# Class Example - WordCounts
# Internal Data Structure: dictionary of frequencies of words
#
# Many extra comments added for explanation purposes.
# Not necessarily an example of good style.
# Modified to take a list of words instead of filename as input
# to the constructor so this code can be cut and pasted into
# the Python Tutor.

class WordCounts:
  """Represents the words in a file."""
  # Internal representation:
  # variable wordcounts is a dictionary from words to their frequency

  def __init__(self, word_list):
    """Create a WordCounts object from the given word list"""
    # This is a constructor, gets called when we do:
    # var_name = WordCounts(list_of_words)
    # WordCount objects will have one data attribute (aka "field")
    # This is created here: wordcounts 
    self.wordcounts = {}
    for w in word_list:
      self.wordcounts.setdefault(w, 0)
      self.wordcounts[w] += 1
    
  def word_count(self, word):
    """Return the count of the given word"""
    if self.wordcounts.has_key(word):
      return self.wordcounts[word]
    else:
      return 0

  def topk(self, k=10):
    """Return a list of the top k most frequent words in order"""
    scores_with_words = [(c,w) for (w,c) in self.wordcounts.items()]
    scores_with_words.sort(reverse=True)
    return scores_with_words[0:k]

  def total_words(self):
    """Return the total number of words in the file"""
    return sum(self.wordcounts.values())
  

#---------------------------------------------------------
# Sample client program
#
# The EXACT same sample program is used with both versions
# of the WordCount class because the client interface is  
# the same for both versions of the classes.
words = ["the", "lazy", "brown", "fox", "fox", "the", "the"]
words2 = ["the", "lazy", "brown", "fox", "fox", "hippo", "hippo", "hippo"]

# Create two instances of the class WordCount
# wc and wc2 are both objects of type WordCount
wc = WordCounts(words)
wc2 = WordCounts(words2)
print "type(wc) is", type(wc)
print "type(wc2) is", type(wc2)
print "type(WordCounts) is:", type(WordCounts)

k=5
print "wc top", k, ":", wc.topk(k)
print "wc2 top", k, ":", wc2.topk(k)

word = "the"
print "wc of", word, ":", wc.word_count(word)
print "wc2 of", word, ":", wc2.word_count(word)

# A few more commands:
# List attributes of the class
#print "dir on WordCounts class:", dir(WordCounts)
# List attributes of an object
#print "dir on wc object:", dir(wc)
# Get docstring for the class and all its methods
#help(WordCounts)