GitLab

Many language do not have a short form with 2-letters (in fact the most common
short-form appears to be 3-letters, see
http://www.loc.gov/standards/iso639-2/php/code_list.php). So store the dict of
concept's preferred labels by full language code along with short form.

Closes #12397683.

So that this matches with `through_cw_source` (inlined) relation and guest
users do not get an Unauthorized when requesting the primary view of a
SKOSSource.

Closes #12437063.

beside the fact that this is not necessary (no appobjects in there), this causes
import error with pre 3.21

also fix two errors in pre 3.21 specific module:

* store_skos_extentities should have a extid_as_cwuri parameter, not
  use_extid_as_cwuri

* fix arguments given to ext_dump_relations

use limit defined by navigation.page-size to swith to paginated list rather than some arbitrary hard-coded value. Closes #11088567

Monkeypatch reintroduced in 7b861a9c0a8b probably by mistake

fake_form was backported in CW in version 3.20.5

Closes #9133216

including monkey-patches for https://www.cubicweb.org/ticket/10468967 until cw
3.23 is released. Now when one ask for synchronization of a skos source, he gets
redirected to the forthcoming import log.

Fix a select ambiguity with cw >= 3.22 along the way.

* make some functions public (they were used anyway)
* improve docstring
* deprecate _skos_import_lcsv and _skos_import_rdf

add a wrapper around dump_relations to move inlined relations into the extentity (or raise an error)

relevant patches have been integrated there

According to the SKOS specification, closeMatch is not symmetric. This means
that telling is close to another doesn't imply the opposite. So it sounds good
to hide this relation from the relation's object (also it's really ambiguous
right now, eg difference between "close match" and "close match of").

* add example to attributes description in schema

* add an introduction to the SKOSSource creation form

* link to associated schemes in the SKOSSource primary view, linking to the
  synchronization action if there is no scheme yet

This first implementation allows to choose store type and RDF implementation.

Related to #10430145

We're not ready for this and I cannot run tests with tox with this change.

And save results as XML.

These are not compatible with test runners other than logilab.common.pytest.

It is used at least in cubicweb-datacat.

It is not automatically imported by the 'import *'.

We loose an assert in the process, but for stores that don't actually insert
entities into database when ``prepare_insert_entity()`` is called (eg.
MassiveStore), there is no easy way to get the actual entity: ``entity_from_eid``
will fail.

Closes #5689852

fake_form was backported in CW in version 3.20.5

Related to #9133216