack("NNN")
+          offset += (data.bytesize + 3) & ~3
+        end
+        sorted.to_h.each_value do |data|
+          out << data.b
+          out << ("\x00".b * ((4 - (out.bytesize % 4)) % 4))
+        end
+        out
+      end
+
+      def checksum(data)
+        padded = data + ("\x00".b * ((4 - (data.bytesize % 4)) % 4))
+        sum = 0
+        padded.unpack("N*").each { |w| sum = (sum + w) & 0xFFFFFFFF }
+        sum
+      end
+    end
+  end
+end

data/lib/pdfrb/font/true_type/subsetter.rb

@@ -310,58 +310,7 @@ module Pdfrb
 
         # rubocop:disable-next Metrics/MethodLength
         def assemble_ttf(tables)
-          checksum_placeholder = [0].pack("N")
-          num = tables.length
-          search_range = power_of_2_floor(num) * 16
-          entry_selector = Math.log2(search_range / 16).to_i
-          range_shift = (num * 16) - search_range
-
-          header_size = 12
-          dir_size = num * 16
-          data_offset = header_size + dir_size
-
-          # Pad data_offset to 4 bytes
-          data_offset = (data_offset + 3) & ~3
-
-          out = +""
-          out << SFNT_VERSION.pack("N") # sfnt version
-          out << [num].pack("n")
-          out << [search_range].pack("n")
-          out << [entry_selector].pack("n")
-          out << [range_shift].pack("n")
-
-          # Sort tables by tag
-          sorted = tables.sort_by { |tag, _| tag }
-
-          # Calculate offsets
-          current = data_offset
-          offsets = {}
-          sorted.each do |tag, data|
-            offsets[tag] = current
-            current += data.bytesize
-            current = (current + 3) & ~3
-
-            # Directory
-            out << tag.to_s
-            out << checksum_placeholder # checksum (skip)
-            out << [offsets[tag]].pack("N")
-            out << [data.bytesize].pack("N")
-          end
-
-          # Pad to data_offset
-          while out.bytesize < data_offset
-            out << "\x00"
-          end
-
-          # Table data
-          sorted.each_value do |data|
-            out << data
-            while (out.bytesize % 4).nonzero?
-              out << "\x00"
-            end
-          end
-
-          out.force_encoding(::Encoding::BINARY)
+          Pdfrb::Font::Sfnt.rebuild(SFNT_VERSION.pack("N"), tables)
         end
 
         # rubocop:disable Naming/MethodName
@@ -382,11 +331,6 @@ module Pdfrb
         def pad_to_even(str); str << "\x00" if (str.bytesize % 2).nonzero?; end
 
         # rubocop:enable Naming/MethodName
-        def power_of_2_floor(n)
-          p = 1
-          while p * 2 <= n; p *= 2; end
-          p
-        end
       end
     end
   end

data/lib/pdfrb/font/true_type/wrapper.rb

@@ -16,7 +16,7 @@ module Pdfrb
           tu_stream = build_tounicode(document, ttf)
           font_file = embed_font_file(document, ttf_data)
 
-          fd.value[:FontFile2] = Pdfrb::Model::Reference.new(font_file.oid, font_file.gen)
+          fd.value[:FontFile2] = font_file.ref
 
           document.add(
             {
@@ -26,8 +26,8 @@ module Pdfrb
               FirstChar: 0,
               LastChar: 255,
               Widths: widths,
-              FontDescriptor: Pdfrb::Model::Reference.new(fd.oid, fd.gen),
-              ToUnicode: Pdfrb::Model::Reference.new(tu_stream.oid, tu_stream.gen),
+              FontDescriptor: fd.ref,
+              ToUnicode: tu_stream.ref,
               Encoding: :WinAnsiEncoding,
             },
             type: Pdfrb::Model::Type::FontTrueType

data/lib/pdfrb/font.rb

@@ -10,6 +10,7 @@ module Pdfrb
     autoload :CMap, "pdfrb/font/cmap"
     autoload :TrueType, "pdfrb/font/true_type"
     autoload :CFF, "pdfrb/font/cff"
+    autoload :Sfnt, "pdfrb/font/sfnt"
     autoload :Type1, "pdfrb/font/type1"
     autoload :Metrics, "pdfrb/font/metrics_helper"
     autoload :MetricsHelper, "pdfrb/font/metrics_helper"

