Apply default null handling to @JsonAnySetter

[Dongnyoung]

Problem

Mapper-level default null handling was not applied to properties handled by @JsonAnySetter.

For regular bean properties, valueNulls(Nulls.SKIP) is propagated through property metadata and resolved to a skipping NullValueProvider, so an input VALUE_NULL skips assignment.

@JsonAnySetter uses a separate path: its synthetic BeanProperty was created with PropertyMetadata.STD_OPTIONAL, without applying default null-handling configuration. As a result, null-valued unknown properties were still passed to the any-setter.

Issue fixed is #6170, discussed in #6169;

Fix

Apply the existing null-handling flow to @JsonAnySetter:

• Resolve null-handling metadata for the synthetic any-setter BeanProperty using the existing setter-info logic.
• Resolve and attach a NullValueProvider to SettableAnyProperty.
• Skip any-setter assignment when the input token is VALUE_NULL and Nulls.SKIP is configured.
• Apply the same behavior to buffered and creator-based any-setter paths.
• Preserve existing custom value deserializer getNullValue() behavior when null skipping is not configured.

The skip decision is based on the input token being VALUE_NULL, not on the deserialized Java value being null. This preserves behavior for non-null input tokens that deserialize into Java null.

Tests

Added regression coverage for:

• mapper-level valueNulls(Nulls.SKIP) with method-based @JsonAnySetter
• buffered method-based any-setter handling
• creator-based any-setter handling
• non-null input tokens that deserialize to Java null
• existing custom getNullValue() behavior

Fixes: #6170

[github-actions]

🧪 Code Coverage Report

Metric	Coverage	Change
Instructions	81.89%	📉 -0.030%
Branches	75.54%	📉 -0.010%

Coverage data generated from JaCoCo test results

[github-actions]

🧪 Code Coverage Report

Metric	Coverage	Change
Instructions	81.92%	📈 +0.000%
Branches	75.56%	📈 +0.010%

Coverage data generated from JaCoCo test results

[skrcode]

I ran JAIPilot Cloud against this exact PR head. It proposed a small follow-up that consolidates the new null-provider bookkeeping into one private helper without changing public behavior.

The cloud run passed 26 focused tests and the clean 6,188-test repository suite. This is a maintainability cleanup, not a performance claim.

Cloud-generated draft and evidence: skrcode#2
PR directly onto this source branch: Dongnyoung#1

Feel free to merge or cherry-pick if the consolidation is useful.

[Dongnyoung]

@skrcode Thanks for the detailed follow-up. I reviewed the diff and the invariants around the JsonNode subclasses, and the cleanup looks good to me. I'll incorporate this into my PR branch.

[github-actions]

🧪 Code Coverage Report

Metric	Coverage	Change
Instructions	81.93%	📈 +0.020%
Branches	75.59%	📈 +0.030%

Coverage data generated from JaCoCo test results

[github-actions]

🧪 Code Coverage Report

Metric	Coverage	Change
Instructions	82.36%	📈 +0.000%
Branches	75.99%	📈 +0.030%

Coverage data generated from JaCoCo test results

[github-actions]

🧪 Code Coverage Report

Metric	Coverage	Change
Instructions	82.37%	📈 +0.030%
Branches	76.06%	📈 +0.060%

Coverage data generated from JaCoCo test results

[github-actions]

🧪 Code Coverage Report

Metric	Coverage	Change
Instructions	82.43%	📈 +0.010%
Branches	76.13%	📈 +0.040%

Coverage data generated from JaCoCo test results

[github-actions]

🧪 Code Coverage Report

Metric	Coverage
Instructions	82.44%
Branches	76.13%

Coverage data generated from JaCoCo test results

[cowtowncoder]

This is HUGE pr for sort of niche case... I wonder if there is any way to simplify code.

But on plus side, fixes look correct.

EDIT: I did some minor simplifying, am content with implementation now.

[github-actions]

🧪 Code Coverage Report

Metric	Coverage	Change
Instructions	82.45%	📈 +0.150%
Branches	76.14%	📈 +0.250%

Coverage data generated from JaCoCo test results

[gitar-bot]

Code Review ✅ Approved 1 closed / 1 findings

🟡 Medium risk · Any-setter deserialization now honors configured null skip/fail behavior across creator paths.

Applies mapper-level default null handling to @JsonAnySetter properties by resolving null-handling metadata and attaching a NullValueProvider to skip null-valued unknown properties when configured. The fix addresses inconsistent null handling between method and field any-setters and covers buffered and creator-based paths, with regression tests for mapper-level valueNulls(Nulls.SKIP) and custom getNullValue() behavior.

✅ 1 closed

✅ Edge Case: Null-handling type differs between method vs field any-setters

📄 src/main/java/tools/jackson/databind/deser/BeanDeserializerFactory.java:877-891 📄 src/main/java/tools/jackson/databind/deser/BeanDeserializerFactory.java:977-988 📄 src/main/java/tools/jackson/databind/deser/BasicDeserializerFactory.java:653-664 📄 src/main/java/tools/jackson/databind/deser/bean/BeanDeserializerBase.java:657-662
_constructAnySetterProperty is passed the element valueType for method-based any-setters but the container fieldType/paramType for field/parameter-based ones. As a result, _getSetterInfo looks up per-type ConfigOverride null-handling (line 654: config.getConfigOverride(prop.getType().getRawClass())) against the value type in one case and against Map/JsonNode in the other, and _findNullProvider receives a prop whose getType() is the container type for field-based any-setters. This can make a ConfigOverride for e.g. Map.class (or Nulls.FAIL error metadata) apply inconsistently across the two any-setter styles. Consider constructing the synthetic property with the element/value type consistently (or documenting the intent) so config-override lookup and null-provider resolution behave the same for method- and field-based any-setters.

Review coverage

📋 Rules No rules evaluated

🧪 Functional validation Not enabled · Set up

Options

Auto-apply is off → Gitar will not commit updates to this branch.
Display: compact → Counting what did not apply, without listing it.

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	82.45%	📈 +0.150%
Branches	76.14%	📈 +0.250%

Coverage data generated from JaCoCo test results