[SPARK-48308] Unify getting data schema without partition columns in FileSourceStrategy

• Type: Bug
• Status: Resolved
• Priority: Trivial
• Resolution: Fixed
• Affects Version/s: 3.5.1
• Fix Version/s: 4.0.0, 3.5.2
• Component/s: Spark Core
• Labels: pull-request-available
• Assignee: Johan Lasperas
• Reporter: Johan Lasperas
• Created: 16/May/24 12:42
• Updated: 25/Jul/24 15:22
• Resolved: 16/May/24 14:38

Description

In FileSourceStrategy, the schema of the data excluding partition columns is computed 2 times in a slightly different way:

 

val dataColumnsWithoutPartitionCols = dataColumns.filterNot(partitionSet.contains) 

vs 

val readDataColumns = dataColumns
  .filterNot(partitionColumns.contains) 

Issue Links

• relates to: SPARK-48241 CSV parsing failure with char/varchar type columns
• links to: GitHub Pull Request #46619
• links to: GitHub Pull Request #47483