Fix conversion of types when mapping Aggregation pipeline. #4723
Conversation
This change makes sure to apply conversion to non native mongo types when the context does not expose fields.
| @@ -62,7 +62,7 @@ static List<Document> toDocument(List<AggregationOperation> operations, Aggregat | |||
| if (operation instanceof InheritsFieldsAggregationOperation || exposedFieldsOperation.inheritsFields()) { | |||
| contextToUse = contextToUse.inheritAndExpose(fields); | |||
| } else { | |||
| contextToUse = fields.exposesNoFields() ? DEFAULT_CONTEXT | |||
| contextToUse = fields.exposesNoFields() ? ConverterAwareNoOpContext.instance(rootContext) | |||
There was a problem hiding this comment.
I'm not sure I follow why we have this distinction here. Switching to contextToUse = contextToUse.expose(fields); also lets all the tests pass.
There was a problem hiding this comment.
it might cause the mapping to consider filenames of follow up stages to be object to field name mapping which should not be done. if no tests fail, we should make sure that we have coverage there and do not hit a blank spot.
There was a problem hiding this comment.
Okay, then this one needs to wait for the next service release as we supposedly do not have sufficient test coverage.
There was a problem hiding this comment.
confirmed, we're missing a test here - like if we have an operation that forces field mapping to back off (like the replaceRoot one) a strict context would fail because of the missing exposure.
This PR makes sure to apply conversion to non native mongo types when the context does not expose fields.
Resolves: #4722