def remove_expired_data(provider):
"""Remove expired data for provider"""
print("Removing expired content for provider: %s" % provider.get("_id", "Detached items"))
minutes_to_keep_content = provider.get("content_expiry", superdesk.app.config["INGEST_EXPIRY_MINUTES"])
expiration_date = utcnow() - timedelta(minutes=minutes_to_keep_content)
ingest_service = superdesk.get_resource_service("ingest")
items = get_expired_items(provider, expiration_date)
ids = [item["_id"] for item in items]
items.rewind()
file_ids = [
rend.get("media")
for item in items
for rend in item.get("renditions", {}).values()
if not item.get("archived") and rend.get("media")
]
if ids:
print("Removing items %s" % ids)
ingest_service.delete({"_id": {"$in": ids}})
for file_id in file_ids:
print("Deleting file: ", file_id)
superdesk.app.media.delete(file_id)
stats.incr("ingest.expired_items", len(ids))
print(
"Removed expired content for provider: {0} count: {1}".format(provider.get("_id", "Detached items"), len(ids))
)
def remove_expired_data(provider):
"""Remove expired data for provider"""
print('Removing expired content for provider: %s' % provider.get('_id', 'Detached items'))
minutes_to_keep_content = provider.get('content_expiry', INGEST_EXPIRY_MINUTES)
expiration_date = utcnow() - timedelta(minutes=minutes_to_keep_content)
ingest_service = superdesk.get_resource_service('ingest')
items = get_expired_items(provider, expiration_date)
ids = [item['_id'] for item in items]
file_ids = [rend.get('media')
for item in items
for rend in item.get('renditions', {}).values()
if not item.get('archived') and rend.get('media')]
if ids:
print('Removing items %s' % ids)
ingest_service.delete({'_id': {'$in': ids}})
for file_id in file_ids:
print('Deleting file: ', file_id)
superdesk.app.media.delete(file_id)
stats.incr('ingest.expired_items', len(ids))
print('Removed expired content for provider: {0} count: {1}'
.format(provider.get('_id', 'Detached items'), len(ids)))
def remove_expired_data(provider):
"""Remove expired data for provider"""
logger.info('Removing expired content for provider: %s' %
provider.get('_id', 'Detached items'))
ingest_service = superdesk.get_resource_service('ingest')
items = get_expired_items(provider)
ids = [item['_id'] for item in items]
items.rewind()
file_ids = [
rend.get('media') for item in items
for rend in item.get('renditions', {}).values()
if not item.get('archived') and rend.get('media')
]
if ids:
logger.info('Removing items %s' % ids)
ingest_service.delete({'_id': {'$in': ids}})
push_expired_notification(ids)
for file_id in file_ids:
logger.info('Deleting file: %s' % file_id)
superdesk.app.media.delete(file_id)
stats.incr('ingest.expired_items', len(ids))
logger.info('Removed expired content for provider: {0} count: {1}'.format(
provider.get('_id', 'Detached items'), len(ids)))
remove_expired_from_elastic()
def remove_expired_data(provider):
"""Remove expired data for provider"""
print('Removing expired content for provider: %s' %
provider.get('_id', 'Detached items'))
minutes_to_keep_content = provider.get(
'content_expiry', superdesk.app.config['INGEST_EXPIRY_MINUTES'])
expiration_date = utcnow() - timedelta(minutes=minutes_to_keep_content)
ingest_service = superdesk.get_resource_service('ingest')
items = get_expired_items(provider, expiration_date)
ids = [item['_id'] for item in items]
items.rewind()
file_ids = [
rend.get('media') for item in items
for rend in item.get('renditions', {}).values()
if not item.get('archived') and rend.get('media')
]
if ids:
print('Removing items %s' % ids)
ingest_service.delete({'_id': {'$in': ids}})
for file_id in file_ids:
print('Deleting file: ', file_id)
superdesk.app.media.delete(file_id)
stats.incr('ingest.expired_items', len(ids))
print('Removed expired content for provider: {0} count: {1}'.format(
provider.get('_id', 'Detached items'), len(ids)))
def remove_expired_data(provider):
"""Remove expired data for provider"""
logger.info('Removing expired content for provider: %s' % provider.get('_id', 'Detached items'))
ingest_service = superdesk.get_resource_service('ingest')
items = get_expired_items(provider)
ids = [item['_id'] for item in items]
items.rewind()
file_ids = [rend.get('media')
for item in items
for rend in item.get('renditions', {}).values()
if not item.get('archived') and rend.get('media')]
if ids:
logger.info('Removing items %s' % ids)
ingest_service.delete({'_id': {'$in': ids}})
push_expired_notification(ids)
for file_id in file_ids:
logger.info('Deleting file: %s' % file_id)
superdesk.app.media.delete(file_id)
stats.incr('ingest.expired_items', len(ids))
logger.info('Removed expired content for provider: {0} count: {1}'
.format(provider.get('_id', 'Detached items'), len(ids)))
remove_expired_from_elastic()
def update_provider(provider, rule_set=None, routing_scheme=None):
"""
Fetches items from ingest provider as per the configuration, ingests them into Superdesk and
updates the provider.
