mcp4rdf / validator.py
RDF Validation Deployment
Initial deployment of RDF validation app to new mcp4rdf space
e344fcd
import os
import logging
import pandas as pd
import rdflib
from rdflib import Namespace, Literal, BNode, RDF, RDFS
from pyshacl import validate
# Set up basic logging (use DEBUG level to see detailed output)
logging.basicConfig(level=logging.DEBUG, format='%(asctime)s %(levelname)s: %(message)s')
BASE_DIR = os.path.join(os.path.dirname(__file__), "MonographDCTAP")
TSV_FILES = [
"MonographDCTAP/Monograph_Work_Text.tsv",
"MonographDCTAP/Monograph_AdminMetadata.tsv",
"MonographDCTAP/Monograph_Instance_Print.tsv",
"electronic_MonographDCTAP/Monograph_Instance_Electronic.tsv",
]
PREFIX_FILE = "./MonographDCTAP/Monograph_Prefixes.tsv"
# Add a global constant for fixed prefixes.
FIXED_PREFIXES = {
"bf": "http://id.loc.gov/ontologies/bibframe/",
"bflc": "http://id.loc.gov/ontologies/bflc/",
"rdfs": "http://www.w3.org/2000/01/rdf-schema#",
"big": "https://example.org/"
}
# Replace load_prefixes() with a simplified function:
def load_prefixes(prefixes_file):
logging.info("Using hardcoded prefixes:")
for p, ns in FIXED_PREFIXES.items():
logging.info(f"{p} -> {ns}")
return FIXED_PREFIXES
# Optionally simplify register_prefixes and _bind_namespaces:
def register_prefixes(graph, prefixes):
for prefix, uri in prefixes.items():
graph.bind(prefix, Namespace(uri), override=True)
def _bind_namespaces(graph: rdflib.Graph):
# Hard-code the fixed namespaces as well.
graph.namespace_manager.bind("bf", Namespace(FIXED_PREFIXES["bf"]))
graph.namespace_manager.bind("bflc", Namespace(FIXED_PREFIXES["bflc"]))
graph.namespace_manager.bind("rdfs", Namespace(FIXED_PREFIXES["rdfs"]))
graph.namespace_manager.bind("big", Namespace(FIXED_PREFIXES["big"]))
def _prop_id_to_uri(property_id, prefixes):
if ":" in property_id:
prefix, suffix = property_id.split(":", 1)
ns = prefixes.get(prefix.strip())
if ns:
return rdflib.URIRef(ns + suffix.strip())
if property_id.startswith("http"):
return rdflib.URIRef(property_id)
return Literal(property_id)
def add_shape_from_row(graph, row, prefixes):
shape_uri = rdflib.URIRef(row['shapeID'])
logging.info(f"Processing shape: {shape_uri}")
if (shape_uri, RDF.type, rdflib.URIRef("http://www.w3.org/ns/shacl#NodeShape")) not in graph:
graph.add((shape_uri, RDF.type, rdflib.URIRef("http://www.w3.org/ns/shacl#NodeShape")))
graph.add((shape_uri, RDFS.label, Literal(row['shapeLabel'])))
logging.info(f"Added NodeShape: {shape_uri} with label {row['shapeLabel']}")
targets = [t.strip() for t in str(row['target']).split(";")]
for target in targets:
target_uri = _prop_id_to_uri(target, prefixes)
graph.add((shape_uri, rdflib.URIRef("http://www.w3.org/ns/shacl#targetClass"), target_uri))
logging.info(f"Added target '{target_uri}' to shape {shape_uri}")
