#
# Copyright (c) 2018, 2023 Oracle and/or its affiliates. All rights reserved.
#
# Licensed under the Universal Permissive License v 1.0 as shown at
# https://oss.oracle.com/licenses/upl/
#
from json import loads
from logging import FileHandler, Formatter, WARNING, getLogger
from os import mkdir, path
from ssl import SSLContext, SSLError, create_default_context
from sys import argv
from .client import Client
from .common import CheckValue, UserInfo
from .config import NoSQLHandleConfig
from .exception import IllegalArgumentException, IllegalStateException
from .iam import SignatureProvider
from .kv import StoreAccessTokenProvider
from .operations import (
DeleteRequest, GetIndexesRequest, GetRequest, GetTableRequest,
ListTablesRequest, MultiDeleteRequest, PrepareRequest, PutRequest,
QueryRequest, SystemRequest, SystemStatusRequest, TableRequest,
TableUsageRequest, WriteMultipleRequest, QueryIterableResult)
[docs]
class NoSQLHandle(object):
"""
NoSQLHandle is a handle that can be used to access Oracle NoSQL tables. To
create a connection represented by NoSQLHandle, request an instance using
:py:class:`NoSQLHandleConfig`, which allows an application to specify
default values and other configuration information to be used by the handle.
The same interface is available to both users of the Oracle NoSQL Database
Cloud Service and the on-premise Oracle NoSQL Database; however, some
methods and/or parameters are specific to each environment. The
documentation has notes about whether a class, method, or parameter is
environment-specific. Unless otherwise noted they are applicable to both
environments.
A handle has memory and network resources associated with it. Consequently,
the :py:meth:`close` method must be invoked to free up the resources when
the application is done using the handle. To minimize network activity as
well as resource allocation and deallocation overheads, it's best to avoid
repeated creation and closing of handles. For example, creating and closing
a handle around each operation, would incur large resource allocation
overheads resulting in poor application performance.
A handle permits concurrent operations, so a single handle is sufficient to
access tables in a multi-threaded application. The creation of multiple
handles incurs additional resource overheads without providing any
performance benefit.
With the exception of :py:meth:`close` the operations on this interface
follow a similar pattern. They accept a Request object containing
parameters, both required and optional. They return a Result object
containing results. Operation failures throw exceptions. Unique subclasses
of Request and Result exist for most operations, containing information
specific to the operation. All of these operations result in remote calls
across a network.
All Request instances support specification of parameters for the operation
as well as the ability to override default parameters which may have been
specified in :py:class:`NoSQLHandleConfig`, such as request timeouts, etc.
Objects returned by methods of this interface can only be used safely by
one thread at a time unless synchronized externally. Request objects are not
copied and must not be modified by the application while a method on this
interface is using them.
For Error and Exception Handling, on success all methods in this interface
return Result objects. Errors are thrown as exceptions. Exceptions that may
be retried may succeed on retry. These are instances of
:py:class:`RetryableException`. Exceptions that may not be retried and if
retried, will fail again. Exceptions that may be retried return True for
:py:meth:`RetryableException.ok_to_retry` while those that may not will
return False. Examples of retryable exceptions are those which indicate
resource consumption violations such as
:py:class:`OperationThrottlingException`. Examples of exceptions that should
not be retried are :py:class:`IllegalArgumentException`,
:py:class:`TableNotFoundException`, and any other exception indicating a
syntactic or semantic error.
Instances of NoSQLHandle are thread-safe and expected to be shared among
threads.
:param config: an instance of NoSQLHandleConfig.
:type config: NoSQLHandleConfig
:raises IllegalArgumentException: raises the exception if config is not an
instance of NoSQLHandleConfig.
"""
def __init__(self, config):
if not isinstance(config, NoSQLHandleConfig):
raise IllegalArgumentException(
'config must be an instance of NoSQLHandleConfig.')
logger = self._get_logger(config)
# config SSLContext first, on-prem authorization provider will reuse the
# context in NoSQLHandleConfig
self._config_ssl_context(config)
self._config_auth_provider(config, logger)
self._client = Client(config, logger)
[docs]
def delete(self, request):
"""
Deletes a row from a table. The row is identified using a primary key
value supplied in :py:meth:`DeleteRequest.set_key`.
