@netwerk-digitaal-erfgoed/network-of-terms-query 6.2.9 → 6.2.11

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@netwerk-digitaal-erfgoed/network-of-terms-query",
-  "version": "6.2.9",
+  "version": "6.2.11",
   "description": "Engine for querying sources in the Network of Terms",
   "keywords": [],
   "homepage": "https://github.com/netwerk-digitaal-erfgoed/network-of-terms#readme",
@@ -30,19 +30,19 @@
   "main": "./dist/index.js",
   "types": "./dist/index.d.ts",
   "dependencies": {
-    "@comunica/query-sparql": "^4.3.0",
-    "@comunica/types": "^4.2.0",
-    "@comunica/utils-bindings-factory": "^4.2.0",
+    "@comunica/query-sparql": "^5.1.2",
+    "@comunica/types": "^5.1.0",
+    "@comunica/utils-bindings-factory": "^5.1.0",
     "@hapi/hoek": "^11.0.7",
     "@opentelemetry/api": "^1.9.0",
-    "@opentelemetry/exporter-metrics-otlp-proto": "0.207.0",
+    "@opentelemetry/exporter-metrics-otlp-proto": "0.208.0",
     "@opentelemetry/resources": "2.2.0",
     "@opentelemetry/sdk-metrics": "2.2.0",
-    "@opentelemetry/semantic-conventions": "1.37.0",
+    "@opentelemetry/semantic-conventions": "1.38.0",
     "@rdfjs/types": "^2.0.1",
     "env-schema": "^6.0.1",
     "jest-dev-server": "11.0.0",
-    "joi": "^17.13.3",
+    "joi": "^18.0.2",
     "nock": "^14.0.10",
     "pino": "^9.7.0",
     "pretty-ms": "^9.2.0",
@@ -51,7 +51,7 @@
     "tslib": "^2.3.0"
   },
   "devDependencies": {
-    "@comunica/query-sparql-file": "^4.3.0",
+    "@comunica/query-sparql-file": "^5.1.2",
     "asynciterator": "^3.9.0"
   }
 }

package/src/query.ts

@@ -4,7 +4,6 @@ import { LoggerPino } from './helpers/logger-pino.js';
 import Pino from 'pino';
 import PrettyMilliseconds from 'pretty-ms';
 import * as RDF from '@rdfjs/types';
-import { Bindings } from '@rdfjs/types';
 import { Term, TermsTransformer } from './terms.js';
 import { QueryMode, queryVariants } from './search/query-mode.js';
 import { Dataset, Distribution, IRI } from './catalog.js';
@@ -14,6 +13,50 @@ import { DataFactory } from 'rdf-data-factory';
 import { sourceQueriesHistogram } from './instrumentation.js';
 import { config } from './config.js';
 
+/**
+ * Check if a query requires string substitution instead of initialBindings.
+ * Workaround for Comunica v5 traqula bug that crashes with:
+ * - SERVICE clauses
+ * - VALUES combination
+ */
+function requiresStringSubstitution(query: string): boolean {
+  const hasService = /\bSERVICE\b/i.test(query);
+  const hasValues = /\bVALUES\b/i.test(query);
+  return hasService || hasValues;
+}
+
+/**
+ * Substitute bindings directly into a SPARQL query string.
+ * This is a workaround for Comunica v5's initialBindings bug with SERVICE clauses.
+ */
+function substituteBindings(
+  query: string,
+  bindings: Record<string, RDF.Term>,
+): string {
+  let result = query;
+  for (const [name, term] of Object.entries(bindings)) {
+    const pattern = new RegExp(`\\?${name}\\b`, 'g');
+    if (term.termType === 'NamedNode') {
+      result = result.replace(pattern, `<${term.value}>`);
+    } else if (term.termType === 'Literal') {
+      const literal = term as RDF.Literal;
+      const datatype = literal.datatype?.value;
+      if (
+        datatype &&
+        datatype !== 'http://www.w3.org/2001/XMLSchema#string' &&
+        datatype !== 'http://www.w3.org/1999/02/22-rdf-syntax-ns#langString'
+      ) {
+        result = result.replace(pattern, `"${term.value}"^^<${datatype}>`);
+      } else if (literal.language) {
+        result = result.replace(pattern, `"${term.value}"@${literal.language}`);
+      } else {
+        result = result.replace(pattern, `"${term.value}"`);
+      }
+    }
+  }
+  return result;
+}
+
 export type TermsResult = Terms | TimeoutError | ServerError;
 
