[SPARK-12300][SQL][PYSPARK] fix schema inferance on local collections

[holdenk]

Current schema inference for local python collections halts as soon as there are no NullTypes. This is different than when we specify a sampling ratio of 1.0 on a distributed collection. This could result in incomplete schema information.

[SparkQA]

Test build #47608 has finished for PR 10275 at commit 6e463fd.

• This patch fails PySpark unit tests.
• This patch merges cleanly.
• This patch adds no public classes.

[SparkQA]

Test build #47611 has finished for PR 10275 at commit b863cc9.

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

[holdenk]

cc @nchammas - this is the fix for the local version of SPARK-2870

[holdenk]

ping @davies if you have a chance to look at this.

[SparkQA]

Test build #48491 has finished for PR 10275 at commit 3f2b825.

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

[davies]

LGTM, merging this into master and 1.6, thanks!

[gatorsmile]

@holdenk Could you take a look at the test failure? I also hit this issue in my local environment without any code change.

https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/48605/consoleFull
https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/48609/consoleFull

It sounds like this is caused by the inferred schema:

py4j.protocol.Py4JJavaError: An error occurred while calling o1979.count.
: org.apache.spark.SparkException: Job aborted due to stage failure: Task 2 in stage 211.0 failed 1 times, most recent failure: Lost task 2.0 in stage 211.0 (TID 886, localhost): java.lang.ArrayIndexOutOfBoundsException: 1
    at org.apache.spark.sql.catalyst.expressions.GenericInternalRow.genericGet(rows.scala:227)
    at org.apache.spark.sql.catalyst.expressions.BaseGenericInternalRow$class.getAs(rows.scala:35)
    at org.apache.spark.sql.catalyst.expressions.BaseGenericInternalRow$class.isNullAt(rows.scala:36)
    at org.apache.spark.sql.catalyst.expressions.GenericInternalRow.isNullAt(rows.scala:221)
    at org.apache.spark.sql.catalyst.expressions.GeneratedClass$SpecificUnsafeProjection.apply(Unknown Source)
    at org.apache.spark.sql.catalyst.expressions.GeneratedClass$SpecificUnsafeProjection.apply(Unknown Source)
    at scala.collection.Iterator$$anon$11.next(Iterator.scala:328)
    at scala.collection.Iterator$$anon$11.next(Iterator.scala:328)

Thanks!

[holdenk]

I'm on vacation with less than great internet but I'll try and repro locally.

[holdenk]

ok reproed and I've got a fix (seems just a test issue from a parallel change in how missing fields can be done).