Writing into array with same name but with and without leading * merges both array into one

[TobiasNx]

In principal it is possible to create elements and array with leading * as part of the name.

But if you write an array on the same level with same name but one with and one without leading *, the two disctinct arrays are merged:

See in example in playground.

Fix:

add_field("test.$append","TEST1")
add_field("*test.$append","TEST2")

Result:

test: "TEST1"
test: "TEST2"

Expected:

test: "TEST1"
*test: "TEST2"

This came up in #589 when transforming marc data to edm, since rdf-macro expects * as refrence marker.

[blackwinter]

The asterisk (*) is a wildcard, so *test matches test.

[TobiasNx]

Interestingly:

add_field("*test2","TEST2")
add_field("test4.*test.$append","TEST6")

works

---
'*test2': "TEST2"
test4:
  '*test': "TEST6"

I am also not sure if we should fix this?

[blackwinter]

I would consider it a bug if the wildcard becomes part of the field name just because it didn't match any preexisting field.

[TobiasNx]

Morph was able to create such elements e.g. here. This was the precondition of the marker * which came up in #589 and try to make configurable in #657

[blackwinter]

Morph was able to create such elements

That's definitely a surprise. Would have to look into it more thoroughly.

[blackwinter]

The reason that it works in Metamorph is that it doesn't evaluate wildcards in the target field (name), only in the source field (source). Metafix, on the other hand, supports wildcards in both source and target fields.

The only solution I can think of right now is to allow escaping of special characters in WildcardTrie (*, ?, |) and SimpleRegexTrie ([, ]). This might also be valuable in its own right.

[TobiasNx]

In my opinion this is not urgent since we enabled this: #657