clownface-shacl-path

clownface-shacl-path

Provides functions to work with SHACL Property Paths

Install

yarn add clownface-shacl-path

Usage

findNodes

Find nodes in RDF/JS graphs by following SHACL Property Paths using clownface graph traversal library.

The exported function takes two parameters:

1. starting graph pointer node
2. graph pointer to a SHACL Property Path

import { findNodes } from 'clownface-shacl-path'
import fetch from '@rdfjs/fetch'
import $rdf from 'rdf-ext'
import clownface from 'clownface'
import { sh } from '@tpluscode/rdf-ns-builders'

// prepare a clownface Graph Pointer
const response = await fetch('http://zazuko.github.io/tbbt-ld/dist/tbbt.nt', { factory: $rdf })
const amy = clownface({ dataset: await response.dataset() })
    .namedNode('http://localhost:8080/data/person/amy-farrah-fowler')

// prepare a SHACL Property Path structure as clownface
const path = clownface({ dataset: $rdf.dataset() }).blankNode()

/*
  sh:path [
    sh:alternativePath ( # find both
      [ sh:inversePath schema:spouse ] # Sheldon, who is Amy's spouse
      [ sh:inversePath schema:knows ] # Leonard, who knows Amy
    )
  ]
*/
path.addList(sh.alternativePath, [
  path.blankNode().addOut(sh.inversePath, schema.spouse),
  path.blankNode().addOut(sh.inversePath, schema.knows)
])

// find nodes connected by the path
findNodes(amy, path)

toSparql

Converts a SHACL Property Path to SPARQL Property Path string template object. Use the property path with @tpluscode/sparql-builder

import type {GraphPointer} from 'clownface'
import { toSparql } from 'clownface-shacl-path'
import { SELECT } from '@tpluscode/sparql-builder'

/*
 [ sh:path 
   [
     sh:alternativePath (
       ( schema:knows schema:name )
       ( foaf:knows foaf:name )
     )
   ]
 ]
 */
let path: GraphPointer

/*
  SELECT ?friendName
  WHERE {
    ?person a <http://schema.org/Person> .
    ?person (schema:knows|schema:name)|(foaf:knows|foaf:name) ?friendName
  }
 */
SELECT`?friendName`
  .WHERE`
    ?person a <http://schema.org/Person> .
    ?person ${toSparql(path)} ?friendName .
  `.build()

toSparql.sequence

In cases when the intermediate nodes of a Sequence Path are important, that path can be split, so that authors can create and capture variables for all the nodes.

For that purpose, call toSparql.sequence()

import type {GraphPointer} from 'clownface'
import { toSparql } from 'clownface-shacl-path'
import { SELECT } from '@tpluscode/sparql-builder'
import $rdf from 'rdf-ext'

/*
 [ sh:path ( schema:employee schema:spouse schema:name ) ]
 */
let path: GraphPointer

const sequence = toSparql.sequence(path)

/*
  SELECT *
  WHERE {
    ?path0 schema:employee ?path1 .
    ?path1 schema:spouse ?path2 .
    ?path2 schema:name ?path3 .
  }
 */
const query = sequence.reduce((query, segment, index) => {
  const subject = $rdf.variable(`path${index}`)
  const object = $rdf.variable(`path${index + 1}`)
    
  return query.WHERE`${subject} ${segment} ${object}`
}, SELECT.ALL)

Advanced options

Allow Named Node Sequence Paths

The SHACL specification requires that lists in Sequence Paths are blank nodes. However, some implementations may use Named Nodes instead. To allow that, you can manually create the SHACL Property Path object from a graph pointer and pass it to findNodes or toSparql:

import type { GraphPointer } from 'clownface'
import { findNodes, fromNode } from 'clownface-shacl-path'

let pathNode: GraphPointer
let startNode: GraphPointer

const path = fromNode(pathNode, { allowNamedNodeSequencePaths: true })
const nodes = findNodes(startNode, path)

Advanced Property Path handling

If it is necessary to implement a custom logic for processing of Property Paths, create a class extending from PathVisitor.

import * as Path from 'clownface-shacl-path'
import type { GraphPointer } from 'clownface'

class MyVisitor extends Path.PathVisitor<TOut, TArg> {
  visitAlternativePath(path: Path.AlternativePath, arg?: TArg): TOut {
  }

  visitInversePath(path: Path.InversePath, arg?: TArg): TOut {
  }

  visitOneOrMorePath(path: Path.OneOrMorePath, arg?: TArg): TOut {
  }

  visitPredicatePath(path: Path.PredicatePath, arg?: TArg): TOut {
  }

  visitSequencePath(path: Path.SequencePath, arg?: TArg): TOut {
  }

  visitZeroOrMorePath(path: Path.ZeroOrMorePath, arg?: TArg): TOut {
  }

  visitZeroOrOnePath(path: Path.ZeroOrOnePath, arg?: TArg): TOut {
  }
}

The type arguments are optional. TOut defaults to void and TArg defaults to unknown.

See the classes ToSparqlPropertyPath and FindNodesVisitor for inspiration

To start visiting path nodes:

let pathNode: GraphPointer
const visitor = new MyVisitor()
  .visit(Path.fromPointer(pathNode)/*, optional initial arg */)

Changelog

2.1.0

Minor Changes

• daef669: Add option to allow sequence paths which are blank nodes