def get_contents(coll_scope=None, coll_name=None, request_id=None, workload_id=None, transform_id=None,
relation_type=None, status=None, to_json=False, session=None):
"""
Get contents with collection scope, collection name, request id, workload id and relation type.
:param coll_scope: scope of the collection.
:param coll_name: name the the collection.
:param request_id: the request id.
:param workload_id: The workload_id of the request.
:param transform_id: The transform id related to this collection.
:param relation_type: The relation type between the collection and transform: input, outpu, logs and etc.
:param to_json: return json format.
:param session: The database session in use.
:returns: dict of contents
"""
collections = get_collections(scope=coll_scope, name=coll_name, request_id=request_id,
workload_id=workload_id, transform_id=transform_id,
relation_type=relation_type, to_json=to_json, session=session)
coll_ids = [coll['coll_id'] for coll in collections]
if coll_ids:
rets = orm_contents.get_contents(coll_id=coll_ids, status=status, to_json=to_json, session=session)
else:
rets = []
return rets
def get_output_contents_by_request_id_status(request_id,
name,
content_status,
limit,
transform_id=None,
to_json=False,
session=None):
"""
Get output content by request_id and content name
:param request_id: requestn id.
:param name: the content name.
:param content_status: The content status.
:param limit: limit number of contents.
:param to_json: return json format.
:param session: The database session in use.
:returns: content of the output collection.
"""
transform_ids = orm_transforms.get_transform_ids(request_id,
session=session)
found_transform_id = None
if transform_ids:
if len(transform_ids) == 1:
found_transform_id = transform_ids[0]
elif len(transform_ids) > 1 and transform_id is None:
raise "Number of the transforms(%s) is bigger than 1 and transform id is not provided" % len(
transform_ids)
else:
for tf_id in transform_ids:
if tf_id == transform_id:
found_transform_id = tf_id
break
coll_id = None
if found_transform_id:
coll_id = orm_collections.get_collection_id(
transform_id=found_transform_id,
relation_type=CollectionRelationType.Output,
session=session)
contents = []
if coll_id:
contents = orm_contents.get_contents(coll_id=coll_id,
status=content_status,
to_json=to_json,
session=session)
if name:
new_contents = []
for content in contents:
if str(content['name']) == str(name):
new_contents.append(content)
contents = new_contents
if contents and limit and len(contents) > limit:
contents = contents[:limit]
return contents
def get_contents(coll_scope=None,
coll_name=None,
request_id=None,
workload_id=None,
relation_type=None,
session=None):
"""
Get contents with collection scope, collection name, request id, workload id and relation type.
:param coll_scope: scope of the collection.
:param coll_name: name the the collection.
:param request_id: the request id.
:param workload_id: The workload_id of the request.
:param relation_type: The relation type between the collection and transform: input, outpu, logs and etc.
:param session: The database session in use.
:returns: dict of contents
"""
if request_id is None and workload_id is None:
raise exceptions.WrongParameterException(
"Either request_id or workload_id should not be None")
req_transfomr_collections = get_collections(scope=coll_scope,
name=coll_name,
request_id=request_id,
workload_id=workload_id,
session=session)
rets = {}
for request_id in req_transfomr_collections:
rets[request_id] = {}
for transform_id in req_transfomr_collections[request_id]:
rets[request_id][transform_id] = {}
for collection in req_transfomr_collections[request_id][
transform_id]:
if relation_type is not None:
if isinstance(relation_type, CollectionRelationType):
relation_type = relation_type.value
if relation_type is None or collection[
'relation_type'].value == relation_type:
scope = collection['scope']
name = collection['name']
coll_id = collection['coll_id']
coll_relation_type = collection['relation_type']
scope_name = '%s:%s' % (scope, name)
contents = orm_contents.get_contents(coll_id=coll_id,
session=session)
rets[request_id][transform_id][scope_name] = {
'collection': collection,
'relation_type': coll_relation_type,
'contents': contents
}
return rets
def get_contents(coll_scope=None,
coll_name=None,
request_id=None,
workload_id=None,
transform_id=None,
relation_type=None,
to_json=False,
session=None):
"""
Get contents with collection scope, collection name, request id, workload id and relation type.
:param coll_scope: scope of the collection.
:param coll_name: name the the collection.
:param request_id: the request id.
:param workload_id: The workload_id of the request.
:param transform_id: The transform id related to this collection.
:param relation_type: The relation type between the collection and transform: input, outpu, logs and etc.
:param to_json: return json format.
:param session: The database session in use.
:returns: dict of contents
"""
req_transfomr_collections = get_collections(scope=coll_scope,
name=coll_name,
request_id=request_id,
workload_id=workload_id,
transform_id=transform_id,
relation_type=relation_type,
to_json=to_json,
session=session)
rets = {}
for request_id in req_transfomr_collections:
rets[request_id] = {}
for transform_id in req_transfomr_collections[request_id]:
rets[request_id][transform_id] = {}
for collection in req_transfomr_collections[request_id][
transform_id]:
scope = collection['scope']
name = collection['name']
coll_id = collection['coll_id']
coll_relation_type = collection['relation_type']
scope_name = '%s:%s' % (scope, name)
contents = orm_contents.get_contents(coll_id=coll_id,
to_json=to_json,
session=session)
rets[request_id][transform_id][scope_name] = {
'collection': collection,
'relation_type': coll_relation_type,
'contents': contents
}
return rets
def get_contents_by_coll_id_status(coll_id, status=None, to_json=False, session=None):
"""
Get contents or raise a NoObject exception.
