pdf-core 0.4.0 → 0.10.0

data/lib/pdf/core/page_geometry.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 # Describes PDF page geometries
 #
@@ -8,119 +8,63 @@
 
 module PDF
   module Core
-
     # Dimensions pulled from PDF::Writer, rubyforge.org/projects/ruby-pdf
-    #
-    # All of these dimensions are in PDF Points (1/72 inch)
-    #
-    # ===Inbuilt Sizes:
-    #
-    #
-    # 4A0:: => 4767.87 x 6740.79
-    # 2A0:: => 3370.39 x 4767.87
-    # A0:: => 2383.94 x 3370.39
-    # A1:: => 1683.78 x 2383.94
-    # A2:: => 1190.55 x 1683.78
-    # A3:: => 841.89 x 1190.55
-    # A4:: => 595.28 x 841.89
-    # A5:: => 419.53 x 595.28
-    # A6:: => 297.64 x 419.53
-    # A7:: => 209.76 x 297.64
-    # A8:: => 147.40 x 209.76
-    # A9:: => 104.88 x 147.40
-    # A10:: => 73.70 x 104.88
-    # B0:: => 2834.65 x 4008.19
-    # B1:: => 2004.09 x 2834.65
-    # B2:: => 1417.32 x 2004.09
-    # B3:: => 1000.63 x 1417.32
-    # B4:: => 708.66 x 1000.63
-    # B5:: => 498.90 x 708.66
-    # B6:: => 354.33 x 498.90
-    # B7:: => 249.45 x 354.33
-    # B8:: => 175.75 x 249.45
-    # B9:: => 124.72 x 175.75
-    # B10:: => 87.87 x 124.72
-    # C0:: => 2599.37 x 3676.54
-    # C1:: => 1836.85 x 2599.37
-    # C2:: => 1298.27 x 1836.85
-    # C3:: => 918.43 x 1298.27
-    # C4:: => 649.13 x 918.43
-    # C5:: => 459.21 x 649.13
-    # C6:: => 323.15 x 459.21
-    # C7:: => 229.61 x 323.15
-    # C8:: => 161.57 x 229.61
-    # C9:: => 113.39 x 161.57
-    # C10:: => 79.37 x 113.39
-    # RA0:: => 2437.80 x 3458.27
-    # RA1:: => 1729.13 x 2437.80
-    # RA2:: => 1218.90 x 1729.13
-    # RA3:: => 864.57 x 1218.90
-    # RA4:: => 609.45 x 864.57
-    # SRA0:: => 2551.18 x 3628.35
-    # SRA1:: => 1814.17 x 2551.18
-    # SRA2:: => 1275.59 x 1814.17
-    # SRA3:: => 907.09 x 1275.59
-    # SRA4:: => 637.80 x 907.09
-    # EXECUTIVE:: => 521.86 x 756.00
-    # FOLIO:: => 612.00 x 936.00
-    # LEGAL:: => 612.00 x 1008.00
-    # LETTER:: => 612.00 x 792.00
-    # TABLOID:: => 792.00 x 1224.00
-    #
     module PageGeometry
-
-      SIZES = { "4A0" => [4767.87, 6740.79],
-                "2A0" => [3370.39, 4767.87],
-                 "A0" => [2383.94, 3370.39],
-                 "A1" => [1683.78, 2383.94],
-                 "A2" => [1190.55, 1683.78],
-                 "A3" => [841.89, 1190.55],
-                 "A4" => [595.28, 841.89],
-                 "A5" => [419.53, 595.28],
-                 "A6" => [297.64, 419.53],
-                 "A7" => [209.76, 297.64],
-                 "A8" => [147.40, 209.76],
-                 "A9" => [104.88, 147.40],
-                "A10" => [73.70, 104.88],
-                 "B0" => [2834.65, 4008.19],
-                 "B1" => [2004.09, 2834.65],
-                 "B2" => [1417.32, 2004.09],
-                 "B3" => [1000.63, 1417.32],
-                 "B4" => [708.66, 1000.63],
-                 "B5" => [498.90, 708.66],
-                 "B6" => [354.33, 498.90],
-                 "B7" => [249.45, 354.33],
-                 "B8" => [175.75, 249.45],
-                 "B9" => [124.72, 175.75],
-                "B10" => [87.87, 124.72],
-                 "C0" => [2599.37, 3676.54],
-                 "C1" => [1836.85, 2599.37],
-                 "C2" => [1298.27, 1836.85],
-                 "C3" => [918.43, 1298.27],
