# Copyright 2020 QuantumBlack Visual Analytics Limited
#
# Licensed 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
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
# EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES
# OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, AND
# NONINFRINGEMENT. IN NO EVENT WILL THE LICENSOR OR OTHER CONTRIBUTORS
# BE LIABLE FOR ANY CLAIM, DAMAGES, OR OTHER LIABILITY, WHETHER IN AN
# ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF, OR IN
# CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
#
# The QuantumBlack Visual Analytics Limited ("QuantumBlack") name and logo
# (either separately or in combination, "QuantumBlack Trademarks") are
# trademarks of QuantumBlack. The License does not grant you any right or
# license to the QuantumBlack Trademarks. You may not use the QuantumBlack
# Trademarks or any confusingly similar mark as a trademark for your product,
# or use the QuantumBlack Trademarks in any other manner that might cause
# confusion in the marketplace, including but not limited to in advertising,
# on websites, or on software.
#
# See the License for the specific language governing permissions and
# limitations under the License.
"""``AbstractDataSet`` implementation to access Spark dataframes using
``pyspark`` on Apache Hive.
"""
import pickle
import uuid
from typing import Any, Dict, List
from pyspark.sql import DataFrame, SparkSession # pylint: disable=import-error
from pyspark.sql.functions import ( # pylint: disable=import-error,no-name-in-module
coalesce,
col,
lit,
)
from kedro.io.core import AbstractDataSet, DataSetError
class StagedHiveDataSet:
"""
Provides a context manager for temporarily writing data to a staging hive table, for example
where you want to replace the contents of a hive table with data which relies on the data
currently present in that table.
Once initialised, the ``staged_data`` ``DataFrame`` can be queried and underlying tables used to
define the initial dataframe can be modified without affecting ``staged_data``.
Upon exiting this object it will drop the redundant staged table.
"""
def __init__(
self, data: DataFrame, stage_table_name: str, stage_database_name: str
):
"""
Creates a new instance eof `StagedHiveDataSet`.
Args:
data: The spark dataframe to be staged
stage_table_name: the database destination for the staged data
stage_database_name: the table destination for the staged data
"""
self.staged_data = None
self._data = data
self._stage_table_name = stage_table_name
self._stage_database_name = stage_database_name
self._spark_session = SparkSession.builder.getOrCreate()
def __enter__(self):
self._data.createOrReplaceTempView("tmp")
_table = f"{self._stage_database_name}.{self._stage_table_name}"
self._spark_session.sql(
f"create table {_table} as select * from tmp" # nosec
).take(1)
self.staged_data = self._spark_session.sql(f"select * from {_table}") # nosec
return self
def __exit__(self, exc_type, exc_val, exc_tb):
self._spark_session.sql(
f"drop table {self._stage_database_name}.{self._stage_table_name}" # nosec
)
[docs]class SparkHiveDataSet(AbstractDataSet):
"""``SparkHiveDataSet`` loads and saves Spark dataframes stored on Hive.
This data set also handles some incompatible file types such as using partitioned parquet on
hive which will not normally allow upserts to existing data without a complete replacement
of the existing file/partition.
This DataSet has some key assumptions:
- Schemas do not change during the pipeline run (defined PKs must be present for the
duration of the pipeline)
- Tables are not being externally modified during upserts. The upsert method is NOT ATOMIC
to external changes to the target table while executing.
Example:
::
>>> from pyspark.sql import SparkSession
>>> from pyspark.sql.types import (StructField, StringType,
>>> IntegerType, StructType)
>>>
>>> from kedro.extras.datasets.spark import SparkHiveDataSet
>>>
>>> schema = StructType([StructField("name", StringType(), True),
>>> StructField("age", IntegerType(), True)])
>>>
>>> data = [('Alex', 31), ('Bob', 12), ('Clarke', 65), ('Dave', 29)]
>>>
>>> spark_df = SparkSession.builder.getOrCreate().createDataFrame(data, schema)
>>>
>>> data_set = SparkHiveDataSet(database="test_database", table="test_table",
>>> write_mode="overwrite")
>>> data_set.save(spark_df)
>>> reloaded = data_set.load()
>>>
>>> reloaded.take(4)
"""
[docs] def __init__(
self, database: str, table: str, write_mode: str, table_pk: List[str] = None
) -> None:
"""Creates a new instance of ``SparkHiveDataSet``.
