def TestSpamFilter(driver):
bayes = driver.manager.classifier_data.bayes
nspam = bayes.nspam
nham = bayes.nham
original_bayes = copy.copy(driver.manager.classifier_data.bayes)
# for each watch folder, create a spam message, and do the training thang
for msf_watch, folder_watch in driver.GetWatchFolderGenerator():
print "Performing Spam test on watch folder '%s'..." % msf_watch.GetFQName()
# Create a spam message in the Inbox - it should get immediately filtered
msg, words = driver.CreateTestMessageInFolder(SPAM, folder_watch)
# sleep to ensure filtering.
WaitForFilters()
# It should no longer be in the Inbox.
driver.CheckMessageFilteredFrom(folder_watch)
# It should be in the "sure is spam" folder.
spam_msg = driver.FindTestMessage(driver.folder_spam)
if spam_msg is None:
TestFailed("The test message vanished from the Inbox, but didn't appear in Spam")
# Check that none of the above caused training.
if nspam != bayes.nspam:
TestFailed("Something caused a new spam message to appear")
if nham != bayes.nham:
TestFailed("Something caused a new ham message to appear")
check_words(words, bayes, 0, 0)
# Now move the message back to the inbox - it should get trained.
store_msg = driver.manager.message_store.GetMessage(spam_msg)
import train
if train.been_trained_as_ham(store_msg, driver.manager.classifier_data):
TestFailed("This new spam message should not have been trained as ham yet")
if train.been_trained_as_spam(store_msg, driver.manager.classifier_data):
TestFailed("This new spam message should not have been trained as spam yet")
spam_msg.Move(folder_watch)
WaitForFilters()
spam_msg = driver.FindTestMessage(folder_watch)
if spam_msg is None:
TestFailed("The message appears to have been filtered out of the watch folder")
store_msg = driver.manager.message_store.GetMessage(spam_msg)
need_untrain = True
try:
if nspam != bayes.nspam:
TestFailed("There were not the same number of spam messages after a re-train")
if nham+1 != bayes.nham:
TestFailed("There was not one more ham messages after a re-train")
if train.been_trained_as_spam(store_msg, driver.manager.classifier_data):
TestFailed("This new spam message should not have been trained as spam yet")
if not train.been_trained_as_ham(store_msg, driver.manager.classifier_data):
TestFailed("This new spam message should have been trained as ham now")
# word infos should have one extra ham
check_words(words, bayes, 0, 1)
# Now move it back to the Spam folder.
# This should see the message un-trained as ham, and re-trained as Spam
spam_msg.Move(driver.folder_spam)
WaitForFilters()
spam_msg = driver.FindTestMessage(driver.folder_spam)
if spam_msg is None:
TestFailed("Could not find the message in the Spam folder")
store_msg = driver.manager.message_store.GetMessage(spam_msg)
if nspam +1 != bayes.nspam:
TestFailed("There should be one more spam now")
if nham != bayes.nham:
TestFailed("There should be the same number of hams again")
if not train.been_trained_as_spam(store_msg, driver.manager.classifier_data):
TestFailed("This new spam message should have been trained as spam by now")
if train.been_trained_as_ham(store_msg, driver.manager.classifier_data):
TestFailed("This new spam message should have been un-trained as ham")
# word infos should have one extra spam, no extra ham
check_words(words, bayes, 1, 0)
# Move the message to another folder, and make sure we still
# identify it correctly as having been trained.
# Move to the "unsure" folder, just cos we know about it, and
# we know that no special watching of this folder exists.
spam_msg.Move(driver.folder_unsure)
spam_msg = driver.FindTestMessage(driver.folder_unsure)
if spam_msg is None:
TestFailed("Could not find the message in the Unsure folder")
store_msg = driver.manager.message_store.GetMessage(spam_msg)
if not train.been_trained_as_spam(store_msg, driver.manager.classifier_data):
TestFailed("Message was not identified as Spam after moving")
# word infos still be 'spam'
check_words(words, bayes, 1, 0)
# Now undo the damage we did.
was_spam = train.untrain_message(store_msg, driver.manager.classifier_data)
if not was_spam:
TestFailed("Untraining this message did not indicate it was spam")
if train.been_trained_as_spam(store_msg, driver.manager.classifier_data) or \
train.been_trained_as_ham(store_msg, driver.manager.classifier_data):
TestFailed("Untraining this message kept it has ham/spam")
need_untrain = False
finally:
if need_untrain:
train.untrain_message(store_msg, driver.manager.classifier_data)
# Check all the counts are back where we started.
if nspam != bayes.nspam:
TestFailed("Spam count didn't get back to the same")
if nham != bayes.nham:
TestFailed("Ham count didn't get back to the same")
check_words(words, bayes, 0, 0)
if bayes.wordinfo != original_bayes.wordinfo:
TestFailed("The bayes object's 'wordinfo' did not compare the same at the end of all this!")
