pick

#! /usr/bin/env python3
'''
Use -h for help
'''

import argparse
import hashlib
from math import log

# Argument parsing
parser = argparse.ArgumentParser(description='Do RFC3797 randomization')
group = parser.add_mutually_exclusive_group(required=True)
group.add_argument('-s', '--seedfile',
        help='File containing the seeds',
        metavar='seeds', type=open, required=False)
group.add_argument('-t', '--seedtext',
        help='Text string to use as seed',
        metavar='seedtext', type=ascii, required=False)
parser.add_argument('-n', '--names',
        help='File containing the list of names',
        metavar='names', type=open, required=True)
parser.add_argument('-c', '--count',
        help='Number of entries to print',
        default=12, metavar='N', type=int)
parser.add_argument('-d', '--debug',
        action='store_true', help='Print some debugging information')
parser.add_argument('-x', '--extra',
        required=False, help='An additional seed for the input')
args = parser.parse_args()

# How many bits of entropy needed to select N from P
#                  P!
#    log  ( ----------------- )
#       2       N! * (P-N)!
def entropy_needed(P, N):
    # Not picking anything, or picking everyone?
    if N < 1 or N >= P: return 0

    result = 0.0
    for i in range(P, P - N, -1):
        result += log(i, 2)
    for i in range(N, 1, -1):
        result -= log(i, 2)
    return int(result) + 0.5

# Read the names, print the count
names = [ line.strip() for line in args.names if line.strip() != "" ]
numnames = len(names)
if args.debug: print('Read', numnames, 'names')
if numnames > 0xFFFF:
    raise SystemError("Too many names (limit is 0xFFFF)")

if args.debug:
    print('Need approximately',
            entropy_needed(numnames, args.count),
            'bits of entropy');

# We could directly shrink the names list (such as by
# using `names.pop(k)`) but then the indices would vary.
selected = [ i + 1 for i in range(0, numnames) ]

# Read the seeds or get the seed text.
keystring = ''
if args.seedfile:
    seeds = []
    for line in args.seedfile:
        if line.strip() == "" or line.strip()[0:1] == '#': continue
        values = [ int(x) for x in line.split() ]
        values.sort()
        seeds.append(values)

    # Generate the key string and convert to bytes
    for s in seeds:
        keystring += '.'.join([ str(n) for n in s ])
        keystring += './'
else:
    keystring = args.seedtext
if args.extra is not None: keystring += args.extra + './'
if args.debug:  print('Key string is', keystring, '\n')
keybytes = keystring.encode()

# Do the work
prefix = 0
div = numnames
print(' index        hex value of MD5        div  selected')
for i in range(0, args.count):
    bracket = prefix.to_bytes(2, 'big')
    result = hashlib.md5(bracket + keybytes + bracket).hexdigest().upper()
    k = int(result, 16) % div
    for j in range(0, numnames):
        if selected[j]:
            k -= 1
            if k < 0: break
    assert k < 0
    print(' %4d  %s  %2d  -> %2d <- %s' % \
            (i + 1, result, div, selected[j], names[j]))
    selected[j] = 0
    div -= 1
    prefix += 1