# This program finds the distribution of leading nonzero digits in a set of
# numbers.  The program is useful for exploring the phenomenon known as
# Benford's Law.  It simply ignores tokens that don't have a nonzero digit in
# them, so you can feed it input files that have extra text like labels.

from string import *

# returns the first nonzero digit of a string or 0 if no such digit exists
def firstDigitOf(str):
    for ch in str:
        if ch >= "1" and ch <= "9":
            return atoi(ch)
    return 0

# open input file
print "Let's count those leading digits..."
name = raw_input("input file name? ")
f = open(name, 'r')
print

# count occurrences of digits
count = []
for i in range(10):
    count.append(0)
for line in f:
    for token in split(line):
        digit = firstDigitOf(token)
        count[digit] += 1

# compute overall sum
total = 0
for n in count:
    total += n
if count[0] > 0:
   print "excluding", count[0], "tokens"
total -= count[0]

# print table
print "Digit Count Percent"
for i in range(1, 10):
    pct = count[i] * 100.0 / total
    print "%5d %5d %6.2f" % (i, count[i], pct)
print "Total %5d %6.2f" % (total, 100.0)