def commit_handler(self, msg):
"""
Handle a dist-git commit message and update Neo4j if necessary.
:param dict msg: a message to be processed
"""
repo = DistGitRepo.get_or_create({
'namespace':
msg['headers']['namespace'],
'name':
msg['headers']['repo']
})[0]
# Get the username from the email if the email is a Red Hat email
email = msg['headers']['email'].lower()
if email.endswith('@redhat.com'):
username = email.split('@redhat.com')[0]
else:
username = email
author = User.create_or_update({
'username': username,
'email': email
})[0]
commit_message = msg['body']['msg']['message']
commit = DistGitCommit.create_or_update({
'hash_':
msg['headers']['rev'],
'log_message':
commit_message,
'author_date':
timestamp_to_datetime(msg['body']['msg']['author_date']),
'commit_date':
timestamp_to_datetime(msg['body']['msg']['commit_date'])
})[0]
bug_rel_mapping = self.parse_bugzilla_bugs(commit_message)
for bug_id in bug_rel_mapping['resolves']:
bug = BugzillaBug.get_or_create({'id_': bug_id})[0]
commit.resolved_bugs.connect(bug)
for bug_id in bug_rel_mapping['related']:
bug = BugzillaBug.get_or_create({'id_': bug_id})[0]
commit.related_bugs.connect(bug)
for bug_id in bug_rel_mapping['reverted']:
bug = BugzillaBug.get_or_create({'id_': bug_id})[0]
commit.reverted_bugs.connect(bug)
commit.conditional_connect(commit.author, author)
repo.commits.connect(commit)
def _update_neo4j(neo4j_url, total_results, counter_and_results):
"""
Update Neo4j results via mapping with multiprocessing.
:param str neo4j_url: database url for Neo4j
:param int total_results: the total number of results that will be processed. This is used
for a logging statement about progress.
:param tuple counter_and_results: a tuple where the first index is the current counter and
the second index is a list of dictionaries representing results from Teiid
"""
try:
previous_total = counter_and_results[0]
results = counter_and_results[1]
# Since _update_neo4j will be run in a separate process, we must configure the database
# URL every time the method is run.
neomodel_config.DATABASE_URL = neo4j_url
# Create a thread pool with 4 threads to speed up queries to cgit
pool = ThreadPool(4)
counter = 0
for result in results:
if counter % 200 == 0:
until = counter + 200
if until > len(results):
until = len(results)
# Because of the joins in the SQL query, we end up with several rows with the
# same commit hash and we only want to query cgit once per commit
unique_commits = set([(c['module'], c['sha'])
for c in results[counter:until]])
log.debug(
'Getting the author email addresses from cgit in parallel '
'for results {0} to {1}'.format(counter, until))
repos_info = {
r['commit']: r
for r in pool.map(DistGitScraper._get_repo_info,
unique_commits)
}
# This is no longer needed so it can be cleared to save RAM
del unique_commits
counter += 1
log.info('Processing commit entry {0}/{1}'.format(
previous_total + counter, total_results))
repo_info = repos_info[result['sha']]
if not repo_info.get('namespace'):
log.info(
'Skipping nodes creation with commit ID {0}'.format(
result['commit_id']))
continue
log.debug(
'Creating nodes associated with commit ID {0}'.format(
result['commit_id']))
repo = DistGitRepo.get_or_create({
'namespace':
repo_info['namespace'],
'name':
result['module']
})[0]
commit = DistGitCommit.create_or_update({
'author_date':
result['author_date'],
'commit_date':
result['commit_date'],
'hash_':
result['sha'],
# In case we get unicode characters in Python 2
'log_message':
bytes(result['log_message'], 'utf-8').decode()
})[0]
bug = BugzillaBug.get_or_create({'id_':
result['bugzilla_id']})[0]
log.debug(
'Creating the user nodes associated with commit ID {0}'.
format(result['commit_id']))
author = User.create_or_update({
'username':
repo_info['author_username'],
'email':
repo_info['author_email']
})[0]
log.debug(
'Creating the relationships associated with commit ID {0}'.
format(result['commit_id']))
repo.commits.connect(commit)
commit.conditional_connect(commit.author, author)
if result['bugzilla_type'] == 'related':
commit.related_bugs.connect(bug)
elif result['bugzilla_type'] == 'resolves':
commit.resolved_bugs.connect(bug)
elif result['bugzilla_type'] == 'reverted':
commit.reverted_bugs.connect(bug)
# This is no longer needed so it can be cleared to save RAM
del repo_info
finally:
# Close the DB connection after this is done processing
db.driver.close()
def update_neo4j(self, results):
"""
Update Neo4j with the dist-git commit and push information from Teiid.
