rhulse

# This code is designed to cleanup Word HTML to a very great extent.
# It is used at Radio NZ to cleanup documents created in Word prior to
# parsing them, line by line, to extract content.

# The Word content is pasted into WYSIWYG which is then HTTP POSTed into 
# the app where the string is cleaned by this code.

# We add new lines after block elements because the next stage is a line-based 
# parser

rhulse

def convert_date_to_maori(date_string)
  tereo_date = date_string.clone
  h = {
        /January/   => 'Kohi-tātea',
        /February/	=> 'Hui-tanguru',
        /March/			=> 'Poutū-te-rangi',
        /April/			=> 'Paengawhā-whā',
        /May/				=> 'Haratua',
        /June/			=> 'Pipiri',
        /July/			=> 'Hōngongoi',