Coverage for pyspark/sql/observation.py : 97%

# 

# 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. 

# 

from pyspark.sql import column 

from pyspark.sql.column import Column 

from pyspark.sql.dataframe import DataFrame 

__all__ = ["Observation"] 

class Observation: 

"""Class to observe (named) metrics on a :class:`DataFrame`. 

Metrics are aggregation expressions, which are applied to the DataFrame while is is being 

processed by an action. 

The metrics have the following guarantees: 

- It will compute the defined aggregates (metrics) on all the data that is flowing through 

the Dataset during the action. 

- It will report the value of the defined aggregate columns as soon as we reach the end of 

the action. 

The metrics columns must either contain a literal (e.g. lit(42)), or should contain one or 

more aggregate functions (e.g. sum(a) or sum(a + b) + avg(c) - lit(1)). Expressions that 

contain references to the input Dataset's columns must always be wrapped in an aggregate 

function. 

An Observation instance collects the metrics while the first action is executed. Subsequent 

actions do not modify the metrics returned by `Observation.get`. Retrieval of the metric via 

`Observation.get` blocks until the first action has finished and metrics become available. 

.. versionadded:: 3.3.0 

Notes 

----- 

This class does not support streaming datasets. 

Examples 

-------- 

>>> from pyspark.sql.functions import col, count, lit, max 

>>> from pyspark.sql import Observation 

>>> df = spark.createDataFrame([["Alice", 2], ["Bob", 5]], ["name", "age"]) 

>>> observation = Observation("my metrics") 

>>> observed_df = df.observe(observation, count(lit(1)).alias("count"), max(col("age"))) 

>>> observed_df.count() 

2 

>>> observation.get 

{'count': 2, 'max(age)': 5} 

""" 

def __init__(self, name=None): 

"""Constructs a named or unnamed Observation instance. 

Parameters 

---------- 

name : str, optional 

default is a random UUID string. This is the name of the Observation and the metric. 

""" 

if name is not None: 

if not isinstance(name, str): 

raise TypeError("name should be a string") 

if name == '': 

raise ValueError("name should not be empty") 

self._name = name 

self._jvm = None 

self._jo = None 

def _on(self, df, *exprs): 

"""Attaches this observation to the given :class:`DataFrame` to observe aggregations. 

Parameters 

---------- 

df : :class:`DataFrame` 

the :class:`DataFrame` to be observed 

exprs : list of :class:`Column` 

column expressions (:class:`Column`). 

Returns 

------- 

:class:`DataFrame` 

the observed :class:`DataFrame`. 

""" 

assert exprs, "exprs should not be empty" 

assert all(isinstance(c, Column) for c in exprs), "all exprs should be Column" 

assert self._jo is None, "an Observation can be used with a DataFrame only once" 

self._jvm = df._sc._jvm 

cls = self._jvm.org.apache.spark.sql.Observation 

self._jo = cls(self._name) if self._name is not None else cls() 

observed_df = self._jo.on(df._jdf, 

exprs[0]._jc, 

column._to_seq(df._sc, [c._jc for c in exprs[1:]])) 

return DataFrame(observed_df, df.sql_ctx) 

@property 

def get(self): 

"""Get the observed metrics. 

Waits until the observed dataset finishes its first action. Only the result of the 

first action is available. Subsequent actions do not modify the result. 

Returns 

------- 

dict 

the observed metrics 

""" 

assert self._jo is not None, 'call DataFrame.observe' 

jmap = self._jo.getAsJava() 

# return a pure Python dict, not jmap which is a py4j JavaMap 

return {k: v for k, v in jmap.items()} 

def _test(): 

import doctest 

import sys 

from pyspark.context import SparkContext 

from pyspark.sql import SparkSession 

import pyspark.sql.observation 

globs = pyspark.sql.observation.__dict__.copy() 

sc = SparkContext('local[4]', 'PythonTest') 

globs['spark'] = SparkSession(sc) 

(failure_count, test_count) = doctest.testmod(pyspark.sql.observation, globs=globs) 

sc.stop() 

138 ↛ 139line 138 didn't jump to line 139, because the condition on line 138 was never true if failure_count: 

sys.exit(-1) 

if __name__ == "__main__": 

_test()