spark-instrumented-optimizer/python/pyspark
hyukjinkwon d492cc5a21 [SPARK-19507][SPARK-21296][PYTHON] Avoid per-record type dispatch in schema verification and improve exception message
## What changes were proposed in this pull request?
**Context**

While reviewing https://github.com/apache/spark/pull/17227, I realised here we type-dispatch per record. The PR itself is fine in terms of performance as is but this prints a prefix, `"obj"` in exception message as below:

```
from pyspark.sql.types import *
schema = StructType([StructField('s', IntegerType(), nullable=False)])
spark.createDataFrame([["1"]], schema)
...
TypeError: obj.s: IntegerType can not accept object '1' in type <type 'str'>
```

I suggested to get rid of this but during investigating this, I realised my approach might bring a performance regression as it is a hot path.

Only for SPARK-19507 and https://github.com/apache/spark/pull/17227, It needs more changes to cleanly get rid of the prefix and I rather decided to fix both issues together.

**Propersal**

This PR tried to

  - get rid of per-record type dispatch as we do in many code paths in Scala  so that it improves the performance (roughly ~25% improvement) - SPARK-21296

    This was tested with a simple code `spark.createDataFrame(range(1000000), "int")`. However, I am quite sure the actual improvement in practice is larger than this, in particular, when the schema is complicated.

   - improve error message in exception describing field information as prose - SPARK-19507

## How was this patch tested?

Manually tested and unit tests were added in `python/pyspark/sql/tests.py`.

Benchmark - codes: https://gist.github.com/HyukjinKwon/c3397469c56cb26c2d7dd521ed0bc5a3
Error message - codes: https://gist.github.com/HyukjinKwon/b1b2c7f65865444c4a8836435100e398

**Before**

Benchmark:
  - Results: https://gist.github.com/HyukjinKwon/4a291dab45542106301a0c1abcdca924

Error message
  - Results: https://gist.github.com/HyukjinKwon/57b1916395794ce924faa32b14a3fe19

**After**

Benchmark
  - Results: https://gist.github.com/HyukjinKwon/21496feecc4a920e50c4e455f836266e

Error message
  - Results: https://gist.github.com/HyukjinKwon/7a494e4557fe32a652ce1236e504a395

Closes #17227

Author: hyukjinkwon <gurwls223@gmail.com>
Author: David Gingrich <david@textio.com>

Closes #18521 from HyukjinKwon/python-type-dispatch.
2017-07-04 20:45:58 +08:00
..
ml [SPARK-19852][PYSPARK][ML] Python StringIndexer supports 'keep' to handle invalid data 2017-07-02 16:17:03 +08:00
mllib [SPARK-20862][MLLIB][PYTHON] Avoid passing float to ndarray.reshape in LogisticRegressionModel 2017-05-24 22:55:38 +08:00
sql [SPARK-19507][SPARK-21296][PYTHON] Avoid per-record type dispatch in schema verification and improve exception message 2017-07-04 20:45:58 +08:00
streaming [SPARK-20285][TESTS] Increase the pyspark streaming test timeout to 30 seconds 2017-04-10 14:06:49 -07:00
__init__.py [MINOR] Fix some typo of the document 2017-06-19 20:35:58 +01:00
accumulators.py [SPARK-8652] [PYSPARK] Check return value for all uses of doctest.testmod() 2015-06-26 08:12:22 -07:00
broadcast.py [SPARK-19505][PYTHON] AttributeError on Exception.message in Python3 2017-04-11 12:18:31 -07:00
cloudpickle.py [SPARK-19505][PYTHON] AttributeError on Exception.message in Python3 2017-04-11 12:18:31 -07:00
conf.py [SPARK-18447][DOCS] Fix the markdown for Note:/NOTE:/Note that across Python API documentation 2016-11-22 11:40:18 +00:00
context.py [SPARK-21125][PYTHON] Extend setJobDescription to PySpark and JavaSpark APIs 2017-06-21 10:51:45 -07:00
daemon.py [SPARK-4897] [PySpark] Python 3 support 2015-04-16 16:20:57 -07:00
files.py [SPARK-3309] [PySpark] Put all public API in __all__ 2014-09-03 11:49:45 -07:00
find_spark_home.py [SPARK-1267][SPARK-18129] Allow PySpark to be pip installed 2016-11-16 14:22:15 -08:00
heapq3.py [SPARK-8652] [PYSPARK] Check return value for all uses of doctest.testmod() 2015-06-26 08:12:22 -07:00
java_gateway.py [SPARK-1267][SPARK-18129] Allow PySpark to be pip installed 2016-11-16 14:22:15 -08:00
join.py [SPARK-14202] [PYTHON] Use generator expression instead of list comp in python_full_outer_jo… 2016-03-28 14:51:36 -07:00
profiler.py [SPARK-8652] [PYSPARK] Check return value for all uses of doctest.testmod() 2015-06-26 08:12:22 -07:00
rdd.py [SPARK-19507][SPARK-21296][PYTHON] Avoid per-record type dispatch in schema verification and improve exception message 2017-07-04 20:45:58 +08:00
rddsampler.py [SPARK-4897] [PySpark] Python 3 support 2015-04-16 16:20:57 -07:00
resultiterable.py [SPARK-3074] [PySpark] support groupByKey() with single huge key 2015-04-09 17:07:23 -07:00
serializers.py Revert "[SPARK-13534][PYSPARK] Using Apache Arrow to increase performance of DataFrame.toPandas" 2017-06-28 14:28:40 +08:00
shell.py [SPARK-19570][PYSPARK] Allow to disable hive in pyspark shell 2017-04-12 10:54:50 -07:00
shuffle.py [SPARK-10710] Remove ability to disable spilling in core and SQL 2015-09-19 21:40:21 -07:00
statcounter.py [SPARK-6919] [PYSPARK] Add asDict method to StatCounter 2015-09-29 13:38:15 -07:00
status.py [SPARK-4172] [PySpark] Progress API in Python 2015-02-17 13:36:43 -08:00
storagelevel.py [SPARK-13992][CORE][PYSPARK][FOLLOWUP] Update OFF_HEAP semantics for Java api and Python api 2016-04-12 23:06:55 -07:00
taskcontext.py [SPARK-18576][PYTHON] Add basic TaskContext information to PySpark 2016-12-20 15:51:21 -08:00
tests.py [SPARK-19872] [PYTHON] Use the correct deserializer for RDD construction for coalesce/repartition 2017-03-15 10:17:18 -07:00
traceback_utils.py [SPARK-1087] Move python traceback utilities into new traceback_utils.py file. 2014-09-15 19:28:17 -07:00
util.py [SPARK-19505][PYTHON] AttributeError on Exception.message in Python3 2017-04-11 12:18:31 -07:00
version.py [MINOR] Bump SparkR and PySpark version to 2.3.0. 2017-06-19 11:13:03 +01:00
worker.py [SPARK-20685] Fix BatchPythonEvaluation bug in case of single UDF w/ repeated arg. 2017-05-10 16:50:57 -07:00