PySpark Utils

The missing PySpark utils.

Usage

To install:

pip install pyspark-utils

helper

import pyspark_utils.helper as spark_helper

# Nicely show rdd count and 3 items.
rdd = spark_helper.cache_and_log('MyRDD', rdd, 3)

op

import pyspark_utils.op as spark_op

# RDD<key, value>  ->  RDD<new_key, value>
pair_rdd.map(spark_op.do_key(lambda key: new_key))

# RDD<key, value>  ->  RDD<result>
pair_rdd.map(spark_op.do_tuple(lambda key, value: result))

# RDD<key, value>  ->  RDD<value, key>
pair_rdd.map(spark_op.swap_kv())

# RDD<key, value>  ->  RDD<key, value> if func(key)
pair_rdd.filter(spark_op.filter_key(lambda key: true_or_false))

# RDD<key, value>  ->  RDD<key, value> if func(value)
pair_rdd.filter(spark_op.filter_value(lambda value: true_or_false))

# RDD<iteratable>  ->  RDD<tuple_or_list> with transformed values.
rdd.map(spark_op.do_tuple_elems(lambda elem: new_elem))
rdd.map(spark_op.do_list_elems(lambda elem: new_elem))

# RDD<path>  ->  RDD<path> if path matches any given fnmatch-style patterns
rdd.filter(spark_op.filter_path(['*.txt', '*.csv', 'path/a.???']))

# RDD<element>  ->  RDD<element, element>
rdd.keyBy(spark_op.identity)

# RDD<key, value>   ->   RDD<key, value> with keys in key_rdd
spark_op.filter_keys(pair_rdd, key_rdd)

# RDD<key, value>   ->   RDD<key, value> with keys in whitelist and not in blacklist
spark_op.filter_keys(pair_rdd, whitelist_key_rdd, blacklist_key_rdd)

# RDD<key, value>   ->   RDD<key, value> with keys not in key_rdd
spark_op.substract_keys(pair_rdd, key_rdd)

# RDD<element>   ->   RDD<element> where element is not None
rdd.filter(spark_op.not_none)

# RDD<key>   ->   RDD<key, value>
rdd.map(spark_op.value_by(lambda key: value))