:param list results: a list of dictionaries
"""
pool = Pool(processes=8)
counter = 0
for result in results:
if counter % 200 == 0:
until = counter + 200
if until > len(results):
until = len(results)
# Because of the joins in the SQL query, we end up with several rows with the same
# commit hash and we only want to query cgit once per commit
unique_commits = set([(c['module'], c['sha'])
for c in results[counter:until]])
log.debug(
'Getting the author and committer email addresses from cgit in parallel '
'for results {0} to {1}'.format(counter, until))
repos_info = {}
for _r in pool.map(DistGitScraper._get_repo_info,
unique_commits):
r = json.loads(_r)
repos_info[r['commit']] = r
# This is no longer needed so it can be cleared to save RAM
del unique_commits
# A lot of RAM was allocated or used up, so let's call gc.collect() to ensure it
# is removed
gc.collect()
counter += 1
log.info('Processing commit and push entry {0}/{1}'.format(
str(counter), str(len(results))))
repo_info = repos_info[result['sha']]
if not repo_info.get('namespace'):
log.info(
'Skipping nodes creation with commit ID {0} and push ID {1}'
.format(result['commit_id'], result['push_id']))
continue
log.debug(
'Creating nodes associated with commit ID {0} and push ID {1}'.
format(result['commit_id'], result['push_id']))
repo = DistGitRepo.get_or_create({
'namespace':
repo_info['namespace'],
'name':
result['module']
})[0]
branch_name = result['ref'].rsplit('/', 1)[1]
branch = DistGitBranch.get_or_create({
'name':
branch_name,
'repo_namespace':
repo_info['namespace'],
'repo_name':
result['module']
})[0]
commit = DistGitCommit.create_or_update({
'author_date':
result['author_date'],
'commit_date':
result['commit_date'],
'hash_':
result['sha'],
# In case we get unicode characters in Python 2
'log_message':
bytes(result['log_message'], 'utf-8').decode()
})[0]
push = DistGitPush.get_or_create({
'id_': result['push_id'],
'push_date': result['push_date'],
'push_ip': result['push_ip']
})[0]
bug = BugzillaBug.get_or_create({'id_': result['bugzilla_id']})[0]
log.debug(
'Creating the user nodes associated with commit ID {0} and push ID {1}'
.format(result['commit_id'], result['push_id']))
author = User.create_or_update({
'username':
repo_info['author_username'],
'email':
repo_info['author_email']
})[0]
committer = User.create_or_update({
'username':
repo_info['committer_username'],
'email':
repo_info['committer_email']
})[0]
pusher = User.get_or_create({'username': result['pusher']})[0]
log.debug(
'Creating the relationships associated with commit ID {0} and push ID {1}'
.format(result['commit_id'], result['push_id']))
repo.contributors.connect(author)
repo.contributors.connect(committer)
repo.contributors.connect(pusher)
repo.commits.connect(commit)
repo.pushes.connect(push)
repo.branches.connect(branch)
branch.contributors.connect(author)
branch.contributors.connect(committer)
branch.contributors.connect(pusher)
branch.commits.connect(commit)
branch.pushes.connect(push)
push.conditional_connect(push.pusher, pusher)
push.commits.connect(commit)
commit.conditional_connect(commit.author, author)
commit.conditional_connect(commit.committer, committer)
if repo_info['parent']:
parent_commit = DistGitCommit.get_or_create(
{'hash_': repo_info['parent']})[0]
commit.conditional_connect(commit.parent, parent_commit)
if result['bugzilla_type'] == 'related':
commit.related_bugs.connect(bug)
elif result['bugzilla_type'] == 'resolves':
commit.resolved_bugs.connect(bug)
elif result['bugzilla_type'] == 'reverted':
commit.reverted_bugs.connect(bug)
# This is no longer needed so it can be cleared to save RAM
del repo_info