Skip to content

Parquet pruning will be incorrect if field names are repeated #8335

@alamb

Description

@alamb

Describe the bug

Basically if you have a parquet file with a nested field that has the same name as a top level field, the datafusion parquet reader will read statistics for the nested field instead of the top level field

To Reproduce

I added a test in #8294 struct_and_non_struct

Basically it shows:

struct_field: struct { 
  int_field: int32,     <-- 'struct_field.int_field'
}
int_field: int32.       <-- just 'int_field'

If you then ask for statistics (e.g. have a predicate on) int_field, the DataFusion reader will produce the statistics for struct_field.int_field rather than the top level int_field

Expected behavior

No response

Additional context

No response

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't working

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions