setop

#!/usr/bin/python2.7
# Multiset operations
# Set operations
import argparse
import collections
import sys
from functools import partial, reduce
import operator

def binary_file_op(string, file1, file2):
    f = tempfile.NamedTemporaryFile()
    subprocess.call(string, file1.name, file2.name, ">", f.name)
    return f

def set_op(op, files):
    sets = map(set, files)
    poss = reduce(operator.or_, sets)
    for line in poss:
        if op(*[line in fn for fn in sets]):
            pass
            print line,

def multiset_op(op, files):
    bags = map(collections.Counter, files)
    poss = reduce(operator.or_, map(set, bags))
    for line in poss:
        for _ in range(op(*[fn.get(line, 0) for fn in bags])):
            print line,

def binary_set_op(op, f1, f2):
    return set_op(op, [f1, f2])

def binary_multiset_op(op, f1, f2):
    return multiset_op(op, [f1, f2])

def main():
    parser = argparse.ArgumentParser(description='Perform simple set operations with unordered files.')
    subparsers = parser.add_subparsers()
    # Set options 
    set_parser = subparsers.add_parser("set", help='perform set operation ("set -h" for more help)')
  
    setops = set_parser.add_mutually_exclusive_group(required=True)
    for opname, help, opfunc in [
            ('union', 'print all lines which appear in either file', operator.or_),
            ('intersection', 'print all lines which appear in either file', operator.and_),
            ('difference', 'print all lines which appear in the first file but not the second', lambda a,b: a and not b),
            ('symmetric-difference', 'print all lines which appear in exactly one of the two files', operator.xor),
            ]:
        setops.add_argument('--'+opname, dest='func', help=help, 
            action='store_const', const=partial(binary_set_op, opfunc),)
    
    set_parser.add_argument('file1', type=file)
    set_parser.add_argument('file2', type=file)
    
    # Multiset options
    multiset_parser = subparsers.add_parser("ms", help='perform multiset operation ("ms -h" for more help)')

    msops = multiset_parser.add_mutually_exclusive_group(required=True)
    for opname, help, opfunc in [
            ('union', 'print whichever count of lines appears more', max),
            ('intersection', 'print which count of lines appears less', min),
            ('difference', 'print <n> copies of a line which appear <m+n> times in the first file but only <m> times in the second; lines which appear more times in the second file are not printed', lambda a,b: max(a-b,0)),
            ('sym_difference', 'print <n> copies of a lines which appears  <m+n> times in one file and <m> times in the other', lambda a,b: abs(m-n)),
            ('add', 'combine the two files', operator.add),
            ]:
        msops.add_argument('--'+opname, dest='func', help=help, 
            action='store_const', const=partial(binary_multiset_op, opfunc),)
    
    multiset_parser.add_argument('file1', type=file)
    multiset_parser.add_argument('file2', type=file)

    options = parser.parse_args()
    options.func(options.file1, options.file2)

if __name__=="__main__":
    main()