mmhistogram

#!/usr/bin/env python2
import argparse
import itertools
import math
import sys

parser = argparse.ArgumentParser(description='Print log-2 histogram, like systemtap')
parser.add_argument('-t', '--title', default="Values",
                    help='title to print')
parser.add_argument('-c', '--columns', type=int, default="50",
                    help='number of colums for')
parser.add_argument('-b', '--base', type=int, default="2",
                    help='log base')
parser.add_argument('-l', '--linear', action='store_true',
                    help='do linear, not log')
parser.add_argument('-j', '--justval', action='store_true',
                    help='ignore bounds, use values from input')
parser.add_argument('-p', '--percentage', action='store_true',
                    help='Print percentage instead of counts')
args = parser.parse_args()


M = []
for line in sys.stdin:
    line = line.strip()
    v = float(line)
    M.append( v )

M.sort(reverse=True)

totalcount = len(M)
minval = M[-1]
maxval = M[0]
avgval = sum(M) / float(len(M))
devval = math.sqrt(sum([(m - avgval)**2 for m in M]) / float(len(M)))

medval = M[len(M)/2]

KV = []

for i in itertools.count():
    if args.justval:
        bound=M[-1]
        boundb=M[-1]
    elif not args.linear:
        if i > 0:
            boundb = args.base**(i-1)
        else:
            boundb = 0
        bound = args.base**i
    else:
        boundb = args.base*(i)
        bound = args.base*(i+1)

    c = 0
    while len(M) > 0 and bound >= M[-1] :
        c += 1
        M.pop()
    KV.append( (boundb, bound, c) )

    if not M:
        break

maxcount = float(max(c for _, _, c in KV))
maxbound = KV[-1][0]
boundl = max(len(str(maxbound)), len('value'))

print "%s min:%.2f avg:%.2f med=%.2f max:%.2f dev:%.2f count:%d" % (
    args.title,
    minval,
    avgval,
    medval,
    maxval,
    devval,
    totalcount,
)
print "%s:" % (
    args.title,
)
print "%*s |%*s %s" % (
    boundl+1,
    "value",
    args.columns,
    "-" * args.columns,
    "count"
)

for boundb, bound, c in KV:
    if args.percentage:
        cp = "%5.2f%%" % ((c / float(totalcount))*100.0,)
    else:
        cp = "%d" % (c,)
    print "%*d |%*s %s" % (
        boundl + 1,
        boundb,
        args.columns,
        "*" * int(args.columns * (c/maxcount)),
        cp
    )