data/lib/pdfrb/font_loader/true_type.rb

@@ -45,12 +45,12 @@ module Pdfrb
               Descent: -200,
               CapHeight: 700,
               StemV: 80,
-              FontFile2: Pdfrb::Model::Reference.new(font_stream.oid, 0)
+              FontFile2: font_stream.ref
             },
             type: Pdfrb::Model::Type::FontDescriptor
           )
           font_dict.value[:FontDescriptor] =
-            Pdfrb::Model::Reference.new(descriptor.oid, 0)
+            descriptor.ref
         end
 
         font_dict

data/lib/pdfrb/font_loader/type3.rb

@@ -40,7 +40,7 @@ module Pdfrb
           stream = document.add({ Length: glyph.procedure.bytesize },
                                 type: Pdfrb::Model::Cos::Stream)
           stream.stream = glyph.procedure
-          char_procs[glyph.name.to_sym] = Pdfrb::Model::Reference.new(stream.oid, stream.gen)
+          char_procs[glyph.name.to_sym] = stream.ref
           widths[code] = glyph.width
           encoding[:Differences] << code << glyph.name.to_sym
         end

data/lib/pdfrb/importer.rb

@@ -38,10 +38,10 @@ module Pdfrb
 
     def import_reference(ref, source_doc)
       cached = @mappings[ref.oid]
-      return Pdfrb::Model::Reference.new(cached.oid, cached.gen) if cached
+      return cached.ref if cached
 
       in_progress = @in_progress[ref.oid]
-      return Pdfrb::Model::Reference.new(in_progress.oid, in_progress.gen) if in_progress
+      return in_progress.ref if in_progress
 
       source_obj = source_doc.object(ref)
       return nil unless source_obj
@@ -58,7 +58,7 @@ module Pdfrb
       end
       @mappings[ref.oid] = stub
       @in_progress.delete(ref.oid)
-      Pdfrb::Model::Reference.new(stub.oid, stub.gen)
+      stub.ref
     end
 
     def import_hash(hash, source_doc)

data/lib/pdfrb/linearization/writer.rb

@@ -239,7 +239,7 @@ module Pdfrb
       def build_trailer(xref_offset)
         root = @document.catalog
         root_ref = if root.is_a?(Pdfrb::Model::Object) && root.indirect?
-                     Pdfrb::Model::Reference.new(root.oid, root.gen)
+                     root.ref
                    end
         size = [@document.next_oid || 1, 2].max
 

data/lib/pdfrb/model/object.rb

@@ -27,6 +27,13 @@ module Pdfrb
         @document = document
       end
 
+      # The indirect reference denoting this object. The single
+      # seam for "a reference to this" — callers never construct
+      # Reference.new(obj.oid, obj.gen) by hand.
+      def ref
+        Pdfrb::Model::Reference.new(oid, gen)
+      end
+
       # True when this object is referenced indirectly (oid > 0).
       def indirect?
         @oid > 0

data/lib/pdfrb/model/type/action_hide.rb

@@ -17,7 +17,7 @@ module Pdfrb
 
           arr = targets.is_a?(Array) ? targets : [targets]
           arr.each do |ref|
-            obj = ref.is_a?(Pdfrb::Model::Reference) ? document.object(ref) : ref
+            obj = document.resolve(ref)
             yield obj if obj
           end
         end

data/lib/pdfrb/model/type/action_sound.rb

@@ -23,7 +23,7 @@ module Pdfrb
           ref = value[:Sound]
           return nil unless ref && document
 
-          ref.is_a?(Pdfrb::Model::Reference) ? document.object(ref) : ref
+          document.resolve(ref)
         end
 
         # /Volume — optional, -1.0 to 1.0 (default 1.0).
@@ -52,7 +52,7 @@ module Pdfrb
           ref = value[:Next]
           return nil unless ref && document
 