"""
if ingest_for_provider_is_already_running(provider):
return
try:
update = {
LAST_UPDATED: utcnow()
}
for items in providers[provider.get('type')].update(provider):
ingest_items(items, provider, rule_set, routing_scheme)
stats.incr('ingest.ingested_items', len(items))
if items:
update[LAST_ITEM_UPDATE] = utcnow()
ingest_service = superdesk.get_resource_service('ingest_providers')
ingest_service.system_update(provider[superdesk.config.ID_FIELD], update, provider)
if LAST_ITEM_UPDATE not in update and get_is_idle(provider):
notify_and_add_activity(
ACTIVITY_EVENT,
'Provider {{name}} has gone strangely quiet. Last activity was on {{last}}',
resource='ingest_providers',
user_list=ingest_service._get_administrators(),
name=provider.get('name'),
last=provider[LAST_ITEM_UPDATE].replace(tzinfo=timezone.utc).astimezone(tz=None).strftime("%c"))
logger.info('Provider {0} updated'.format(provider[superdesk.config.ID_FIELD]))
push_notification('ingest:update', provider_id=str(provider[superdesk.config.ID_FIELD]))
finally:
mark_provider_as_not_running(provider)
def update_provider(provider, rule_set=None, routing_scheme=None):
"""Fetch items from ingest provider, ingest them into Superdesk and update the provider.
:param provider: Ingest Provider data
:param rule_set: Translation Rule Set if one is associated with Ingest Provider.
:param routing_scheme: Routing Scheme if one is associated with Ingest Provider.
"""
lock_name = get_lock_id('ingest', provider['name'],
provider[superdesk.config.ID_FIELD])
if not lock(lock_name, expire=1810):
return
try:
feeding_service = registered_feeding_services[
provider['feeding_service']]
feeding_service = feeding_service.__class__()
update = {LAST_UPDATED: utcnow()}
for items in feeding_service.update(provider, update):
ingest_items(items, provider, feeding_service, rule_set,
routing_scheme)
stats.incr('ingest.ingested_items', len(items))
if items:
update[LAST_ITEM_UPDATE] = utcnow()
# Some Feeding Services update the collection and by this time the _etag might have been changed.
# So it's necessary to fetch it once again. Otherwise, OriginalChangedError is raised.
ingest_provider_service = superdesk.get_resource_service(
'ingest_providers')
provider = ingest_provider_service.find_one(
req=None, _id=provider[superdesk.config.ID_FIELD])
ingest_provider_service.system_update(
provider[superdesk.config.ID_FIELD], update, provider)
if LAST_ITEM_UPDATE not in update and get_is_idle(provider):
admins = superdesk.get_resource_service(
'users').get_users_by_user_type('administrator')
notify_and_add_activity(
ACTIVITY_EVENT,
'Provider {{name}} has gone strangely quiet. Last activity was on {{last}}',
resource='ingest_providers',
user_list=admins,
name=provider.get('name'),
last=provider[LAST_ITEM_UPDATE].replace(
tzinfo=timezone.utc).astimezone(tz=None).strftime("%c"))
logger.info('Provider {0} updated'.format(
provider[superdesk.config.ID_FIELD]))
if LAST_ITEM_UPDATE in update: # Only push a notification if there has been an update
push_notification('ingest:update',
provider_id=str(
provider[superdesk.config.ID_FIELD]))
finally:
unlock(lock_name)
def update_provider(self, provider, rule_set=None, routing_scheme=None):
"""
Fetches items from ingest provider as per the configuration, ingests them into Superdesk and
updates the provider.
