def maybe_ingest_app_info(now_offset=0):
now_timestamp = time.time() + now_offset
app_ids_needing_ingestion = []
# Atomically find next app ids to check.
redis = redis_wrap.client()
while True:
with redis.pipeline() as pipe:
try:
pipe.watch(APP_INFO_INGESTION_ZSET_KEY)
app_ids_needing_ingestion = pipe.zrangebyscore(APP_INFO_INGESTION_ZSET_KEY,
0, now_timestamp,
start=0, num=30)
if app_ids_needing_ingestion:
rescored_app_ids = []
for app_id in app_ids_needing_ingestion:
rescored_app_ids.append(next_ingestion_time())
rescored_app_ids.append(app_id)
pipe.multi()
pipe.zadd(APP_INFO_INGESTION_ZSET_KEY, *rescored_app_ids)
pipe.execute()
break
except WatchError:
logging.info('Interrupted while fetching app ids to ingest...')
if not app_ids_needing_ingestion:
return
logging.info('Ingesting app infos for %d app(s)...', len(app_ids_needing_ingestion))
for app_id in app_ids_needing_ingestion:
try:
app = AppStoreApp.objects.get(pk=app_id)
except AppStoreApp.DoesNotExist:
logging.warn('Deleted app id, removing from ingestion queue: %s', app_id)
redis.zrem(APP_INFO_INGESTION_ZSET_KEY, app_id)
continue
# NOTE: This should be unique but it isn't right now for whatever reason.
unique_countries = set(app.app_info_countries)
for country in unique_countries:
try:
fetch_info_and_maybe_update_app(app, country)
except appstore_fetch.RateLimitedError:
# This is not likely to fix itself soon. Give up.
logging.info('Rate limited by App Store, giving up for now...')
time.sleep(15)
return
except Exception:
logging.exception('Problem ingesting app info for app id: %s - %s (%s - %s)',
app.id, country, app.itunes_id, app.bundle_id)
app.app_info_ingestion_time = datetime.now()
app.save(update_fields=['app_info_ingestion_time'])
time.sleep(0.3333)
def finish_auth(user, token, verifier):
auth = get_auth_handler()
redis = redis_wrap.client()
redis_key = get_redis_key(token)
token_secret = redis.get(redis_key)
auth.request_token = {'oauth_token': token, 'oauth_token_secret': token_secret}
try:
auth.get_access_token(verifier)
except tweepy.TweepError as e:
logging.error('tweepy error during finish auth: %s', e)
return False, None
access_token = auth.access_token
access_token_secret = auth.access_token_secret
twitter_handle = auth.get_username()
TwitterAccessToken.objects.get_or_create(
handle=twitter_handle,
token=access_token,
token_secret=access_token_secret,
user=user
)
return True, twitter_handle
def finish_auth(user, token, verifier):
auth = get_auth_handler()
redis = redis_wrap.client()
redis_key = get_redis_key(token)
token_secret = redis.get(redis_key)
auth.request_token = {
'oauth_token': token,
'oauth_token_secret': token_secret
}
try:
auth.get_access_token(verifier)
except tweepy.TweepError as e:
logging.error('tweepy error during finish auth: %s', e)
return False, None
access_token = auth.access_token
access_token_secret = auth.access_token_secret
twitter_handle = auth.get_username()
TwitterAccessToken.objects.get_or_create(handle=twitter_handle,
token=access_token,
token_secret=access_token_secret,
user=user)
return True, twitter_handle
def update_labels_for_user(sdk_user, future_days_offset=0):
old_labels = set(sdk_user.labels or [])
new_labels = labels_for_user(sdk_user, future_days_offset=future_days_offset)
if old_labels == new_labels:
return False
change_events = []
increment_label_by = collections.defaultdict(lambda: 0)
removed_labels = old_labels - new_labels
for label in removed_labels:
change_event = SDKUserLabelChangeEvent(app_id=sdk_user.app_id, sdk_user_id=sdk_user.id, user_id=sdk_user.user_id,
label=label, kind='removed')
increment_label_by[label] -= 1
change_events.append(change_event)
added_labels = new_labels - old_labels
for label in added_labels:
change_event = SDKUserLabelChangeEvent(app_id=sdk_user.app_id, sdk_user_id=sdk_user.id, user_id=sdk_user.user_id,
label=label, kind='added')
increment_label_by[label] += 1
change_events.append(change_event)
