def setUp(self):
self.subject = "dummySubject"
self.body = "dummyBody"
self.analyzer = EmailAnalyzer()
self.clean_subject = [
"best", "quick", "netco"
] # données pour mocker "return_value" du "clean_text"
self.clean_body = [
"prescription", "drug", "overview", "operations"
] # données pour mocker "return_value" du "clean_text"
self.spam_ham_body_prob_true = (
0,
0,
) # données pour mocker "return_value" du "spam_ham_body_prob"
self.subject_spam_ham_prob_true = (
0,
0,
) # données pour mocker "return_value" du "subject_spam_ham_prob"
self.spam_ham_body_prob_false = (
0,
0,
) # données pour mocker "return_value" du "spam_ham_body_prob"
self.subject_spam_ham_prob_false = (
0,
0,
) # données pour mocker "return_value" du "subject_spam_ham_prob"
self.vocab = (
{
"spam_sub": {
"best": 1 / 4,
"online": 1 / 4,
"medicine": 1 / 4,
"here": 1 / 4,
},
"ham_sub": {
"netco": 1 / 3,
"due": 1 / 3,
"diligence": 1 / 3
},
"spam_body": {
"prescription": 1 / 5,
"drug": 1 / 5,
"simple": 1 / 5,
"quick": 1 / 5,
"affordable": 1 / 5
},
"ham_body": {
"big": 1 / 6,
"pig": 1 / 6,
"met": 1 / 6,
"today": 1 / 6,
"overview": 1 / 6,
"operations": 1 / 6
}
}
) # vocabulaire avec les valeurs de la probabilité pour mocker "return_value" du "load_dict"
self.spam_ham_body_prob_expected = 0, 0 # valeurs de la probabilité attendus
self.subject_spam_ham_prob_expected = 0, 0 # valeurs de la probabilité attendus
def test_body_spam_ham_prob_Returns_expected_probability(
self, mock_load_vocab):
"""
Il faut mocker la fonction "load_dict"
Il faut vérifier que probabilité est calculée correctement donné le "body" à l'entrée
(ces probabilites devron etre calcule selon l'enonce dans le TP1 )
"""
email_analyzer = EmailAnalyzer()
mock_load_vocab.return_value = self.vocab
self.assertEqual(email_analyzer.body_spam_ham_prob(self.body_true),
self.body_spam_ham_prob_expected)
def test_is_spam_Returns_True_if_spam_prob_is_higher(
self, mock_subject_spam_ham_prob, mock_body_spam_ham_prob,
mock_clean_text):
"""
Il faut mocker les fonctions "spam_ham_body_prob" et "subject_spam_ham_prob".
La sortie de la fonction doit être True si probabilité spam > probabilité ham
(ces probabilites devron etre calcule selon l'enonce dans le TP1 )
"""
mock_clean_text.return_value = self.clean_subject_true
mock_subject_spam_ham_prob.return_value = self.subject_spam_ham_prob_true
mock_clean_text.return_value = self.clean_body_true
mock_body_spam_ham_prob.return_value = self.body_spam_ham_prob_true
email_analyzer = EmailAnalyzer()
self.assertEqual(
email_analyzer.is_spam(self.subject_true, self.body_true), True)
def evaluate():
tp = 0
tn = 0
fp = 0
fn = 0
total = 0
analyzer = EmailAnalyzer()
with open("200-mails.json") as email_file:
new_emails = json.load(email_file)
counter = 0
for e_mail in new_emails["dataset"]:
counter += 1
print(counter)
new_email = e_mail["mail"]
subject = new_email["Subject"]
body = new_email["Body"]
spam = new_email["Spam"]
if ((analyzer.is_spam(subject, body))) and (spam == "true"):
tp += 1
if (not (analyzer.is_spam(subject, body))) and (spam == "false"):
tn += 1
if ((analyzer.is_spam(subject, body))) and (spam == "false"):
fp += 1
if (not (analyzer.is_spam(subject, body))) and (spam == "true"):
fn += 1
total += 1
print("Accuracy: ", (tp + tn) / (tp + tn + fp + fn))
print("Precision: ", tp / (tp + fp))
print("Recall: ", tp / (tp + fn))
return True
def evaluate(is_log_estimation, is_log_combination, clean_text_mode, k):
tp = 0
tn = 0
fp = 0
fn = 0
total = 0
analyzer = EmailAnalyzer()
with open("test_set.json") as email_file:
new_emails = json.load(email_file)
i = 0
email_count = len(new_emails["dataset"])
print("Evaluating emails ")
for e_mail in new_emails["dataset"]:
i += 1
print("\rEmail " + str(i) + "/" + str(email_count), end="")
new_email = e_mail["mail"]
subject = new_email["Subject"]
body = new_email["Body"]
spam = new_email["Spam"]
if ((analyzer.is_spam(subject, body, is_log_estimation,