 export class TermsResponse {
@@ -141,10 +184,21 @@ export class QueryTermsService {
     const logger = new LoggerPino({ logger: this.logger });
     // Extract HTTP credentials if the distribution URL contains any.
     const url = new URL(distribution.endpoint.toString());
-    this.logger.info(`Querying "${url}" with "${query}"...`);
-    const quadStream = await this.engine.queryQuads(query, {
+
+    // Workaround for https://github.com/comunica/comunica/issues/1655, so use
+    // string substitution instead of initialBindings for:
+    // - SERVICE clauses crash with initialBindings
+    // - VALUES crashes in some combinations
+    const useStringSubstitution = requiresStringSubstitution(query);
+    const finalQuery = useStringSubstitution
+      ? substituteBindings(query, bindings)
+      : query;
+
+    this.logger.info(`Querying "${url}" with "${finalQuery}"...`);
+    const quadStream = await this.engine.queryQuads(finalQuery, {
       log: logger,
-      httpAuth: url.username === '' ? '' : url.username + ':' + url.password,
+      httpAuth:
+        url.username === '' ? undefined : url.username + ':' + url.password,
       httpTimeout: timeoutMs,
       noCache: true,
       sources: [
@@ -153,9 +207,10 @@ export class QueryTermsService {
           value: url.origin + url.pathname,
         },
       ],
-      initialBindings: bindingsFactory.fromRecord(
-        bindings,
-      ) as unknown as Bindings,
+      // Only pass initialBindings when NOT using string substitution
+      ...(useStringSubstitution
+        ? {}
+        : { initialBindings: bindingsFactory.fromRecord(bindings) }),
     });
 
     return new Promise((resolve) => {

package/src/test-utils.ts

@@ -187,7 +187,7 @@ export const testCatalog = (port: number) =>
         new SparqlDistribution(
           'https://data.beeldengeluid.nl/id/datadownload/0026',
           'https://username:password@gtaa.apis.beeldengeluid.nl/sparql',
-          'CONSTRUCT { ?s ?p ?o } WHERE { ?s ?p ?o }',
+          'CONSTRUCT { ?s ?p ?o } WHERE { ?s skos:inScheme ?datasetUri ; ?p ?o }',
           'CONSTRUCT { ?s ?p ?o } WHERE { ?s ?p ?o }',
         ),
       ],

package/test/query.test.ts

@@ -21,11 +21,12 @@ describe('Query', () => {
     vi.clearAllMocks();
   });
   it('passes dataset IRI query parameter to Comunica', async () => {
+    // Use GTAA which doesn't have VALUES, so uses initialBindings (not string substitution)
     const config = await query(
-      'https://data.netwerkdigitaalerfgoed.nl/rkd/rkdartists/sparql',
+      'https://data.beeldengeluid.nl/id/datadownload/0026',
     );
     expect(config.initialBindings.get('datasetUri')?.value).toEqual(
-      'https://data.rkd.nl/rkdartists',
+      'http://data.beeldengeluid.nl/gtaa/Persoonsnamen',
     );
   });
 

package/vite.config.ts

@@ -16,11 +16,11 @@ export default defineConfig(() => ({
       provider: 'v8' as const,
       thresholds: {
         autoUpdate: true,
-        lines: 58.24,
-        functions: 40,
-        branches: 96.36,
-        statements: 58.24,
+        lines: 56.81,
+        functions: 40.25,
+        branches: 91.22,
+        statements: 56.81,
       },
     },
   },
-}));
+}));