def delete_owner(self, *, table_uri: str, owner: str) -> None:
"""
:param table_uri:
:param owner:
:return:
"""
table = self._get_table_entity(table_uri=table_uri)
table_entity = table.entity
if table_entity[self.REL_ATTRS_KEY].get("ownedBy"):
try:
active_owners = filter(
lambda item: item['relationshipStatus'] == Status.ACTIVE
and item['displayText'] == owner,
table_entity[self.REL_ATTRS_KEY]['ownedBy'])
if list(active_owners):
self._driver.relationship_guid(
next(active_owners).get('relationshipGuid')).delete()
else:
raise BadRequest('You can not delete this owner.')
except NotFound as ex:
LOGGER.exception(
'Error while removing table data owner. {}'.format(
str(ex)))
def get_table(self, *, table_uri: str) -> Table:
"""
Gathers all the information needed for the Table Detail Page.
:param table_uri:
:return: A Table object with all the information available
or gathered from different entities.
"""
entity = self._get_table_entity(table_uri=table_uri)
table_details = entity.entity
try:
attrs = table_details[self.ATTRS_KEY]
programmatic_descriptions = self._get_programmatic_descriptions(
attrs.get('parameters'))
table_qn = parse_table_qualified_name(
qualified_name=attrs.get(self.QN_KEY))
tags = []
# Using or in case, if the key 'classifications' is there with a None
for classification in table_details.get(
"classifications") or list():
tags.append(
Tag(tag_name=classification.get('typeName'),
tag_type="default"))
columns = self._serialize_columns(entity=entity)
reports_guids = [
report.get("guid")
for report in attrs.get("reports") or list()
]
table = Table(
database=table_details.get('typeName'),
cluster=table_qn.get('cluster_name', ''),
schema=table_qn.get('db_name', ''),
name=attrs.get('name') or table_qn.get("table_name", ''),
tags=tags,
description=attrs.get('description') or attrs.get('comment'),
owners=[User(email=attrs.get('owner'))],
resource_reports=self._get_reports(guids=reports_guids),
columns=columns,
table_readers=self._get_readers(attrs.get(self.QN_KEY)),
last_updated_timestamp=self._parse_date(
table_details.get('updateTime')),
programmatic_descriptions=programmatic_descriptions)
return table
except KeyError as ex:
LOGGER.exception(
'Error while accessing table information. {}'.format(str(ex)))
raise BadRequest(
'Some of the required attributes '
'are missing in : ( {table_uri} )'.format(table_uri=table_uri))
def get_popular_tables(self,
*,
num_entries: int = 10) -> List[PopularTable]:
"""
FixMe: For now it simply returns ALL the tables available,
Need to generate the formula for popular tables only.
:param num_entries:
:return:
"""
popular_tables = list()
params = {
'typeName': self.TABLE_ENTITY,
'excludeDeletedEntities': True,
self.ATTRS_KEY: [self.DB_ATTRIBUTE]
}
try:
# Fetch all the Popular Tables
_table_collection = self._driver.search_basic.create(data=params)
# Inflate the table entities
table_entities = _table_collection.entities
except BadRequest as ex:
LOGGER.exception(
f'Please make sure you have assigned the appropriate '
f'self.TABLE_ENTITY entity to your atlas tables. {ex}')
raise BadRequest('Unable to fetch popular tables. '
'Please check your configurations.')
# Make a dictionary of Database Entities to avoid multiple DB calls
dbs_dict = self._get_rel_attributes_dict(entities=table_entities,
attribute=self.DB_ATTRIBUTE)
# Make instances of PopularTable
for entity in table_entities:
attrs = entity.attributes
# DB would be available in attributes
# because it is in the request parameter.
db_id = attrs.get(self.DB_ATTRIBUTE, {}).get('guid')
db_entity = dbs_dict.get(db_id)
if db_entity:
db_attrs = db_entity.attributes
db_name = db_attrs.get(self.NAME_ATTRIBUTE)
db_cluster = db_attrs.get('clusterName')
else:
db_name = ''
db_cluster = ''
popular_table = PopularTable(database=entity.typeName,
cluster=db_cluster,
schema=db_name,
name=attrs.get(self.NAME_ATTRIBUTE),
description=attrs.get('description'))
popular_tables.append(popular_table)
return popular_tables
def get_table(self, *, table_id: str, table_info: Dict) -> Table:
"""
Gathers all the information needed for the Table Detail Page.
