gistfile1.rb

#!/usr/bin/env ruby

require 'iconv'
require 'tempfile'

unless ARGV.length == 2
  STDERR.puts "Usage: <INPUT-PDF> <OUTPUT-PDF>"
  exit 1
end

field_mapping = {
  'Text1' => 'departmental-reference-number',
  'Text2' => 'departmental-request-received-by',
  'Text3' => 'departmental-request-received-date',
  'Text4' => 'departmental-request-received-place',
  'Text5a' => 'request-fee',
  'Text5b' => 'deposit',
  'Text5c' => 'access-fee',
  'Text7' => 'full-names-and-surname-a',
  'Text7a' => 'id-a-01',
  'Text7b' => 'id-a-02',
  'Text7c' => 'id-a-03',
  'Text7d' => 'id-a-04',
  'Text7e' => 'id-a-05',
  'Text7f' => 'id-a-06',
  'Text7g' => 'id-a-07',
  'Text7h' => 'id-a-08',
  'Text7i' => 'id-a-09',
  'Text7j' => 'id-a-10',
  'Text7k' => 'id-a-11',
  'Text7l' => 'id-a-12',
  'Text7m' => 'id-a-13',
  'Text8' => 'full-names-and-surname-b',
  'Text8a' => 'id-b-01',
  'Text8b' => 'id-b-02',
  'Text8c' => 'id-b-03',
  'Text8d' => 'id-b-04',
  'Text8e' => 'id-b-05',
  'Text8f' => 'id-b-06',
  'Text8g' => 'id-b-07',
  'Text8h' => 'id-b-08',
  'Text8i' => 'id-b-09',
  'Text8j' => 'id-b-10',
  'Text8k' => 'id-b-11',
  'Text8l' => 'id-b-12',
  'Text8m' => 'id-b-13',
  'Text9' => 'description-of-record-1',
  'Text9a' => 'description-of-record-2',
  # TODO: add mappings for the other keys...
}

values_to_fill_in = {
  'description-of-record-1' => 'just like to say',
  'description-of-record-2' => 'helloooo',
}

template_pdf_filename, output_pdf_filename = ARGV

fdf_tempfile = Tempfile.new(['paia-', '.fdf'])
fdf_tempfile.close
fdf_filename = fdf_tempfile.path

$utf_16_to_utf_8 = Iconv.new "UTF-8", "UTF-16BE"
$utf_8_to_utf_16 = Iconv.new "UTF-16BE", "UTF-8"

def decode_key(s)
  without_bom = s.gsub(/^\xFE\xFF/, '')
  $utf_16_to_utf_8.iconv without_bom
end

def encode_value(s)
  without_bom = $utf_8_to_utf_16.iconv s
  "\xFE\xFF#{without_bom}"
end

# Generate the FDF file from the PDF:
unless system('pdftk',
              template_pdf_filename,
              'generate_fdf',
              'output',
              fdf_filename)
  STDERR.puts "Generating FDF from #{template_pdf_filename} failed"
  exit(1)
end

lines = File.open(fdf_filename, 'rb') { |f|
  f.readlines
}

puts "fdf_filename is: #{fdf_filename}"

line_number_to_replacement_line = {}

lines.each_with_index do |line, i|

  # We assume that the keys and values come in pairs of lines, with
  # the value before the key, like:
  #
  # /V (<FE><FF>)
  # /T (<FE><FF>^@T^@e^@x^@t^@9^@a)

  if line =~ /T \((.*)\)$/
    key = decode_key $1
    puts "found key: #{key}"
    useful_key_name = field_mapping[key]
    if useful_key_name
      # FIXME: just for figuring out IDs, fill in the key even if there
      # isn't a value for it...
      # if values_to_fill_in.has_key?(useful_key_name)
        encoded_value = encode_value values_to_fill_in[useful_key_name] || useful_key_name
        line_number_to_replacement_line[i - 1] = "/V (#{encoded_value})"
      # end
    end
  end

end

File.open(fdf_filename, 'w') { |f|
  lines.each_with_index do |line, i|
    replacment_line = line_number_to_replacement_line[i]
    f.puts replacment_line || line
  end
}

# Output a PDF with those fields filled in:
unless system('pdftk',
              template_pdf_filename,
              'fill_form',
              fdf_filename,
              'output',
              output_pdf_filename)
  STDERR.puts "Generating completed PDF #{output_pdf_filename} failed"
  exit 1
end