-                 "C4" => [649.13, 918.43],
-                 "C5" => [459.21, 649.13],
-                 "C6" => [323.15, 459.21],
-                 "C7" => [229.61, 323.15],
-                 "C8" => [161.57, 229.61],
-                 "C9" => [113.39, 161.57],
-                "C10" => [79.37, 113.39],
-                "RA0" => [2437.80, 3458.27],
-                "RA1" => [1729.13, 2437.80],
-                "RA2" => [1218.90, 1729.13],
-                "RA3" => [864.57, 1218.90],
-                "RA4" => [609.45, 864.57],
-               "SRA0" => [2551.18, 3628.35],
-               "SRA1" => [1814.17, 2551.18],
-               "SRA2" => [1275.59, 1814.17],
-               "SRA3" => [907.09, 1275.59],
-               "SRA4" => [637.80, 907.09],
-          "EXECUTIVE" => [521.86, 756.00],
-              "FOLIO" => [612.00, 936.00],
-              "LEGAL" => [612.00, 1008.00],
-             "LETTER" => [612.00, 792.00],
-            "TABLOID" => [792.00, 1224.00] }
-
+      # Named page sizes.
+      #
+      # All of these dimensions are in PDF Points (1/72 inch).
+      SIZES = {
+        '4A0' => [4767.87, 6740.79],
+        '2A0' => [3370.39, 4767.87],
+        'A0' => [2383.94, 3370.39],
+        'A1' => [1683.78, 2383.94],
+        'A2' => [1190.55, 1683.78],
+        'A3' => [841.89, 1190.55],
+        'A4' => [595.28, 841.89],
+        'A5' => [419.53, 595.28],
+        'A6' => [297.64, 419.53],
+        'A7' => [209.76, 297.64],
+        'A8' => [147.40, 209.76],
+        'A9' => [104.88, 147.40],
+        'A10' => [73.70, 104.88],
+        'B0' => [2834.65, 4008.19],
+        'B1' => [2004.09, 2834.65],
+        'B2' => [1417.32, 2004.09],
+        'B3' => [1000.63, 1417.32],
+        'B4' => [708.66, 1000.63],
+        'B5' => [498.90, 708.66],
+        'B6' => [354.33, 498.90],
+        'B7' => [249.45, 354.33],
+        'B8' => [175.75, 249.45],
+        'B9' => [124.72, 175.75],
+        'B10' => [87.87, 124.72],
+        'C0' => [2599.37, 3676.54],
+        'C1' => [1836.85, 2599.37],
+        'C2' => [1298.27, 1836.85],
+        'C3' => [918.43, 1298.27],
+        'C4' => [649.13, 918.43],
+        'C5' => [459.21, 649.13],
+        'C6' => [323.15, 459.21],
+        'C7' => [229.61, 323.15],
+        'C8' => [161.57, 229.61],
+        'C9' => [113.39, 161.57],
+        'C10' => [79.37, 113.39],
+        'RA0' => [2437.80, 3458.27],
+        'RA1' => [1729.13, 2437.80],
+        'RA2' => [1218.90, 1729.13],
+        'RA3' => [864.57, 1218.90],
+        'RA4' => [609.45, 864.57],
+        'SRA0' => [2551.18, 3628.35],
+        'SRA1' => [1814.17, 2551.18],
+        'SRA2' => [1275.59, 1814.17],
+        'SRA3' => [907.09, 1275.59],
+        'SRA4' => [637.80, 907.09],
+        'EXECUTIVE' => [521.86, 756.00],
+        'FOLIO' => [612.00, 936.00],
+        'LEGAL' => [612.00, 1008.00],
+        'LETTER' => [612.00, 792.00],
+        'TABLOID' => [792.00, 1224.00],
+      }.freeze
     end
   end
 end
-

data/lib/pdf/core/pdf_object.rb

@@ -1,95 +1,131 @@
-# encoding: utf-8
-#
-# pdf_object.rb : Handles Ruby to PDF object serialization
-#
-# Copyright April 2008, Gregory Brown.  All Rights Reserved.
-#
-# This is free software. Please see the LICENSE and COPYING files for details.
+# frozen_string_literal: true
 
