@frink-okn/actor-query-source-identify-kgf
v0.1.1
Published
Native Knowledge Graph Fragments query source for Comunica
Downloads
32
Readme
@frink-okn/actor-query-source-identify-kgf
A native KGF JSON triple source for Comunica 5. Most users should use the
@frink-okn/kgf-sparql engine, which registers
this actor alongside Comunica's standard source types.
Configure sources as { type: 'kgf', value: datasetOrVersionedFragmentUrl }.
Dataset /latest/fragment aliases are also accepted and pinned to the current
release through the dataset descriptor, without fetching a fragment page.
The actor identifies sources without network access. Query planning then discovers
service limits and requests /count, once per pattern per query however often the query
plans it; fragment pages remain lazy. Filter bindings
use QUERY, with POST fallback on HTTP 405/501, and follow opaque cursors. A call with
filter bindings is a bind join's probe: it is not counted, and reports an unknown
cardinality.
Pages default to 1,000 rows. Bindings default to the advertised max_bindings
(200 if absent), subject to the request-byte limit; Comunica may supply smaller
batches. Exported context keys KEY_KGF_PAGE_SIZE, KEY_KGF_BINDINGS_SIZE, and
KEY_KGF_HTTP_METHOD override these choices. KEY_KGF_VOID_STATISTICS disables
the lazy VoID provider when set to false. The provider reads only relevant schema
partitions during join planning and memoizes them per query.
The source implements Comunica's filterBindings semantics, removing repeated
matches from overlapping input rows while leaving SPARQL join multiplicities to
Comunica. KGF's union is exposed as the default graph; named graphs are empty.
Graph-scoped operations and quoted triple terms are not implemented.
To register the actor in another Components.js configuration, provide
mediatorHttp and mediatorMergeBindingsContext, and put it before the generic
hypermedia identification actor via beforeActors. See the engine's
config/config-default.json for a complete configuration.
License
MIT
