rdf_library.pl -- RDF Library Manager

This module manages an ontology library. Such a library consists of a directory with manifest files named Manifest.rdf or Manifest.ttl (Turtle). The manifest files define ontologies appearing in the library as well as namespace mnemonics and dependencies.

The typical usage scenario is

?- rdf_attach_library('/some/directory').
?- rdf_load_library(my_ontology).

author: - Jan Wielemaker
To be done: - Add caching info; - Allow using Manifests on HTTP servers

rdf_load_library(+Id) is det

rdf_load_library(+Id, +Options) is det

Load ontologies from the library. A library must first be attached using rdf_attach_library/1. Defined Options are:

import(Bool): If true (default), also load ontologies that are explicitely imported.
base_uri(URI): BaseURI used for loading RDF. Local definitions in ontologies overrule this option.
claimed_source(URL): URL from which we claim to have loaded the data.
not_found(+Level): The system does a pre-check for the existence of all references RDF databases. If Level is error it reports missing databases as an error and fails. If warning it prints them, but continues. If silent, no checks are preformed. Default is error.
concurrent(Threads): Perform the load concurrently using N threads. If not specified, the number is determined by guess_concurrency/2.
load(+Bool): If false, to all the preparation, but do not execute the actual loading. See also rdf_list_library/2.

rdf_library_source(+Id, -Source) is nondet

True of Source is the URL that is part of the given library Id. This predicate finds all indirect dependencies. It does not check whether the source exists or is valid.

See also: - uri_file_name/2 for converting file:// URLs to a filename.

find_conflicts(+LoadCommands) is semidet[private]

Find possibly conflicting options for loading the same source

no_source_with_different_options(+Commands) is semidet[private]

True if there are not multiple calls to load the same graph, but with different load-options. Prints a warning and fails otherwise.

no_sources_in_same_graph(+Commands) is semidet[private]

True if there are not two load commands referring to the same graph.

check_existence(+CommandsIn, -Commands, +Options) is det[private]

Report existence errors. Fail if at least one source does not exist. and the not_found level is not silent.

Errors: - existence_error(urls, ListOfUrls)

guess_concurrency(+Commands, -Threads) is det[private]

How much concurrency to use? Set to the number of CPUs if all input comes from files or 5 if network based loading is demanded.

load_commands(+Id, +Options, -Pairs:list(Level-Command)) is det[private]

Commands are the RDF commands to execute for rdf_load_library/2. Splitting in command collection and execution allows for concurrent execution as well as forward checking of possible problems.

To be done: - Fix poor style; avoid assert/retract.

import(+URL, +Level, +Options) is det[private]

rdf_list_library(+Id) is det

rdf_list_library(+Id, +Options) is det

Print library dependency tree to the terminal. Options include options for rdf_load_library/2 and

show_source(+Boolean): If true (default), show location we are loading
show_graph(+Boolean): If true (default false), show name of graph
show_virtual(+Boolean): If false (default true), do not show virtual repositories.
indent(Atom): Atom repeated for indentation levels

rdf_list_library

Prints known RDF library identifiers to current output.

rdf_library_index(?Id, ?Facet) is nondet

Query the content of the library. Defined facets are:

source(URL): Location from which to load the ontology
title(Atom): Title used for the ontology
comment(Atom): Additional comments for the ontology
version(Atom): Version information on the ontology
imports(Type, URL): URLs needed by this ontology. May succeed multiple times. Type is one of ontology, schema or instances.
base_uri(BaseURI): Base URI to use when loading documents. If BaseURI ends in /, the actual filename is attached.
claimed_source(Source): URL from which we claim to have loaded the RDF. If Source ends in /, the actual filename is attached.
blank_nodes(Share): Defines how equivalent blank nodes are handled, where Share is one of share or noshare. Default is to share.
format(Format): Format of the resource. Can be used to overrule if the format as derived from the HTTP content type is wrong.
provides_ns(URL): Ontology provides definitions in the namespace URL. The formal definition of this is troublesome, but in practice it means the ontology has triples whose subjects are in the given namespace.
uses_ns(URL): The ontology depends on the given namespace. Normally means it contains triples that have predicates or objects in the given namespace.
manifest(URL): URL of the manifest in which this ontology is defined.
virtual: Entry is virtual (cannot be loaded)

rdf_attach_library(+Source)

Attach manifest from Source. Source is one of

URL: Load single manifest from this URL
File: Load single manifest from this file
Directory: Scan all subdirectories and load all Manifest.ttl or Manifest.rdf found. If Directory is a path-alias (e.g., ontology(.)), all referenced directories are scanned for manifest files.

Encountered namespaces are registered using rdf_register_ns/2. Encountered ontologies are added to the index. If a manifest was already loaded it will be reloaded if the modification time has changed.

rdf_update_library_index[private]

Reload all Manifest files.

process_manifest(+Location) is det[private]

Process a manifest file, registering encountered namespaces and creating clauses for library/3. No op if manifest was loaded and not changed. Removes old data if the manifest was changed.

Arguments:

Location

- is either a path name or a URL.

extract_namespace(+Triples, -Mnemonic, -NameSpace)[private]

True if Mnemonic is an abbreviation of NameSpace.

extract_ontology(+Triples, -Ontology) is nondet[private]

Extract definition of an ontology

facet(+Triples, +File, -Facet) is nondet[private]

Enumerate facets about File from Triples. Facets are described with rdf_library_index/2.

edge(+Triples, ?S, ?P, ?O) is nondet[private]

Like rdf_has/3 over a list of Triples.

source_time(+Source, -Modified) is semidet[private]

Modified is the last modification time of Source.

Errors: - existence_error(Type, Source).

read_triples(+URL, -Triples) is det[private]

Read RDF/XML or Turtle file into a list of triples.

assert_ontology(+Manifest, +Term:library(Name,File,Facets)) is det[private]

Add ontology to our library.

To be done: - Proper behaviour of re-definition?

library(?Id, ?URL, ?Facets)[private]

Access DB for library information.

normalize_url(+URL, -Normalized)[private]

Like uri_normalized/2, but we also need (platform dependent) filename canonization.

define_namespace(NS:ns(Mnemonic,Namespace)) is det[private]

Add namespace declaration for Mnemonic.

rdf_current_manifest(-URL) is nondet

True if URL is the URL of a currently loaded manifest file.