Attributeerror: 'str' object has no attribute 'copy' in input nltk Python

lines_seen = set() # holds lines already seen

#check for duplicates
with open("male.txt", "r+") as f:
    d = f.readlines()
    f.seek(0)
    for i in d:
        if i not in lines_seen:
            f.write(i)
            lines_seen.add(i)
    f.truncate()

with open("female.txt", "r+") as f:
    d = f.readlines()
    f.seek(0)
    for i in d:
        if i not in lines_seen:
            f.write(i)
            lines_seen.add(i)
    f.truncate()

def gender_features(word):
	return {'last_letter': word[-1]}

# output : {'last_letter': 'r'}
# importing libraries
import random
from nltk.corpus import names
import nltk
#nltk.download()

def gender_features(word):
	return {'last_letter':word[-1]}

# preparing a list of examples and corresponding class labels.
labeled_names = ([(name, 'male') for name in names.words('male.txt')]+
			[(name, 'female') for name in names.words('female.txt')])

random.shuffle(labeled_names)

# we use the feature extractor to process the names data.
featuresets = [(gender_features(n), gender)
			for (n, gender)in labeled_names]

# Divide the resulting list of feature
# sets into a training set and a test set.
train_set, test_set = featuresets[500:], featuresets[:500]

# The training set is used to
# train a new "naive Bayes" classifier.
classifier = nltk.NaiveBayesClassifier.train(train_set)
x2b = input("name: ")
print(classifier.classify(x2b))

# output should be 'male'
print(nltk.classify.accuracy(classifier, train_set))

# it shows the accuracy of our classifier and
# train_set. which must be more than 99 %
classifier.show_most_informative_features(10)