# Create change events and update labels on user.
SDKUserLabelChangeEvent.objects.bulk_create(change_events)
labels = list(sorted(new_labels))
sdk_user.labels = labels
sdk_user.save(update_fields=['labels'])
# Now track the changes we've made in our counts.
for l1, l2 in TRACK_LABEL_COMBINATIONS:
had_label = l1 in old_labels and l2 in old_labels
has_label = l1 in new_labels and l2 in new_labels
if had_label != has_label:
if had_label:
# removed label
increment = -1
else:
# added label
increment = 1
combined_label = '%s-%s' % (l1, l2)
increment_label_by[combined_label] = increment
redis = redis_wrap.client()
redis_key = CURRENT_COUNT_REDIS_KEY_FORMAT % sdk_user.app_id
with redis.pipeline() as pipe:
for label, increment in increment_label_by.items():
pipe.hincrby(redis_key, label, increment)
pipe.execute()
# logging.info('Updated sdk user %s labels: %s', sdk_user.id, labels)
return True
def process_sdkuser_labels_periodically():
redis = redis_wrap.client()
max_id = redis.get(LAST_SDKUSER_LABELS_PROCESSED_KEY) or 0
with transaction.atomic():
sdk_users = list(
SDKUser.objects
.select_for_update()
.filter(id__gt=max_id)
.order_by('id')[:PROCESS_SDKUSERS_LABELS_LIMIT]
)
if sdk_users:
new_max_id = sdk_users[-1].id
updates = 0
for sdk_user in sdk_users:
if update_labels_for_user(sdk_user):
updates += 1
if updates:
logging.info('Updated %s sdk user labels periodically', updates)
else:
logging.info('No more users to update labels, starting over')
new_max_id = 0
redis.set(LAST_SDKUSER_LABELS_PROCESSED_KEY, new_max_id)
def add_all_apps_to_ingestion_queue():
apps_countries = list(AppStoreAppReviewTracker.objects.all().values_list('app_id', 'country'))
redis = redis_wrap.client()
for app_id, country in apps_countries:
app = AppStoreApp.objects.get(pk=app_id)
appstore_app_info.decorate_app(app, country)
target_time = next_ingestion_time(app, country)
redis.zadd(APP_REVIEWS_INGESTION_ZSET_KEY, target_time, '%s:%s' % (app_id, country))
def process_sdkuser_days_active():
redis = redis_wrap.client()
max_id = redis.get(LAST_SDKUSER_PROCESSED_KEY) or 0
new_max_id = process_sdkuser_days_active_from_id(max_id)
if new_max_id:
logging.info('New max id: %s', new_max_id)
redis.set(LAST_SDKUSER_PROCESSED_KEY, new_max_id)
def process_sdkuser_days_active():
redis = redis_wrap.client()
max_id = redis.get(LAST_SDKUSER_PROCESSED_KEY) or 0
new_max_id = process_sdkuser_days_active_from_id(max_id)
if new_max_id:
logging.info('New max id: %s', new_max_id)
redis.set(LAST_SDKUSER_PROCESSED_KEY, new_max_id)
def add_all_apps_to_ingestion_queue():
apps_countries = list(AppStoreAppReviewTracker.objects.all().values_list(
'app_id', 'country'))
redis = redis_wrap.client()
for app_id, country in apps_countries:
app = AppStoreApp.objects.get(pk=app_id)
appstore_app_info.decorate_app(app, country)
target_time = next_ingestion_time(app, country)
redis.zadd(APP_REVIEWS_INGESTION_ZSET_KEY, target_time,
'%s:%s' % (app_id, country))
def mark_app_needs_ingestion(app, country, force=False):
t = AppStoreAppReviewTracker.objects.get(app_id=app.id, country=country)
if not force and t.last_ingestion_time:
return
# If it needs ingestion, do it now.
target_time = time.time()
# ADD app.id TO ZSET AT target_time
redis = redis_wrap.client()
redis.zadd(APP_REVIEWS_INGESTION_ZSET_KEY, target_time, '%s:%s' % (app.id, country))
def get_auth_redirect(app_id=None, custom_cb=None):
auth = get_auth_handler(app_id, custom_cb)
try:
redirect_url = auth.get_authorization_url()
except tweepy.TweepError as e:
logging.error('tweepy error getting auth url: %s', e)
return None
redis = redis_wrap.client()
redis_key = get_redis_key(auth.request_token['oauth_token'])
redis.setex(redis_key, 3600, auth.request_token['oauth_token_secret'])
return redirect_url
def mark_app_needs_ingestion(app, country, force=False):
t = AppStoreAppReviewTracker.objects.get(app_id=app.id, country=country)
if not force and t.last_ingestion_time:
return
# If it needs ingestion, do it now.
target_time = time.time()
# ADD app.id TO ZSET AT target_time
redis = redis_wrap.client()
redis.zadd(APP_REVIEWS_INGESTION_ZSET_KEY, target_time,
'%s:%s' % (app.id, country))
def get_auth_redirect(app_id=None, custom_cb=None):
auth = get_auth_handler(app_id, custom_cb)
try:
redirect_url = auth.get_authorization_url()
except tweepy.TweepError as e:
logging.error('tweepy error getting auth url: %s', e)
return None
redis = redis_wrap.client()
redis_key = get_redis_key(auth.request_token['oauth_token'])
redis.setex(redis_key, 3600, auth.request_token['oauth_token_secret'])
return redirect_url
def apps_countries_to_ingest(limit, now_offset=0):
now_timestamp = time.time() + now_offset
apps_needing_ingestion = []
countries = []
# Atomically find next app ids to check.
redis = redis_wrap.client()
while True:
with redis.pipeline() as pipe:
try:
pipe.watch(APP_REVIEWS_INGESTION_ZSET_KEY)
app_ids_countries_needing_ingestion = pipe.zrangebyscore(
APP_REVIEWS_INGESTION_ZSET_KEY,
0,
now_timestamp,
start=0,
num=limit)
if app_ids_countries_needing_ingestion:
app_ids = []
countries = []
for ac in app_ids_countries_needing_ingestion:
app_id, country = ac.split(':')
app_id = long(app_id)
app_ids.append(app_id)
countries.append(country)
apps_by_id = dict(
(a.id, a)
for a in AppStoreApp.objects.filter(id__in=app_ids))
apps_needing_ingestion = [
apps_by_id[app_id] for app_id in app_ids
]
rescored_app_ids = []
for app, country in zip(apps_needing_ingestion, countries):
rescored_app_ids.append(
next_ingestion_time(app, country))
rescored_app_ids.append('%s:%s' % (app.id, country))
if rescored_app_ids:
pipe.multi()
pipe.zadd(APP_REVIEWS_INGESTION_ZSET_KEY,
*rescored_app_ids)
pipe.execute()
break
except WatchError:
logging.info('Interrupted while fetching app ids to ingest...')
for app, country in zip(apps_needing_ingestion, countries):
yield app, country
def mark_sdk_user_labels_dirty(sdk_user_ids):
if not sdk_user_ids:
return
t = time.time()
rescored_user_ids = []
user_ids_uniq = set(sdk_user_ids)
for user_id in user_ids_uniq:
rescored_user_ids.append(t)
rescored_user_ids.append(user_id)
redis = redis_wrap.client()
redis.zadd(DIRTY_SDK_USER_LABELS_REDIS_KEY, *rescored_user_ids)
def mark_sdk_user_labels_dirty(sdk_user_ids):
if not sdk_user_ids:
return
t = time.time()
rescored_user_ids = []
user_ids_uniq = set(sdk_user_ids)
for user_id in user_ids_uniq:
rescored_user_ids.append(t)
rescored_user_ids.append(user_id)
redis = redis_wrap.client()
redis.zadd(DIRTY_SDK_USER_LABELS_REDIS_KEY, *rescored_user_ids)
def label_counts_by_app_ids(app_ids):
redis = redis_wrap.client()
results = None
with redis.pipeline() as pipe:
for app_id in app_ids:
redis_key = CURRENT_COUNT_REDIS_KEY_FORMAT % app_id
pipe.hgetall(redis_key)
results = pipe.execute()
response = {}
for app_id, result in zip(app_ids, results):
response[app_id] = {k: int(v) for k, v in result.items()}
return response
def label_counts_by_app_ids(app_ids):
redis = redis_wrap.client()
results = None
with redis.pipeline() as pipe:
for app_id in app_ids:
redis_key = CURRENT_COUNT_REDIS_KEY_FORMAT % app_id
pipe.hgetall(redis_key)
results = pipe.execute()
response = {}
for app_id, result in zip(app_ids, results):
response[app_id] = {k: int(v) for k, v in result.items()}
return response
def apps_countries_to_ingest(limit, now_offset=0):
now_timestamp = time.time() + now_offset
apps_needing_ingestion = []
countries = []
