[
https://issues.apache.org/jira/browse/BEAM-7246?focusedWorklogId=367169&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-367169
]
ASF GitHub Bot logged work on BEAM-7246:
----------------------------------------
Author: ASF GitHub Bot
Created on: 07/Jan/20 02:18
Start Date: 07/Jan/20 02:18
Worklog Time Spent: 10m
Work Description: chamikaramj commented on pull request #9606:
[BEAM-7246] Add Google Spanner IO Read on Python SDK
URL: https://github.com/apache/beam/pull/9606#discussion_r363566806
##########
File path: sdks/python/apache_beam/io/gcp/experimental/spannerio.py
##########
@@ -0,0 +1,565 @@
+#
+# Licensed to the Apache Software Foundation (ASF) under one or more
+# contributor license agreements. See the NOTICE file distributed with
+# this work for additional information regarding copyright ownership.
+# The ASF licenses this file to You under the Apache License, Version 2.0
+# (the "License"); you may not use this file except in compliance with
+# the License. You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+"""Google Cloud Spanner IO
+
+Experimental; no backwards-compatibility guarantees.
+
+This is an experimental module for reading and writing data from Google Cloud
+Spanner. Visit: https://cloud.google.com/spanner for more details.
+
+To read from Cloud Spanner apply ReadFromSpanner transformation. It will
+return a PCollection, where each element represents an individual row returned
+from the read operation. Both Query and Read APIs are supported.
+
+ReadFromSpanner relies on the ReadOperation objects which is exposed by the
+SpannerIO API. ReadOperation holds the immutable data which is responsible to
+execute batch and naive reads on Cloud Spanner. This is done for more
+convenient programming.
+
+ReadFromSpanner reads from Cloud Spanner by providing either an 'sql' param
+in the constructor or 'table' name with 'columns' as list. For example:::
+
+ records = (pipeline
+ | ReadFromSpanner(PROJECT_ID, INSTANCE_ID, DB_NAME,
+ sql='Select * from users'))
+
+ records = (pipeline
+ | ReadFromSpanner(PROJECT_ID, INSTANCE_ID, DB_NAME,
+ table='users', columns=['id', 'name', 'email']))
+
+You can also perform multiple reads by providing a list of ReadOperations
+to the ReadFromSpanner transform constructor. ReadOperation exposes two static
+methods. Use 'query' to perform sql based reads, 'table' to perform read from
+table name. For example:::
+
+ read_operations = [
+ ReadOperation.table(table='customers', columns=['name',
+ 'email']),
+ ReadOperation.table(table='vendors', columns=['name',
+ 'email']),
+ ]
+ all_users = pipeline | ReadFromSpanner(PROJECT_ID, INSTANCE_ID, DB_NAME,
+ read_operations=read_operations)
+
+ ...OR...
+
+ read_operations = [
+ ReadOperation.query(sql='Select name, email from
+ customers'),
+ ReadOperation.query(
+ sql='Select * from users where id <= @user_id',
+ params={'user_id': 100},
+ params_type={'user_id': param_types.INT64}
+ ),
+ ]
+ # `params_types` are instance of `google.cloud.spanner.param_types`
+ all_users = pipeline | ReadFromSpanner(PROJECT_ID, INSTANCE_ID, DB_NAME,
+ read_operations=read_operations)
+
+For more information, please review the docs on class ReadOperation.
+
+User can also able to provide the ReadOperation in form of PCollection via
+pipeline. For example:::
+
+ users = (pipeline
+ | beam.Create([ReadOperation...])
+ | ReadFromSpanner(PROJECT_ID, INSTANCE_ID, DB_NAME))
+
+User may also create cloud spanner transaction from the transform called
+`create_transaction` which is available in the SpannerIO API.
+
+The transform is guaranteed to be executed on a consistent snapshot of data,
+utilizing the power of read only transactions. Staleness of data can be
+controlled by providing the `read_timestamp` or `exact_staleness` param values
+in the constructor.
+
+This transform requires root of the pipeline (PBegin) and returns PTransform
+which is passed later to the `ReadFromSpanner` constructor. `ReadFromSpanner`
+pass this transaction PTransform as a singleton side input to the
+`_NaiveSpannerReadDoFn` containing 'session_id' and 'transaction_id'.
+For example:::
+
+ transaction = (pipeline | create_transaction(TEST_PROJECT_ID,
+ TEST_INSTANCE_ID,
+ DB_NAME))
+
+ users = pipeline | ReadFromSpanner(PROJECT_ID, INSTANCE_ID, DB_NAME,
+ sql='Select * from users', transaction=transaction)
+
+ tweets = pipeline | ReadFromSpanner(PROJECT_ID, INSTANCE_ID, DB_NAME,
+ sql='Select * from tweets', transaction=transaction)
+
+For further details of this transform, please review the docs on the
+:meth:`create_transaction` method available in the SpannerIO API.
+
+ReadFromSpanner takes this transform in the constructor and pass this to the
+read pipeline as the singleton side input.
