validate-examples

#!/usr/bin/env python

import argparse
import os
from pyshacl import validate
from rdflib import Graph

def load_rdfs_files(directory):
    rdfs_graph = Graph()
    for root, _, files in os.walk(directory):
        for filename in files:
            if filename.endswith(".ttl"):
                print(f"Loading RDFS file: {filename}")
                rdfs_file = os.path.join(root, filename)
                rdfs_graph.parse(rdfs_file, format="turtle")
    return rdfs_graph

def load_context_files(context_files):
    context_graph = Graph()
    for file in context_files:
        print(f"Loading context file: {file}")
        context_graph.parse(file, format="turtle")
    return context_graph

def validate_turtle_against_shacl(turtle_file, shacl_file, rdfs_directory=None, context_files=None):
    # Load Turtle file
    data_graph = Graph()
    data_graph.parse(turtle_file, format="turtle")

    # Load SHACL file
    shacl_graph = Graph()
    shacl_graph.parse(shacl_file, format="turtle")

    # Load RDFS files if directory is provided
    if rdfs_directory:
        rdfs_graph = load_rdfs_files(rdfs_directory)
    else:
        rdfs_graph = Graph()

    # Load context files
    if context_files:
        context_graph = load_context_files(context_files)
    else:
        context_graph = Graph()

    # Validate
    conforms, results_graph, results_text = validate(data_graph=data_graph+context_graph+rdfs_graph, shacl_graph=shacl_graph)

    return conforms, results_text

def validate_turtle_files_in_directory(directory, shacl_file, rdfs_directory=None, context_files=None):
    for root, _, files in os.walk(directory):
        for filename in files:
            if filename.endswith(".ttl"):
                turtle_file = os.path.join(root, filename)
                print(f"Validating {turtle_file} against SHACL...")
                conforms, results_text = validate_turtle_against_shacl(turtle_file, shacl_file, rdfs_directory, context_files)
                print(results_text)
                print(f"{turtle_file} conforms to SHACL constraints: {conforms}")
                print("=" * 50)

def main():
    parser = argparse.ArgumentParser(description="Validate Turtle files against a SHACL file")
    parser.add_argument("directory", help="Directory containing Turtle files")
    parser.add_argument("shacl_file", help="SHACL file")
    parser.add_argument("--rdfs-directory", help="Directory containing RDFS files for inference (optional)")
    parser.add_argument("--context-files", nargs="+", help="Arbitrary list of context Turtle files for inference (optional)")
    args = parser.parse_args()

    validate_turtle_files_in_directory(args.directory, args.shacl_file, args.rdfs_directory, args.context_files)

if __name__ == "__main__":
    main()