@lde/pipeline 0.30.13 → 0.30.15

package/README.md

@@ -357,7 +357,7 @@ new Pipeline({
 
 #### `provenancePlugin()`
 
-Appends [PROV-O](https://www.w3.org/TR/prov-o/) provenance quads (`prov:Entity`, `prov:Activity`, `prov:startedAtTime`, `prov:endedAtTime`) to every stage’s output.
+Appends [PROV-O](https://www.w3.org/TR/prov-o/) provenance quads (`prov:Entity`, `prov:Activity`, `prov:startedAtTime`, `prov:endedAtTime`) to every stage’s output. The `prov:Activity` is a stable IRI keyed on `(dataset, stage)`, not a blank node, so activities stay distinct – and a re-run stays idempotent – when per-dataset outputs are merged into one graph (blank-node labels are not unique across separately serialised documents and would fuse unrelated activities).
 
 #### `schemaOrgNormalizationPlugin(options?)`
 

package/dist/distribution/report.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"report.d.ts","sourceRoot":"","sources":["../../src/distribution/report.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,sBAAsB,CAAC;AACzD,OAAO,EAAe,KAAK,IAAI,EAAE,MAAM,IAAI,CAAC;AAC5C,OAAO,EAIL,KAAK,eAAe,EACrB,MAAM,yBAAyB,CAAC;AAUjC;;;;;;;;;GASG;AACH,wBAAuB,mBAAmB,CACxC,YAAY,EAAE,eAAe,EAAE,EAC/B,UAAU,EAAE,MAAM,EAClB,YAAY,CAAC,EAAE,YAAY,GAC1B,aAAa,CAAC,IAAI,CAAC,CA2CrB"}
+{"version":3,"file":"report.d.ts","sourceRoot":"","sources":["../../src/distribution/report.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,sBAAsB,CAAC;AAEzD,OAAO,EAAe,KAAK,IAAI,EAAE,MAAM,IAAI,CAAC;AAC5C,OAAO,EAIL,KAAK,eAAe,EACrB,MAAM,yBAAyB,CAAC;AAUjC;;;;;;;;;GASG;AACH,wBAAuB,mBAAmB,CACxC,YAAY,EAAE,eAAe,EAAE,EAC/B,UAAU,EAAE,MAAM,EAClB,YAAY,CAAC,EAAE,YAAY,GAC1B,aAAa,CAAC,IAAI,CAAC,CAiDrB"}

package/dist/distribution/report.js

@@ -1,6 +1,7 @@
+import { hashSuffix, skolemIri } from '@lde/dataset';
 import { DataFactory } from 'n3';
 import { NetworkError, SparqlProbeResult, } from '@lde/distribution-probe';
-const { quad, namedNode, blankNode, literal } = DataFactory;
+const { quad, namedNode, literal } = DataFactory;
 const RDF = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#';
 const SCHEMA = 'https://schema.org/';
 const VOID = 'http://rdfs.org/ns/void#';
@@ -17,10 +18,16 @@ const HTTP_STATUS = 'https://www.w3.org/2011/http-statusCodes#';
  * whose `schema:target` matches the failed distribution's access URL.
  */
 export async function* probeResultsToQuads(probeResults, datasetIri, importResult) {
-    // Track blank nodes per URL so import errors can reference the right action.
+    // Track each action node per URL so import errors can reference the right
+    // action. Each action is a deterministic IRI keyed on (dataset, URL), not a
+    // blank node: this output is merged with other datasets' into one cat-built
+    // graph where blank-node labels are not unique across documents and would
+    // fuse unrelated actions into one node (see issue #474). The
+    // `.well-known/schema#action-<hash>` shape mirrors the linkset skolem.
+    const actionBase = `${datasetIri}/.well-known/schema#action`;
     const actionsByUrl = new Map();
     for (const result of probeResults) {
-        const action = blankNode();
+        const action = namedNode(skolemIri(actionBase, hashSuffix(result.url)));
         actionsByUrl.set(result.url, action);
         yield quad(action, namedNode(`${RDF}type`), namedNode(`${SCHEMA}Action`));
         yield quad(action, namedNode(`${SCHEMA}target`), namedNode(result.url));

