[ 
https://issues.apache.org/jira/browse/WW-5642?focusedWorklogId=1030004&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-1030004
 ]

ASF GitHub Bot logged work on WW-5642:
--------------------------------------

                Author: ASF GitHub Bot
            Created on: 13/Jul/26 06:23
            Start Date: 13/Jul/26 06:23
    Worklog Time Spent: 10m 
      Work Description: lukaszlenart commented on PR #1774:
URL: https://github.com/apache/struts/pull/1774#issuecomment-4955031334

   Thanks for the thorough follow-up — `0c93ea2fa` closes the completeness gaps 
I raised. I re-ran both suites locally against the PR head:
   
   - `ParameterAuthorizationContextTest` (core) — green
   - `ParameterAuthorizingModuleTest` (rest) — 23/23 green
   
   The `RedactionAwareDeserializer` + redaction-scope stack is a clean way to 
handle the construction-failure case: substituting a stand-in for a dropped 
creator-bound component and then treating a resulting construction failure as 
"drop the whole object" is consistent with how a rejected non-creator nested 
property already behaves. Good that 
`testValidatingRecord_genuineClientErrorStillPropagates` pins down the other 
side of it — a construction failure with nothing dropped still propagates, so 
genuine client/data errors aren't masked. Coverage across records, 
static-factory `@JsonCreator`, `@ConstructorProperties`, top-level records, 
3-level nesting, and `List`/`Map` creator params is exactly the matrix I was 
after.
   
   Three small, non-blocking notes:
   
   1. **Array-typed creator param.** `prefixForNested` handles 
`type.isArray()`, but only `List`/`Map` element paths are exercised in tests — 
an `Item[]` creator-param case would round out the collection matrix.
   2. **`FAIL_ON_NULL_FOR_PRIMITIVES` off (the default).** With it disabled, a 
dropped primitive creator component silently becomes the type default 
(`0`/`false`) rather than dropping the object. That's fine — the point is the 
client value never lands — but worth a one-line comment so it reads as a 
deliberate choice rather than an oversight.
   3. **Co-located failures.** If the same object both had a property dropped 
*and* hit an unrelated `JsonMappingException`, the current scope is marked, so 
the unrelated error is folded into "object dropped." Harmless in outcome, just 
slightly less informative to the caller — fine to leave, worth being aware of.
   
   None of these block. Nice work tightening it up.
   




Issue Time Tracking
-------------------

    Worklog Id:     (was: 1030004)
    Time Spent: 1h 10m  (was: 1h)

> @StrutsParameter authorization bypassed for record/creator-bound REST body 
> properties
> -------------------------------------------------------------------------------------
>
>                 Key: WW-5642
>                 URL: https://issues.apache.org/jira/browse/WW-5642
>             Project: Struts 2
>          Issue Type: Bug
>          Components: Plugin - REST
>    Affects Versions: 7.2.1
>         Environment: Reproducible with struts2-rest-plugin, 
> struts.parameters.requireAnnotations
> enabled, and any REST action whose request body binds a Java record
> (or @JsonCreator/@ConstructorProperties type) either at the top level
> or nested. JDK 17, Jackson 2.22.0.
>            Reporter: Gouri Sankar A
>            Priority: Major
>             Fix For: 7.3.0
>
>          Time Spent: 1h 10m
>  Remaining Estimate: 0h
>
> ParameterAuthorizingModule enforces @StrutsParameter on REST/JSON body
> deserialization by wrapping each Jackson property's deserializeAndSet /
> deserializeSetAndReturn (AuthorizingSettableBeanProperty). Jackson never
> calls either method for creator-bound properties — Java records,
> @JsonCreator constructors, @ConstructorProperties — it calls the final
> SettableBeanProperty#deserialize directly, which the existing wrapper
> cannot intercept.
> With struts.parameters.requireAnnotations enabled, any record-typed
> field in a REST action's request body is populated with no
> @StrutsParameter check at all, silently defeating the protection for
> that entire subtree.
> Fix: wrap the property's value deserializer (the only non-final
> interception point for the creator-bound path), scoped to
> CreatorProperty so existing setter/field/builder paths are unaffected.
> Fixed by: https://github.com/apache/struts/pull/1774



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to