By default a delete operation is unconditional and will succeed if the
specified row exists. Delete operations can be made conditional based
on whether the :py:class:`Version` of an existing row matches that
supplied by :py:meth:`DeleteRequest.set_match_version`.
It is also possible, on failure, to return information about the
existing row. The row, including it's :py:class:`Version` can be
optionally returned if a delete operation fails because of a Version
mismatch. The existing row information will only be returned if
:py:meth:`DeleteRequest.set_return_row` is True and the operation fails
because :py:meth:`DeleteRequest.set_match_version` is used and the
operation fails because the row exists and its version does not match.
Use of :py:meth:`DeleteRequest.set_return_row` may result in additional
consumed read capacity. If the operation is successful there will be no
information returned about the previous row.
:param request: the input parameters for the operation.
:type request: DeleteRequest
:returns: the result of the operation.
:rtype: DeleteResult
:raises IllegalArgumentException: raises the exception if request is not
an instance of :py:class:`DeleteRequest`.
:raises NoSQLException: raises the exception if the operation cannot be
performed for any other reason.
"""
if not isinstance(request, DeleteRequest):
raise IllegalArgumentException(
'The parameter should be an instance of DeleteRequest.')
return self._execute(request)
[docs]
def get(self, request):
"""
Gets the row associated with a primary key. On success the value of the
row is available using the :py:meth:`GetResult.get_value` operation. If
there are no matching rows that method will return None.
The default consistency used for the operation is Consistency.EVENTUAL
unless an explicit value has been set using
:py:meth:`NoSQLHandleConfig.set_consistency` or
:py:meth:`GetRequest.set_consistency`. Use of Consistency.ABSOLUTE may
affect latency of the operation and may result in additional cost for
the operation.
:param request: the input parameters for the operation.
:type request: GetRequest
:returns: the result of the operation.
:rtype: GetResult
:raises IllegalArgumentException: raises the exception if request is not
an instance of :py:class:`GetRequest`.
:raises NoSQLException: raises the exception if the operation cannot be
performed for any other reason.
"""
if not isinstance(request, GetRequest):
raise IllegalArgumentException(
'The parameter should be an instance of GetRequest.')
return self._execute(request)
[docs]
def get_indexes(self, request):
"""
Returns information about and index, or indexes on a table. If no index
name is specified in the :py:class:`GetIndexesRequest`, then information
on all indexes is returned.
:param request: the input parameters for the operation.
:type request: GetIndexesRequest
:returns: the result of the operation.
:rtype: GetIndexesResult
:raises IllegalArgumentException: raises the exception if request is not
an instance of :py:class:`GetIndexesRequest`.
:raises NoSQLException: raises the exception if the operation cannot be
performed for any other reason.
"""
if not isinstance(request, GetIndexesRequest):
raise IllegalArgumentException(
'The parameter should be an instance of GetIndexesRequest.')
return self._execute(request)
[docs]
def get_table(self, request):
"""
Gets static information about the specified table including its state,
provisioned throughput and capacity and schema. Dynamic information such
as usage is obtained using :py:meth:`get_table_usage`. Throughput,
capacity and usage information is only available when using the Cloud
Service and will be None or not defined on-premise.
:param request: the input parameters for the operation.
:type request: GetTableRequest
:returns: the result of the operation.
:rtype: TableResult
:raises IllegalArgumentException: raises the exception if request is not
an instance of :py:class:`GetTableRequest`.
:raises TableNotFoundException: raises the exception if the specified
table does not exist.
:raises NoSQLException: raises the exception if the operation cannot be
performed for any other reason.
"""
if not isinstance(request, GetTableRequest):
raise IllegalArgumentException(
'The parameter should be an instance of GetTableRequest.')
res = self._execute(request)
# Update rate limiters, if table has limits.
self._client.update_rate_limiters(
res.get_table_name(), res.get_table_limits())
return res
[docs]
def get_table_usage(self, request):
"""
Cloud service only.
Gets dynamic information about the specified table such as the current
throughput usage. Usage information is collected in time slices and
returned in individual usage records. It is possible to specify a
time-based range of usage records using input parameters.
:param request: the input parameters for the operation.
:type request: TableUsageRequest
:returns: the result of the operation.