-          ref.is_a?(Pdfrb::Model::Reference) ? document.object(ref) : ref
+          document.resolve(ref)
         end
       end
     end

data/lib/pdfrb/model/type/add_action_catalog.rb

@@ -41,7 +41,7 @@ module Pdfrb
           ref = value[key]
           return nil unless ref && document
 
-          ref.is_a?(Pdfrb::Model::Reference) ? document.object(ref) : ref
+          document.resolve(ref)
         end
       end
     end

data/lib/pdfrb/model/type/add_action_form_field.rb

@@ -34,7 +34,7 @@ module Pdfrb
           ref = value[key]
           return nil unless ref && document
 
-          ref.is_a?(Pdfrb::Model::Reference) ? document.object(ref) : ref
+          document.resolve(ref)
         end
       end
     end

data/lib/pdfrb/model/type/add_action_page_object.rb

@@ -24,7 +24,7 @@ module Pdfrb
           ref = value[key]
           return nil unless ref && document
 
-          ref.is_a?(Pdfrb::Model::Reference) ? document.object(ref) : ref
+          document.resolve(ref)
         end
       end
     end

data/lib/pdfrb/model/type/add_action_screen_annotation.rb

@@ -54,7 +54,7 @@ module Pdfrb
           ref = value[key]
           return nil unless ref && document
 
-          ref.is_a?(Pdfrb::Model::Reference) ? document.object(ref) : ref
+          document.resolve(ref)
         end
       end
     end

data/lib/pdfrb/model/type/add_action_widget_annotation.rb

@@ -64,7 +64,7 @@ module Pdfrb
           ref = value[key]
           return nil unless ref && document
 
-          ref.is_a?(Pdfrb::Model::Reference) ? document.object(ref) : ref
+          document.resolve(ref)
         end
       end
     end

data/lib/pdfrb/model/type/alternate_image.rb

@@ -16,7 +16,7 @@ module Pdfrb
           ref = value[:Image]
           return nil unless ref && document
 
-          ref.is_a?(Pdfrb::Model::Reference) ? document.object(ref) : ref
+          document.resolve(ref)
         end
 
         # /DefaultForPrinting — optional, default false. When true,
@@ -31,7 +31,7 @@ module Pdfrb
           ref = value[:OC]
           return nil unless ref && document
 
-          ref.is_a?(Pdfrb::Model::Reference) ? document.object(ref) : ref
+          document.resolve(ref)
         end
       end
     end

data/lib/pdfrb/model/type/bead.rb

@@ -19,7 +19,7 @@ module Pdfrb
           ref = value[:T]
           return nil unless ref && document
 
-          ref.is_a?(Pdfrb::Model::Reference) ? document.object(ref) : ref
+          document.resolve(ref)
         end
 
         # /N — required indirect next Bead.
@@ -27,7 +27,7 @@ module Pdfrb
           ref = value[:N]
           return nil unless ref && document
 
-          ref.is_a?(Pdfrb::Model::Reference) ? document.object(ref) : ref
+          document.resolve(ref)
         end
 
         # /V — required indirect previous Bead.
@@ -35,7 +35,7 @@ module Pdfrb
           ref = value[:V]
           return nil unless ref && document
 
-          ref.is_a?(Pdfrb::Model::Reference) ? document.object(ref) : ref
+          document.resolve(ref)
         end
 
         # /P — required indirect Page the bead sits on.
@@ -43,7 +43,7 @@ module Pdfrb
           ref = value[:P]
           return nil unless ref && document
 
-          ref.is_a?(Pdfrb::Model::Reference) ? document.object(ref) : ref
+          document.resolve(ref)
         end
 
         # /R — required rectangle on the page.

data/lib/pdfrb/model/type/catalog.rb

@@ -123,7 +123,7 @@ module Pdfrb
           return unless arr.is_a?(Array) || arr.is_a?(Pdfrb::Model::PdfArray)
 
           arr.each do |entry|
-            obj = entry.is_a?(Pdfrb::Model::Reference) ? document.object(entry) : entry
+            obj = document.resolve(entry)
             yield obj if obj
           end
         end

data/lib/pdfrb/model/type/color_space.rb

@@ -168,49 +168,23 @@ module Pdfrb
       # defaults /DefaultGray, /DefaultRGB, /DefaultCMYK.
       class ColorSpaceMap < Pdfrb::Model::Cos::Dictionary
         arlington_object "ColorSpaceMap"
-
-        def [](name)
-          value[name.to_sym] || value[name.to_s]
-        end
-
-        def add(name, color_space)
-          value[name.to_sym] = color_space
-          name.to_sym
-        end
+        include NameMap
 
         def default_gray; self[:DefaultGray]; end
         def default_rgb; self[:DefaultRGB]; end
         def default_cmyk; self[:DefaultCMYK]; end
 