+"""
+from __future__ import absolute_import
+
+import typing
+from collections import namedtuple
+
+from apache_beam import Create
+from apache_beam import DoFn
+from apache_beam import ParDo
+from apache_beam import Reshuffle
+from apache_beam.pvalue import AsSingleton
+from apache_beam.pvalue import PBegin
+from apache_beam.transforms import PTransform
+from apache_beam.transforms import ptransform_fn
+from apache_beam.transforms.display import DisplayDataItem
+from apache_beam.typehints import with_input_types
+from apache_beam.typehints import with_output_types
+from apache_beam.utils.annotations import experimental
+
+try:
+ from google.cloud.spanner import Client
+ from google.cloud.spanner import KeySet
+ from google.cloud.spanner_v1.database import BatchSnapshot
+except ImportError:
+ Client = None
+ KeySet = None
+ BatchSnapshot = None
+
+__all__ = ['create_transaction', 'ReadFromSpanner', 'ReadOperation']
+
+
+class ReadOperation(namedtuple("ReadOperation", ["is_sql", "is_table",
+ "read_operation", "kwargs"])):
+ """
+ Encapsulates a spanner read operation.
+ """
+
+ __slots__ = ()
+
+ @classmethod
+ def query(cls, sql, params=None, param_types=None):
+ """
+ A convenient method to construct ReadOperation from sql query.
+
+ Args:
+ sql: SQL query statement
+ params: (optional) values for parameter replacement. Keys must match the
+ names used in sql
+ param_types: (optional) maps explicit types for one or more param values;
+ required if parameters are passed.
+ """
+
+ if params:
+ assert param_types is not None
+
+ return cls(
+ is_sql=True,
+ is_table=False,
+ read_operation="process_query_batch",
+ kwargs={'sql': sql, 'params': params, 'param_types': param_types}
+ )
+
+ @classmethod
+ def table(cls, table, columns, index="", keyset=None):
+ """
+ A convenient method to construct ReadOperation from table.
+
+ Args:
+ table: name of the table from which to fetch data.
+ columns: names of columns to be retrieved.
+ index: (optional) name of index to use, rather than the table's primary
+ key.
+ keyset: (optional) `KeySet` keys / ranges identifying rows to be
+ retrieved.
+ """
+ keyset = keyset or KeySet(all_=True)
+ if not isinstance(keyset, KeySet):
+ raise ValueError("keyset must be an instance of class "
+ "google.cloud.spanner.KeySet")
+ return cls(
+ is_sql=False,
+ is_table=True,
+ read_operation="process_read_batch",
+ kwargs={'table': table, 'columns': columns, 'index': index,
+ 'keyset': keyset}
+ )
+
+
+class _BeamSpannerConfiguration(namedtuple(
+ "_BeamSpannerConfiguration", ["project", "instance", "database",
+ "credentials", "pool",
+ "snapshot_read_timestamp",
+ "snapshot_exact_staleness"])):
+ """
+ A namedtuple holds the immutable data of the connection string to the cloud
+ spanner.
+ """
+
+ @property
+ def snapshot_options(self):
+ snapshot_options = {}
+ if self.snapshot_exact_staleness:
+ snapshot_options['exact_staleness'] = self.snapshot_exact_staleness
+ if self.snapshot_read_timestamp:
+ snapshot_options['read_timestamp'] = self.snapshot_read_timestamp
+ return snapshot_options
+
+@with_input_types(ReadOperation, typing.Dict[typing.Any, typing.Any])
+@with_output_types(typing.List[typing.Any])
+class _NaiveSpannerReadDoFn(DoFn):
+
+ def __init__(self, spanner_configuration):
+ """
+ A naive version of Spanner read which uses the transaction API of the
+ cloud spanner.
+ https://googleapis.dev/python/spanner/latest/transaction-api.html
+ In Naive reads, this transform performs single reads, where as the
+ Batch reads use the spanner partitioning query to create batches.
+
+ Args:
+ spanner_configuration: (_BeamSpannerConfiguration) Connection details to
+ connect with cloud spanner.
+ """
+ self._spanner_configuration = spanner_configuration
+ self._snapshot = None
+ self._session = None
+
+ def _get_session(self):
+ if self._session is None:
+ session = self._session = self._database.session()
+ session.create()
+ return self._session
+
+ def _close_session(self):
+ if self._session is not None:
+ self._session.delete()
+
+ def setup(self):
+ # setting up client to connect with cloud spanner
+ spanner_client = Client(self._spanner_configuration.project)
+ instance = spanner_client.instance(self._spanner_configuration.instance)
+ self._database = instance.database(self._spanner_configuration.database,
+ pool=self._spanner_configuration.pool)
+
+ def process(self, element, transaction_info):
+ # We used batch snapshot to reuse the same transaction passed through the
+ # side input
+ self._snapshot = BatchSnapshot.from_dict(self._database, transaction_info)
Review comment:
How can we make sure that what's passed by transaction_info is consistent
with what BatchSnapshot.from_dict() expects ? (for example, index). Can we
introduce some sort of validation before this call ?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
Issue Time Tracking
-------------------
Worklog Id: (was: 367169)
Time Spent: 10.5h (was: 10h 20m)
> Create a Spanner IO for Python
> ------------------------------
>
> Key: BEAM-7246
> URL: https://issues.apache.org/jira/browse/BEAM-7246
> Project: Beam
> Issue Type: Bug
> Components: io-py-gcp
> Reporter: Reuven Lax
> Assignee: Shehzaad Nakhoda
> Priority: Major
> Time Spent: 10.5h
> Remaining Estimate: 0h
>
> Add I/O support for Google Cloud Spanner for the Python SDK (Batch Only).
> Testing in this work item will be in the form of DirectRunner tests and
> manual testing.
> Integration and performance tests are a separate work item (not included
> here).
> See https://beam.apache.org/documentation/io/built-in/. The goal is to add
> Google Clound Spanner to the Database column for the Python/Batch row.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