:rtype: TableUsageResult
:raises IllegalArgumentException: raises the exception if request is not
an instance of :py:class:`TableUsageRequest`.
:raises TableNotFoundException: raises the exception if the specified
table does not exist.
:raises NoSQLException: raises the exception if the operation cannot be
performed for any other reason.
"""
if not isinstance(request, TableUsageRequest):
raise IllegalArgumentException(
'The parameter should be an instance of TableUsageRequest.')
return self._execute(request)
[docs]
def list_tables(self, request):
"""
Lists tables, returning table names. If further information about a
specific table is desired the :py:meth:`get_table` interface may be
used. If a given identity has access to a large number of tables the
list may be paged using input parameters.
:param request: the input parameters for the operation.
:type request: ListTablesRequest
:returns: the result of the operation.
:rtype: ListTablesResult
:raises IllegalArgumentException: raises the exception if request is not
an instance of :py:class:`ListTablesRequest`.
:raises NoSQLException: raises the exception if the operation cannot be
performed for any other reason.
"""
if not isinstance(request, ListTablesRequest):
raise IllegalArgumentException(
'The parameter should be an instance of ListTablesRequest.')
return self._execute(request)
[docs]
def multi_delete(self, request):
"""
Deletes multiple rows from a table in an atomic operation. The key used
may be partial but must contain all of the fields that are in the shard
key. A range may be specified to delete a range of keys.
:param request: the input parameters for the operation.
:type request: MultiDeleteRequest
:returns: the result of the operation.
:rtype: MultiDeleteResult
:raises IllegalArgumentException: raises the exception if request is not
an instance of :py:class:`MultiDeleteRequest`.
:raises NoSQLException: raises the exception if the operation cannot be
performed for any other reason.
"""
if not isinstance(request, MultiDeleteRequest):
raise IllegalArgumentException(
'The parameter should be an instance of MultiDeleteRequest.')
return self._execute(request)
[docs]
def prepare(self, request):
"""
Prepares a query for execution and reuse. See :py:meth:`query` for
general information and restrictions. It is recommended that prepared
queries are used when the same query will run multiple times as
execution is much more efficient than starting with a query string every
time. The query language and API support query variables to assist with
re-use.
:param request: the input parameters for the operation.
:type request: PrepareRequest
:returns: the result of the operation.
:rtype: PrepareResult
:raises IllegalArgumentException: raises the exception if request is not
an instance of :py:class:`PrepareRequest`.
:raises NoSQLException: raises the exception if the operation cannot be
performed for any other reason.
"""
if not isinstance(request, PrepareRequest):
raise IllegalArgumentException(
'The parameter should be an instance of PrepareRequest.')
return self._execute(request)
[docs]
def put(self, request):
"""
Puts a row into a table. This method creates a new row or overwrites
an existing row entirely. The value used for the put is in the
:py:class:`PutRequest` object and must contain a complete primary key
and all required fields.
It is not possible to put part of a row.
Any fields that are not provided will be defaulted, overwriting any
existing value. Fields that are not noneable or defaulted must be
provided or an exception will be thrown.
By default a put operation is unconditional, but put operations can be
conditional based on existence, or not, of a previous value as well as
conditional on the :py:class:`Version` of the existing value.
Use PutOption.IF_ABSENT to do a put only if there is no existing row
that matches the primary key.\n
Use PutOption.IF_PRESENT to do a put only if there is an existing
row that matches the primary key.\n
Use PutOption.IF_VERSION to do a put only if there is an existing
row that matches the primary key and its :py:class:`Version` matches
that provided.
It is also possible, on failure, to return information about the
existing row. The row, including it's :py:class:`Version` can be
optionally returned if a put operation fails because of a Version
mismatch or if the operation fails because the row already exists.
The existing row information will only be returned if
:py:meth:`PutRequest.set_return_row` is True and one of the following
occurs:
The PutOption.IF_ABSENT is used and the operation fails because the
row already exists.\n
The PutOption.IF_VERSION is used and the operation fails because the
row exists and its version does not match.
Use of :py:meth:`PutRequest.set_return_row` may result in additional
consumed read capacity. If the operation is successful there will be no
information returned about the previous row.
:param request: the input parameters for the operation.
:type request: PutRequest
:returns: the result of the operation.
:rtype: PutResult
:raises IllegalArgumentException: raises the exception if request is not
an instance of :py:class:`PutRequest`.
