HyukjinKwon commented on a change in pull request #27109: [SPARK-30434][PYTHON][SQL] Move pandas related functionalities into 'pandas' sub-package URL: https://github.com/apache/spark/pull/27109#discussion_r367208121
########## File path: python/pyspark/sql/pandas/conversion.py ########## @@ -0,0 +1,431 @@ +# +# 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. +# +import sys +import warnings +if sys.version >= '3': + basestring = unicode = str + xrange = range +else: + from itertools import izip as zip + +from pyspark import since +from pyspark.rdd import _load_from_socket +from pyspark.sql.pandas.serializers import ArrowCollectSerializer +from pyspark.sql.types import IntegralType +from pyspark.sql.types import * +from pyspark.traceback_utils import SCCallSiteSync +from pyspark.util import _exception_message + + +class PandasConversionMixin(object): + """ + Min-in for the conversion from Spark to pandas. Currently, only :class:`DataFrame` + can use this class. + """ + + @since(1.3) + def toPandas(self): + """ + Returns the contents of this :class:`DataFrame` as Pandas ``pandas.DataFrame``. + + This is only available if Pandas is installed and available. + + .. note:: This method should only be used if the resulting Pandas's :class:`DataFrame` is + expected to be small, as all the data is loaded into the driver's memory. + + .. note:: Usage with spark.sql.execution.arrow.pyspark.enabled=True is experimental. + + >>> df.toPandas() # doctest: +SKIP + age name + 0 2 Alice + 1 5 Bob + """ + from pyspark.sql.dataframe import DataFrame + + assert isinstance(self, DataFrame) + + from pyspark.sql.pandas.utils import require_minimum_pandas_version + require_minimum_pandas_version() + + import numpy as np + import pandas as pd + + if self.sql_ctx._conf.pandasRespectSessionTimeZone(): + timezone = self.sql_ctx._conf.sessionLocalTimeZone() + else: + timezone = None + + if self.sql_ctx._conf.arrowPySparkEnabled(): + use_arrow = True + try: + from pyspark.sql.pandas.types import to_arrow_schema + from pyspark.sql.pandas.utils import require_minimum_pyarrow_version + + require_minimum_pyarrow_version() + to_arrow_schema(self.schema) + except Exception as e: + + if self.sql_ctx._conf.arrowPySparkFallbackEnabled(): + msg = ( + "toPandas attempted Arrow optimization because " + "'spark.sql.execution.arrow.pyspark.enabled' is set to true; however, " + "failed by the reason below:\n %s\n" + "Attempting non-optimization as " + "'spark.sql.execution.arrow.pyspark.fallback.enabled' is set to " + "true." % _exception_message(e)) + warnings.warn(msg) + use_arrow = False + else: + msg = ( + "toPandas attempted Arrow optimization because " + "'spark.sql.execution.arrow.pyspark.enabled' is set to true, but has " + "reached the error below and will not continue because automatic fallback " + "with 'spark.sql.execution.arrow.pyspark.fallback.enabled' has been set to " + "false.\n %s" % _exception_message(e)) + warnings.warn(msg) + raise + + # Try to use Arrow optimization when the schema is supported and the required version + # of PyArrow is found, if 'spark.sql.execution.arrow.pyspark.enabled' is enabled. + if use_arrow: + try: + from pyspark.sql.pandas.types import _check_dataframe_localize_timestamps + import pyarrow + batches = self._collect_as_arrow() + if len(batches) > 0: + table = pyarrow.Table.from_batches(batches) + # Pandas DataFrame created from PyArrow uses datetime64[ns] for date type + # values, but we should use datetime.date to match the behavior with when + # Arrow optimization is disabled. + pdf = table.to_pandas(date_as_object=True) + return _check_dataframe_localize_timestamps(pdf, timezone) + else: + return pd.DataFrame.from_records([], columns=self.columns) + except Exception as e: + # We might have to allow fallback here as well but multiple Spark jobs can + # be executed. So, simply fail in this case for now. + msg = ( + "toPandas attempted Arrow optimization because " + "'spark.sql.execution.arrow.pyspark.enabled' is set to true, but has " + "reached the error below and can not continue. Note that " + "'spark.sql.execution.arrow.pyspark.fallback.enabled' does not have an " + "effect on failures in the middle of " + "computation.