# Atomically find next app ids to check.
redis = redis_wrap.client()
while True:
with redis.pipeline() as pipe:
try:
pipe.watch(APP_REVIEWS_INGESTION_ZSET_KEY)
app_ids_countries_needing_ingestion = pipe.zrangebyscore(APP_REVIEWS_INGESTION_ZSET_KEY,
0, now_timestamp,
start=0, num=limit)
if app_ids_countries_needing_ingestion:
app_ids = []
countries = []
for ac in app_ids_countries_needing_ingestion:
app_id, country = ac.split(':')
app_id = long(app_id)
app_ids.append(app_id)
countries.append(country)
apps_by_id = dict((a.id, a) for a in AppStoreApp.objects.filter(id__in=app_ids))
apps_needing_ingestion = [apps_by_id[app_id] for app_id in app_ids]
rescored_app_ids = []
for app, country in zip(apps_needing_ingestion, countries):
rescored_app_ids.append(next_ingestion_time(app, country))
rescored_app_ids.append('%s:%s' % (app.id, country))
if rescored_app_ids:
pipe.multi()
pipe.zadd(APP_REVIEWS_INGESTION_ZSET_KEY, *rescored_app_ids)
pipe.execute()
break
except WatchError:
logging.info('Interrupted while fetching app ids to ingest...')
for app, country in zip(apps_needing_ingestion, countries):
yield app, country
def process_dirty_sdk_user_labels():
now_timestamp = time.time()
sdk_user_ids_needing_attention = []
# Atomically find next app ids to check.
redis = redis_wrap.client()
while True:
with redis.pipeline() as pipe:
try:
pipe.watch(DIRTY_SDK_USER_LABELS_REDIS_KEY)
sdk_user_ids_needing_attention = pipe.zrangebyscore(
DIRTY_SDK_USER_LABELS_REDIS_KEY,
0,
now_timestamp,
start=0,
num=50)
if sdk_user_ids_needing_attention:
pipe.multi()
pipe.zrem(DIRTY_SDK_USER_LABELS_REDIS_KEY,
*sdk_user_ids_needing_attention)
pipe.execute()
break
except WatchError:
logging.info('Interrupted while fetching app ids to ingest...')
updates = 0
with transaction.atomic():
# Select these in a single transaction instead of individually to avoid lock contention.
sdk_users_needing_attention = list(
SDKUser.objects.filter(id__in=sdk_user_ids_needing_attention).
select_for_update().order_by('id'))
for sdk_user in sdk_users_needing_attention:
if update_labels_for_user(sdk_user):
updates += 1
if updates:
logging.info('Updated %s sdk user rows labels...', updates)
def process_dirty_sdk_user_labels():
now_timestamp = time.time()
sdk_user_ids_needing_attention = []
# Atomically find next app ids to check.
redis = redis_wrap.client()
while True:
with redis.pipeline() as pipe:
try:
pipe.watch(DIRTY_SDK_USER_LABELS_REDIS_KEY)
sdk_user_ids_needing_attention = pipe.zrangebyscore(DIRTY_SDK_USER_LABELS_REDIS_KEY,
0, now_timestamp,
start=0, num=50)
if sdk_user_ids_needing_attention:
pipe.multi()
pipe.zrem(DIRTY_SDK_USER_LABELS_REDIS_KEY, *sdk_user_ids_needing_attention)
pipe.execute()
break
except WatchError:
logging.info('Interrupted while fetching app ids to ingest...')
updates = 0
with transaction.atomic():
# Select these in a single transaction instead of individually to avoid lock contention.
sdk_users_needing_attention = list(
SDKUser.objects
.filter(id__in=sdk_user_ids_needing_attention)
.select_for_update()
.order_by('id')
)
for sdk_user in sdk_users_needing_attention:
if update_labels_for_user(sdk_user):
updates += 1
if updates:
logging.info('Updated %s sdk user rows labels...', updates)
def passes_health_check():
try:
users = list(User.objects.all()[:1])
except Exception as e:
# Note: This won't usually happen because
# session middleware will try accessing the DB cursor before we
# get here, which will cause an error farther up the stack.
return 'Database unreachable with error: %s' % e
if not users or not users[0].date_joined:
return 'Could not find a user, database must be FUBAR'
redis = redis_wrap.client()
if not redis:
return 'Redis unreachable, see log for more detail'
try:
redis.echo('Hello, world!')
except Exception as e:
return 'Redis unreachable with error: %s' % e
return None
def process_sdkuser_labels_periodically():
redis = redis_wrap.client()
max_id = redis.get(LAST_SDKUSER_LABELS_PROCESSED_KEY) or 0
with transaction.atomic():
sdk_users = list(SDKUser.objects.select_for_update().filter(
id__gt=max_id).order_by('id')[:PROCESS_SDKUSERS_LABELS_LIMIT])
if sdk_users:
new_max_id = sdk_users[-1].id
updates = 0
for sdk_user in sdk_users:
if update_labels_for_user(sdk_user):
updates += 1
if updates:
logging.info('Updated %s sdk user labels periodically',
updates)
else:
logging.info('No more users to update labels, starting over')
new_max_id = 0
redis.set(LAST_SDKUSER_LABELS_PROCESSED_KEY, new_max_id)
def mark_app_needs_info_ingestion(app): target_time = time.time() # ADD app.id TO ZSET AT target_time redis = redis_wrap.client() redis.zadd(APP_INFO_INGESTION_ZSET_KEY, target_time, app.id)
def maybe_ingest_app_info(now_offset=0):
now_timestamp = time.time() + now_offset
app_ids_needing_ingestion = []