# If the property is mandatory, add a SPARQLTarget to force evaluation of nodes missing the property.
if str(row['mandatory']).strip().lower() == "true":
property_uri = _prop_id_to_uri(row['propertyID'], prefixes)
target_uris = [ _prop_id_to_uri(t, prefixes) for t in targets ]
union_clause = " UNION ".join([f"{{ ?this a <{uri}> }}" for uri in target_uris])
query = f"SELECT ?this WHERE {{ {union_clause} FILTER NOT EXISTS {{ ?this <{property_uri}> ?o }} }}"
bnode = BNode()
sh = rdflib.URIRef("http://www.w3.org/ns/shacl#")
# Use RDF.type triple to mark the bnode as a SPARQLTarget
graph.add((bnode, RDF.type, rdflib.URIRef("http://www.w3.org/ns/shacl#SPARQLTarget")))
graph.add((bnode, rdflib.URIRef("http://www.w3.org/ns/shacl#select"), Literal(query)))
graph.add((shape_uri, rdflib.URIRef("http://www.w3.org/ns/shacl#target"), bnode))
logging.info(f"Added SPARQLTarget with query: {query} to shape {shape_uri}")
property_bnode = BNode()
graph.add((shape_uri, rdflib.URIRef("http://www.w3.org/ns/shacl#property"), property_bnode))
graph.add((property_bnode, RDF.type, rdflib.URIRef("http://www.w3.org/ns/shacl#PropertyShape")))
graph.add((property_bnode, RDFS.label, Literal(row['propertyLabel'])))
path_uri = _prop_id_to_uri(row['propertyID'], prefixes)
graph.add((property_bnode, rdflib.URIRef("http://www.w3.org/ns/shacl#path"), path_uri))
logging.info(f"Added property shape for property {row['propertyID']} with label {row['propertyLabel']}")
if str(row['mandatory']).strip().lower() == "true":
graph.add((property_bnode, rdflib.URIRef("http://www.w3.org/ns/shacl#minCount"), Literal(1)))
logging.info(f"Set minCount 1 for property {row['propertyID']}")
if str(row['repeatable']).strip().lower() == "false":
graph.add((property_bnode, rdflib.URIRef("http://www.w3.org/ns/shacl#maxCount"), Literal(1)))
logging.info(f"Set maxCount 1 for property {row['propertyID']}")
severity = str(row.get("severity", "")).strip()
if severity:
sev_ns = rdflib.URIRef("http://www.w3.org/ns/shacl#")
if severity == "Violation":
graph.add((property_bnode, rdflib.URIRef("http://www.w3.org/ns/shacl#severity"),
rdflib.URIRef(sev_ns + "Violation")))
elif severity == "Warning":
graph.add((property_bnode, rdflib.URIRef("http://www.w3.org/ns/shacl#severity"),
rdflib.URIRef(sev_ns + "Warning")))
else:
graph.add((property_bnode, rdflib.URIRef("http://www.w3.org/ns/shacl#severity"),
rdflib.URIRef(sev_ns + "Info")))
logging.info(f"Set severity {severity} for property {row['propertyID']}")
if pd.notna(row.get("valueShape")) and row["valueShape"].strip():
value_shape_uri = _prop_id_to_uri(row["valueShape"], prefixes)
graph.add((property_bnode, rdflib.URIRef("http://www.w3.org/ns/shacl#node"), value_shape_uri))
logging.info(f"Linked valueShape {value_shape_uri} for property {row['propertyID']}")
return graph
def build_shacl_graphs():
logging.info("Building individual SHACL graphs from TSV files")
module_graphs = {} # Initialize the dictionary for module graphs
prefixes = load_prefixes(PREFIX_FILE)
for tsv in TSV_FILES:
tsv_path = tsv # already an absolute path
if not os.path.exists(tsv_path):
logging.error(f"TSV file not found: {tsv_path}")
logging.info(f"Processing TSV file: {tsv_path}")
graph = rdflib.Graph()
register_prefixes(graph, prefixes)
_bind_namespaces(graph) # Bind fixed namespaces for the SHACL graph
df = pd.read_csv(tsv_path, sep='\t', comment='/')
for _, row in df.iterrows():
if pd.isna(row.get("shapeID")):
continue
add_shape_from_row(graph, row, prefixes)
module_graphs[tsv] = graph
logging.info("Completed building individual SHACL graphs")
return module_graphs
def parse_results_text(results_text: str) -> str:
"""
Parse and reformat raw results_text for nicer display.