is_log_combination, clean_text_mode,
k))) and (spam == "true"):
tp += 1
if (not (analyzer.is_spam(subject, body, is_log_estimation,
is_log_combination, clean_text_mode,
k))) and (spam == "false"):
tn += 1
if ((analyzer.is_spam(subject, body, is_log_estimation,
is_log_combination, clean_text_mode,
k))) and (spam == "false"):
fp += 1
if (not (analyzer.is_spam(subject, body, is_log_estimation,
is_log_combination, clean_text_mode,
k))) and (spam == "true"):
fn += 1
total += 1
print("")
print("\nAccuracy: ", round((tp + tn) / (tp + tn + fp + fn), 2))
if (tp + fp == 0):
print("Precision: ", 0)
else:
print("Precision: ", round(tp / (tp + fp), 2))
print("Recall: ", round(tp / (tp + fn), 2))
return True
def evaluate(is_log_estimation, is_log_combo, calculation_mode):
tp = 0
tn = 0
fp = 0
fn = 0
total = 0
analyzer = EmailAnalyzer()
with open("200-mails.json") as email_file:
new_emails = json.load(email_file)
counter = 0
for e_mail in new_emails["dataset"]:
counter += 1
print(counter)
new_email = e_mail["mail"]
subject = new_email["Subject"]
body = new_email["Body"]
spam = new_email["Spam"]
if ((analyzer.is_spam(subject, body, is_log_estimation, False,
0))) and (spam == "true"):
tp += 1
if (not (analyzer.is_spam(subject, body, False, False,
0))) and (spam == "false"):
tn += 1
if ((analyzer.is_spam(subject, body, False, False,
0))) and (spam == "false"):
fp += 1
if (not (analyzer.is_spam(subject, body, False, False,
0))) and (spam == "true"):
fn += 1
total += 1
accuracy = (tp + tn) / (tp + tn + fp + fn)
precision = tp / (tp + fp)
recall = tp / (tp + fn)
print("Accuracy: ", accuracy)
print("Precision: ", tp / (tp + fp))
print("Recall: ", tp / (tp + fn))
return accuracy, precision, recall
def evaluate(estimation_option, combination_option):
tp = 0
tn = 0
fp = 0
fn = 0
total = 0
analyzer = EmailAnalyzer()
with open("test-emails.json") as email_file:
new_emails = json.load(email_file)
for e_mail in new_emails["dataset"]:
new_email = e_mail["mail"]
subject = new_email["Subject"]
body = new_email["Body"]
spam = new_email["Spam"]
if (analyzer.is_spam_with_params(subject, body, estimation_option,
combination_option, cleaning_mode,
0.3)) and (spam == "true"):
tp += 1
if (not (analyzer.is_spam_with_params(
subject, body, estimation_option, combination_option,
cleaning_mode, 0.3))) and (spam == "false"):
tn += 1
if (analyzer.is_spam_with_params(subject, body, estimation_option,
combination_option, cleaning_mode,
0.3)) and (spam == "false"):
fp += 1
if (not (analyzer.is_spam_with_params(
subject, body, estimation_option, combination_option,
cleaning_mode, 0.3))) and (spam == "true"):
fn += 1
total += 1
print("Accuracy: ", (tp + tn) / (tp + tn + fp + fn))
print("Precision: ", tp / (tp + fp))
print("Recall: ", tp / (tp + fn))
return True
def test_is_spam_function_two_returns_true_vns_test_four(self):
return_val = EmailAnalyzer.is_spam_function_two(False, 20, 65)
self.assertTrue(return_val)
def test_is_spam_function_two_returns_true_vns_test_three(self):
return_val = EmailAnalyzer.is_spam_function_two(True, 20, 80)
self.assertTrue(return_val)
def test_is_spam_function_two_returns_false_pic_test_one(self):
return_val = EmailAnalyzer.is_spam_function_two(False, 80, 65)
self.assertFalse(return_val)
def test_is_spam_function_one_returns_false_icc_test_twenty(self):
return_val = EmailAnalyzer.is_spam_function_one(True, 35, 80, 60)
self.assertTrue(return_val)
def test_is_spam_function_one_returns_true_icc_test_nineteen(self):
return_val = EmailAnalyzer.is_spam_function_one(True, 35, 65, 60)
self.assertTrue(return_val)
def test_is_spam_function_one_returns_false_icc_test_seventeen(self):
return_val = EmailAnalyzer.is_spam_function_one(False, 35, 65, 80)
self.assertFalse(return_val)
class TestEmailAnalyzer(unittest.TestCase):
def setUp(self):
self.analyzer = EmailAnalyzer()
self.subject_true = " no more outdated software ! upgrade !"
