Faster way to do this?

[Niklas Frykholm]

Here is yet another take on the problem. It is quite a bit faster for 
most "sane" word lengths:

	niklas% python wordfinder.py somnabulistexplos
	Read 234937 words in 1.39s
	Method 1: found 3566 words in 17.20s
	Built dict in 17.73s
	Dict method found 3566 words in 0.45s
	
---- BEGIN CODE
	
def main():
	
	# ... Read in wordlist and words here
	
	start = time.time()
	dict = Dict(words)
	print 'Built dict in %.2fs' % (time.time() - start)

	start = time.time()
	found = dict.count(word)
	print 'Dict method found %d words in %.2fs' % (found, time.time() - start)
	
class Dict:
	def __init__(self, words):
		self.dict = {}
		for w in words:
			w = self.sort_letters(w)
			for i in range(1,len(w)):
				sub = w[0:i]
				if not self.dict.has_key(sub):
					self.dict[sub] = 0
			self.dict[w] = self.dict.get(w,0) + 1
	
	def sort_letters(self, w):
		w = [c for c in w]
		w.sort()
		return ''.join(w)
		
	def count(self, word):
		self.counted = {}
		word = self.sort_letters(word)
		return self.count1('', word)
		
	def count1(self, start, rem):
		sum = self.dict.get(start, 0)
		if sum > 0:
			if not self.counted.has_key(start):
				self.counted[start] = 1
			else:
				sum = 0
		for i in range(len(rem)):
			next = start + rem[i]
			if self.dict.has_key(next):
				sum += self.count1(next, rem[i+1:])
		return sum
		
---- END CODE