Adjust the logic to suit your output format.
"""
lines = results_text.strip().splitlines()
formatted_lines = []
for line in lines:
line = line.strip()
if line.startswith("==="):
# Start of a module section
formatted_lines.append("\n" + line)
elif line.startswith("Validation Result"):
# Start a new violation
formatted_lines.append("\n" + line)
else:
formatted_lines.append("\t" + line)
return "\n".join(formatted_lines)
def validate_rdf(rdf_data, template):
logging.info("Starting validation")
data_graph = rdflib.Graph()
logging.info("Parsing RDF data")
try:
data_graph.parse(data=rdf_data, format='xml')
except Exception as e:
logging.error(f"Error parsing RDF data: {e}")
raise e
logging.info(f"Data graph has {len(data_graph)} triples.")
# Bind known namespaces explicitly from the input RDF/XML
namespaces = {
"bf": "http://id.loc.gov/ontologies/bibframe/",
"bflc": "http://id.loc.gov/ontologies/bflc/",
"bfsimple": "http://id.loc.gov/ontologies/bfsimple/",
"cc": "http://creativecommons.org/ns#",
"datatypes": "http://id.loc.gov/datatypes/",
"dcterms": "http://purl.org/dc/terms/",
"foaf": "http://xmlns.com/foaf/0.1/",
"lcc": "http://id.loc.gov/ontologies/lcc#",
"lclocal": "http://id.loc.gov/ontologies/lclocal/",
"madsrdf": "http://www.loc.gov/mads/rdf/v1#",
"mnotetype": "http://id.loc.gov/vocabulary/mnotetype/",
"mstatus": "https://id.loc.gov/vocabulary/mstatus/",
"owl": "http://www.w3.org/2002/07/owl#",
"pmo": "http://performedmusicontology.org/ontology/",
"rdf": "http://www.w3.org/1999/02/22-rdf-syntax-ns#",
"rdfs": "http://www.w3.org/2000/01/rdf-schema#",
"skos": "http://www.w3.org/2004/02/skos/core#",
"vartitletype": "http://id.loc.gov/vocabulary/vartitletype/",
"void": "http://rdfs.org/ns/void#",
"xsd": "http://www.w3.org/2001/XMLSchema#"
}
for prefix, uri in namespaces.items():
data_graph.bind(prefix, uri)
logging.info(f"Data graph has {len(data_graph)} triples.")
for s, p, o in list(data_graph)[:10]:
logging.debug(f"Parsed triple: {s} {p} {o}")
# New: Log the full RDF graph in turtle format
serialized_graph = data_graph.serialize(format='turtle')
logging.info("Full RDF graph:\n" + (serialized_graph.decode('utf-8') if isinstance(serialized_graph, bytes) else serialized_graph))
# Extra debugging: log all rdf:type values from the data graph
classes = set()
for s, o in data_graph.subject_objects(RDF.type):
classes.add(o)
logging.debug(f"Data graph contains these types: {list(classes)}")
# === Added debugging to check expected target class URIs ===
prefixes = load_prefixes(PREFIX_FILE)
# List your expected target class identifiers as they are used in your TSV
expected_targets = ["https:Agent", "big:Contribution"]
expanded_targets = [ _prop_id_to_uri(t, prefixes) for t in expected_targets ]
logging.debug(f"Expected target classes per TSV: {expanded_targets}")
if template.lower() == 'monograph':
logging.info("Using Monograph template; processing individual TSV modules")
module_graphs = build_shacl_graphs()
