[NIFI-8259] CSVSchemaInference should infer fields as strings if no records are present

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 1.14.0, 1.13.1
• Component/s: Extensions
• Assignee: Matt Burgess
• Reporter: Matt Burgess
• Created: 25/Feb/21 15:22
• Updated: 25/Feb/21 17:48
• Resolved: 25/Feb/21 17:48

Description

Even though the field names may be present in a flow file whose schema is to be inferred by CSVReader, if there are no records in the flow file the data types of the fields cannot be ascertained. This causes the schema to be empty and can result in unwanted behavior in processors such as QueryRecord and ConvertRecord.

Instead the schema inference should assume all fields are of string types, such that a "best-effort" schema is generated for use by the processors.

Issue Links

• links to: GitHub Pull Request #4845