Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In
`@application-engine/src/main/groovy/com/netgrif/application/engine/migration/helpers/AbstractMigrationHelper.groovy`:
- Around line 53-58: Update the field Javadoc for migrationErrors in
AbstractMigrationHelper to reflect that it is a List<MigrationError> (not a
map): describe that migrationErrors is a collection holding MigrationError
entries encountered during migration, each representing an individual migration
failure; remove references to keys, string identifiers, and lists-per-key, and
adjust any thread-safety wording to match the actual implementation (e.g., note
whether the list is synchronized or not) so the comment accurately matches the
migrationErrors field.
- Around line 192-194: The Javadoc block in AbstractMigrationHelper is malformed
because the first line of the comment lacks the leading "*" and runs into the
next line; update the Javadoc for the method (the comment starting at the block
above the iterator/bulk-operation method in class AbstractMigrationHelper) by
adding the missing "*" at the start of the first line so each line begins with
"*" and the comment renders correctly as a standard Javadoc block.
- Around line 226-266: The outer catch in the iterate loop is re-invoking
handleMigrationError for exceptions that have already been routed (and possibly
re-thrown) by prepareOperations or effectiveProcessOperations, which causes
double-handling and masks original errors; modify the catch block around the
cursor iteration so it checks for MigrationErrorException (or a marker that
handleMigrationError already threw) and rethrows it immediately without calling
handleMigrationError again, and only call handleMigrationError for other
unexpected exceptions (so keep references to prepareOperations,
effectiveProcessOperations, handleMigrationError, MigrationErrorException and
finishMigrationErrorPolicy to locate the logic).
- Around line 317-321: The THROW_AFTER_LIMIT branch incorrectly throws
immediately when policy.maxErrors == 0 and bypasses throwOriginal; change the
logic in the MigrationErrorHandlingMode.THROW_AFTER_LIMIT case to first treat
maxErrors==0 as "no limit" (i.e. only apply the limit when policy.maxErrors > 0)
and when the limit is reached call the central throwError(...) path instead of
directly throwing new MigrationErrorException so throwOriginal and cacheErrors
behavior are honored; specifically update the condition around
getErrors().size() and policy.maxErrors to require policy.maxErrors > 0 and
route handling through throwError(...) (respecting cacheErrors and existing
error caching) rather than constructing/throwing MigrationErrorException inline.

In
`@application-engine/src/main/groovy/com/netgrif/application/engine/migration/helpers/CaseMigrationHelper.groovy`:
- Around line 448-450: The error logging and error handler call in removeCase
are mislabelled: update the message string and the handleMigrationError
operation parameter to reference "removeCase" (or "removeCase: delete case"
phrasing) instead of "migratePetriNet"; specifically modify the local variable
message and the handleMigrationError(errorPolicy, "migratePetriNet", type,
useCase.stringId, message) invocation in the removeCase method so both
accurately reflect removeCase while keeping errorPolicy, type and
useCase.stringId unchanged.

In
`@application-engine/src/main/java/com/netgrif/application/engine/configuration/properties/MigrationProperties.java`:
- Around line 143-147: The docs say maxErrors == 0 means "no limit" but the
handler throws immediately because it checks getErrors().size() >=
policy.maxErrors; update AbstractMigrationHelper.handleMigrationError to treat 0
as unlimited by changing the throw condition to only apply when policy.maxErrors
> 0 (e.g., require policy.maxErrors > 0 && getErrors().size() >=
policy.maxErrors), keeping existing behavior for positive limits and preserving
references to getErrors(), policy.maxErrors, and the THROW_AFTER_LIMIT policy.

In
`@application-engine/src/main/java/com/netgrif/application/engine/utils/MongodbUtils.java`:
- Around line 9-15: MongodbUtils currently exposes only the static method
toQuery but is instantiable; make the intent explicit by adding a private
constructor to MongodbUtils (e.g., private MongodbUtils() { throw new
AssertionError("No instances."); } or simply private MongodbUtils() {} ) so
callers cannot instantiate the class; update class MongodbUtils to include this
private constructor while leaving the static toQuery(MongoTemplate, Class<T>,
Predicate...) unchanged.

In
`@application-engine/src/test/groovy/com/netgrif/application/engine/migration/MigrationTest.groovy`:
- Around line 69-104: The beforeEach setup resets the DBs but not the
centralized migration error cache, causing flaky tests; add a call to clear the
error cache (e.g., migrationHelper.clearErrors() or the appropriate
clearErrors() method on the migration helper) immediately after
testHelper.truncateDbs() inside the beforeEach method so the shared error state
is reset before importing nets and creating cases, ensuring
migrationHelper.hasErrors() checks in other tests are not impacted by prior
runs.

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In
`@application-engine/src/main/groovy/com/netgrif/application/engine/migration/helpers/CaseMigrationHelper.groovy`:
- Line 343: The JavaDoc comment for the toRemove parameter in the
CaseMigrationHelper class incorrectly describes the operation as "add" when it
should describe the operation as "remove". Update the `@param` toRemove
documentation to accurately reflect that the parameter contains a map of field
IDs to lists of choices to remove from the data field, not add.

---

Outside diff comments:
In
`@application-engine/src/main/groovy/com/netgrif/application/engine/migration/helpers/CaseMigrationHelper.groovy`:
- Around line 190-219: The null check at line 194 in the elasticIndex method
executes too late to prevent a NullPointerException. The issue is that
petriNetService.get(useCase.petriNetObjectId) is passed directly to
PetriNetMigrationHelper.setPetriNet at line 193, and if the service returns
null, setPetriNet will throw an exception before the null check can run. Move
the null check before calling setPetriNet by first storing the result of
petriNetService.get(useCase.petriNetObjectId) in a variable, checking if it is
null, and only then calling PetriNetMigrationHelper.setPetriNet with that
non-null result. Keep the existing error handling logic intact.

In
`@application-engine/src/main/groovy/com/netgrif/application/engine/migration/helpers/PetriNetMigrationHelper.groovy`:
- Around line 300-307: The warning log message in the updateTransitionRoles
method is logging incorrect information. The code checks whether a ProcessRole
with the given roleImportId exists, but the warning message incorrectly states
"Transition with importId $transitionId not found". Update the log.warn message
to correctly indicate that the role with roleImportId was not found, not the
transition. Change the message to reference roleImportId and clarify that it is
the role that could not be found in the net.