"""File:     word_list.py
   Purpose:  Build a list of the words on a web page that
             that has a simple format.

   Run:      python word_list.py

   Input:    name of html file
   Output:   list of the words on the page

   Note:     Each line of the html file is a tag
             or a word.
"""
#----------------------------------------------------------------------
def get_words(ifile):
   """Extract words from an html page"""
   wlist = []
   for line in ifile:
      line = line.strip() # Get rid of leading and trailing whitespace
      if len(line) > 0:   # If this is a nonempty line . . .
         if line[0] != '<' and not line in wlist:
            wlist.append(line)
   return wlist


#----------------------------------------------------------------------
def print_words(filename, word_list):
   """Print the words in a list"""
   print "The words in", filename, "are:"
   print "  ",
   for word in word_list:
      print word,
   print

#----------------------------------------------------------------------
filename = raw_input("What's the name of the HTML file?\n   ")
ifile = open(filename, "r")
word_list = get_words(ifile)
print_words(filename, word_list)
ifile.close()