def update_pr(pr_number):
logging.debug("Updating pull request %i" % pr_number)
pr = Issue.get_or_create(pr_number)
issue_response = raw_github_request(get_pulls_base() + '/%i' % pr_number,
oauth_token=oauth_token, etag=pr.etag)
if issue_response is None:
logging.debug("PR %i hasn't changed since last visit; skipping" % pr_number)
return "Done updating pull request %i (nothing changed)" % pr_number
pr.pr_json = json.loads(issue_response.content)
pr.etag = issue_response.headers["ETag"]
pr.state = pr.pr_json['state']
pr.user = pr.pr_json['user']['login']
pr.updated_at = \
parse_datetime(pr.pr_json['updated_at']).astimezone(tz.tzutc()).replace(tzinfo=None)
for issue_number in pr.parsed_title['jiras']:
try:
link_issue_to_pr("%s-%s" % (app.config['JIRA_PROJECT'], issue_number), pr)
except:
logging.exception("Exception when linking to JIRA issue %s-%s" %
(app.config['JIRA_PROJECT'], issue_number))
try:
start_issue_progress("%s-%s" % (app.config['JIRA_PROJECT'], issue_number))
except:
logging.exception(
"Exception when starting progress on JIRA issue %s-%s" %
(app.config['JIRA_PROJECT'], issue_number))
pr.put() # Write our modifications back to the database
subtasks = [".update_pr_comments", ".update_pr_review_comments", ".update_pr_files"]
for task in subtasks:
taskqueue.add(url=url_for(task, pr_number=pr_number), queue_name='fresh-prs')
return "Done updating pull request %i" % pr_number
def update_pr_files(pr_number):
pr = Issue.get(pr_number)
files_response = paginated_github_request(get_pulls_base() + "/%i/files" % pr_number,
oauth_token=oauth_token, etag=pr.files_etag)
if files_response is None:
return "Files for PR %i are up-to-date" % pr_number
else:
pr.files_json, pr.files_etag = files_response
pr.put() # Write our modifications back to the database
return "Done updating files for PR %i" % pr_number
def update_pr_files(pr_number):
pr = Issue.get(pr_number)
files_response = paginated_github_request(get_pulls_base() + "/%i/files" % pr_number,
oauth_token=oauth_token, etag=pr.files_etag)
if files_response is None:
return "Files for PR %i are up-to-date" % pr_number
else:
pr.files_json, pr.files_etag = files_response
pr.put() # Write our modifications back to the database
return "Done updating files for PR %i" % pr_number
def update_pr_review_comments(pr_number):
pr = Issue.get(pr_number)
pr_comments_response = paginated_github_request(get_pulls_base() + '/%i/comments' % pr_number,
oauth_token=oauth_token)
# TODO: after fixing #32, re-enable etags here: etag=self.pr_review_comments_etag)
if pr_comments_response is None:
return "Review comments for PR %i are up-to-date" % pr_number
else:
pr.pr_comments_json, pr.pr_comments_etag = pr_comments_response
pr.cached_commenters = pr._compute_commenters()
pr.put() # Write our modifications back to the database
return "Done updating review comments for PR %i" % pr_number
def update_pr_review_comments(pr_number):
pr = Issue.get(pr_number)
pr_comments_response = paginated_github_request(get_pulls_base() + '/%i/comments' % pr_number,
oauth_token=oauth_token)
# TODO: after fixing #32, re-enable etags here: etag=self.pr_review_comments_etag)
if pr_comments_response is None:
return "Review comments for PR %i are up-to-date" % pr_number
else:
pr.pr_comments_json, pr.pr_comments_etag = pr_comments_response
pr.cached_commenters = pr._compute_commenters()
pr.put() # Write our modifications back to the database
return "Done updating review comments for PR %i" % pr_number
def update_pr(pr_number):
logging.debug("Updating pull request %i" % pr_number)
pr = Issue.get_or_create(pr_number)
try:
issue_response = raw_github_request(get_pulls_base() + '/%i' % pr_number,
oauth_token=oauth_token, etag=pr.etag)
except HTTPError as e:
if e.code == 404:
logging.debug("Pull request %i has been deleted" % pr_number)
pr.state = "deleted"
pr.put()
return "Done updating pull request %i (PR deleted)" % pr_number
else:
raise
if issue_response is None:
logging.debug("PR %i hasn't changed since last visit; skipping" % pr_number)
return "Done updating pull request %i (nothing changed)" % pr_number
pr.pr_json = json.loads(issue_response.content)
pr.etag = issue_response.headers["ETag"]
pr.state = pr.pr_json['state']
pr.user = pr.pr_json['user']['login']
pr.updated_at = \
parse_datetime(pr.pr_json['updated_at']).astimezone(tz.tzutc()).replace(tzinfo=None)
for issue_number in pr.parsed_title['jiras']:
try:
link_issue_to_pr("%s-%s" % (app.config['JIRA_PROJECT'], issue_number), pr)
except:
logging.exception("Exception when linking to JIRA issue %s-%s" %
(app.config['JIRA_PROJECT'], issue_number))
try:
start_issue_progress("%s-%s" % (app.config['JIRA_PROJECT'], issue_number))
except:
logging.exception(
"Exception when starting progress on JIRA issue %s-%s" %
(app.config['JIRA_PROJECT'], issue_number))
pr.put() # Write our modifications back to the database
subtasks = [".update_pr_comments", ".update_pr_review_comments", ".update_pr_files"]
for task in subtasks:
taskqueue.add(url=url_for(task, pr_number=pr_number), queue_name='fresh-prs')
return "Done updating pull request %i" % pr_number
def update_github_prs():
last_update_time = KVS.get("issues_since")
if last_update_time:
last_update_time = \
parse_datetime(last_update_time).astimezone(tz.tzutc()).replace(tzinfo=None)
else:
# If no update has ever run successfully, store "now" as the watermark. If this update
# task fails (because there are too many old PRs to load / backfill) then there's a chance
# that this initial timestamp won't be the true watermark. If we are trying to bulk-load
# old data then this should be done by calling /github/backfill-prs instead.
last_update_time = datetime.min
KVS.put('issues_since',
datetime.utcnow().strftime("%Y-%m-%dT%H:%M:%SZ"))
def fetch_and_process(url):
logging.debug("Following url %s" % url)
response = raw_github_request(url, oauth_token=oauth_token)
prs = json.loads(response.content)
now = datetime.utcnow()
should_continue_loading = True
update_time = last_update_time
for pr in prs:
updated_at = \
parse_datetime(pr['updated_at']).astimezone(tz.tzutc()).replace(tzinfo=None)
update_time = max(update_time, updated_at)
if updated_at < last_update_time:
should_continue_loading = False
break
is_fresh = (now - updated_at
).total_seconds() < app.config['FRESHNESS_THRESHOLD']
queue_name = ("fresh-prs" if is_fresh else "old-prs")
taskqueue.add(url=url_for(".update_pr", pr_number=pr['number']),
queue_name=queue_name)
if should_continue_loading:
link_header = parse_link_header(response.headers.get('Link', ''))
for link in link_header.links:
if link.rel == 'next':
fetch_and_process(link.href)
return update_time
update_time = \
fetch_and_process(get_pulls_base() + "?sort=updated&state=all&direction=desc&per_page=100")
KVS.put('issues_since', update_time.strftime("%Y-%m-%dT%H:%M:%SZ"))
return "Done fetching updated GitHub issues"
def update_github_prs():
last_update_time = KVS.get("issues_since")
if last_update_time:
last_update_time = \
parse_datetime(last_update_time).astimezone(tz.tzutc()).replace(tzinfo=None)
else:
# If no update has ever run successfully, store "now" as the watermark. If this update
# task fails (because there are too many old PRs to load / backfill) then there's a chance
# that this initial timestamp won't be the true watermark. If we are trying to bulk-load
# old data then this should be done by calling /github/backfill-prs instead.
last_update_time = datetime.min
KVS.put('issues_since', datetime.utcnow().strftime("%Y-%m-%dT%H:%M:%SZ"))
def fetch_and_process(url):
logging.debug("Following url %s" % url)
response = raw_github_request(url, oauth_token=oauth_token)
prs = json.loads(response.content)
now = datetime.utcnow()
should_continue_loading = True
update_time = last_update_time
for pr in prs:
updated_at = \
parse_datetime(pr['updated_at']).astimezone(tz.tzutc()).replace(tzinfo=None)
update_time = max(update_time, updated_at)
if updated_at < last_update_time:
should_continue_loading = False
break
is_fresh = (now - updated_at).total_seconds() < app.config['FRESHNESS_THRESHOLD']
queue_name = ("fresh-prs" if is_fresh else "old-prs")
taskqueue.add(url=url_for(".update_pr", pr_number=pr['number']), queue_name=queue_name)
if should_continue_loading:
link_header = parse_link_header(response.headers.get('Link', ''))
for link in link_header.links:
if link.rel == 'next':
fetch_and_process(link.href)
return update_time
update_time = \
fetch_and_process(get_pulls_base() + "?sort=updated&state=all&direction=desc&per_page=100")
KVS.put('issues_since', update_time.strftime("%Y-%m-%dT%H:%M:%SZ"))
return "Done fetching updated GitHub issues"
def backfill_prs():
"""
This method attempts to update every PR ever opened against the repository.
As a result, this method should only be invoked by admins when trying to bootstrap a new
deployment of the PR board.
"""
# Determine the number of PRs:
url = get_pulls_base() + "?sort=created&state=all&direction=desc"
response = raw_github_request(url, oauth_token=oauth_token)
latest_prs = json.loads(response.content)
latest_pr_number = int(latest_prs[0]['number'])
queue = taskqueue.Queue('old-prs')
update_tasks = []
for num in reversed(xrange(1, latest_pr_number + 1)):
update_tasks.append(taskqueue.Task(url=url_for(".update_pr", pr_number=num)))
# Can only enqueue up to 100 tasks per API call
async_call_results = []
for group_of_tasks in chunked(update_tasks, 100):
async_call_results.append(queue.add_async(group_of_tasks))
# Block until the async calls are finished:
for r in async_call_results:
r.get_result()
return "Enqueued tasks to backfill %i PRs" % latest_pr_number
def backfill_prs():
"""
This method attempts to update every PR ever opened against the repository.
As a result, this method should only be invoked by admins when trying to bootstrap a new
deployment of the PR board.
"""
# Determine the number of PRs:
url = get_pulls_base() + "?sort=created&state=all&direction=desc"
response = raw_github_request(url, oauth_token=oauth_token)
latest_prs = json.loads(response.content)
latest_pr_number = int(latest_prs[0]['number'])
queue = taskqueue.Queue('old-prs')
update_tasks = []
for num in reversed(xrange(1, latest_pr_number + 1)):
update_tasks.append(taskqueue.Task(url=url_for(".update_pr", pr_number=num)))
# Can only enqueue up to 100 tasks per API call
async_call_results = []
for group_of_tasks in chunked(update_tasks, 100):
async_call_results.append(queue.add_async(group_of_tasks))
# Block until the async calls are finished:
for r in async_call_results:
r.get_result()
return "Enqueued tasks to backfill %i PRs" % latest_pr_number