if bayes.probcache != original_bayes.probcache:
TestFailed("The bayes object's 'probcache' did not compare the same at the end of all this!")
spam_msg.Delete()
print "Created a Spam message, and saw it get filtered and trained."
def TestSpamFilter(driver):
bayes = driver.manager.classifier_data.bayes
nspam = bayes.nspam
nham = bayes.nham
original_bayes = copy.copy(driver.manager.classifier_data.bayes)
for msf_watch, folder_watch in driver.GetWatchFolderGenerator():
print("Performing Spam test on watch folder '%s'..." % msf_watch.GetFQName())
msg, words = driver.CreateTestMessageInFolder(SPAM, folder_watch)
WaitForFilters()
driver.CheckMessageFilteredFrom(folder_watch)
spam_msg = driver.FindTestMessage(driver.folder_spam)
if spam_msg is None:
TestFailed("The test message vanished from the Inbox, but didn't appear in Spam")
if nspam != bayes.nspam:
TestFailed("Something caused a new spam message to appear")
if nham != bayes.nham:
TestFailed("Something caused a new ham message to appear")
check_words(words, bayes, 0, 0)
store_msg = driver.manager.message_store.GetMessage(spam_msg)
driver.manager.classifier_data.message_db.load_msg(store_msg)
import train
if train.been_trained_as_ham(store_msg):
TestFailed("This new spam message should not have been trained as ham yet")
if train.been_trained_as_spam(store_msg):
TestFailed("This new spam message should not have been trained as spam yet")
spam_msg.Move(folder_watch)
WaitForFilters()
spam_msg = driver.FindTestMessage(folder_watch)
if spam_msg is None:
TestFailed("The message appears to have been filtered out of the watch folder")
store_msg = driver.manager.message_store.GetMessage(spam_msg)
driver.manager.classifier_data.message_db.load_msg(store_msg)
need_untrain = True
try:
if nspam != bayes.nspam:
TestFailed("There were not the same number of spam messages after a re-train")
if nham+1 != bayes.nham:
TestFailed("There was not one more ham messages after a re-train")
if train.been_trained_as_spam(store_msg):
TestFailed("This new spam message should not have been trained as spam yet")
if not train.been_trained_as_ham(store_msg):
TestFailed("This new spam message should have been trained as ham now")
check_words(words, bayes, 0, 1)
spam_msg.Move(driver.folder_spam)
WaitForFilters()
spam_msg = driver.FindTestMessage(driver.folder_spam)
if spam_msg is None:
TestFailed("Could not find the message in the Spam folder")
store_msg = driver.manager.message_store.GetMessage(spam_msg)
driver.manager.classifier_data.message_db.load_msg(store_msg)
if nspam +1 != bayes.nspam:
TestFailed("There should be one more spam now")
if nham != bayes.nham:
TestFailed("There should be the same number of hams again")
if not train.been_trained_as_spam(store_msg):
TestFailed("This new spam message should have been trained as spam by now")
if train.been_trained_as_ham(store_msg):
TestFailed("This new spam message should have been un-trained as ham")
check_words(words, bayes, 1, 0)
spam_msg.Move(driver.folder_unsure)
spam_msg = driver.FindTestMessage(driver.folder_unsure)
if spam_msg is None:
TestFailed("Could not find the message in the Unsure folder")
store_msg = driver.manager.message_store.GetMessage(spam_msg)
driver.manager.classifier_data.message_db.load_msg(store_msg)
if not train.been_trained_as_spam(store_msg):
TestFailed("Message was not identified as Spam after moving")
check_words(words, bayes, 1, 0)
was_spam = train.untrain_message(store_msg, driver.manager.classifier_data)
driver.manager.classifier_data.message_db.load_msg(store_msg)
if not was_spam:
TestFailed("Untraining this message did not indicate it was spam")
if train.been_trained_as_spam(store_msg) or \
train.been_trained_as_ham(store_msg):
TestFailed("Untraining this message kept it has ham/spam")
need_untrain = False
finally:
if need_untrain:
train.untrain_message(store_msg, driver.manager.classifier_data)
if nspam != bayes.nspam:
TestFailed("Spam count didn't get back to the same")
if nham != bayes.nham:
TestFailed("Ham count didn't get back to the same")
check_words(words, bayes, 0, 0)
if bayes.wordinfo != original_bayes.wordinfo:
TestFailed("The bayes object's 'wordinfo' did not compare the same at the end of all this!")
if bayes.probcache != original_bayes.probcache:
TestFailed("The bayes object's 'probcache' did not compare the same at the end of all this!")
spam_msg.Delete()
print("Created a Spam message, and saw it get filtered and trained.")