self.body_true = "we get you the best deal ! skip the retail box and save !\namazing special # 1 :\nadobe - photoshop 7 premiere 7 illustrator 10 = only $ 120\namazing special # 2 :\nwindows xp professional + microsoft office xp professional = only $ 80\namazing special # 3 :\nadobe photoshop cs + adobe illustrator cs + adobe indesign cs\namazing special # 4 :\n"
self.clean_subject_true = [
'more', 'oudat', 'software', 'upgrade'
] # données pour mocker "return_value" du "clean_text"
self.clean_body_true = [
'get', 'best', 'deal', 'skip', 'retail', 'box', 'sav', 'amaz',
'special', 'adobe', 'photoshop', 'premiere', 'illustrator', 'only',
'windows', 'xp', 'professional', 'microsoft', 'office', 'cs',
'indesign'
] # données pour mocker "return_value" du "clean_text"
self.subject_false = "re :"
self.body_false = "we are using it for other things . mary joyce and robert have discussed with mcmahon and bowen .\n- - - - - original message - - - - -\nfrom : kitchen louise\nsent : monday december 10 2001 8 : 26 am\nto : oxley david\nsubject :\nwhat happens to the money in wachovia ?\nlouise kitchen\nchief operating \n"
self.clean_subject_false = [
're'
] # données pour mocker "return_value" du "clean_text"
self.clean_body_false = [
'us', 'other', 'thing', 'mary', 'joyce', 'robert', 'discuss',
'mcmahon', 'bowen', 'original', 'message', 'kitchen', 'louise',
'sent', 'monday', "december", 'oxley', 'david', 'subject',
'happen', 'money', 'wachovia'
]
self.spam_ham_body_prob_true = (
1,
(1 / 6),
) # données pour mocker "return_value" du "spam_ham_body_prob"
self.subject_spam_ham_prob_true = (
(2 / 3),
(1 / 6),
) # données pour mocker "return_value" du "subject_spam_ham_prob"
self.spam_ham_body_prob_false = (
(1 / 4),
(2 / 6),
) # données pour mocker "return_value" du "spam_ham_body_prob"
self.subject_spam_ham_prob_false = (
0,
(1 / 2),
) # données pour mocker "return_value" du "subject_spam_ham_prob"
self.vocab = (
{
"p_sub_spam": {
"upgrade": 1 / 3,
"software": 1 / 3
},
"p_sub_ham": {
"re": 1 / 2,
"annoucement": 1 / 6,
"more": 1 / 6
},
"p_body_spam": {
"best": 1 / 4,
"deal": 1 / 4,
"skip": 1 / 4,
"special": 1 / 4,
"money": 1 / 4
},
"p_body_ham": {
"today": 1 / 6,
"professional": 1 / 6,
"meet": 1 / 6,
"discuss": 1 / 6,
"sent": 1 / 6
}
}
) # vocabulaire avec les valeurs de la probabilité pour mocker "return_value" du "load_dict"
# valeurs de la probabilité attendus : (0.5925*1/(256*pow(6,17))), (0.4075*1/pow(6,21))
self.spam_ham_body_prob_expected = (1.3673419333309543e-16,
1.8575963755415577e-17)
# valeurs de la probabilité attendus : (0.5925*1/81, 0.4075*1/6*1/4*1/4*1/4)
self.subject_spam_ham_prob_expected = (0.007314814814814815,
0.0010611979166666665)
def tearDown(self):
pass
@patch("email_analyzer.EmailAnalyzer.clean_text")
@patch("email_analyzer.EmailAnalyzer.spam_ham_body_prob")
@patch("email_analyzer.EmailAnalyzer.spam_ham_subject_prob")
def test_is_spam_Returns_True_if_spam_prob_is_higher(
self, mock_subject_spam_ham_prob, mock_spam_ham_body_prob,
mock_clean_text):
mock_subject_spam_ham_prob.return_value = self.subject_spam_ham_prob_true
mock_spam_ham_body_prob.return_value = self.spam_ham_body_prob_true
return_val = self.analyzer.is_spam(self.subject_true, self.body_true)
self.assertTrue(return_val)
"""
Il faut mocker les fonctions "spam_ham_body_prob" et "subject_spam_ham_prob".
La sortie de la fonction doit être True si probabilité spam > probabilité ham
(ces probabilites devron etre calcule selon l'enonce dans le TP1 )
"""
@patch("email_analyzer.EmailAnalyzer.clean_text")
@patch("email_analyzer.EmailAnalyzer.spam_ham_body_prob")
@patch("email_analyzer.EmailAnalyzer.spam_ham_subject_prob")
def test_is_spam_Returns_False_if_spam_prob_is_lower(
self, mock_subject_spam_ham_prob, mock_spam_ham_body_prob,
mock_clean_text):
mock_subject_spam_ham_prob.return_value = self.subject_spam_ham_prob_false
mock_spam_ham_body_prob.return_value = self.spam_ham_body_prob_false
return_val = self.analyzer.is_spam(self.subject_false, self.body_false)
self.assertFalse(return_val)
"""
Il faut mocker les fonctions "spam_ham_body_prob" et "subject_spam_ham_prob".
