adds FdfHex dataformat

lib/pdf_forms.rb

@@ -5,6 +5,7 @@
 require 'pdf_forms/data_format'
 require 'pdf_forms/fdf'
 require 'pdf_forms/xfdf'
+require 'pdf_forms/fdf_hex'
 require 'pdf_forms/pdf'
 require 'pdf_forms/pdftk_wrapper'
 

lib/pdf_forms/fdf_hex.rb

@@ -0,0 +1,32 @@
+# coding: UTF-8
+
+module PdfForms
+  # Map keys and values to Adobe's FDF format.
+  #
+  # This is a variation of the original Fdf data format, values are encoded in UTF16 hexadesimal
+  # notation to improve compatibility with non ascii charsets.
+  #
+  # Information about hexadesimal FDF values was found here:
+  #
+  # http://stackoverflow.com/questions/6047970/weird-characters-when-filling-pdf-with-pdftk
+  #
+  class FdfHex < Fdf
+    private
+
+    def field(key, value)
+      "<</T(#{key})/V" +
+        (Array === value ? encode_many(value) : encode_value_as_hex(value)) +
+        ">>\n"
+    end
+
+    def encode_many(values)
+      "[#{values.map { |v| encode_value_as_hex(v) }.join}]"
+    end
+
+    def encode_value_as_hex(value)
+      value = value.to_s
+      utf_16 = value.encode('UTF-16BE', :invalid => :replace, :undef => :replace)
+      '<FEFF' + utf_16.unpack('H*').first.upcase + '>'
+    end
+  end
+end

test/fdf_hex_test.rb

@@ -0,0 +1,29 @@
+# coding: UTF-8
+
+require 'test_helper'
+
+# utf16 hex representation:
+# foo: 0066006F006F
+# bar: 006200610072
+# qux: 007100750078
+
+class FdfHexTest < Minitest::Test
+  def test_fdf_generation
+    fdf = PdfForms::FdfHex.new :field1 => 'foo', :other_field => 'bar qux'
+    assert fdf_text = fdf.to_fdf
+    assert_match %r{<</T\(field1\)/V<FEFF0066006F006F>>>}, fdf_text
+    assert_match %r{<</T\(other_field\)/V<FEFF0062006100720020007100750078>>>}, fdf_text
+  end
+
+  def test_multival
+    fdf = PdfForms::FdfHex.new :field1 => %w(foo bar)
+    assert fdf_text = fdf.to_fdf
+    assert_match '<</T(field1)/V[<FEFF0066006F006F><FEFF006200610072>]>>', fdf_text
+  end
+
+  def test_nil
+    fdf = PdfForms::FdfHex.new :field1 => nil
+    assert fdf_text = fdf.to_fdf
+    assert_match '<</T(field1)/V<FEFF>>>', fdf_text
+  end
+end