:param self:
:type self:
:param provider: Ingest Provider Details
:type provider: dict :py:class:`superdesk.io.ingest_provider_model.IngestProviderResource`
:param rule_set: Translation Rule Set if one is associated with Ingest Provider.
:type rule_set: dict :py:class:`apps.rules.rule_sets.RuleSetsResource`
:param routing_scheme: Routing Scheme if one is associated with Ingest Provider.
:type routing_scheme: dict :py:class:`apps.rules.routing_rules.RoutingRuleSchemeResource`
"""
lock_name = get_lock_id('ingest', provider['name'], provider[superdesk.config.ID_FIELD])
host_name = get_host_id(self)
if not lock(lock_name, host_name, expire=1800):
return
try:
feeding_service = registered_feeding_services[provider['feeding_service']]
feeding_service = feeding_service.__class__()
update = {LAST_UPDATED: utcnow()}
for items in feeding_service.update(provider):
ingest_items(items, provider, feeding_service, rule_set, routing_scheme)
stats.incr('ingest.ingested_items', len(items))
if items:
update[LAST_ITEM_UPDATE] = utcnow()
# Some Feeding Services update the collection and by this time the _etag might have been changed.
# So it's necessary to fetch it once again. Otherwise, OriginalChangedError is raised.
ingest_provider_service = superdesk.get_resource_service('ingest_providers')
provider = ingest_provider_service.find_one(req=None, _id=provider[superdesk.config.ID_FIELD])
ingest_provider_service.system_update(provider[superdesk.config.ID_FIELD], update, provider)
if LAST_ITEM_UPDATE not in update and get_is_idle(provider):
admins = superdesk.get_resource_service('users').get_users_by_user_type('administrator')
notify_and_add_activity(
ACTIVITY_EVENT,
'Provider {{name}} has gone strangely quiet. Last activity was on {{last}}',
resource='ingest_providers', user_list=admins, name=provider.get('name'),
last=provider[LAST_ITEM_UPDATE].replace(tzinfo=timezone.utc).astimezone(tz=None).strftime("%c"))
logger.info('Provider {0} updated'.format(provider[superdesk.config.ID_FIELD]))
if LAST_ITEM_UPDATE in update: # Only push a notification if there has been an update
push_notification('ingest:update', provider_id=str(provider[superdesk.config.ID_FIELD]))
finally:
unlock(lock_name, host_name)
def update_provider(self, provider, rule_set=None, routing_scheme=None):
"""
Fetches items from ingest provider as per the configuration, ingests them into Superdesk and
updates the provider.
"""
if provider.get('type') == 'search':
return
if not is_updatable(provider):
return
lock_name = get_lock_id('ingest', provider['name'],
provider[superdesk.config.ID_FIELD])
host_name = get_host_id(self)
if not lock(lock_name, host_name, expire=1800):
return
try:
update = {LAST_UPDATED: utcnow()}
for items in providers[provider.get('type')].update(provider):
ingest_items(items, provider, rule_set, routing_scheme)
stats.incr('ingest.ingested_items', len(items))
if items:
update[LAST_ITEM_UPDATE] = utcnow()
ingest_service = superdesk.get_resource_service('ingest_providers')
ingest_service.system_update(provider[superdesk.config.ID_FIELD],
update, provider)
if LAST_ITEM_UPDATE not in update and get_is_idle(provider):
notify_and_add_activity(
ACTIVITY_EVENT,
'Provider {{name}} has gone strangely quiet. Last activity was on {{last}}',
resource='ingest_providers',
user_list=ingest_service._get_administrators(),
name=provider.get('name'),
last=provider[LAST_ITEM_UPDATE].replace(
tzinfo=timezone.utc).astimezone(tz=None).strftime("%c"))
logger.info('Provider {0} updated'.format(
provider[superdesk.config.ID_FIELD]))
# Only push a notification if there has been an update
if LAST_ITEM_UPDATE in update:
push_notification('ingest:update',
provider_id=str(
provider[superdesk.config.ID_FIELD]))
finally:
unlock(lock_name, host_name)
def update_provider(provider, rule_set=None):
"""
Fetches items from ingest provider as per the configuration, ingests them into Superdesk and
updates the provider.