-# Top level Module
-#
 module PDF
   module Core
     module_function
 
+    # Serializes floating number into a string
+    #
+    # @param num [Numeric]
+    # @return [String]
+    def real(num)
+      result = format('%.5f', num)
+      result.sub!(/((?<!\.)0)+\z/, '')
+      result
+    end
+
+    # Serializes a n array of numbers. This is specifically for use in PDF
+    # content streams.
+    #
+    # @param array [Array<Numeric>]
+    # @return [String]
+    def real_params(array)
+      array.map { |e| real(e) }.join(' ')
+    end
+
+    # Converts string to UTF-16BE encoding as expected by PDF.
+    #
+    # @param str [String]
+    # @return [String]
+    # @api private
     def utf8_to_utf16(str)
-      "\xFE\xFF".force_encoding(::Encoding::UTF_16BE) + str.encode(::Encoding::UTF_16BE)
+      (+"\xFE\xFF").force_encoding(::Encoding::UTF_16BE) <<
+        str.encode(::Encoding::UTF_16BE)
     end
 
-    # encodes any string into a hex representation. The result is a string
+    # Encodes any string into a hex representation. The result is a string
     # with only 0-9 and a-f characters. That result is valid ASCII so tag
-    # it as such to account for behaviour of different ruby VMs
+    # it as such to account for behaviour of different ruby VMs.
+    #
+    # @param str [String]
+    # @return [String]
     def string_to_hex(str)
-      str.unpack("H*").first.force_encoding(::Encoding::US_ASCII)
+      str.unpack1('H*').force_encoding(::Encoding::US_ASCII)
     end
 