La sortie de la fonction doit être False si probabilité spam probabilité ham
(ces probabilites devron etre calcule selon l'enonce dans le TP1 )
"""
@patch("email_analyzer.EmailAnalyzer.load_dict")
def test_spam_ham_body_prob_Returns_expected_probability(
self, mock_load_dict):
mock_load_dict.return_value = self.vocab
self.assertEqual(
self.analyzer.spam_ham_body_prob(self.clean_body_true),
self.spam_ham_body_prob_expected)
"""
Il faut mocker la fonction "load_dict"
Il faut vérifier que probabilité est calculée correctement donné le "body" à l'entrée
(ces probabilites devron etre calcule selon l'enonce dans le TP1 )
"""
@patch("email_analyzer.EmailAnalyzer.load_dict")
def test_subject_spam_ham_prob_Returns_expected_probability(
self, mock_load_dict):
mock_load_dict.return_value = self.vocab
self.assertEqual(
self.analyzer.spam_ham_subject_prob(self.clean_subject_true),
self.subject_spam_ham_prob_expected)
"""
def test_is_spam_function_one_returns_false_acc_test_eight(self):
return_val = EmailAnalyzer.is_spam_function_one(False, 20, 70, 80)
self.assertFalse(return_val)
def test_is_spam_function_one_returns_true_acc_test_five(self):
return_val = EmailAnalyzer.is_spam_function_one(True, 20, 50, 80)
self.assertTrue(return_val)
class TestEmailAnalyzer(unittest.TestCase):
def setUp(self):
self.subject = "dummySubject"
self.body = "dummyBody"
self.analyzer = EmailAnalyzer()
self.clean_subject = ["best", "quick", "netco"] # données pour mocker "return_value" du "clean_text"
self.clean_body = ["prescription", "drug", "overview",
"operations"] # données pour mocker "return_value" du "clean_text"
self.spam_ham_body_prob_true = (
0,
0,
) # données pour mocker "return_value" du "spam_ham_body_prob"
self.subject_spam_ham_prob_true = (
0,
0,
) # données pour mocker "return_value" du "subject_spam_ham_prob"
self.spam_ham_body_prob_false = (
0,
0,
) # données pour mocker "return_value" du "spam_ham_body_prob"
self.subject_spam_ham_prob_false = (
0,
0,
) # données pour mocker "return_value" du "subject_spam_ham_prob"
self.vocab = (
{
"spam_sub": {
"best": 1 / 4,
"online": 1 / 4,
"medicine": 1 / 4,
"here": 1 / 4,
},
"ham_sub": {
"netco": 1 / 3,
"due": 1 / 3,
"diligence": 1 / 3
},
"spam_body": {
"prescription": 1 / 5,
"drug": 1 / 5,
"simple": 1 / 5,
"quick": 1 / 5,
"affordable": 1 / 5
},
"ham_body": {
"big": 1 / 6,
"pig": 1 / 6,
"met": 1 / 6,
"today": 1 / 6,
"overview": 1 / 6,
"operations": 1 / 6
}
}
) # vocabulaire avec les valeurs de la probabilité pour mocker "return_value" du "load_dict"
self.spam_ham_body_prob_expected = 0, 0 # valeurs de la probabilité attendus
self.subject_spam_ham_prob_expected = 0, 0 # valeurs de la probabilité attendus
def tearDown(self):
pass
### Tests pour l'Active clause coverage
def test_is_spam_function_one_returns_true_acc_test_one(self):
return_val = EmailAnalyzer.is_spam_function_one(True, 35, 65, 60)
self.assertTrue(return_val)
def test_is_spam_function_one_returns_false_acc_test_two(self):
return_val = EmailAnalyzer.is_spam_function_one(True, 35, 65, 80)
self.assertFalse(return_val)
def test_is_spam_function_one_returns_true_acc_test_three(self):
return_val = EmailAnalyzer.is_spam_function_one(False, 20, 50, 50)
self.assertTrue(return_val)
def test_is_spam_function_one_returns_false_acc_test_four(self):
return_val = EmailAnalyzer.is_spam_function_one(False, 35, 50, 50)
self.assertFalse(return_val)
def test_is_spam_function_one_returns_true_acc_test_five(self):
return_val = EmailAnalyzer.is_spam_function_one(True, 20, 50, 80)
self.assertTrue(return_val)
def test_is_spam_function_one_returns_false_acc_test_six(self):
return_val = EmailAnalyzer.is_spam_function_one(True, 20, 70, 80)
self.assertFalse(return_val)
def test_is_spam_function_one_returns_true_acc_test_seven(self):
return_val = EmailAnalyzer.is_spam_function_one(False, 20, 70, 50)
self.assertTrue(return_val)
def test_is_spam_function_one_returns_false_acc_test_eight(self):
return_val = EmailAnalyzer.is_spam_function_one(False, 20, 70, 80)
self.assertFalse(return_val)
def test_is_spam_function_one_returns_false_acc_test_nine(self):
return_val = EmailAnalyzer.is_spam_function_one(False, 20, 76, 50)
self.assertFalse(return_val)
# Tests pour l'Inactive Clause Coverage
# P est clause majeure:
def test_is_spam_function_one_returns_false_icc_test_one(self):
return_val = EmailAnalyzer.is_spam_function_one(True, 35, 65, 80)
self.assertFalse(return_val)
def test_is_spam_function_one_returns_false_icc_test_two(self):
return_val = EmailAnalyzer.is_spam_function_one(False, 35, 65, 80)
self.assertFalse(return_val)
def test_is_spam_function_one_returns_true_icc_test_three(self):
return_val = EmailAnalyzer.is_spam_function_one(False, 15, 65, 50)
self.assertTrue(return_val)
def test_is_spam_function_one_returns_false_icc_test_four(self):
return_val = EmailAnalyzer.is_spam_function_one(True, 15, 65, 50)
self.assertTrue(return_val)
# H est clause majeure
def test_is_spam_function_one_returns_true_icc_test_five(self):