:param table_id:
:param table_info: Additional table information (entity, db, cluster, name)
:return: A Table object with all the information available
or gathered from different entities.
"""
table_entity = self._get_table_entity(table_id=table_id)
table_details = table_entity.entity
try:
attrs = table_details[self.ATTRS_KEY]
rel_attrs = table_details[self.REL_ATTRS_KEY]
tags = []
# Using or in case, if the key 'classifications' is there with a None
for classification in table_details.get(
"classifications") or list():
tags.append(
Tag(tag_name=classification.get('typeName'),
tag_type="default"))
columns = []
for column in rel_attrs.get('columns') or list():
col_entity = table_entity.referredEntities[column['guid']]
col_attrs = col_entity[self.ATTRS_KEY]
columns.append(
Column(
name=col_attrs.get(self.NAME_ATTRIBUTE),
description=col_attrs.get('description'),
col_type=col_attrs.get('type')
or col_attrs.get('dataType'),
sort_order=col_attrs.get('position'),
))
table = Table(
database=table_info['entity'],
cluster=table_info['cluster'],
schema=table_info['db'],
name=table_info['name'],
tags=tags,
description=attrs.get('description'),
owners=[User(email=attrs.get('owner'))],
columns=columns,
last_updated_timestamp=table_details.get('updateTime'))
return table
except KeyError as ex:
LOGGER.exception(
'Error while accessing table information. {}'.format(str(ex)))
raise BadRequest(
'Some of the required attributes '
'are missing in : ( {table_id} )'.format(table_id=table_id))
def add_owner(self, *, table_uri: str, owner: str) -> None:
"""
Query on Atlas User entity to find if the entity exist for the
owner string in parameter, if not create one. And then use that User
entity's GUID and add a relationship between Table and User, on ownedBy field.
:param table_uri:
:param owner: Email address of the owner
:return: None, as it simply adds the owner.
"""
# Generating owner_info to validate if the user exists
owner_info = self._get_user_details(owner, fallback=owner)
if not owner_info:
raise NotFoundException(f'User "{owner}" does not exist.')
user_dict = {
"entity": {
"typeName": "User",
"attributes": {
"qualifiedName": owner
},
}
}
# Get or Create a User
user_entity = self._driver.entity_post.create(data=user_dict)
user_guid = next(iter(user_entity.get("guidAssignments").values()))
table = self._get_table_entity(table_uri=table_uri)
entity_def = {
"typeName": "DataSet_Users_Owner",
"end1": {
"guid": table.entity.get("guid"),
"typeName": "Table",
},
"end2": {
"guid": user_guid,
"typeName": "User",
},
}
try:
self._driver.relationship.create(data=entity_def)
except Conflict as ex:
LOGGER.exception(
'Error while adding the owner information. {}'.format(str(ex)))
raise BadRequest(
f'User {owner} is already added as a data owner for '
f'table {table_uri}.')
def get_popular_tables(self,
*,
num_entries: int = 10) -> List[PopularTable]:
"""
FixMe: For now it simply returns ALL the tables available,
Need to generate the formula for popular tables only.
:param num_entries:
:return:
"""
popular_tables = list()
params = {
'typeName': self.TABLE_ENTITY,
'excludeDeletedEntities': True
}
try:
guids = self._get_ids_from_basic_search(params=params)
entity_collection = self._driver.entity_bulk(guid=guids)
except BadRequest as ex:
LOGGER.exception(
f'Please make sure you have assigned the appropriate '
f'self.TABLE_ENTITY entity to your atlas tables. {ex}')
raise BadRequest('Unable to fetch popular tables. '
'Please check your configurations.')