-        def each_color_space(&)
-          return enum_for(:each_color_space) unless block_given?
-
-          value.each(&)
-        end
-
-        def names
-          value.keys
-        end
+        alias each_color_space each_entry
       end
 
       # Colorants dictionary (s8.6.6.5, DeviceN /Colorants). Maps
       # individual colorant names to Separation spaces.
       class ColorantsDict < Pdfrb::Model::Cos::Dictionary
         arlington_object "ColorantsDict"
+        include NameMap
 
-        def [](name)
-          value[name.to_sym] || value[name.to_s]
-        end
-
-        def each_colorant(&)
-          return enum_for(:each_colorant) unless block_given?
-
-          value.each(&)
-        end
-
-        def colorant_names
-          value.keys
-        end
+        alias each_colorant each_entry
+        alias colorant_names names
       end
 
       # Box style dictionary (s7.11.4, BoxColorInfo entries). Guides

data/lib/pdfrb/model/type/crypt_filter.rb

@@ -29,9 +29,9 @@ module Pdfrb
       # names to the default CryptFilter used.
       class CryptFilterMap < Pdfrb::Model::Cos::Dictionary
         arlington_object "CryptFilterMap"
-        def filter_for(stream_name)
-          value[stream_name.to_sym] || value[stream_name.to_s]
-        end
+        include NameMap
+
+        alias filter_for []
 
         def each_filter(&)
           return enum_for(:each_filter) unless block_given?
@@ -53,9 +53,9 @@ module Pdfrb
       # public-key crypt filters.
       class CryptFilterPublicKeyMap < Pdfrb::Model::Cos::Dictionary
         arlington_object "CryptFilterPublicKeyMap"
-        def filter_for(stream_name)
-          value[stream_name.to_sym] || value[stream_name.to_s]
-        end
+        include NameMap
+
+        alias filter_for []
       end
     end
   end

data/lib/pdfrb/model/type/d_part.rb

@@ -21,7 +21,7 @@ module Pdfrb
           ref = value[:DPartRootNode]
           return nil unless ref && document
 
-          resolved = ref.is_a?(Pdfrb::Model::Reference) ? document.object(ref) : ref
+          resolved = document.resolve(ref)
           return nil unless resolved && resolved.value.is_a?(::Hash)
 
           DPart.new(resolved.value)
@@ -57,7 +57,7 @@ module Pdfrb
           ref = value[:Parent]
           return nil unless ref && document
 
-          ref.is_a?(Pdfrb::Model::Reference) ? document.object(ref) : ref
+          document.resolve(ref)
         end
 
         # /DParts — array of arrays of child DParts. Mutually
@@ -71,7 +71,7 @@ module Pdfrb
           ref = value[:Start]
           return nil unless ref && document
 
-          ref.is_a?(Pdfrb::Model::Reference) ? document.object(ref) : ref
+          document.resolve(ref)
         end
 
         # /End — indirect ref to the last Page in this part.
@@ -79,7 +79,7 @@ module Pdfrb
           ref = value[:End]
           return nil unless ref && document
 
-          ref.is_a?(Pdfrb::Model::Reference) ? document.object(ref) : ref
+          document.resolve(ref)
         end
 
         def leaf?

data/lib/pdfrb/model/type/destination.rb

@@ -111,25 +111,9 @@ module Pdfrb
       # map living on the (page or) catalog.
       class DestsMap < Pdfrb::Model::Cos::Dictionary
         arlington_object "DestsMap"
+        include NameMap
 