return_val = EmailAnalyzer.is_spam_function_one(True, 20, 65, 60)
self.assertTrue(return_val)
def test_is_spam_function_one_returns_true_icc_test_six(self):
return_val = EmailAnalyzer.is_spam_function_one(True, 15, 65, 60)
self.assertTrue(return_val)
def test_is_spam_function_one_returns_false_icc_test_seven(self):
return_val = EmailAnalyzer.is_spam_function_one(False, 40, 65, 80)
self.assertFalse(return_val)
def test_is_spam_function_one_returns_false_icc_test_eight(self):
return_val = EmailAnalyzer.is_spam_function_one(False, 15, 65, 80)
self.assertFalse(return_val)
# T1 clause majeure
def test_is_spam_function_one_returns_false_icc_test_nine(self):
return_val = EmailAnalyzer.is_spam_function_one(False, 35, 65, 80)
self.assertFalse(return_val)
def test_is_spam_function_one_returns_false_icc_test_ten(self):
return_val = EmailAnalyzer.is_spam_function_one(False, 35, 50, 80)
self.assertFalse(return_val)
def test_is_spam_function_one_returns_true_icc_test_eleven(self):
return_val = EmailAnalyzer.is_spam_function_one(False, 20, 50, 60)
self.assertTrue(return_val)
def test_is_spam_function_one_returns_false_icc_test_twelve(self):
return_val = EmailAnalyzer.is_spam_function_one(False, 20, 65, 60)
self.assertTrue(return_val)
# T2 clause majeure
def test_is_spam_function_one_returns_false_icc_test_thirteen(self):
return_val = EmailAnalyzer.is_spam_function_one(False, 35, 65, 80)
self.assertFalse(return_val)
def test_is_spam_function_one_returns_false_icc_test_fourteen(self):
return_val = EmailAnalyzer.is_spam_function_one(False, 35, 65, 50)
self.assertFalse(return_val)
def test_is_spam_function_one_returns_true_icc_test_fifteen(self):
return_val = EmailAnalyzer.is_spam_function_one(True, 20, 50, 80)
self.assertTrue(return_val)
def test_is_spam_function_one_returns_false_icc_test_sixteen(self):
return_val = EmailAnalyzer.is_spam_function_one(True, 20, 50, 50)
self.assertTrue(return_val)
# T3 clause majeure
def test_is_spam_function_one_returns_false_icc_test_seventeen(self):
return_val = EmailAnalyzer.is_spam_function_one(False, 35, 65, 80)
self.assertFalse(return_val)
def test_is_spam_function_one_returns_false_icc_test_eighteen(self):
return_val = EmailAnalyzer.is_spam_function_one(False, 35, 80, 80)
self.assertFalse(return_val)
def test_is_spam_function_one_returns_true_icc_test_nineteen(self):
return_val = EmailAnalyzer.is_spam_function_one(True, 35, 65, 60)
self.assertTrue(return_val)
def test_is_spam_function_one_returns_false_icc_test_twenty(self):
return_val = EmailAnalyzer.is_spam_function_one(True, 35, 80, 60)
self.assertTrue(return_val)
# Critère IC
def test_is_spam_function_two_returns_false_ic_test_one(self):
return_val = EmailAnalyzer.is_spam_function_two(False, 80, 80)
self.assertFalse(return_val)
def test_is_spam_function_two_returns_true_ic_test_two(self):
return_val = EmailAnalyzer.is_spam_function_two(True, 20, 65)
self.assertTrue(return_val)
# Critère PIC
def test_is_spam_function_two_returns_false_pic_test_one(self):
return_val = EmailAnalyzer.is_spam_function_two(False, 80, 65)
self.assertFalse(return_val)
def test_is_spam_function_two_returns_false_pic_test_two(self):
return_val = EmailAnalyzer.is_spam_function_two(False, 20, 80)
self.assertFalse(return_val)
def test_is_spam_function_two_returns_true_pic_test_three(self):
return_val = EmailAnalyzer.is_spam_function_two(True, 80, 65)
self.assertTrue(return_val)
def test_is_spam_function_two_returns_true_pic_test_four(self):
return_val = EmailAnalyzer.is_spam_function_two(False, 20, 50)
self.assertTrue(return_val)
# Critère VNS
def test_is_spam_function_two_returns_true_vns_test_one(self):
return_val = EmailAnalyzer.is_spam_function_two(True, 80, 80)
self.assertTrue(return_val)
def test_is_spam_function_two_returns_true_vns_test_two(self):
return_val = EmailAnalyzer.is_spam_function_two(True, 80, 65)
self.assertTrue(return_val)
def test_is_spam_function_two_returns_true_vns_test_three(self):
return_val = EmailAnalyzer.is_spam_function_two(True, 20, 80)
self.assertTrue(return_val)
def test_is_spam_function_two_returns_true_vns_test_four(self):
return_val = EmailAnalyzer.is_spam_function_two(False, 20, 65)
self.assertTrue(return_val)
def test_is_spam_function_two_returns_false_vns_test_five(self):
return_val = EmailAnalyzer.is_spam_function_two(False, 20, 80)
self.assertFalse(return_val)
def test_is_spam_function_two_returns_false_vns_test_six(self):
return_val = EmailAnalyzer.is_spam_function_two(False, 80, 60)
self.assertFalse(return_val)
def test_is_spam_function_two_returns_false_vns_test_seven(self):
return_val = EmailAnalyzer.is_spam_function_two(False, 80, 80)
self.assertFalse(return_val)
@patch("email_analyzer.EmailAnalyzer.subject_spam_ham_prob")
@patch("email_analyzer.EmailAnalyzer.spam_ham_body_prob")
def test_is_spam_Returns_True_if_spam_prob_is_higher(
self, mock_subject_spam_ham_prob, mock_spam_ham_body_prob
):
"""
Il faut mocker les fonctions "spam_ham_body_prob" et "subject_spam_ham_prob".
