Check url redirects to be correct

check_redirects.rb

#!/usr/bin/env ruby
# encoding: utf-8

`chcp 65001`
 
require 'rubygems'
require 'httpclient'
require 'csv'

filename = ARGV[0] || 'data/in.csv'
from_line = ARGV[1] || 0

output_filename = 'data/out.csv'

result_template = {
	old_url: "old_url",
	new_url: "new_url",
	result: "result",
	status: "status",
	location: "location"
}

results = []
results << result_template.clone

result_template.each { |k, v| result_template[k] = '' }

httpc = HTTPClient.new

begin
	CSV.foreach(filename, { col_sep: ","}) do |row|
		if from_line > 0 and from_line > $.
			next
		end
		puts $.

		old_url = row[0].strip.downcase
		new_url = row[1].strip.downcase
		
		next if !old_url || !new_url

		result_array = result_template.clone
		result_array[:old_url] = old_url
		result_array[:new_url] = new_url

		begin
			response = httpc.head(old_url)
		rescue Exception => e
			result_array[:result] = 'error'
			result_array[:location] = e.to_s
			puts old_url
			puts e
			results << result_array
			next
		end

		result_array[:status] = response.status

		if %w(301 302).include? response.status.to_s
			result_array[:location] = response.header['Location'][0].strip.downcase
			if result_array[:location] == new_url
				result_array[:result] = 'ok'
			else
				result_array[:result] = 'error'
			end
		else
			result_array[:result] = 'error'
		end

		unless result_array[:result] == 'ok'
			puts result_array
		end

		results << result_array
	end
rescue Exception => e
	puts "Exception: #{e}"
end

puts "Writing to file..."

if from_line > 0
	type = "ab"
	results.delete_at(0)
else
	type = "wb"
end

CSV.open(output_filename, type) do |csv|
	results.each { |result_row| csv << result_row.values }
end

puts "Finished"