-        def [](name)
-          value[name.to_sym] || value[name.to_s]
-        end
-
-        def add(name, destination)
-          value[name.to_sym] = destination
-          name.to_sym
-        end
-
-        def each_destination(&)
-          return enum_for(:each_destination) unless block_given?
-
-          value.each(&)
-        end
-
-        def names
-          value.keys
-        end
+        alias each_destination each_entry
       end
     end
   end

data/lib/pdfrb/model/type/document_security_store.rb

@@ -45,7 +45,7 @@ module Pdfrb
 
           arr = certs.is_a?(Pdfrb::Model::PdfArray) ? certs.to_a : certs
           arr.each do |ref|
-            obj = ref.is_a?(Pdfrb::Model::Reference) ? document.object(ref) : ref
+            obj = document.resolve(ref)
             yield obj if obj
           end
         end

data/lib/pdfrb/model/type/dss.rb

@@ -73,7 +73,7 @@ module Pdfrb
 
           arr = refs.is_a?(Pdfrb::Model::PdfArray) ? refs.value : refs
           arr = [arr] unless arr.is_a?(::Array)
-          arr.filter_map { |r| r.is_a?(Pdfrb::Model::Reference) ? document.object(r) : r }
+          arr.filter_map { |r| document.resolve(r) }
         end
       end
     end

data/lib/pdfrb/model/type/font_multiple_master.rb

@@ -39,19 +39,10 @@ module Pdfrb
       # Tiling pattern's Resources.
       class CharProcMap < Pdfrb::Model::Cos::Dictionary
         arlington_object "CharProcMap"
-        def entries
-          value.each
-        end
-
-        def each_procedure(&block)
-          return enum_for(:each_procedure) unless block
+        include NameMap
 
-          value.each(&block)
-        end
-
-        def procedure_for(char_name)
-          value[char_name.to_sym] || value[char_name.to_s]
-        end
+        alias each_procedure each_entry
+        alias procedure_for []
 
         def has_glyph?(char_name)
           !!procedure_for(char_name)

data/lib/pdfrb/model/type/function_stitching.rb

@@ -24,7 +24,7 @@ module Pdfrb
 
           arr = sub_functions.is_a?(Pdfrb::Model::PdfArray) ? sub_functions.to_a : sub_functions
           arr.each do |ref|
-            obj = ref.is_a?(Pdfrb::Model::Reference) ? document.object(ref) : ref
+            obj = document.resolve(ref)
             yield obj if obj
           end
         end

data/lib/pdfrb/model/type/generic_appearance.rb

@@ -31,7 +31,7 @@ module Pdfrb
             type: Pdfrb::Model::Cos::Stream
           )
           stream.stream = "q Q\n"
-          annotation[:AP] = { N: Pdfrb::Model::Reference.new(stream.oid, stream.gen) }
+          annotation[:AP] = { N: stream.ref }
           stream
         end
       end

data/lib/pdfrb/model/type/misc_extras.rb

@@ -208,14 +208,9 @@ module Pdfrb
       # colorant name -> ink solidity percentage.
       class Solidities < Pdfrb::Model::Cos::Dictionary
         arlington_object "Solidities"
+        include NameMap
 
-        def [](colorant)
-          value[colorant.to_sym] || value[colorant.to_s]
-        end
-
-        def colorant_names
-          value.keys
-        end
+        alias colorant_names names
       end
 
       # Source information (s14.10.2, /SourceInfo): provenance for

data/lib/pdfrb/model/type/name_map.rb

@@ -0,0 +1,33 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Shared behaviour for name-keyed maps (Resources
+      # sub-dictionaries, role/class maps, VRI entries): lookup
+      # accepts Symbol or String, +add+ normalises the key to a
+      # Symbol, +names+ lists the keys.
+      module NameMap
+        def [](name)
+          key = name.to_sym
+          value.key?(key) ? value[key] : value[name.to_s]
+        end
+
+        def add(name, entry)
+          value[name.to_sym] = entry
+          name.to_sym
+        end
+
+        def names
+          value.keys
+        end
+
+        def each_entry(&)
+          return enum_for(:each_entry) unless block_given?
+
+          value.each(&)
+        end
+      end
+    end
+  end
+end