[SPARK-22814][SQL] Support Date/Timestamp in a JDBC partition column by maropu · Pull Request #21834 · apache/spark

maropu · 2018-07-21T08:16:14Z

What changes were proposed in this pull request?

This pr supported Date/Timestamp in a JDBC partition column (a numeric column is only supported in the master). This pr also modified code to verify a partition column type;

val jdbcTable = spark.read
 .option("partitionColumn", "text")
 .option("lowerBound", "aaa")
 .option("upperBound", "zzz")
 .option("numPartitions", 2)
 .jdbc("jdbc:postgresql:postgres", "t", options)

// with this pr
org.apache.spark.sql.AnalysisException: Partition column type should be numeric, date, or timestamp, but string found.;
  at org.apache.spark.sql.execution.datasources.jdbc.JDBCRelation$.verifyAndGetNormalizedPartitionColumn(JDBCRelation.scala:165)
  at org.apache.spark.sql.execution.datasources.jdbc.JDBCRelation$.columnPartition(JDBCRelation.scala:85)
  at org.apache.spark.sql.execution.datasources.jdbc.JdbcRelationProvider.createRelation(JdbcRelationProvider.scala:36)
  at org.apache.spark.sql.execution.datasources.DataSource.resolveRelation(DataSource.scala:317)

// without this pr
java.lang.NumberFormatException: For input string: "aaa"
  at java.lang.NumberFormatException.forInputString(NumberFormatException.java:65)
  at java.lang.Long.parseLong(Long.java:589)
  at java.lang.Long.parseLong(Long.java:631)
  at scala.collection.immutable.StringLike$class.toLong(StringLike.scala:277)

Closes #19999

How was this patch tested?

Added tests in JDBCSuite.

SparkQA · 2018-07-21T12:15:48Z

Test build #93381 has finished for PR 21834 at commit 87f9be3.

This patch passes all tests.
This patch merges cleanly.
This patch adds no public classes.

maropu · 2018-07-21T12:22:15Z

@gatorsmile

maropu · 2018-07-24T02:09:09Z

ping

gatorsmile · 2018-07-24T20:02:42Z

+      case LogicalRelation(JDBCRelation(_, parts, _), _, _, _) =>
+        val whereClauses = parts.map(_.asInstanceOf[JDBCPartition].whereClause).toSet
+        assert(whereClauses === Set(
+          """"T" < '2018-07-15 20:50:32.5' or "T" is null""",


Add a test case to OracleIntegrationSuite.scala? I am afraid this does not work.

gatorsmile · 2018-07-25T03:15:19Z

@maropu Do we have a log message for users to know the generated where clauses? If not, could you add one?

maropu · 2018-07-25T03:32:39Z

Currently, no. Is it ok that the log level is INFO?

SparkQA · 2018-07-25T07:05:01Z

Test build #93532 has finished for PR 21834 at commit 1041a38.

This patch fails due to an unknown error code, -9.
This patch merges cleanly.
This patch adds no public classes.

maropu · 2018-07-25T07:11:30Z

retest this please

SparkQA · 2018-07-25T11:45:08Z

Test build #93537 has finished for PR 21834 at commit 1041a38.

This patch passes all tests.
This patch merges cleanly.
This patch adds no public classes.

SparkQA · 2018-07-25T17:59:58Z

Test build #93542 has finished for PR 21834 at commit 577f66e.

This patch passes all tests.
This patch merges cleanly.
This patch adds no public classes.

gatorsmile · 2018-07-30T07:28:32Z

My only concern is the time zone issues. However, it will not affect the correctness since the logical partitioning will cover the whole range anyway. The worse case is the partitioning might not start and end in the expected boundaries due to the local/remote time zone issues, if existed.

Since this is a new feature, we can address the issues and improve the solution in the following releases.

gatorsmile · 2018-07-30T07:28:38Z

retest this please

gatorsmile · 2018-07-30T07:28:47Z

LGTM pending tests

SparkQA · 2018-07-30T12:10:46Z

Test build #93778 has finished for PR 21834 at commit 577f66e.

This patch passes all tests.
This patch merges cleanly.
This patch adds no public classes.

gatorsmile · 2018-07-30T14:41:25Z

Thanks! Merged to master.

maropu · 2018-07-31T02:11:09Z

Thanks for the merge!