La sortie de la fonction doit être True si probabilité spam > probabilité ham
(ces probabilites devron etre calcule selon l'enonce dans le TP1 )
"""
mock_subject_spam_ham_prob.return_value = (10, 0)
mock_spam_ham_body_prob.return_value = (10, 0)
is_spam_return_val = self.analyzer.is_spam("dummySubject", "dummyBody")
self.assertTrue(is_spam_return_val)
@patch("email_analyzer.EmailAnalyzer.spam_ham_body_prob")
@patch("email_analyzer.EmailAnalyzer.subject_spam_ham_prob")
def test_is_spam_Returns_False_if_spam_prob_is_lower(
self, mock_subject_spam_ham_prob, mock_spam_ham_body_prob
):
"""
Il faut mocker les fonctions "spam_ham_body_prob" et "subject_spam_ham_prob".
La sortie de la fonction doit être False si probabilité spam probabilité ham
(ces probabilites devron etre calcule selon l'enonce dans le TP1 )
"""
mock_subject_spam_ham_prob.return_value = (0, 10)
mock_spam_ham_body_prob.return_value = (0, 10)
is_spam_return_val = self.analyzer.is_spam("dummySubject", "dummyBody")
self.assertFalse(is_spam_return_val)
@patch("email_analyzer.EmailAnalyzer.clean_text")
@patch("email_analyzer.EmailAnalyzer.calculate_ham_divided_by_email")
@patch("email_analyzer.EmailAnalyzer.calculate_spam_divided_by_email")
@patch("email_analyzer.EmailAnalyzer.load_dict")
def test_spam_ham_body_prob_Returns_expected_probability(self, mock_load_dict, mock_calculate_spam_divided_by_email,
mock_calculate_ham_divided_by_email, mock_clean_text):
"""
Il faut mocker la fonction "load_dict"
Il faut vérifier que probabilité est calculée correctement donné le "body" à l'entrée
(ces probabilites devront etre calcule selon l'enonce dans le TP1 )
"""
mock_load_dict.return_value = self.vocab
mock_calculate_ham_divided_by_email.return_value = 1 / 2
mock_calculate_spam_divided_by_email.return_value = 1 / 2
mock_clean_text.return_value = self.clean_body
expected_return_value = ((0.5 * 0.2 * 0.2), (1 / 2 * 1 / 6 * 1 / 6))
self.assertEqual(self.analyzer.spam_ham_body_prob(self.body), expected_return_value)
@patch("email_analyzer.EmailAnalyzer.clean_text")
@patch("email_analyzer.EmailAnalyzer.calculate_ham_divided_by_email")
@patch("email_analyzer.EmailAnalyzer.calculate_spam_divided_by_email")
@patch("email_analyzer.EmailAnalyzer.load_dict")
def test_subject_spam_ham_prob_Returns_expected_probability(self, mock_load_dict,
mock_calculate_spam_divided_by_email,
mock_calculate_ham_divided_by_email, mock_clean_text):
"""
Il faut mocker la fonction "load_dict"
il faut vérifier que probabilité est calculée correctement donné le "sujet" a l'entrée
(ces probabilites devron etre calcule selon l'enonce dans le TP1 )
"""
mock_load_dict.return_value = self.vocab
mock_calculate_ham_divided_by_email.return_value = 1 / 2
mock_calculate_spam_divided_by_email.return_value = 1 / 2
mock_clean_text.return_value = self.clean_subject
expected_return_value = ((0.5 * 0.25), (0.5*1/3))
self.assertEqual(self.analyzer.subject_spam_ham_prob(self.subject), expected_return_value)
def test_is_spam_function_two_returns_false_vns_test_seven(self):
return_val = EmailAnalyzer.is_spam_function_two(False, 80, 80)
self.assertFalse(return_val)
def __init__(self):
self.email_file = "train_set.json"
self.crud = CRUD()
self.e_mail = EmailAnalyzer()
for dirpath, subdirs, files in os.walk(rootdir):
for file in files:
if not file.startswith('.'): # Ignore hidden files
email_filenames.append(os.path.join(dirpath, file))
"""
Calling the function, and testing if the # of elements are the same
"""
for filename in email_filenames:
with open(filename, "r") as f:
data = f.read()
email_df = json_normalize(json.loads(EmailAnalyzer().parse(data)))
print(email_df)
"""
Attempting to iterate through the files and populate a larger JSON object
"""
dic_tmp = {}
for x in range(len(email_filenames)):
key_name = email_filenames[x]
with open(email_filenames[x], "r") as f:
data2 = f.read()
data2_parsed = EmailAnalyzer().parse(data2)
dic_tmp[key_name] = data2_parsed
def setUp(self):
self.analyzer = EmailAnalyzer()
self.subject_true = " no more outdated software ! upgrade !"