:param coll_id: Collection id.
:param status: Content status or list of content status.
:param to_json: return json format.
:param session: The database session in use.
:raises NoObject: If no content is founded.
:returns: list of contents.
"""
return orm_contents.get_contents(coll_id=coll_id, status=status, to_json=to_json, session=session)
def update_input_collection_with_contents(coll,
parameters,
contents,
returning_id=False,
bulk_size=100,
session=None):
"""
update a collection.
:param coll_id: the collection id.
:param parameters: A dictionary of parameters.
:param contents: dict of contents.
:param returning_id: whether to return id.
:param bulk_size: bulk per insert to db.
:param session: The database session in use.
:raises NoObject: If no request is founded.
:raises DatabaseException: If there is a database error.
:returns new contents
"""
new_files = 0
processed_files = 0
avail_contents = orm_contents.get_contents(coll_id=coll['coll_id'],
session=session)
avail_contents_dict = {}
for content in avail_contents:
key = '%s:%s:%s:%s' % (content['scope'], content['name'],
content['min_id'], content['max_id'])
avail_contents_dict[key] = content
if content['status'] in [
ContentStatus.Mapped, ContentStatus.Mapped.value
]:
processed_files += 1
if content['status'] in [ContentStatus.New, ContentStatus.New.value]:
new_files += 1
to_addes = []
# to_updates = []
for content in contents:
key = '%s:%s:%s:%s' % (content['scope'], content['name'],
content['min_id'], content['max_id'])
if key in avail_contents_dict:
"""
to_update = {'content_id': content['content_id'],
'status': content['status']}
if 'bytes' in content:
to_update['bytes'] = content['bytes']
if 'md5' in content:
to_update['md5'] = content['md5']
if 'adler32' in content:
to_update['adler32'] = content['adler32']
if 'expired_at' in content:
to_update['expired_at'] = content['expired_at']
to_updates.append(to_updated)
# not to do anything, no need to update
"""
pass
else:
to_addes.append(content)
# there are new files
if to_addes:
add_contents(to_addes,
returning_id=returning_id,
bulk_size=bulk_size,
session=session)
parameters['processed_files'] = processed_files
parameters['new_files'] = new_files
if processed_files == coll['total_files']:
parameters['status'] = CollectionStatus.Closed
update_collection(coll['coll_id'], parameters, session=session)
return to_addes
def test_contents_orm(self):
""" Contents (ORM): Test contents """
req_properties = get_request_properties()
trans_properties = get_transform_properties()
coll_properties = get_collection_properties()
content_properties = get_content_properties()
request_id = add_request(**req_properties)
trans_properties['request_id'] = request_id
trans_id = add_transform(**trans_properties)
coll_properties['transform_id'] = trans_id
coll_id = add_collection(**coll_properties)
content_properties['coll_id'] = coll_id
origin_content_id = add_content(**content_properties)
content_properties1 = copy.deepcopy(content_properties)
content_properties1['min_id'] = 101
content_properties1['max_id'] = 200
origin_content_id1 = add_content(**content_properties1)
content_properties2 = copy.deepcopy(content_properties)
content_properties2['min_id'] = 0
content_properties2['max_id'] = 200
origin_content_id2 = add_content(**content_properties2)
content_properties3 = copy.deepcopy(content_properties)
content_properties3['name'] = content_properties3['name'] + '_1'
origin_content_id3 = add_content(**content_properties3)
origin_content_ids = [
origin_content_id, origin_content_id1, origin_content_id2,
origin_content_id3
]
contents = get_contents(coll_id=coll_id)
assert_equal(len(contents), 4)
for content in contents:
assert_in(content['content_id'], origin_content_ids)
contents = get_contents(scope=content_properties['scope'],
name=content_properties['name'],
coll_id=coll_id)
assert_equal(len(contents), 3)
for content in contents:
assert_in(content['content_id'], origin_content_ids)
contents = get_contents(scope=content_properties3['scope'],
name=content_properties3['name'],
coll_id=coll_id)
assert_equal(len(contents), 1)
assert_equal(contents[0]['content_id'], origin_content_id3)
contents = get_match_contents(coll_id=content_properties['coll_id'],
scope=content_properties['scope'],
name=content_properties['name'],
min_id=content_properties['min_id'],
max_id=content_properties['max_id'])
assert_equal(len(contents), 2)
for content in contents:
assert_in(content['content_id'],
[origin_content_id, origin_content_id2])
to_updates = [{
'path': 'test_path1',
'status': ContentStatus.Processing,
'content_id': origin_content_id
}, {
'path': 'test_path2',
'status': ContentStatus.Processing,
'content_id': origin_content_id1
}]
update_contents(to_updates)
content = get_content(content_id=origin_content_id)
assert_equal(content['status'], ContentStatus.Processing)
assert_equal(content['path'], 'test_path1')
content = get_content(content_id=origin_content_id1)
assert_equal(content['status'], ContentStatus.Processing)
assert_equal(content['path'], 'test_path2')
