clean_csv.rb

require 'fastercsv'
require 'csv'
require 'net/http'
require 'sanitize'


#CSV_FILE_PATH = File.join(File.dirname(__FILE__), 'c2i.csv')
CSV_FILE_PATH = File.join('/tmp', 'Computers_Accessories_Master.csv')

=begin
FasterCSV.foreach(CSV_FILE_PATH, {:headers => true, :col_sep => "@@"}) do |line|
  puts line
end
=end
uri = URI.parse('http://wordoff.org/api/clean')

=begin
CSV.foreach(CSV_FILE_PATH, {:headers => true, :col_sep => "@"}) do |line|
  puts line["Product Description*"]
  req = Net::HTTP::Post.new(uri.path)
  req.set_form_data({'html' => line["Product Description*"]})
  http = Net::HTTP.new(uri.host, uri.port)
  resp = http.request(req)
  puts resp.body
end
=end

#csv_in = CSV.new(data, :headers => true)
csv_in = CSV.read(CSV_FILE_PATH, {:headers => true, :col_sep => ","})


CSV.open('/tmp/temp2.csv', 'w', {:col_sep => "|", :quote_char => '\'', :force_quotes => true}) do |csv_out|
 csv_out << csv_in.first.headers
 #csv_in.rewind

 csv_in.each do |line|
  #puts line["Product Description*"]
begin
 clean = Sanitize.clean(line["Product Description*"] , :elements => ['a', 'b', 'blockquote', 'br', 'em', 'i', 'img', 'li', 'ol', 'p', 'span', 'strong', 'ul', 'table', 'td' , 'tr'], :add_attributes => { 'table' => {"width" => "100%", "border" => "1", 
                            "bordercolor" => "#dfdede", "cellpadding" => "0", "cellspacing" => "0"} } )
end

#clean = Sanitize.clean(line["Product Description*"] , Sanitize::Config::RELAXED)
#clean.gsub! /"/, ''
  clean.gsub! /\n/, ''
  line["Product Description*"] = clean
  csv_out << line.fields
 end
end