## What changes were proposed in this pull request? This pr supported Date/Timestamp in a JDBC partition column (a numeric column is only supported in the master). This pr also modified code to verify a partition column type; ``` val jdbcTable = spark.read .option("partitionColumn", "text") .option("lowerBound", "aaa") .option("upperBound", "zzz") .option("numPartitions", 2) .jdbc("jdbc:postgresql:postgres", "t", options) // with this pr org.apache.spark.sql.AnalysisException: Partition column type should be numeric, date, or timestamp, but string found.; at org.apache.spark.sql.execution.datasources.jdbc.JDBCRelation$.verifyAndGetNormalizedPartitionColumn(JDBCRelation.scala:165) at org.apache.spark.sql.execution.datasources.jdbc.JDBCRelation$.columnPartition(JDBCRelation.scala:85) at org.apache.spark.sql.execution.datasources.jdbc.JdbcRelationProvider.createRelation(JdbcRelationProvider.scala:36) at org.apache.spark.sql.execution.datasources.DataSource.resolveRelation(DataSource.scala:317) // without this pr java.lang.NumberFormatException: For input string: "aaa" at java.lang.NumberFormatException.forInputString(NumberFormatException.java:65) at java.lang.Long.parseLong(Long.java:589) at java.lang.Long.parseLong(Long.java:631) at scala.collection.immutable.StringLike$class.toLong(StringLike.scala:277) ``` Closes apache#19999 ## How was this patch tested? Added tests in `JDBCSuite`. Author: Takeshi Yamamuro <yamamuro@apache.org> Closes apache#21834 from maropu/SPARK-22814.

MaxGekk · 2019-01-16T22:24:53Z

+  private def toInternalBoundValue(value: String, columnType: DataType): Long = columnType match {
+    case _: NumericType => value.toLong
+    case DateType => DateTimeUtils.fromJavaDate(Date.valueOf(value)).toLong
+    case TimestampType => DateTimeUtils.fromJavaTimestamp(Timestamp.valueOf(value))


@maropu The Timestamp.valueOf method expects timestamp in the format yyyy-[m]m-[d]d hh:mm:ss[.f...]. Was it selected intentionally or just because it is the default pattern for Timestamp?

For example, it cannot parse time zones as Cast can:

Timestamp.valueOf("1973-02-27 02:30:00.102030+01:00")

Timestamp format must be yyyy-mm-dd hh:mm:ss[.fffffffff] java.lang.IllegalArgumentException: Timestamp format must be yyyy-mm-dd hh:mm:ss[.fffffffff] at java.sql.Timestamp.valueOf(Timestamp.java:251)

I thinks it'd be nice to accept more general pattens for date/timestamp. Can you make a pr to fix that?

Yes, I am going to prepare a PR for that.

AlJohri · 2019-05-04T19:27:47Z

@gatorsmile @maropu this currently does not work with pyspark due to this line:

spark/python/pyspark/sql/readwriter.py

Lines 563 to 564 in d9bcacf

    
           return self._df(self._jreader.jdbc(url, table, column, int(lowerBound), int(upperBound), 
        
                                              int(numPartitions), jprop))

it tries to convert lowerBound and upperBound to an int.

The resulting traceback is:

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-40-2636f0dd1e0a> in <module>
     16                         upperBound=now,
     17                         numPartitions=sc.defaultParallelism,
---> 18                         properties={'driver': 'org.postgresql.Driver'})
     19                    .join(article_metadata, on=['url'], how='left')
     20                    .orderBy('timestamp', ascending=False))

/usr/lib/spark/python/pyspark/sql/readwriter.py in jdbc(self, url, table, column, lowerBound, upperBound, numPartitions, predicates, properties)
    550             assert numPartitions is not None, \
    551                 "numPartitions can not be None when ``column`` is specified"
--> 552             return self._df(self._jreader.jdbc(url, table, column, int(lowerBound), int(upperBound),
    553                                                int(numPartitions), jprop))
    554         if predicates is not None:

TypeError: int() argument must be a string, a bytes-like object or a number, not 'datetime.datetime'

I think just removing the int may fix the issue but I'm not 100% sure.

EDIT: looks like the pyspark docs explicitly say to use an integer column at the moment

shatestest · 2019-05-07T07:05:36Z

When i chose INSERTION_DATE as my partitionColumn with below dates
.option("lowerBound","2002-03-31");
.option("upperBound", "2019-05-01");
.option("dateFormat", "yyyy-mm-dd");

Getting error : ORA-01861: literal does not match format string
How to pass the dates for "lower/upperBound" ???

