Exercice Sam & Max

import unicodedata
from codecs import open
import re
from operator import itemgetter
import sys


# unicodedata.normalize ignores ligature characters...
def decode_str(text):
    a = ''
    descriptors = ['LETTER', 'LIGATURE']
    for c in text:
        char = None
        try:
            desc = unicodedata.name(c)
            for d in descriptors:
                index = desc.find(d)
                if index >= 0:
                    index = index + len(d) + 1  # +1: space
                    end_index = desc[index:].find(' ')  # find next space
                    if end_index >= 0:
                        char = desc[index:index+end_index]
                    else:
                        char = desc[index:]
        except ValueError:
            pass

        a += ' ' if char is None else char.lower()
    return a


def sort_dict(x, y):
    k1, v1 = x
    k2, v2 = y
    if len(v1) != len(v2):  # Sort by number of pos
        return len(v1) - len(v2)
    return v1[0] - v2[0]  # If same num of pos: sort by first pos


def find_positions(filename):
    with open(filename, encoding='utf8') as f:
        data = f.read()
    data = decode_str(data)

    words = {}
    for i, w in enumerate(re.findall('(\w+)', data)):
        words.setdefault(w, []).append(i)

    for w, pos in sorted(words.iteritems(), cmp=sort_dict):
        print '- %s: %s' % (w, ', '.join([str(p) for p in pos]))

if __name__ == '__main__':
    if len(sys.argv) == 2:
        find_positions(sys.argv[1])
    else:
        print 'Usage: %s filename' % sys.argv[0]