#!/usr/bin/python

# P2PU - Python Programming - Chapter 5 - Datatypes
# Chapter 10 - Tuples
print 'Python for Informatics - Chapter 10: Tuples'

# Exercise 10.1
print '\n# Exercise 10.1\n'
fname =  raw_input('Enter file name: ')
try:
    ffile = open(fname)
except:
    print 'File', fname, 'cannot be opened!'
    exit()
mails = dict()
for line in ffile:
    words = line.split()
    if len(words) < 2 or words[0] != 'From': continue
    mails[words[1]] = mails.get(words[1], 0) + 1
lst = list()
for key, val in mails.items():
   lst.append((val,key))
lst.sort(reverse=True)
print lst[0][1], ':', lst[0][0]

# Exercise 10.2
print '\n# Exercise 10.2\n'
fname =  raw_input('Enter file name: ')
try:
    ffile = open(fname)
except:
    print 'File', fname, 'cannot be opened!'
    exit()
hcount = dict()
for line in ffile:
    words = line.split()
    if len(words) < 6 or words[0] != 'From': continue
    h = words[5].split(':')
    hcount[h[0]] = hcount.get(h[0], 0) + 1
lst = list()
for key, val in hcount.items():
   lst.append((key,val))
lst.sort()
for key, val in lst:
   print key, val

# Exercise 10.3
import string
print '\n# Exercise 10.3\n'
def most_frequent(f):
   #small change: function takes a file instead of a string
   letters = dict()
   for line in ffile:
      # remove puntuation
      line = line.translate(None, string.punctuation)
      # remove spaces
      line = line.translate(None, ' ')
      # all lowercase
      line = line.lower()
      # remove numbers
      line = line.translate(None, '0123456789')
      # remove newlines
      line = line.strip()
      for letter in line:
         letters[letter] = letters.get(letter, 0) + 1
   lst = list()
   for key, val in letters.items():
      lst.append((val,key))
   lst.sort(reverse=True)
   return lst

fname =  raw_input('Enter file name: ')
try:
    ffile = open(fname)
except:
    print 'File', fname, 'cannot be opened!'
    exit()
res = most_frequent(ffile)
for key, val in res[0:10]:
   print key, val

# Top 10 letters from newspaper article in
#   English: e t a n s o i r h l
#   French:  e s a n i t l r u o
#   Dutch:   e n t a i r d l o s
#   German:  e n a r i s t h u l