# Debug: inspect declared target classes in each module and query focus nodes.
for tsv, module in module_graphs.items():
logging.debug(f"Module {tsv} declared targets:")
for shape in module.subjects(RDF.type, rdflib.URIRef("http://www.w3.org/ns/shacl#NodeShape")):
for target in module.objects(shape, rdflib.URIRef("http://www.w3.org/ns/shacl#targetClass")):
logging.debug(f"Shape {shape} declares target: {target}")
q = f"SELECT ?x WHERE {{ ?x a <{target}> . }}"
matches = list(data_graph.query(q))
logging.debug(f"Found {len(matches)} focus node(s) for target {target}")
for match in matches:
logging.debug(f"Focus node: {match.x}")
all_results = []
overall_conforms = True
for tsv, graph in module_graphs.items():
shacl_text = graph.serialize(format='turtle')
logging.info(f"Module {tsv} SHACL shapes:")
logging.info(shacl_text.decode('utf-8') if isinstance(shacl_text, bytes) else shacl_text)
conforms, results_graph, results_text = validate(data_graph, shacl_graph=graph, inference='rdfs', debug=True)
# Override conform status if any violation has severity sh:Violation.
violation_query = """
PREFIX sh: <http://www.w3.org/ns/shacl#>
SELECT ?severity WHERE {
?vr a sh:ValidationResult ;
sh:resultSeverity ?severity .
}
"""
severities = [str(row.severity) for row in results_graph.query(violation_query)]
module_conforms = False if any("http://www.w3.org/ns/shacl#Violation" in s for s in severities) else True
logging.info(f"Module {tsv} - Overridden Conforms: {module_conforms}")
# Build a nicely formatted summary of the results.
query_formatted = """
PREFIX sh: <http://www.w3.org/ns/shacl#>
SELECT ?component ?severity ?sourceShape ?focus ?resultPath ?message
WHERE {
?vr a sh:ValidationResult ;
sh:sourceConstraintComponent ?component ;
sh:resultSeverity ?severity ;
sh:sourceShape ?sourceShape ;
sh:focusNode ?focus ;
sh:resultPath ?resultPath ;
sh:resultMessage ?message .
}
ORDER BY ?component
"""
formatted_results = ""
count = 0
for row in results_graph.query(query_formatted):
count += 1
formatted_results += f"Validation Result in {row.component}:\n"
formatted_results += f"\tSeverity: {row.severity}\n"
formatted_results += f"\tSource Shape: {row.sourceShape}\n"
formatted_results += f"\tFocus Node: {row.focus}\n"
formatted_results += f"\tResult Path: {row.resultPath}\n"
formatted_results += f"\tMessage: {row.message}\n"
formatted_results = f"Results ({count}):\n" + formatted_results
# Assemble module output.
module_output = (
f"\n=== Module: {tsv} ===\n"
f"Overridden Conforms: {module_conforms}\n"
f"{formatted_results}\n"
"------------------------\n"
)
all_results.append(module_output)
if not module_conforms:
overall_conforms = False
combined_results = "\n".join(all_results)
# Optionally, parse the combined results for easier display.
combined_results = parse_results_text(combined_results)
return overall_conforms, combined_results
else:
logging.info("Using default SHACL template")
shacl_text = """
@prefix sh: <http://www.w3.org/ns/shacl#> .
@prefix ex: <http://example.org/> .
ex:DefaultShape a sh:NodeShape ;
sh:targetNode ex:SomeNode ;
sh:property [
sh:path ex:someProperty ;
sh:datatype xsd:string ;
] .
"""
shacl_graph = rdflib.Graph()
shacl_graph.parse(data=shacl_text, format='turtle')
conforms, results_graph, results_text = validate(data_graph, shacl_graph=shacl_graph, inference='rdfs', debug=True)
logging.info(f"Validation completed; Conforms: {conforms}")
logging.info("Results text:")
logging.info(results_text)
serialized_results = results_graph.serialize(format='turtle')
logging.info("Detailed results graph:")
logging.info(serialized_results.decode('utf-8') if isinstance(serialized_results, bytes) else serialized_results)
combined_results = (f"{results_text.strip()}\nDetailed Results:\n"
f"{serialized_results.decode('utf-8') if isinstance(serialized_results, bytes) else serialized_results}")
return conforms, combined_results