+    # Characters to escape in name objects
+    # @api private
+    ESCAPED_NAME_CHARACTERS = (1..32).to_a + [35, 40, 41, 47, 60, 62] + (127..255).to_a
+
+    # How to escape special characters in literal strings
+    # @api private
+    STRING_ESCAPE_MAP = { '(' => '\(', ')' => '\)', '\\' => '\\\\', "\r" => '\r' }.freeze
+
     # Serializes Ruby objects to their PDF equivalents.  Most primitive objects
     # will work as expected, but please note that Name objects are represented
-    # by Ruby Symbol objects and Dictionary objects are represented by Ruby hashes
-    # (keyed by symbols)
+    # by Ruby Symbol objects and Dictionary objects are represented by Ruby
+    # hashes (keyed by symbols)
     #
-    #  Examples:
+    # Examples:
     #
-    #     PdfObject(true)      #=> "true"
-    #     PdfObject(false)     #=> "false"
-    #     PdfObject(1.2124)    #=> "1.2124"
-    #     PdfObject("foo bar") #=> "(foo bar)"
-    #     PdfObject(:Symbol)   #=> "/Symbol"
-    #     PdfObject(["foo",:bar, [1,2]]) #=> "[foo /bar [1 2]]"
+    #     pdf_object(true)      #=> "true"
+    #     pdf_object(false)     #=> "false"
+    #     pdf_object(1.2124)    #=> "1.2124"
+    #     pdf_object('foo bar') #=> "(foo bar)"
+    #     pdf_object(:Symbol)   #=> "/Symbol"
+    #     pdf_object(['foo',:bar, [1,2]]) #=> "[foo /bar [1 2]]"
     #
-    def PdfObject(obj, in_content_stream = false)
-      case(obj)
-      when NilClass   then "null"
-      when TrueClass  then "true"
-      when FalseClass then "false"
+    # @param obj [nil, Boolean, Numeric, Array, Hash, Time, Symbol, String,
+    #   PDF::Core::ByteString, PDF::Core::LiteralString,
+    #   PDF::Core::NameTree::Node, PDF::Core::NameTree::Value,
+    #   PDF::Core::OutlineRoot, PDF::Core::OutlineItem, PDF::Core::Reference]
+    #   Object to serialise
+    # @param in_content_stream [Boolean] Specifies whther to use content stream
+    #   format or object format
+    # @return [String]
+    # @raise [PDF::Core::Errors::FailedObjectConversion]
+    def pdf_object(obj, in_content_stream = false)
+      case obj
+      when NilClass then 'null'
+      when TrueClass then 'true'
+      when FalseClass then 'false'
       when Numeric
-        if (str = String(obj)) =~ /e/i
-          # scientific notation is not supported in PDF
-          sprintf("%.16f", obj).gsub(/\.?0+\z/, "")
-        else
-          str
-        end
+        num_string = obj.is_a?(Integer) ? String(obj) : real(obj)
+
+        # Truncate trailing fraction zeroes
+        num_string.sub!(/(\d*)((\.0*$)|(\.0*[1-9]*)0*$)/, '\1\4')
+        num_string
       when Array
-        "[" << obj.map { |e| PdfObject(e, in_content_stream) }.join(' ') << "]"
+        "[#{obj.map { |e| pdf_object(e, in_content_stream) }.join(' ')}]"
       when PDF::Core::LiteralString
-        obj = obj.gsub(/[\\\n\r\t\b\f\(\)]/n) { |m| "\\#{m}" }
+        obj = obj.gsub(/[\\\r()]/, STRING_ESCAPE_MAP)
         "(#{obj})"
       when Time
-        obj = obj.strftime("D:%Y%m%d%H%M%S%z").chop.chop + "'00'"
-        obj = obj.gsub(/[\\\n\r\t\b\f\(\)]/n) { |m| "\\#{m}" }
+        obj = "#{obj.strftime('D:%Y%m%d%H%M%S%z').chop.chop}'00'"
+        obj = obj.gsub(/[\\\r()]/, STRING_ESCAPE_MAP)
         "(#{obj})"
       when PDF::Core::ByteString
-        "<" << obj.unpack("H*").first << ">"
+        "<#{obj.unpack1('H*')}>"
       when String
         obj = utf8_to_utf16(obj) unless in_content_stream
-        "<" << string_to_hex(obj) << ">"
-       when Symbol
-         "/" + obj.to_s.unpack("C*").map { |n|
-          if n < 33 || n > 126 || [35,40,41,47,60,62].include?(n)
-            "#" + n.to_s(16).upcase
+        "<#{string_to_hex(obj)}>"
+      when Symbol
+        (@symbol_str_cache ||= {})[obj] ||= (+'/') << obj.to_s.unpack('C*').map { |n|
+          if ESCAPED_NAME_CHARACTERS.include?(n)
+            "##{n.to_s(16).upcase}"
           else
-            [n].pack("C*")
+            n.chr
           end
-         }.join
+        }.join
       when ::Hash
-        output = "<< "
-        obj.each do |k,v|
-          unless String === k || Symbol === k
+        output = +'<< '
+        obj
+          .sort_by { |k, _v| k.to_s }
+          .each do |(k, v)|
+          unless k.is_a?(String) || k.is_a?(Symbol)
             raise PDF::Core::Errors::FailedObjectConversion,
-              "A PDF Dictionary must be keyed by names"
+              'A PDF Dictionary must be keyed by names'
           end
-          output << PdfObject(k.to_sym, in_content_stream) << " " <<
-                    PdfObject(v, in_content_stream) << "\n"
+          output << pdf_object(k.to_sym, in_content_stream) << ' ' <<
+            pdf_object(v, in_content_stream) << "\n"
         end
-        output << ">>"
+        output << '>>'
       when PDF::Core::Reference
         obj.to_s
-      when PDF::Core::NameTree::Node
-        PdfObject(obj.to_hash)
+      when PDF::Core::NameTree::Node, PDF::Core::OutlineRoot, PDF::Core::OutlineItem
+        pdf_object(obj.to_hash)
       when PDF::Core::NameTree::Value
-        PdfObject(obj.name) + " " + PdfObject(obj.value)
-      when PDF::Core::OutlineRoot, PDF::Core::OutlineItem
-        PdfObject(obj.to_hash)
+        "#{pdf_object(obj.name)} #{pdf_object(obj.value)}"
       else
         raise PDF::Core::Errors::FailedObjectConversion,
           "This object cannot be serialized to PDF (#{obj.inspect})"

data/lib/pdf/core/reference.rb

@@ -1,78 +1,118 @@
-# encoding: utf-8
-
-# reference.rb : Implementation of PDF indirect objects
-#
-# Copyright April 2008, Gregory Brown.  All Rights Reserved.
-#
-# This is free software. Please see the LICENSE and COPYING files for details.
+# frozen_string_literal: true
 
