[NIFI-13557] Record Inference Code (org.apache.nifi.util.SchemaInferenceUtil.getDataType) Does not Allow Single Digit Months

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 1.28.0, 2.0.0
• Assignee: Daniel Stieglitz
• Reporter: Daniel Stieglitz
• Created: 17/Jul/24 19:56
• Updated: 22/Jul/24 20:10
• Resolved: 19/Jul/24 22:01

Description

When running tests using the ExcelReader and inferring the schema, even though the date pattern 'M/d/yy' was specified (which per this post should allow single digit months), dates like '1/1/18' are not inferred to be DATE:M/d/yy but rather STRING. The source for this is
org.apache.nifi.util.text.RegexDateTimeMatcher whose regular expression for months requires two digits. It should be relaxed to allow for one or two digits.

Issue Links

• Discovered while testing: NIFI-12491 Add Use Starting Row Strategy to Excel Reader
• relates to: NIFI-13565 Record Date Time Properties should mention DateTimeFormatter instead of SimpleDateFormat
• relates to: NIFI-13569 Improve TestRegexDateTimeMatcher.java by making it JUnit 5 Parameterized Test
• links to: GitHub Pull Request #9094