Turn Santander UK 'xls' export files (actually HTML) into CSV

santander-html-to-csv

#!/usr/bin/env ruby

def to_csv(file)
	lines = []
	file_content = File.open(file, 'r', :encoding => 'iso-8859-1') {|f| f.read}.gsub(/\n/, '')
	file_content.scan(/<tr.*?>(.+?)<\/tr>/) do |match|
		next unless match[0][/\d{2}\/\d{2}\/\d{4}/] and not match[0][/XXXX/]
		fields = []
		match[0].scan(/>([^<>]+?)</) do |field|
			fields.push(field[0].strip.encode('utf-8').gsub(/[£, ]/, ''))
		end
		lines.push(fields.join(','))
	end
	lines.join("\n")
end

def print_usage_and_exit()
	puts "Usage: #{__FILE__} <filename of rubbish Santander 'xls' file aka HTML>"
	exit 1
end

if __FILE__ == $0
	print_usage_and_exit if ARGV.empty?
	puts to_csv(ARGV[0])
end

And yes, I know ...