Language codes not always stripped from descriptions

[Twan Goosen]

When there are multiple description fields, the language code prefixes do not get stripped out in the view (in VLO-3.2 in production).

See for example ​http://catalog.clarin.eu/vlo/record?fq=organisation:Austrian+Academy+of+Sciences&docId=http_58__47__47_hdl.handle.net_47_11022_47_0000-0000-001B-2

[Jörg Knappen]

Just a note: The language codes in the the example aren't ISO 639 language codes, but IETF language tags ( ​https://en.wikipedia.org/wiki/IETF_language_tag ).

AFAIK, language tags aren't supported by the VLO yet; should they? Or is it a case for metadata curation?

EDIT: The usage of a language tag for xml:lang is legal by the XML specification (see: ​http://www.w3.org/TR/REC-xml/ ) --- I tend to ask for support of it.

[Twan Goosen]

As mentioned by Jörg, the language codes in this case were not lower case ISO 639 three letter codes whereas the web app was assuming that they always are in this context. The responsible regex is now a bit more relaxed, see r6404.

It would be nice, but not critical, to make sure that the xml:lang based content language prefix for description fields would always be converted to ISO639 by the importer, as it does for values in the languageCode field. I will make a separate ticket and close this one.

[Twan Goosen]

Follow-up: #780