[SPARK-8093] Spark 1.4 branch's new JSON schema inference has changed the behavior of handling inner empty JSON object.

• Type: Bug
• Status: Resolved
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: 1.4.0
• Fix Version/s: 1.4.1, 1.5.0
• Component/s: SQL
• Target Version/s: 1.4.1, 1.5.0
• Assignee: Nathan Howell
• Reporter: Harish Butani
• Shepherd: Yin Huai
• Created: 04/Jun/15 01:04
• Updated: 19/Jun/15 23:28
• Resolved: 19/Jun/15 23:23

Description

This is similar to SPARK-3365. Sample json is attached. Code to reproduce

var jsonDF = read.json("/tmp/t1.json")
jsonDF.write.parquet("/tmp/t1.parquet")

The 'integration' object is empty in the json.
StackTrace:

....
Caused by: java.io.IOException: Could not read footer: java.lang.IllegalStateException: Cannot build an empty group
	at parquet.hadoop.ParquetFileReader.readAllFootersInParallel(ParquetFileReader.java:238)
	at org.apache.spark.sql.parquet.ParquetRelation2$MetadataCache.refresh(newParquet.scala:369)
	at org.apache.spark.sql.parquet.ParquetRelation2.org$apache$spark$sql$parquet$ParquetRelation2$$metadataCache$lzycompute(newParquet.scala:154)
	at org.apache.spark.sql.parquet.ParquetRelation2.org$apache$spark$sql$parquet$ParquetRelation2$$metadataCache(newParquet.scala:152)
	at org.apache.spark.sql.parquet.ParquetRelation2.refresh(newParquet.scala:197)
	at org.apache.spark.sql.sources.InsertIntoHadoopFsRelation.insert(commands.scala:134)
	... 69 more
Caused by: java.lang.IllegalStateException: Cannot build an empty group

Issue Links

• links to: [Github] Pull Request #6799 (NathanHowell)