# Atomically find next app ids to check.
redis = redis_wrap.client()
while True:
with redis.pipeline() as pipe:
try:
pipe.watch(APP_INFO_INGESTION_ZSET_KEY)
app_ids_needing_ingestion = pipe.zrangebyscore(
APP_INFO_INGESTION_ZSET_KEY,
0,
now_timestamp,
start=0,
num=30)
if app_ids_needing_ingestion:
rescored_app_ids = []
for app_id in app_ids_needing_ingestion:
rescored_app_ids.append(next_ingestion_time())
rescored_app_ids.append(app_id)
pipe.multi()
pipe.zadd(APP_INFO_INGESTION_ZSET_KEY, *rescored_app_ids)
pipe.execute()
break
except WatchError:
logging.info('Interrupted while fetching app ids to ingest...')
if not app_ids_needing_ingestion:
return
logging.info('Ingesting app infos for %d app(s)...',
len(app_ids_needing_ingestion))
for app_id in app_ids_needing_ingestion:
try:
app = AppStoreApp.objects.get(pk=app_id)
except AppStoreApp.DoesNotExist:
logging.warn('Deleted app id, removing from ingestion queue: %s',
app_id)
redis.zrem(APP_INFO_INGESTION_ZSET_KEY, app_id)
continue
# NOTE: This should be unique but it isn't right now for whatever reason.
unique_countries = set(app.app_info_countries)
for country in unique_countries:
try:
fetch_info_and_maybe_update_app(app, country)
except appstore_fetch.RateLimitedError:
# This is not likely to fix itself soon. Give up.
logging.info('Rate limited by App Store, giving up for now...')
time.sleep(15)
return
except Exception:
logging.exception(
'Problem ingesting app info for app id: %s - %s (%s - %s)',
app.id, country, app.itunes_id, app.bundle_id)
app.app_info_ingestion_time = datetime.now()
app.save(update_fields=['app_info_ingestion_time'])
time.sleep(0.3333)
def mark_app_needs_info_ingestion(app):
target_time = time.time()
# ADD app.id TO ZSET AT target_time
redis = redis_wrap.client()
redis.zadd(APP_INFO_INGESTION_ZSET_KEY, target_time, app.id)
def update_labels_for_user(sdk_user, future_days_offset=0):
old_labels = set(sdk_user.labels or [])
new_labels = labels_for_user(sdk_user,
future_days_offset=future_days_offset)
if old_labels == new_labels:
return False
change_events = []
increment_label_by = collections.defaultdict(lambda: 0)
removed_labels = old_labels - new_labels
for label in removed_labels:
change_event = SDKUserLabelChangeEvent(app_id=sdk_user.app_id,
sdk_user_id=sdk_user.id,
user_id=sdk_user.user_id,
label=label,
kind='removed')
increment_label_by[label] -= 1
change_events.append(change_event)
added_labels = new_labels - old_labels
for label in added_labels:
change_event = SDKUserLabelChangeEvent(app_id=sdk_user.app_id,
sdk_user_id=sdk_user.id,
user_id=sdk_user.user_id,
label=label,
kind='added')
increment_label_by[label] += 1
change_events.append(change_event)
# Create change events and update labels on user.
SDKUserLabelChangeEvent.objects.bulk_create(change_events)
labels = list(sorted(new_labels))
sdk_user.labels = labels
sdk_user.save(update_fields=['labels'])
# Now track the changes we've made in our counts.
for l1, l2 in TRACK_LABEL_COMBINATIONS:
had_label = l1 in old_labels and l2 in old_labels
has_label = l1 in new_labels and l2 in new_labels
if had_label != has_label:
if had_label:
# removed label
increment = -1
else:
# added label
increment = 1
combined_label = '%s-%s' % (l1, l2)
increment_label_by[combined_label] = increment
redis = redis_wrap.client()
redis_key = CURRENT_COUNT_REDIS_KEY_FORMAT % sdk_user.app_id
with redis.pipeline() as pipe:
for label, increment in increment_label_by.items():
pipe.hincrby(redis_key, label, increment)
pipe.execute()
# logging.info('Updated sdk user %s labels: %s', sdk_user.id, labels)
return True