Args:
database: The name of the hive database.
table: The name of the table within the database.
write_mode: ``insert``, ``upsert`` or ``overwrite`` are supported.
table_pk: If performing an upsert, this identifies the primary key columns used to
resolve preexisting data. Is required for ``write_mode="upsert"``.
Raises:
DataSetError: Invalid configuration supplied
"""
valid_write_modes = ["insert", "upsert", "overwrite"]
if write_mode not in valid_write_modes:
valid_modes = ", ".join(valid_write_modes)
raise DataSetError(
f"Invalid `write_mode` provided: {write_mode}. "
f"`write_mode` must be one of: {valid_modes}"
)
if write_mode == "upsert" and not table_pk:
raise DataSetError("`table_pk` must be set to utilise `upsert` read mode")
self._write_mode = write_mode
self._table_pk = table_pk or []
self._database = database
self._table = table
self._stage_table = "_temp_" + table
# self._table_columns is set up in _save() to speed up initialization
self._table_columns = [] # type: List[str]
def _describe(self) -> Dict[str, Any]:
return dict(
database=self._database,
table=self._table,
write_mode=self._write_mode,
table_pk=self._table_pk,
)
@staticmethod
def _get_spark() -> SparkSession:
return SparkSession.builder.getOrCreate()
def _create_empty_hive_table(self, data):
data.createOrReplaceTempView("tmp")
self._get_spark().sql(
f"create table {self._database}.{self._table} select * from tmp limit 1" # nosec
)
self._get_spark().sql(f"truncate table {self._database}.{self._table}") # nosec
def _load(self) -> DataFrame:
if not self._exists():
raise DataSetError(
f"Requested table not found: {self._database}.{self._table}"
)
return self._get_spark().sql(
f"select * from {self._database}.{self._table}" # nosec
)
def _save(self, data: DataFrame) -> None:
if not self._exists():
self._create_empty_hive_table(data)
self._table_columns = data.columns
else:
self._table_columns = self._load().columns
if self._write_mode == "upsert":
non_existent_columns = set(self._table_pk) - set(self._table_columns)
if non_existent_columns:
colnames = ", ".join(sorted(non_existent_columns))
raise DataSetError(
f"Columns [{colnames}] selected as primary key(s) not found in "
f"table {self._database}.{self._table}"
)
self._validate_save(data)
write_methods = {
"insert": self._insert_save,
"upsert": self._upsert_save,
"overwrite": self._overwrite_save,
}
write_methods[self._write_mode](data)
def _insert_save(self, data: DataFrame) -> None:
data.createOrReplaceTempView("tmp")
columns = ", ".join(self._table_columns)
self._get_spark().sql(
f"insert into {self._database}.{self._table} select {columns} from tmp" # nosec
)
def _upsert_save(self, data: DataFrame) -> None:
if self._load().rdd.isEmpty():
self._insert_save(data)
else:
joined_data = data.alias("new").join(
self._load().alias("old"), self._table_pk, "outer"
)
upsert_dataset = joined_data.select(
[ # type: ignore
coalesce(f"new.{col_name}", f"old.{col_name}").alias(col_name)
for col_name in set(data.columns)
- set(self._table_pk) # type: ignore
]
+ self._table_pk
)
temporary_persisted_tbl_name = f"temp_{uuid.uuid4().int}"
with StagedHiveDataSet(
upsert_dataset,
stage_database_name=self._database,
stage_table_name=temporary_persisted_tbl_name,
) as temp_table:
self._overwrite_save(temp_table.staged_data)
def _overwrite_save(self, data: DataFrame) -> None:
self._get_spark().sql(f"truncate table {self._database}.{self._table}") # nosec
self._insert_save(data)
def _validate_save(self, data: DataFrame):
hive_dtypes = set(self._load().dtypes)
data_dtypes = set(data.dtypes)
if data_dtypes != hive_dtypes:
new_cols = data_dtypes - hive_dtypes
missing_cols = hive_dtypes - data_dtypes
raise DataSetError(
f"Dataset does not match hive table schema.\n"
f"Present on insert only: {sorted(new_cols)}\n"
f"Present on schema only: {sorted(missing_cols)}"
)
def _exists(self) -> bool:
if (
self._get_spark()
.sql("show databases")
.filter(col("databaseName") == lit(self._database))
.take(1)
):
self._get_spark().sql(f"use {self._database}")
if (
self._get_spark()
.sql("show tables")
.filter(col("tableName") == lit(self._table))
.take(1)
):
return True
return False
def __getstate__(self) -> None:
raise pickle.PicklingError("PySpark datasets can't be serialized")