"""
superdesk.get_resource_service('ingest_providers').update(provider['_id'], {
LAST_UPDATED: utcnow(),
# Providing the _etag as system updates to the documents shouldn't override _etag.
app.config['ETAG']: provider.get(app.config['ETAG'])
})
for items in providers[provider.get('type')].update(provider):
ingest_items(items, provider, rule_set)
stats.incr('ingest.ingested_items', len(items))
logger.info('Provider {0} updated'.format(provider['_id']))
push_notification('ingest:update')
def update_provider(self, provider, rule_set=None, routing_scheme=None):
"""
Fetches items from ingest provider as per the configuration, ingests them into Superdesk and
updates the provider.
"""
if provider.get('type') == 'search':
return
if not is_updatable(provider):
return
lock_name = get_lock_id('ingest', provider['name'], provider[superdesk.config.ID_FIELD])
host_name = get_host_id(self)
if not lock(lock_name, host_name, expire=1800):
return
try:
update = {
LAST_UPDATED: utcnow()
}
for items in providers[provider.get('type')].update(provider):
ingest_items(items, provider, rule_set, routing_scheme)
stats.incr('ingest.ingested_items', len(items))
if items:
update[LAST_ITEM_UPDATE] = utcnow()
ingest_service = superdesk.get_resource_service('ingest_providers')
ingest_service.system_update(provider[superdesk.config.ID_FIELD], update, provider)
if LAST_ITEM_UPDATE not in update and get_is_idle(provider):
notify_and_add_activity(
ACTIVITY_EVENT,
'Provider {{name}} has gone strangely quiet. Last activity was on {{last}}',
resource='ingest_providers',
user_list=ingest_service._get_administrators(),
name=provider.get('name'),
last=provider[LAST_ITEM_UPDATE].replace(tzinfo=timezone.utc).astimezone(tz=None).strftime("%c"))
logger.info('Provider {0} updated'.format(provider[superdesk.config.ID_FIELD]))
# Only push a notification if there has been an update
if LAST_ITEM_UPDATE in update:
push_notification('ingest:update', provider_id=str(provider[superdesk.config.ID_FIELD]))
finally:
unlock(lock_name, host_name)
def remove_expired_data(provider):
"""Remove expired data for provider"""
print('Removing expired content for provider: %s' % provider['_id'])
days_to_keep_content = provider.get('days_to_keep', DAYS_TO_KEEP)
expiration_date = utcnow() - timedelta(days=days_to_keep_content)
items = get_expired_items(str(provider['_id']), expiration_date)
if items.count() > 0:
for item in items:
print('Removing item %s' % item['_id'])
superdesk.get_resource_service('ingest').delete_action({'_id': str(item['_id'])})
if not item.get('archived'):
for file_id in [rend.get('media') for rend in item.get('renditions', {}).values()
if rend.get('media')]:
superdesk.app.media.delete(file_id)
stats.incr('ingest.expired_items', items.count())
print('Removed expired content for provider: %s' % provider['_id'])
def update_provider(provider, rule_set=None, routing_scheme=None):
"""
Fetches items from ingest provider as per the configuration, ingests them into Superdesk and
updates the provider.
"""
if is_task_running(
provider["name"], provider[superdesk.config.ID_FIELD], provider.get("update_schedule", UPDATE_SCHEDULE_DEFAULT)
):
return
if provider.get("type") == "search":
return
if not is_updatable(provider):
return
try:
update = {LAST_UPDATED: utcnow()}
for items in providers[provider.get("type")].update(provider):
ingest_items(items, provider, rule_set, routing_scheme)
stats.incr("ingest.ingested_items", len(items))
if items:
update[LAST_ITEM_UPDATE] = utcnow()
ingest_service = superdesk.get_resource_service("ingest_providers")
ingest_service.system_update(provider[superdesk.config.ID_FIELD], update, provider)
if LAST_ITEM_UPDATE not in update and get_is_idle(provider):
notify_and_add_activity(
ACTIVITY_EVENT,
"Provider {{name}} has gone strangely quiet. Last activity was on {{last}}",
resource="ingest_providers",
user_list=ingest_service._get_administrators(),
name=provider.get("name"),
last=provider[LAST_ITEM_UPDATE].replace(tzinfo=timezone.utc).astimezone(tz=None).strftime("%c"),
)
logger.info("Provider {0} updated".format(provider[superdesk.config.ID_FIELD]))
push_notification("ingest:update", provider_id=str(provider[superdesk.config.ID_FIELD]))
finally:
mark_task_as_not_running(provider["name"], provider[superdesk.config.ID_FIELD])
def update_provider(provider, rule_set=None):
"""
Fetches items from ingest provider as per the configuration, ingests them into Superdesk and
updates the provider.
