Change Details

Right now, WDQS assumes for it's default prefix/URL system that entity and predicate prefixes come from the same host. This influences both default prefixes and more importantly, URIs for dictionaries and inlining. We should enable WDQS to be able to handle scheme where entity URIs may come from different places, e.g. mixing entity with `wd:` prefix for wikidata and `sdc:` prefix for commons. For default prefixes, we can just add a config, but we also may want the inlining to work for both URI schemes, which may require some additional support.

Right now, WDQS assumes for it's default prefix/URL system that entity and predicate prefixes come from the same host. This influences both default prefixes and more importantly, URIs for dictionaries and inlining. We should enable WDQS to be able to handle scheme where entity URIs may come from different places, e.g. mixing entity with `wd:` prefix for wikidata and `sdc:` prefix for commons. For default prefixes, we can just add a config, but we also may want the inlining to work for both URI schemes, which may require some additional support. The goals to achieve are: 1. Have `sdc:` items and data triples recognized by Munger as proper entities. 2. Have all SDC URIs be properly inlined. 3. Have Updater handle all statements properly 4. Have `sdc:` and `sdcs:` prefixes to work by default. We still need to have single URI for: * statement prefix (Updater assumes there's single statement prefix) Needs refactoring: * Entity ID <-> URI conversion, can be different for different URIs.

Right now, WDQS assumes for it's default prefix/URL system that entity and predicate prefixes come from the same host. This influences both default prefixes and more importantly, URIs for dictionaries and inlining. We should enable WDQS to be able to handle scheme where entity URIs may come from different places, e.g. mixing entity with `wd:` prefix for wikidata and `sdc:` prefix for commons. For default prefixes, we can just add a config, but we also may want the inlining to work for both URI schemes, which may require some additional support. The goals to achieve are: 1. Have `sdc:` items and data triples recognized by Munger as proper entities. 2. Have all SDC URIs be properly inlined. 3. Have Updater handle all statements properly 4. Have `sdc:` and `sdcs:` prefixes to work by default. We still need to have single URI for: * statement prefix (Updater assumes there's single statement prefix) Needs refactoring: * Entity ID <-> URI conversion, can be different for different URIs.