quick thing to parse two cobiont files

cobiont_stats.cr

#!/bin/env crystal
# cobiont_stats.cr contamination_file1 contamination_file_2

# returns a Array of String
def read_contamination_file(file : String)
	ids = Array(String).new
	File.each_line(file){|line|
		ids << $1 if /REMOVE\s+(\S+)/.match(line)
	}
	return ids.uniq
end

file1=ARGV[0]
file2=ARGV[1]
x1 = read_contamination_file(file1)
x2 = read_contamination_file(file2)

# do intersection / x1 - x2 / x2 - x1
puts "#{file1} & #{file2}: #{(x1 & x2 ).size}"
puts "#{file1} - #{file2}: #{(x1 - x2).size} #{x1-x2}"
puts "#{file2} - #{file1}: #{(x2 - x1).size} #{x2-x1}"

cobiont_stats.py

#!/bin/env python
# cobiont_stats.py contamination_file1 contamination_file_2

import sys
import re

# returns a set
def read_contamination_file(file):
	ids = set()
	for line in open(file,'r'):
		if match := re.search('REMOVE\s+(\S+)',line):
			ids.add(match.group(1))
	return ids  

file1=sys.argv[1]
file2=sys.argv[2]
x1 = read_contamination_file(file1)
x2 = read_contamination_file(file2)

# do intersection / x1 - x2 / x2 - x1
print(f"{file1} & {file2}: {len(x1.intersection(x2))}")
print(f"{file1} - {file2}: {len(x1.difference(x2))} {x1.difference(x2)}")
print(f"{file2} - {file1}: {len(x2.difference(x1))} {x2.difference(x1)}")