\n %s" % _exception_message(e)) + warnings.warn(msg) + raise + + # Below is toPandas without Arrow optimization. + pdf = pd.DataFrame.from_records(self.collect(), columns=self.columns) + + dtype = {} + for field in self.schema: + pandas_type = PandasConversionMixin._to_corrected_pandas_type(field.dataType) + # SPARK-21766: if an integer field is nullable and has null values, it can be + # inferred by pandas as float column. Once we convert the column with NaN back + # to integer type e.g., np.int16, we will hit exception. So we use the inferred + # float type, not the corrected type from the schema in this case. + if pandas_type is not None and \ + not(isinstance(field.dataType, IntegralType) and field.nullable and + pdf[field.name].isnull().any()): + dtype[field.name] = pandas_type + # Ensure we fall back to nullable numpy types, even when whole column is null: + if isinstance(field.dataType, IntegralType) and pdf[field.name].isnull().any(): + dtype[field.name] = np.float64 + if isinstance(field.dataType, BooleanType) and pdf[field.name].isnull().any(): + dtype[field.name] = np.object + + for f, t in dtype.items(): + pdf[f] = pdf[f].astype(t, copy=False) + + if timezone is None: + return pdf + else: + from pyspark.sql.pandas.types import _check_series_convert_timestamps_local_tz + for field in self.schema: + # TODO: handle nested timestamps, such as ArrayType(TimestampType())? + if isinstance(field.dataType, TimestampType): + pdf[field.name] = \ + _check_series_convert_timestamps_local_tz(pdf[field.name], timezone) + return pdf + + @staticmethod + def _to_corrected_pandas_type(dt): + """ + When converting Spark SQL records to Pandas :class:`DataFrame`, the inferred data type + may be wrong. This method gets the corrected data type for Pandas if that type may be + inferred incorrectly. + """ + import numpy as np + if type(dt) == ByteType: + return np.int8 + elif type(dt) == ShortType: + return np.int16 + elif type(dt) == IntegerType: + return np.int32 + elif type(dt) == LongType: + return np.int64 + elif type(dt) == FloatType: + return np.float32 + elif type(dt) == DoubleType: + return np.float64 + elif type(dt) == BooleanType: + return np.bool + elif type(dt) == TimestampType: + return np.datetime64 + else: + return None + + def _collect_as_arrow(self): + """ + Returns all records as a list of ArrowRecordBatches, pyarrow must be installed + and available on driver and worker Python environments. + + .. note:: Experimental. + """ + from pyspark.sql.dataframe import DataFrame + + assert isinstance(self, DataFrame) + + with SCCallSiteSync(self._sc): + port, auth_secret, jsocket_auth_server = self._jdf.collectAsArrowToPython() + + # Collect list of un-ordered batches where last element is a list of correct order indices + try: + results = list(_load_from_socket((port, auth_secret), ArrowCollectSerializer())) + finally: + # Join serving thread and raise any exceptions from collectAsArrowToPython + jsocket_auth_server.getResult() + + # Separate RecordBatches from batch order indices in results + batches = results[:-1] + batch_order = results[-1] + + # Re-order the batch list using the correct order + return [batches[i] for i in batch_order] + + +class SparkConversionMixin(object): + """ + Min-in for the conversion from pandas to Spark. Currently, only :class:`SparkSession` + can use this class. + """ + def createDataFrame(self, data, schema=None, samplingRatio=None, verifySchema=True): Review comment: I did the current way so that `SparkConversionMixin.createDataFrame` can work alone by itself. I didn't like the second way because it's weird to return both `schema` and `data`. ---------------------------------------------------------------- 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: [email protected] With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