"""
superdesk.get_resource_service('ingest_providers').update(
provider['_id'],
{
LAST_UPDATED:
utcnow(),
# Providing the _etag as system updates to the documents shouldn't override _etag.
app.config['ETAG']:
provider.get(app.config['ETAG'])
})
for items in providers[provider.get('type')].update(provider):
ingest_items(items, provider, rule_set)
stats.incr('ingest.ingested_items', len(items))
logger.info('Provider {0} updated'.format(provider['_id']))
push_notification('ingest:update')
def remove_expired_data(provider):
"""Remove expired data for provider"""
print('Removing expired content for provider: %s' % provider['_id'])
days_to_keep_content = provider.get('days_to_keep', DAYS_TO_KEEP)
expiration_date = utcnow() - timedelta(days=days_to_keep_content)
items = get_expired_items(str(provider['_id']), expiration_date)
if items.count() > 0:
for item in items:
print('Removing item %s' % item['_id'])
superdesk.get_resource_service('ingest').delete_action(
{'_id': str(item['_id'])})
if not item.get('archived'):
for file_id in [
rend.get('media')
for rend in item.get('renditions', {}).values()
if rend.get('media')
]:
superdesk.app.media.delete(file_id)
stats.incr('ingest.expired_items', items.count())
print('Removed expired content for provider: %s' % provider['_id'])
def remove_expired_data(provider):
"""Remove expired data for provider"""
logger.info('Removing expired content for provider: %s' % provider.get('_id', 'Detached items'))
try:
feeding_service = registered_feeding_services[provider['feeding_service']]
feeding_service = feeding_service.__class__()
ingest_collection = feeding_service.service if hasattr(feeding_service, 'service') else 'ingest'
except KeyError:
ingest_collection = 'ingest'
ingest_service = superdesk.get_resource_service(ingest_collection)
items = get_expired_items(provider, ingest_collection)
ids = [item['_id'] for item in items]
items.rewind()
file_ids = [rend.get('media')
for item in items
for rend in item.get('renditions', {}).values()
if not item.get('archived') and rend.get('media')]
if ids:
logger.info('Removing items %s' % ids)
ingest_service.delete({'_id': {'$in': ids}})
push_expired_notification(ids)
for file_id in file_ids:
logger.info('Deleting file: %s' % file_id)
superdesk.app.media.delete(file_id)
stats.incr('ingest.expired_items', len(ids))
logger.info('Removed expired content for provider: {0} count: {1}'
.format(provider.get('_id', 'Detached items'), len(ids)))
remove_expired_from_elastic(ingest_collection)
def remove_expired_data(provider):
"""Remove expired data for provider"""
logger.info('Removing expired content for provider: %s' % provider.get('_id', 'Detached items'))
try:
feeding_service = registered_feeding_services[provider['feeding_service']]
feeding_service = feeding_service.__class__()
ingest_collection = feeding_service.service if hasattr(feeding_service, 'service') else 'ingest'
except KeyError:
ingest_collection = 'ingest'
ingest_service = superdesk.get_resource_service(ingest_collection)
items = get_expired_items(provider, ingest_collection)
ids = [item['_id'] for item in items]
items.rewind()
file_ids = [rend.get('media')
for item in items
for rend in item.get('renditions', {}).values()
if not item.get('archived') and rend.get('media')]
if ids:
logger.info('Removing items %s' % ids)
ingest_service.delete({'_id': {'$in': ids}})
push_expired_notification(ids)
for file_id in file_ids:
logger.info('Deleting file: %s' % file_id)
superdesk.app.media.delete(file_id)
stats.incr('ingest.expired_items', len(ids))
logger.info('Removed expired content for provider: {0} count: {1}'
.format(provider.get('_id', 'Detached items'), len(ids)))
remove_expired_from_elastic(ingest_collection)
