[cowtowncoder]

A token stream that ends right after a property name, which is possible with TokenBuffer, leaves value deserializers with no current token. StdDeserializer._parseString() called p.currentToken().isScalarValue() without a null check, so String and other values going through that method failed with a DatabindException wrapping an NPE, instead of reporting a problem with the input.

Changes

• _parseString() now checks for a null token, and falls through to the existing ctxt.handleUnexpectedToken(...) call, which throws MismatchedInputException.

Tests

EndOfInputAfterPropertyNameTest (new) has 52 cases covering primitives and wrappers, String, enums, dates, POJOs, records, collections, maps, arrays, Optional, Object and JsonNode. Before the fix 5 cases failed, all at this line; the other 47 already threw MismatchedInputException.

The full suite passes locally on the default JDK: 7725 tests, 1 skipped, no failures.

Found while working on FasterXML/jackson-modules-base#371, where Afterburner matches this stock behavior.

🤖 Generated with Claude Code

---

Summary by Gitar

• Null token checks:
  • BeanDeserializer and BuilderBasedDeserializer skip external type id handling when no token remains
  • ByteBufferDeserializer reports an unexpected token when the parser has no current token
• TokenBuffer:
  • copyCurrentStructure() throws UnexpectedEndOfInputException when input ends after a property name
• Enum errors:
  • FactoryBasedEnumDeserializer passes the token itself into mismatch messages instead of t.name()
• Release notes:
  • Added an entry for Report end of input after property name as MismatchedInputException #6376 in release-notes/VERSION

_{This will update automatically on new commits.}

[github-actions]

🧪 Code Coverage Report

Metric	Coverage	Change
Instructions	83.02%	📈 +0.010%
Branches	77.18%	📈 +0.070%

Coverage data generated from JaCoCo test results

[github-actions]

🧪 Code Coverage Report

Metric	Coverage	Change
Instructions	83.02%	📈 +0.010%
Branches	77.16%	📈 +0.050%

Coverage data generated from JaCoCo test results

[gitar-bot]

Code Review ✅ Approved

🟡 Medium risk · Handling missing tokens in deserialization paths changes malformed-input exceptions, which could affect callers that distinguish exception types.

Reports end of input after a property name as MismatchedInputException instead of a wrapped NPE, with null-token guards added to _parseString(), BeanDeserializer, BuilderBasedDeserializer, and ByteBufferDeserializer, and TokenBuffer.copyCurrentStructure() now throwing UnexpectedEndOfInputException in that case. No issues found.

Verification

Rules No rules evaluated

Functional validation Not enabled

Set up

_{For more verification details, comment gitar display:verbose.}

Options

Auto-apply is off → Gitar will not commit updates to this branch.
Display: compact → Showing compact verification results.

Comment with these commands to change the behavior for this request:

Auto-apply	Compact
gitar auto-apply:on	gitar display:verbose

_{Was this helpful? React with 👍 / 👎 | Powered by Gitar — free for open source}

[github-actions]

🧪 Code Coverage Report

Metric	Coverage	Change
Instructions	83.03%	📈 +0.020%
Branches	77.20%	📈 +0.090%

Coverage data generated from JaCoCo test results