:raises NoSQLException: raises the exception if the operation cannot be
performed for any other reason.
"""
if not isinstance(request, PutRequest):
raise IllegalArgumentException(
'The parameter should be an instance of PutRequest.')
return self._execute(request)
[docs]
def query(self, request):
"""
Queries a table based on the query statement specified in the
:py:class:`QueryRequest`.
Queries that include a full shard key will execute much more efficiently
than more distributed queries that must go to multiple shards.
Table and system-style queries such as "CREATE TABLE ..." or "DROP TABLE
..." are not supported by these interfaces. Those operations must be
performed using :py:meth:`table_request` or :py:meth:`system_request` as
appropriate.
The amount of data read by a single query request is limited by a system
default and can be further limited using
:py:meth:`QueryRequest.set_max_read_kb`. This limits the amount of data
*read* and not the amount of data *returned*, which means that a query
can return zero results but still have more data to read. This situation
is detected by checking if the :py:class:`QueryRequest` is done using
:py:meth:`QueryRequest.is_done`. For this reason queries should always
operate in a loop, acquiring more results, until
:py:meth:`QueryRequest.is_done` returns True, indicating that the query
is done.
:param request: the input parameters for the operation.
:type request: QueryRequest
:returns: the result of the operation.
:rtype: QueryResult
:raises IllegalArgumentException: raises the exception if request is not
an instance of :py:class:`QueryRequest`.
:raises NoSQLException: raises the exception if the operation cannot be
performed for any other reason.
"""
if not isinstance(request, QueryRequest):
raise IllegalArgumentException(
'The parameter should be an instance of QueryRequest.')
return self._execute(request)
[docs]
def query_iterable(self, request):
# type: (QueryRequest) -> QueryIterableResult
"""
Queries a table based on the query statement specified in the
:py:class:`QueryRequest`.
Queries that include a full shard key will execute much more efficiently
than more distributed queries that must go to multiple shards.
Table and system-style queries such as "CREATE TABLE ..." or "DROP TABLE
..." are not supported by these interfaces. Those operations must be
performed using :py:meth:`table_request` or :py:meth:`system_request` as
appropriate.
The amount of data read by a single query request is limited by a system
default and can be further limited using
:py:meth:`QueryRequest.set_max_read_kb`. This limits the amount of data
*read* and not the amount of data *returned*.
:param request: the input parameters for the operation.
:type request: QueryRequest
:returns: the result of the operation.
:rtype: QueryResult
:raises IllegalArgumentException: raises the exception if request is not
an instance of :py:class:`QueryRequest`.
:raises NoSQLException: raises the exception if the operation cannot be
performed for any other reason.
:versionadded: 5.3.6
"""
if not isinstance(request, QueryRequest):
raise IllegalArgumentException(
'The parameter should be an instance of QueryRequest.')
return QueryIterableResult(request, self)
[docs]
def system_request(self, request):
"""
On-premise only.
Performs a system operation on the system, such as administrative
operations that don't affect a specific table. For table-specific
operations use :py:meth:`table_request` or :py:meth:`do_table_request`.
Examples of statements in the :py:class:`SystemRequest` passed to this
method include:
CREATE NAMESPACE mynamespace\n
CREATE USER some_user IDENTIFIED BY password\n
CREATE ROLE some_role\n
GRANT ROLE some_role TO USER some_user
This operation is implicitly asynchronous. The caller must poll using
methods on :py:class:`SystemResult` to determine when it has completed.
:param request: the input parameters for the operation.
:type request: SystemRequest
:returns: the result of the operation.
:rtype: SystemResult
:raises IllegalArgumentException: raises the exception if request is not
an instance of :py:class:`SystemRequest`.
:raises NoSQLException: raises the exception if the operation cannot be
performed for any other reason.
"""
if not isinstance(request, SystemRequest):
raise IllegalArgumentException(
'The parameter should be an instance of SystemRequest.')
return self._execute(request)
[docs]
def system_status(self, request):
"""
On-premise only.
Checks the status of an operation previously performed using
:py:meth:`system_request`.
:param request: the input parameters for the operation.
:type request: SystemStatusRequest
:returns: the result of the operation.