self.body_true = "we get you the best deal ! skip the retail box and save !\namazing special # 1 :\nadobe - photoshop 7 premiere 7 illustrator 10 = only $ 120\namazing special # 2 :\nwindows xp professional + microsoft office xp professional = only $ 80\namazing special # 3 :\nadobe photoshop cs + adobe illustrator cs + adobe indesign cs\namazing special # 4 :\n"
self.clean_subject_true = [
'more', 'oudat', 'software', 'upgrade'
] # données pour mocker "return_value" du "clean_text"
self.clean_body_true = [
'get', 'best', 'deal', 'skip', 'retail', 'box', 'sav', 'amaz',
'special', 'adobe', 'photoshop', 'premiere', 'illustrator', 'only',
'windows', 'xp', 'professional', 'microsoft', 'office', 'cs',
'indesign'
] # données pour mocker "return_value" du "clean_text"
self.subject_false = "re :"
self.body_false = "we are using it for other things . mary joyce and robert have discussed with mcmahon and bowen .\n- - - - - original message - - - - -\nfrom : kitchen louise\nsent : monday december 10 2001 8 : 26 am\nto : oxley david\nsubject :\nwhat happens to the money in wachovia ?\nlouise kitchen\nchief operating \n"
self.clean_subject_false = [
're'
] # données pour mocker "return_value" du "clean_text"
self.clean_body_false = [
'us', 'other', 'thing', 'mary', 'joyce', 'robert', 'discuss',
'mcmahon', 'bowen', 'original', 'message', 'kitchen', 'louise',
'sent', 'monday', "december", 'oxley', 'david', 'subject',
'happen', 'money', 'wachovia'
]
self.spam_ham_body_prob_true = (
1,
(1 / 6),
) # données pour mocker "return_value" du "spam_ham_body_prob"
self.subject_spam_ham_prob_true = (
(2 / 3),
(1 / 6),
) # données pour mocker "return_value" du "subject_spam_ham_prob"
self.spam_ham_body_prob_false = (
(1 / 4),
(2 / 6),
) # données pour mocker "return_value" du "spam_ham_body_prob"
self.subject_spam_ham_prob_false = (
0,
(1 / 2),
) # données pour mocker "return_value" du "subject_spam_ham_prob"
self.vocab = (
{
"p_sub_spam": {
"upgrade": 1 / 3,
"software": 1 / 3
},
"p_sub_ham": {
"re": 1 / 2,
"annoucement": 1 / 6,
"more": 1 / 6
},
"p_body_spam": {
"best": 1 / 4,
"deal": 1 / 4,
"skip": 1 / 4,
"special": 1 / 4,
"money": 1 / 4
},
"p_body_ham": {
"today": 1 / 6,
"professional": 1 / 6,
"meet": 1 / 6,
"discuss": 1 / 6,
"sent": 1 / 6
}
}
) # vocabulaire avec les valeurs de la probabilité pour mocker "return_value" du "load_dict"
# valeurs de la probabilité attendus : (0.5925*1/(256*pow(6,17))), (0.4075*1/pow(6,21))
self.spam_ham_body_prob_expected = (1.3673419333309543e-16,
1.8575963755415577e-17)
# valeurs de la probabilité attendus : (0.5925*1/81, 0.4075*1/6*1/4*1/4*1/4)
self.subject_spam_ham_prob_expected = (0.007314814814814815,
0.0010611979166666665)
class RENEGE:
"""Class pour realiser le filtrage du spam en utilisant vocabular.json file et
CRUD et EmalAnalyze classes"""
def __init__(self):
self.email_file = "800-mails.json"
self.crud = CRUD()
self.e_mail = EmailAnalyzer()
def calculate_user_trust(self, user_id):
#extracting json data
date_of_first_seen_message = self.crud.get_user_data(user_id, "Date_of_first_seen_message")
date_of_last_seen_message = self.crud.get_user_data(user_id, "Date_of_last_seen_message")
n_ham = self.crud.get_user_data(user_id, "HamN")
n_spam = self.crud.get_user_data(user_id, "SpamN")
groups = self.crud.get_user_data(user_id, "Groups")
#calculate the sum of trust values of all groups
sum_trust = 0
for group in groups :
group_id = self.crud.get_group_id(group)
sum_trust += self.crud.get_group_data(group_id, 'Trust')
#now that we have all the needed vars, calculate trust1, trust2 and trust
trust1 = (date_of_last_seen_message * n_ham) / (date_of_first_seen_message * (n_ham + n_spam))
trust2 = sum_trust / len(groups)
trust = (trust1 + trust2) / 2
if trust2 < 50:
trust = trust2
if trust1 > 100:
trust = 100
#before returning a value, check if trust is between 0 and 100
if trust < 0:
trust = 0
elif trust > 100:
trust = 100
return trust
def classify_emails(self, calculation_mode, is_log_est, is_log_combo):
'''
fonction deja implemente
Description: fonction pour commencer l'analyse des e-mails.
