bare-bones uniq for cmake trace profiling

cmake_prof.py

#!/usr/bin/python

# usage:
#  > cmake --trace > foo
#  > cmake_prof.py foo


# Return a sorted list of (hitcount, line) for lines in a given file
#   dramatically faster than running `sort | uniq` on large files.

def file_line_hitcount(fname):
    with open(fname) as f:
        h = {}
        for l in f.readlines(): h[l] = h.get(l, 0) + 1

        rev_paired = [(h[key],key) for key in h.keys()]
        return sorted(rev_paired, cmp=lambda x,y: y[0] - x[0])

if __name__ == '__main__' and not '__file__' in globals():
    import sys
    assert(len(sys.argv) == 2)
    for c,l in file_line_hitcount(sys.argv[1]):
        s1,s2 = l.strip().split(':')
        print s1 + ": " + str(c) + " -- " + s2