:rtype: SystemResult
:raises IllegalArgumentException: raises the exception if request is not
an instance of :py:class:`SystemStatusRequest`.
:raises NoSQLException: raises the exception if the operation cannot be
performed for any other reason.
"""
if not isinstance(request, SystemStatusRequest):
raise IllegalArgumentException(
'The parameter should be an instance of SystemStatusRequest.')
return self._execute(request)
[docs]
def table_request(self, request):
"""
Performs an operation on a table. This method is used for creating and
dropping tables and indexes as well as altering tables. Only one
operation is allowed on a table at any one time.
This operation is implicitly asynchronous. The caller must poll using
methods on :py:class:`TableResult` to determine when it has completed.
:param request: the input parameters for the operation.
:type request: TableRequest
:returns: the result of the operation.
:rtype: TableResult
:raises IllegalArgumentException: raises the exception if request is not
an instance of :py:class:`TableRequest`.
:raises NoSQLException: raises the exception if the operation cannot be
performed for any other reason.
"""
if not isinstance(request, TableRequest):
raise IllegalArgumentException(
'The parameter should be an instance of TableRequest.')
res = self._execute(request)
# Update rate limiters, if table has limits.
self._client.update_rate_limiters(
res.get_table_name(), res.get_table_limits())
return res
[docs]
def write_multiple(self, request):
"""
Executes a sequence of operations associated with a table that share the
same shard key portion of their primary keys, all the specified
operations are executed within the scope of a single transaction.
There are some size-based limitations on this operation:
The max number of individual operations (put, delete) in a single
WriteMultipleRequest is 50.\n
The total request size is limited to 25MB.
:param request: the input parameters for the operation.
:type request: WriteMultipleRequest
:returns: the result of the operation.
:rtype: WriteMultipleResult
:raises IllegalArgumentException: raises the exception if request is not
an instance of :py:class:`WriteMultipleRequest`.
:raises RowSizeLimitException: raises the exception if data size in an
operation exceeds the limit.
:raises BatchOperationNumberLimitException: raises the exception if the
number of operations exceeds this limit.
:raises NoSQLException: raises the exception if the operation cannot be
performed for any other reason.
"""
if not isinstance(request, WriteMultipleRequest):
raise IllegalArgumentException(
'The parameter should be an instance of WriteMultipleRequest.')
return self._execute(request)
[docs]
def close(self):
"""
Close the NoSQLHandle.
"""
if self._client is not None:
self._client.shut_down()
self._client = None
[docs]
def do_system_request(self, statement, timeout_ms=30000,
poll_interval_ms=1000):
"""
On-premise only.
A convenience method that performs a SystemRequest and waits for
completion of the operation. This is the same as calling
:py:meth:`system_request` then calling
:py:meth:`SystemResult.wait_for_completion`. If the operation fails an
exception is thrown.
System requests are those related to namespaces and security and are
generally independent of specific tables. Examples of statements include
CREATE NAMESPACE mynamespace\n
CREATE USER some_user IDENTIFIED BY password\n
CREATE ROLE some_role\n
GRANT ROLE some_role TO USER some_user
:param statement: the system statement for the operation.
:type statement: str
:param timeout_ms: the amount of time to wait for completion, in
milliseconds.
:type timeout_ms: int
:param poll_interval_ms: the polling interval for the wait operation.
:type poll_interval_ms: int
:returns: the result of the system request.
:rtype: SystemResult
:raises IllegalArgumentException: raises the exception if any of the
parameters are invalid or required parameters are missing.
:raises RequestTimeoutException: raises the exception if the operation
times out.
:raises NoSQLException: raises the exception if the operation cannot be
performed for any other reason.
"""
CheckValue.check_str(statement, 'statement')
req = SystemRequest().set_statement(statement)
res = self.system_request(req)
res.wait_for_completion(self, timeout_ms, poll_interval_ms)
return res
[docs]
def do_table_request(self, request, timeout_ms, poll_interval_ms):
"""
A convenience method that performs a TableRequest and waits for
completion of the operation. This is the same as calling
:py:meth:`table_request` then calling
:py:meth:`TableResult.wait_for_completion`. If the operation fails an
exception is thrown. All parameters are required.
:param request: the :py:class:`TableRequest` to perform.
:type request: TableRequest
:param timeout_ms: the amount of time to wait for completion, in
milliseconds.
:type timeout_ms: int
:param poll_interval_ms: the polling interval for the wait operation.
:type poll_interval_ms: int
:returns: the result of the table request.
:rtype: TableResult
:raises IllegalArgumentException: raises the exception if any of the
parameters are invalid or required parameters are missing.
:raises RequestTimeoutException: raises the exception if the operation
times out.
:raises NoSQLException: raises the exception if the operation cannot be
performed for any other reason.
"""
if not isinstance(request, TableRequest):
raise IllegalArgumentException(
'The parameter should be an instance of TableRequest.')
res = self.table_request(request)
res.wait_for_completion(self, timeout_ms, poll_interval_ms)
return res
[docs]
def list_namespaces(self):
"""
On-premise only.
Returns the namespaces in a store as a list of string.
:returns: the namespaces, or None if none are found.
:rtype: list(str)
"""
res = self.do_system_request('show as json namespaces')
json_res = res.get_result_string()
if json_res is None:
return None
root = loads(json_res)
namespaces = root.get('namespaces')
if namespaces is None:
return None
results = list()
for namespace in namespaces:
results.append(namespace)
return results
[docs]
def list_roles(self):
"""
On-premise only.
Returns the roles in a store as a list of string.
:returns: the list of roles, or None if none are found.
:rtype: list(str)
"""
res = self.do_system_request('show as json roles')
json_res = res.get_result_string()
if json_res is None:
return None
root = loads(json_res)
roles = root.get('roles')
if roles is None:
return None
results = list()
for role in roles:
results.append(role['name'])
return results
[docs]
def list_users(self):
"""
On-premise only.
Returns the users in a store as a list of :py:class:`UserInfo`.
:returns: the list of users, or None if none are found.
:rtype: list(UserInfo)
"""
res = self.do_system_request('show as json users')
json_res = res.get_result_string()
if json_res is None:
return None
root = loads(json_res)
users = root.get('users')
if users is None:
return None
results = list()
for user in users:
results.append(UserInfo(user['id'], user['name']))
return results
[docs]
def get_client(self):
# For testing use
return self._client
def _execute(self, request):
# Ensure that the client exists and hasn't been closed.
if self._client is None:
raise IllegalStateException('NoSQLHandle has been closed.')
return self._client.execute(request)
def _get_logger(self, config):
"""
Returns the logger used for the driver. If no logger is specified,
create one based on this class name.
"""
if config.get_logger() is None and config.is_default_logger():
logger = getLogger(self.__class__.__name__)
logger.setLevel(WARNING)
log_dir = path.join(path.abspath(path.dirname(argv[0])), 'logs')
if not path.exists(log_dir):
mkdir(log_dir)
handler = FileHandler(path.join(log_dir, 'driver.log'))
formatter = Formatter('%(asctime)s [%(levelname)s] %(message)s')
handler.setFormatter(formatter)
logger.addHandler(handler)
else:
logger = config.get_logger()
return logger
@staticmethod
def _config_auth_provider(config, logger):
provider = config.get_authorization_provider()
if provider.get_logger() is None:
provider.set_logger(logger)
if (isinstance(provider, StoreAccessTokenProvider) and
provider.is_secure()):
if provider.get_endpoint() is None:
endpoint = config.get_service_url().geturl()
if endpoint.endswith('/'):
endpoint = endpoint[:len(endpoint) - 1]
provider.set_endpoint(endpoint)
provider.set_ssl_context(config.get_ssl_context())
elif isinstance(provider, SignatureProvider):
provider.set_service_url(config)
@staticmethod
def _config_ssl_context(config):
if config.get_ssl_context() is not None:
return
if config.get_service_url().scheme == 'https':
try:
if config.get_ssl_protocol() is None:
ctx = create_default_context()
else:
ctx = SSLContext(config.get_ssl_protocol())
if config.get_ssl_cipher_suites() is not None:
ctx.set_ciphers(config.get_ssl_cipher_suites())
if config.get_ssl_ca_certs() is not None:
ctx.load_verify_locations(config.get_ssl_ca_certs())
config.set_ssl_context(ctx)
except (SSLError, ValueError) as err:
raise IllegalArgumentException(str(err))
[docs]
def get_stats_control(self):
return self._client.get_stats_control()