Strip path from domain, off from sitemap xml file.

path_stripper.rb

def path_strip(input_file, domain, output_file)
  raise "domain must be string form" unless domain.is_a? String
  raise "invalid input file name" unless input_file.is_a? String
  raise "invalid output file name" unless output_file.is_a? String

  file = File.open(input_file, "r")
  data = file.read
  file.close
  data_lines = data.lines
  cleared_arr = []

  result_file = File.open(output_file, "w")

  num = 0
  data_lines.each do |line|
    if line =~ /<loc>http:\/\/#{Regexp.quote(domain)}\/.*<\/loc>/
      num += 1
      result_file.puts "link_#{num},#{line[5..-8].sub("http://#{domain}/", '')}"
    end
  end
  result_file.close
  p num
end

# example:
path_strip "./sitemap.xml", "hunj.github.io", "./result.csv"

Calling File.open twice is alright.

You can refactor lines 6 to 9 to this:

data_lines = File.readlines(input_file)

I guess you could prevent creating unnecessary variables by doing something like this:
(A shorter version)

def path_strip(input_file, domain, output_file)
  raise "domain must be string form" unless domain.is_a? String
  raise "invalid input file name" unless input_file.is_a? String
  raise "invalid output file name" unless output_file.is_a? String

  data = File.read(input_file).scan(/<loc>http:\/\/#{Regexp.quote(domain)}\/.*<\/loc>/)

  data.map!.with_index do |x, index|
    "link_#{index + 1},#{x[5..-8].sub("http://#{domain}/", '')}"
  end

  File.write(output_file, data.join("\n"))
  p data.length
end

# example:
path_strip "./sitemap.xml", "hunj.github.io", "./result.csv"

By the way, with reference to your code (L12, L21), what you're doing is: open a File, process the lines one by one, then close the File instance. Only do that if you have a lot of lines (and you think that the lines will take up a lot of memory).

An alternative would be to store the processed lines in a String (concatenate) or Array, and write it to the output_file all at one go. This method would be faster, but you need memory to store your data.

Cheers,
Jay