feat(rdf-api): Add a general-purpose SHACL validation utility (DSP-930) #1762
Conversation
| * Recursively loads graphs of SHACL shapes from a directory and its subdirectories. | ||
| * | ||
| * @param baseDir the base directory that SHACL graphs are loaded from. | ||
| * @param dir the base directory or a subdirectory. |
There was a problem hiding this comment.
It seems like the Shacl shapes are actually loaded from dir directory and its subdirectories, and baseDir is only used to get its path to make the relativePath. Can you please change the comments for baseDir and dir in scaladocs?
I also wonder, if only the path of the baseDir is needed here, why not just using its path as the parameter of this function, I mean
private def loadShaclGraphs(baseDirPath: Path, dir: File) : Map[Path, ShaclGraphT] = {...}
two directory parameters are a bit confusing.
There was a problem hiding this comment.
It seems like the Shacl shapes are actually loaded from
dirdirectory and its subdirectories, andbaseDiris only used to get its path to make therelativePath.... I also wonder, if only the path of thebaseDiris needed here, why not just using its path as the parameter of this function
Actually, the SHACL shapes are loaded from baseDir and its subdirectories. File and Path are basically the same thing, but File is the old java.io API which I'm more used to, and Path is the newer java.nio API. The consensus in the Java community seems to be that seems that Path should be used instead of File for everything nowadays (https://stackoverflow.com/a/26658436). I guess we should do a PR sometime to use java.nio instead of java.io everywhere. For now, I've refactored AbstractShaclValidator to use java.nio to walk the directory tree, which seems clearer.
| val graphToValidate: jena.graph.Graph = rdfModel.asJenaDataset.asDatasetGraph.getDefaultGraph | ||
|
|
||
| // Validate the data and get Jena's validation report. | ||
| val validationReport: jena.shacl.ValidationReport = shaclValidator.validate(shaclGraph, graphToValidate) |
There was a problem hiding this comment.
Nice! Jena Shacl validate is like pyShacl getting the same parameters and returning report and conforms in the same way.
resolves DSP-930.
ShaclValidator.scala.JenaShaclValidatorRDF4JShaclValidatorRdfFeatureFactory.