MaxGekk · 2019-05-07T07:47:55Z

@shatestest mm means minute-of-hour, see https://docs.oracle.com/javase/8/docs/api/java/time/format/DateTimeFormatter.html . I guess you need to replace mm by MM.

shatestest · 2019-05-07T10:56:08Z

@MaxGekk thanks for reply but same error with .option("dateFormat", "yyyy-MM-dd");

shatestest · 2019-05-07T11:45:22Z

@SparkQA , may I know how you tested partitionColumn for date/timestamp ? any sample test cases ?

MaxGekk · 2019-05-07T12:06:50Z

There are tests in the PR, see https://github.com/apache/spark/pull/21834/files#diff-5e0cadf526662f9281aa26315b3750adR442 . @shatestest Could you open an JIRA ticket, and provide a reproducible example, please.

AlJohri · 2019-05-07T19:21:21Z

If anyone finds themselves here looking to do this in pyspark, until support for this feature is added, here is a workaround:

import datetime
from itertools import tee

def date_range(start, end, intv):
    diff = (end - start) / intv
    for i in range(intv):
        yield start + diff * i
    yield end

def pairwise(iterable):
    a, b = tee(iterable)
    next(b, None)
    return zip(a, b)

partition_column = 'mypartitioncol'
now = datetime.datetime.now(datetime.timezone.utc)
num_partitions = sc.defaultParallelism
lower_bound = now + datetime.timedelta(-30)
upper_bound = now

predicates = []
for start, end in pairwise(date_range(lower_bound, upper_bound, num_partitions)):
    predicates.append(f'{partition_column} >= \'{start.isoformat()}\' AND {partition_column} < \'{end.isoformat()}\'')

df = (spark.read.jdbc(
            url='myjdbcuri',
            table='mytable',
            predicates=predicates,
            properties={'driver': 'org.postgresql.Driver'}))

Associated JIRA comment.

shatestest · 2019-05-15T09:49:39Z

@MaxGekk sorry for delay . raised a ticket ...please let me know if you need any more info .... https://issues.apache.org/jira/browse/SPARK-27723

shatestest · 2019-05-19T13:43:01Z

@MaxGekk any help plz https://stackoverflow.com/questions/56208362/numberformatexception-thrown-when-passed-date-as-lowerbound-upperbound-in-spark

Fix

87f9be3

gatorsmile reviewed Jul 24, 2018

View reviewed changes

maropu added 2 commits July 25, 2018 11:50

Fix

d894459

Add tests

1041a38

Fix

577f66e

asfgit closed this in 47d84e4 Jul 30, 2018

MaxGekk reviewed Jan 16, 2019

View reviewed changes

Conversation

maropu commented Jul 21, 2018

What changes were proposed in this pull request?

How was this patch tested?

Uh oh!

SparkQA commented Jul 21, 2018

Uh oh!

maropu commented Jul 21, 2018

Uh oh!

maropu commented Jul 24, 2018

Uh oh!

gatorsmile Jul 24, 2018

Choose a reason for hiding this comment

Uh oh!

maropu Jul 25, 2018

Choose a reason for hiding this comment

Uh oh!

gatorsmile commented Jul 25, 2018

Uh oh!

maropu commented Jul 25, 2018

Uh oh!

SparkQA commented Jul 25, 2018

Uh oh!

maropu commented Jul 25, 2018

Uh oh!

SparkQA commented Jul 25, 2018

Uh oh!

SparkQA commented Jul 25, 2018

Uh oh!

gatorsmile commented Jul 30, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

gatorsmile commented Jul 30, 2018

Uh oh!

gatorsmile commented Jul 30, 2018

Uh oh!

SparkQA commented Jul 30, 2018

Uh oh!

gatorsmile commented Jul 30, 2018

Uh oh!

maropu commented Jul 31, 2018

Uh oh!

MaxGekk Jan 16, 2019

Choose a reason for hiding this comment

Uh oh!

maropu Jan 17, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

MaxGekk Jan 17, 2019

Choose a reason for hiding this comment

Uh oh!

maropu Jan 17, 2019

Choose a reason for hiding this comment

Uh oh!

AlJohri commented May 4, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

shatestest commented May 7, 2019

Uh oh!

MaxGekk commented May 7, 2019

Uh oh!

shatestest commented May 7, 2019

Uh oh!

shatestest commented May 7, 2019

Uh oh!

MaxGekk commented May 7, 2019

Uh oh!

AlJohri commented May 7, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

shatestest commented May 15, 2019

Uh oh!

shatestest commented May 19, 2019

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

gatorsmile commented Jul 30, 2018 •

edited

Loading

maropu Jan 17, 2019 •

edited

Loading

AlJohri commented May 4, 2019 •

edited

Loading

AlJohri commented May 7, 2019 •

edited

Loading