Sortie: bool, 'True' pour succes, 'False' dans le cas de failure.
'''
try:
self.process_email(self.get_email(), calculation_mode, is_log_est, is_log_combo)
return True
except Exception:
raise Exception
def process_email(self, new_emails, calculation_mode, is_log_est, is_log_combo):
'''
Description: fonction pour analyser chaque nouvel e-mail dans le
dictionare. Elle gere l'ajout des nouveux utilisateurs et/ou modification
de l'information existante sur les utilisateurs et groupes.
Sortie: bool, 'True' pour succes, 'False' dans le cas de failure.
'''
print(type(new_emails))
emails = new_emails["dataset"]
for email in emails:
email_adr = email['mail']['From']
date = email['mail']['Date']
spam = email['mail']['Spam'] == 'true'
subject = email['mail']['Subject']
body = email['mail']['Body']
user_id = self.crud.get_user_id(email_adr)
if user_id:
self.update_user_info(email_adr, date, spam)
else:
self.add_user_info(email_adr, date)
# trust de l'utilisateur
trust = self.crud.get_user_data(user_id, "Trust")
# moyenne du trust de tous les groupes
user_group = self.crud.get_user_data(user_id, "Groups")
sum_trust = 0
groups = self.crud.read_groups_file()
for group in groups :
if group['List_of_members'] == user_group:
sum_trust += group['Trust']
avg_group_trust = sum_trust / len(groups)
# difference de jours entre last_seen_msg et first_seen_msg
user_activity = self.substract_dates(self.crud.get_user_data(email_adr, "Date_of_last_seen_message"),
self.crud.get_user_data(email_adr, "Date_of_first_seen_message"))
if(calculation_mode == 1):
spam = self.e_mail.is_spam_function_one(spam, user_activity, trust, avg_group_trust)
elif(calculation_mode == 2):
spam = self.e_mail.is_spam_function_two(spam, trust, avg_group_trust)
elif(calculation_mode == 0):
spam = self.e_mail.is_spam(subject, body, is_log_est, is_log_combo, 0)
# mettre a jour l'utilisateur avec la nouvelle valeur de spam
self.update_user_info(email_adr, date, spam)
return True
def substract_dates(self, last_seen_msg, first_seen_msg):
# calcul de difference de jours entre le dernier et le premier message
last_seen_msg = str(last_seen_msg).split("-")
first_seen_msg = str(first_seen_msg).split("-")
user_activity = date(int(last_seen_msg[0]), int(last_seen_msg[1]), int(last_seen_msg[2]))
- date(int(first_seen_msg[0]), int(first_seen_msg[1]), int(first_seen_msg[2]))
# convertir la soustraction en int
user_activity = int((str(user_activity).split(" "))[0])
return user_activity
def test_is_spam_function_one_returns_false_acc_test_two(self):
return_val = EmailAnalyzer.is_spam_function_one(True, 35, 65, 80)
self.assertFalse(return_val)
def test_is_spam_function_one_returns_true_icc_test_three(self):
return_val = EmailAnalyzer.is_spam_function_one(False, 15, 65, 50)
self.assertTrue(return_val)
def test_is_spam_function_one_returns_false_acc_test_four(self):
return_val = EmailAnalyzer.is_spam_function_one(False, 35, 50, 50)
self.assertFalse(return_val)
def test_is_spam_function_one_returns_false_icc_test_four(self):
return_val = EmailAnalyzer.is_spam_function_one(True, 15, 65, 50)
self.assertTrue(return_val)
def test_is_spam_function_one_returns_true_acc_test_seven(self):
return_val = EmailAnalyzer.is_spam_function_one(False, 20, 70, 50)
self.assertTrue(return_val)
def test_is_spam_function_one_returns_false_icc_test_twelve(self):
return_val = EmailAnalyzer.is_spam_function_one(False, 20, 65, 60)
self.assertTrue(return_val)
def __init__(self):
self.email_file = "800-mails.json"
self.crud = CRUD()
self.e_mail = EmailAnalyzer()
def test_is_spam_function_one_returns_false_icc_test_sixteen(self):
return_val = EmailAnalyzer.is_spam_function_one(True, 20, 50, 50)
self.assertTrue(return_val)
