def fetch_bugs(self):
logger.info('Fetching Bugzilla bugs ...')
lower, upper = self._get_bounds()
if lower > upper:
logger.info('Found cache that says bugs already fetched. Skipping.')
return
total = upper - lower
x = {'failed': 0, 'completed': 0}
def write(bug):
if bug == 'failed':
x['failed'] += 1
elif bug:
self._on_bug(bug)
x['completed'] += 1
print_progressbar(x['completed'], total - x['failed'])
print_progressbar(0, total)
self._run(lower, upper, write)
logger.info('\n--> completed with ' + str(x['failed']) + ' failed bug fetches: ')
def write(bug):
if bug == 'failed':
x['failed'] += 1
elif bug:
self._on_bug(bug)
x['completed'] += 1
print_progressbar(x['completed'], total - x['failed'])
def evaluate(self, prediction_func, test_indices, prefix=''):
top_5_acc = [0.0, 0.0, 0.0, 0.0, 0.0]
ft_change = [0.0, 0.0, 0.0, 0.0, 0.0]
total_ft = 0
predictions_x = []
predictions_y = []
total = len(test_indices)
print('evaluating ' + prefix + ' ...')
for count, index in enumerate(test_indices):
print_progressbar(count, total)
test_x = self._c['data'][index]
test_y = self._c['target'][index]
prediction = prediction_func(test_x)
if test_y in prediction:
hit = prediction.index(test_y)
predictions_x.append(prediction)
predictions_y.append(hit)
for i in range(hit, len(top_5_acc)):
top_5_acc[i] += 1.0
fix_time = self._ft_estimator.actual_ft(test_x['bug_id'])
total_ft += fix_time
lowest_ft = None
for _index in range(len(ft_change)):
estimation = self._ft_estimator.predict(
prediction[_index], test_x['vector'])
change = fix_time
if _index == 0:
lowest_ft = change
if estimation and prediction[_index] != test_y:
change += fix_time - estimation
if change < lowest_ft:
lowest_ft = change
ft_change[_index] += lowest_ft
top_5_acc_perc = [
str(round(Decimal(acc / total), 2)) for acc in top_5_acc
]
fix_time_perc = [
str(round(Decimal(1 - ft / total_ft), 2)) for ft in ft_change
]
print('\n' + prefix + ' accuracy: ' + ', '.join(top_5_acc_perc))
print(prefix + ' fix time reduction [%]: ' + ', '.join(fix_time_perc))
return predictions_x, predictions_y
def update(self):
bugs = self._storage.get_existing_full()
total_length = len(bugs)
for index, bug in enumerate(bugs):
print_progressbar(index, total_length)
if bug['assignee_email'][0] == '(' and bug['assignee_email'][
-1] == ')':
email = bug['assignee_email'].split(',')[0][1:]
self._storage.update_github(bug['id'], email)
def fetch_all(self):
repo = self._g.get_repo(self._owner + '/' + self._repo)
on_bug = self._get_on_bug_handler()
count = 0
start_time = time.time()
labels = [repo.get_label(label) for label in self._bug_labels]
# get issues starting with the oldest to make the date checkpoints work
issues = repo.get_issues(labels=labels,
sort="updated",
state="all",
direction="asc",
since=self._get_checkpoint())
length = estimate_total_count(issues)
if length:
logger.info('Fetching about ' + str(length) +
' GitHub issues from ' + self._remote_url)
else:
logger.info('Fetching GitHub issues from ' + self._remote_url)
for issue in issues:
if not issue.pull_request:
count += 1
print_progressbar(count, length)
bug = GithubBug(issue)
on_bug(bug)
logger.info('Importing {} issues took {:.1f} seconds'.format(
count,
time.time() - start_time))
def write(self, bug):
self._processed += 1
print_progressbar(self._processed, self._total)
self._callback(bug)