def fscore (precision, recall):
	if (precision+recall!=0):
		return 2*precision*recall/(precision+recall)
	else:
		return 0

def precision(list1,list2):#lists should be 1,0 lists and same size
#list 2 is (gold standard) hand tagged list
#list 1 is the code generated list
	main_list=[]
	if len(list1)!=len(list2):
		print "Error! Sizes dont match"
		return -1
	else:
		#print list1,list2
		for i in xrange(len(list1)):
			if (list1[i]==1 and list1[i]==list2[i]):
				main_list.append(1);
			else:
				main_list.append(0);
		#print main_list;
		return float(sum(main_list))/sum(list1)		

def recall(list1, list2):
#lists should be 1,1 and same size
#list 2 is hand tagged list
#list 1 is the code generated list
	main_list=[]	
	if len(list1)!=len(list2):
		print "Error! Sizes dont match"
		return -1
	else:
		for i in xrange(len(list1)):
			if (list1[i]==1 and list1[i]==list2[i]):
				main_list.append(1);
			else:
				main_list.append(0);
		return float(sum(main_list))/sum(list2)

def get_binary_list(word):
	bin_list=[]
	flag=0
	for i in xrange(len(word)):
		if(word[i]==' '):
			bin_list.append(1);
			flag=1;
		else:
			if(flag==1):
				flag=0;
			else:
				bin_list.append(0);
	return bin_list

def compare_words(word1,word2):
	list1=get_binary_list(word1);
	list2=get_binary_list(word2);
	print precision(list1,list2), recall(list1,list2),fscore(precision(list1,list2), recall(list1,list2));
i=0
f="testData2_copy.txt"
data = open(f).read()
data = data.decode("utf=8")
data = data.split('\n')
big_list1=[]
big_list2=[]
#with open(f,'r') as f:
for line in data:
	#print line
	#line=line.decode('utf-8')
	[word1,word2]=line.split("\t")
	print word2
	#a = line.split(":")
	#print a[0].encode("utf-8")
	#print line.split(':')
	#word2=word2[:-1]#remove the newline character
	#print word1,word2
	if(len(get_binary_list(word1))!=len(get_binary_list(word2))):
		continue
	else:
		i=i+1
		big_list1.extend(get_binary_list(word1));
		big_list2.extend(get_binary_list(word2));
	#print i
	#print big_list1,big_list2
	#print len(big_list1),len(big_list2)
	prec=precision(big_list1,big_list2)
	rec=recall(big_list1,big_list2)
	print "precision=",prec,"recall=",rec,"fscore=",fscore(prec,rec);