Python native implementation of the Spark RDD interface.
A native Python implementation of Spark’s RDD interface, but instead of being resilient and distributed it is just transient and local; but fast (lower latency than PySpark). It is a drop in replacement for PySpark’s SparkContext and RDD.
Use case: you have a pipeline that processes 100k input documents and converts them to normalized features. They are used to train a local scikit-learn classifier. The preprocessing is perfect for a full Spark task. Now, you want to use this trained classifier in an API endpoint. You need the same pre-processing pipeline for a single document per API call. This does not have to be done in parallel, but there should be only a small overhead in initialization and preferably no dependency on the JVM. This is what pysparkling is for.
pip install pysparkling
- Parallelization via multiprocessing.Pool, concurrent.futures.ThreadPoolExecutor or any other Pool-like objects that have a map(func, iterable) method.
- AWS S3 is supported. Use file paths of the form s3n://bucket_name/filename.txt with Context.textFile(). Specify multiple files separated by comma. Use environment variables AWS_SECRET_ACCESS_KEY and AWS_ACCESS_KEY_ID for auth. Mixed local and S3 files are supported. Glob expressions (filenames with * and ?) are resolved.
- Lazy execution is in development.
- Seamlessly handles .gz and .bz2 compressed files.
- only dependency: boto for AWS S3 access
Count the lines in the *.py files in the tests directory:
import pysparkling context = pysparkling.Context() print(context.textFile('tests/*.py').count())
- __init__(pool=None, serializer=None, deserializer=None, data_serializer=None, data_deserializer=None): takes a pool object (an object that has a map() method, e.g. a multiprocessing.Pool) to parallelize methods. To support functions and lambda functions, specify custom serializers and deserializers, e.g. serializer=dill.dumps, deserializer=dill.loads.
- broadcast(var): returns an instance of Broadcast() and it’s values are accessed with value.
- newRddId(): incrementing number
- textFile(filename): load every line of a text file into a RDD. filename can contain a comma separated list of many files, ? and * wildcards, file paths on S3 (s3n://bucket_name/filename.txt) and local file paths (relative/path/my_text.txt, /absolut/path/my_text.txt or file:///absolute/file/path.txt). If the filename points to a folder containing part* files, those are resolved.
- version: the version of pysparkling
- aggregate(zeroValue, seqOp, combOp): aggregate value in partition with seqOp and combine with combOp
- aggregateByKey(zeroValue, seqFunc, combFunc): aggregate by key
- cache(): execute previous steps and cache result
- cartesian(other): cartesian product
- coalesce(): do nothing
- collect(): return the underlying list
- count(): get length of internal list
- countApprox(): same as count()
- countByKey: input is list of pairs, returns a dictionary
- countByValue: input is a list, returns a dictionary
- context(): return the context
- distinct(): returns a new RDD containing the distinct elements
- filter(func): return new RDD filtered with func
- first(): return first element
- flatMap(func): return a new RDD of a flattened map
- flatMapValues(func): return new RDD
- fold(zeroValue, op): aggregate elements
- foldByKey(zeroValue, op): aggregate elements by key
- foreach(func): apply func to every element in place
- foreachPartition(func): same as foreach()
- getNumPartitions(): number of partitions
- getPartitions(): returns an iterator over the partitions
- groupBy(func): group by the output of func
- groupByKey(): group by key where the RDD is of type [(key, value), …]
- histogram(buckets): buckets can be a list or an int
- id(): currently just returns None
- intersection(other): return a new RDD with the intersection
- isCheckpointed(): returns False
- join(other): join
- keyBy(func): creates tuple in new RDD
- keys(): returns the keys of tuples in new RDD
- leftOuterJoin(other): left outer join
- lookup(key): return list of values for this key
- map(func): apply func to every element and return a new RDD
- mapPartitions(func): apply f to entire partitions
- mapValues(func): apply func to value in (key, value) pairs and return a new RDD
- max(): get the maximum element
- mean(): mean
- min(): get the minimum element
- name(): RDD’s name
- persist(): implemented as synonym for cache()
- pipe(command): pipe the elements through an external command line tool
- reduce(): reduce
- reduceByKey(): reduce by key and return the new RDD
- rightOuterJoin(other): right outer join
- saveAsTextFile(path): save RDD as text file
- subtract(other): return a new RDD without the elements in other
- sum(): sum
- take(n): get the first n elements
- takeSample(n): get n random samples
- value: access the value it stores
The functionality provided by this module is used in Context.textFile() for reading and in RDD.saveAsTextFile() for writing. Normally, you should not have to use this submodule directly.
Infers .gz and .bz2 compressions from the file name.
- File: abstract. Only contains static methods.
- exists(path): check for existance of path
- path_type(path): returns s3, http or local
- resolve_filenames(expr): given a glob-like expression with * and ?, get a list of all matching filenames (either locally or on S3).
- WholeFile(file_name): file_name is either local or S3 or …
- load(): return the contents as BytesIO
- dump(stream): write the stream to the file
- make_public(recursive=False): only for files on S3
- v0.2.5 (2015-05-20)
- improved handling of compressed files (backwards incompatible)
- v0.2.4 (2015-05-19)
- fileio interface changed to dump() and load() methods. Added make_public() for S3.
- v0.2.3 (2015-05-19)
- factor file related operations into fileio submodule
- v0.2.2 (2015-05-18)
- compressions: .gz, .bz2
- v0.2.0 (2015-05-17)
- proper handling of partitions
- custom serializers, deserializers (for functions and data separately)
- more tests for parallelization options
- execution of distributed jobs is such that a chain of map() operations gets executed on workers without sending intermediate results back to the master
- a few more methods for RDDs implemented
- v0.1.1 (2015-05-12)
- implemented a few more RDD methods
- changed handling of context in RDD
- v0.1.0 (2015-05-09)