for _collection in entity_collection:
for entity in _collection.entities:
attrs = entity.attributes
# ToDo (Verdan): Investigate why db is not in referredEntities
database = attrs.get(self.DB_KEY)
if database:
db_entity = self._driver.entity_guid(database['guid'])
db_attrs = db_entity.entity['attributes']
db_name = db_attrs.get(self.NAME_KEY)
db_cluster = db_attrs.get('clusterName')
else:
db_name = ''
db_cluster = ''
popular_table = PopularTable(
database=entity.typeName,
cluster=db_cluster,
schema=db_name,
name=attrs.get(self.NAME_KEY),
description=attrs.get('description'))
popular_tables.append(popular_table)
return popular_tables
def get_popular_tables(self, *, num_entries: int) -> List[PopularTable]:
"""
:param num_entries: Number of popular tables to fetch
:return: A List of popular tables instances
"""
popular_tables = list()
try:
# Fetch the metadata entities based on popularity score
query_metadata_ids = {'query': f'FROM Table SELECT metadata.__guid '
f'ORDERBY popularityScore desc '
f'LIMIT {num_entries}'}
metadata_ids = self._get_flat_values_from_dsl(dsl_param=query_metadata_ids)
metadata_collection = self._driver.entity_bulk(guid=metadata_ids)
except KeyError as ex:
LOGGER.exception(f'DSL Search query failed: {ex}')
raise BadRequest('Unable to fetch popular tables. '
'Please check your configurations.')
if not metadata_collection:
raise NotFoundException('Unable to fetch popular tables. '
'Please check your configurations.')
for _collection in metadata_collection:
metadata_entities = _collection.entities_with_relationships(attributes=["parentEntity"])
for metadata in metadata_entities:
table = metadata.relationshipAttributes.get("parentEntity")
table_attrs = table.get(self.ATTRS_KEY)
_regex_result = self.TABLE_QN_REGEX.match(table_attrs.get(self.QN_KEY))
table_qn = _regex_result.groupdict() if _regex_result else dict()
# Hardcoded empty strings as default, because these values are not optional
table_name = table_attrs.get(self.NAME_ATTRIBUTE) or table_qn.get("table_name", '')
db_name = table_qn.get("db_name", '')
db_cluster = table_qn.get("cluster_name", '')
popular_table = PopularTable(database=table.get("typeName"),
cluster=db_cluster,
schema=db_name,
name=table_name,
description=table_attrs.get('description'))
popular_tables.append(popular_table)
return popular_tables
def get_table(self, *, table_uri: str) -> Table:
"""
Gathers all the information needed for the Table Detail Page.
:param table_uri:
:return: A Table object with all the information available
or gathered from different entities.
"""
entity, table_info = self._get_table_entity(table_uri=table_uri)
table_details = entity.entity
try:
attrs = table_details[self.ATTRS_KEY]
tags = []
# Using or in case, if the key 'classifications' is there with a None
for classification in table_details.get("classifications") or list():
tags.append(
Tag(
tag_name=classification.get('typeName'),
tag_type="default"
)
)
columns = self._serialize_columns(entity=entity)
table = Table(database=table_info['entity'],
cluster=table_info['cluster'],
schema=table_info['db'],
name=table_info['name'],
tags=tags,
description=attrs.get('description'),
owners=[User(email=attrs.get('owner'))],
columns=columns,
last_updated_timestamp=table_details.get('updateTime'))
return table
except KeyError as ex:
LOGGER.exception('Error while accessing table information. {}'
.format(str(ex)))
raise BadRequest('Some of the required attributes '
'are missing in : ( {table_uri} )'
.format(table_uri=table_uri))
def update(self, data):
"""
Updates entities in bulk amount. Data must be a list of instances
"""
LOG.debug(
f"Trying to update {self.__class__.__name__} with the data {data}")
if not isinstance(data, list):
raise BadRequest(
url=self.model_class.path,
method="PUT",
message=
f'Data should be a list of "{self.model_class.data_class}"')
_data = list()
for item in data:
# noinspection PyProtectedMember
_data.append(
self.model_class.data_class(**item).to_dict(
data_key=self.model_class.data_key, ignore_falsy=True))
self.load(self.client.put(self.url, data=_data))
return self._models
def test_get_popular_tables_search_exception(self):
with self.assertRaises(BadRequest):
self.proxy._driver.search_basic = MagicMock(
side_effect=BadRequest('Boom!'))
self.proxy.get_popular_tables(num_entries=2)