+require 'pdf/core/utils'
 
 module PDF
   module Core
-    class Reference #:nodoc:
+    # PDF indirect objects
+    #
+    # @api private
+    class Reference
+      # Object identifier
+      # @return [Integer]
+      attr_accessor :identifier
+
+      # Object generation
+      # @return [Integer]
+      attr_accessor :gen
+
+      # Object data
+      # @return [any]
+      attr_accessor :data
+
+      # Offset of the serialized object in the document
+      # @return [Integer]
+      attr_accessor :offset
 
-      attr_accessor :gen, :data, :offset, :stream, :identifier
+      # Object stream
+      # @return [Stream]
+      attr_accessor :stream
 
+      # In PDF only dict object can have a stream attached. This exception
+      # indicates someone tried to add a stream to another kind of object.
+      class CannotAttachStream < StandardError
+        # @param message [String] Error message
+        def initialize(message = 'Cannot attach stream to a non-dictionary object')
+          super
+        end
+      end
+
+      # @param id [Integer] Object identifier
+      # @param data [any] Object data
       def initialize(id, data)
         @identifier = id
-        @gen        = 0
-        @data       = data
-        @stream     = Stream.new
+        @gen = 0
+        @data = data
+        @stream = Stream.new
       end
 
+      # Serialized PDF object
+      #
+      # @return [String]
       def object
-        output = "#{@identifier} #{gen} obj\n"
-        unless @stream.empty?
-          output << PDF::Core::PdfObject(data.merge @stream.data) << "\n" << @stream.object
+        output = +"#{@identifier} #{gen} obj\n"
+        if @stream.empty?
+          output << PDF::Core.pdf_object(data) << "\n"
         else
-          output << PDF::Core::PdfObject(data) << "\n"
+          output << PDF::Core.pdf_object(data.merge(@stream.data)) <<
+            "\n" << @stream.object
         end
 
         output << "endobj\n"
       end
 
+      # Appends data to object stream
+      #
+      # @param io [String] data
+      # @return [io]
+      # @raise [CannotAttachStream] if object is not a dict
       def <<(io)
-        raise "Cannot attach stream to non-dictionary object" unless @data.is_a?(::Hash)
+        unless @data.is_a?(::Hash)
+          raise CannotAttachStream
+        end
+
         (@stream ||= Stream.new) << io
       end
 
+      # Object reference in PDF format
+      #
+      # @return [String]
       def to_s
         "#{@identifier} #{gen} R"
       end
 
-      # Creates a deep copy of this ref. If +share+ is provided, shares the
-      # given dictionary entries between the old ref and the new.
+      # Creates a deep copy of this ref.
       #
-      def deep_copy(share=[])
+      # @param share [Array<Symbol>] a list of dictionary entries to share
+      #   between the old ref and the new
+      # @return [Reference]
+      def deep_copy(share = [])
         r = dup
 
         case r.data
         when ::Hash
           # Copy each entry not in +share+.
           (r.data.keys - share).each do |k|
-            r.data[k] = Marshal.load(Marshal.dump(r.data[k]))
+            r.data[k] = Utils.deep_clone(r.data[k])
           end
         when PDF::Core::NameTree::Node
           r.data = r.data.deep_copy
         else
-          r.data = Marshal.load(Marshal.dump(r.data))
+          r.data = Utils.deep_clone(r.data)
         end
 
-        r.stream = Marshal.load(Marshal.dump(r.stream))
+        r.stream = Utils.deep_clone(r.stream)
         r
       end
 
       # Replaces the data and stream with that of other_ref.
+      #
+      # @param other_ref [Reference]
+      # @return [void]
       def replace(other_ref)
-        @data   = other_ref.data
+        @data = other_ref.data
         @stream = other_ref.stream
       end
     end
-
-    module_function
-
-    def Reference(*args, &block) #:nodoc:
-      Reference.new(*args, &block)
-    end
   end
 end