package/dist/guard/blankNodes.d.ts

@@ -0,0 +1,38 @@
+import type { Quad } from '@rdfjs/types';
+import type { QuadTransform } from '../stage.js';
+/**
+ * Why this guard exists.
+ *
+ * A file-based served store (e.g. the Dataset Knowledge Graph) rebuilds its
+ * index by concatenating every per-dataset n-quads file and parsing the
+ * concatenation as ONE RDF document (`qlever index` over
+ * `find … -exec cat {} +`). Blank-node labels are only document-scoped, and the
+ * pipeline emits deterministic labels (n3 `DataFactory.blankNode()` → `n3-N`,
+ * the counter resets per dataset/run), so the same label recurs across files and
+ * the indexer fuses those nodes into one — merging unrelated provenance,
+ * measurements and linksets across datasets and runs. Named nodes never fuse.
+ *
+ * The invariant for any quads the pipeline writes into such a store is therefore:
+ * NO blank nodes. Mint stable (skolem) IRIs instead — see `skolemIri` in
+ * `@lde/dataset`. These helpers make that invariant testable and enforceable.
+ *
+ * See ldelements/lde#474 and netwerk-digitaal-erfgoed/dataset-knowledge-graph#352.
+ */
+/**
+ * The distinct blank-node labels appearing in subject, object, or graph position
+ * across `quads`. Empty when the quads are blank-node-free.
+ */
+export declare function blankNodes(quads: Iterable<Quad>): string[];
+/**
+ * Throw if any quad carries a blank node. Use in producer tests to lock in the
+ * no-blank-nodes invariant (see module docs).
+ */
+export declare function assertNoBlankNodes(quads: Iterable<Quad>): void;
+/**
+ * A {@link QuadTransform} that passes quads through unchanged but throws on the
+ * first blank node it sees. Insert it just before the writer to turn the
+ * no-blank-nodes invariant into a hard pipeline failure (e.g. in a CI/staging
+ * run) rather than a per-test opt-in.
+ */
+export declare function failOnBlankNodes<Context>(): QuadTransform<Context>;
+//# sourceMappingURL=blankNodes.d.ts.map

package/dist/guard/blankNodes.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"blankNodes.d.ts","sourceRoot":"","sources":["../../src/guard/blankNodes.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,IAAI,EAAE,MAAM,cAAc,CAAC;AACzC,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,aAAa,CAAC;AAEjD;;;;;;;;;;;;;;;;;GAiBG;AAEH;;;GAGG;AACH,wBAAgB,UAAU,CAAC,KAAK,EAAE,QAAQ,CAAC,IAAI,CAAC,GAAG,MAAM,EAAE,CAU1D;AAED;;;GAGG;AACH,wBAAgB,kBAAkB,CAAC,KAAK,EAAE,QAAQ,CAAC,IAAI,CAAC,GAAG,IAAI,CAU9D;AAED;;;;;GAKG;AACH,wBAAgB,gBAAgB,CAAC,OAAO,KAAK,aAAa,CAAC,OAAO,CAAC,CAelE"}

package/dist/guard/blankNodes.js

@@ -0,0 +1,66 @@
+/**
+ * Why this guard exists.
+ *
+ * A file-based served store (e.g. the Dataset Knowledge Graph) rebuilds its
+ * index by concatenating every per-dataset n-quads file and parsing the
+ * concatenation as ONE RDF document (`qlever index` over
+ * `find … -exec cat {} +`). Blank-node labels are only document-scoped, and the
+ * pipeline emits deterministic labels (n3 `DataFactory.blankNode()` → `n3-N`,
+ * the counter resets per dataset/run), so the same label recurs across files and
+ * the indexer fuses those nodes into one — merging unrelated provenance,
+ * measurements and linksets across datasets and runs. Named nodes never fuse.
+ *
+ * The invariant for any quads the pipeline writes into such a store is therefore:
+ * NO blank nodes. Mint stable (skolem) IRIs instead — see `skolemIri` in
+ * `@lde/dataset`. These helpers make that invariant testable and enforceable.
+ *
+ * See ldelements/lde#474 and netwerk-digitaal-erfgoed/dataset-knowledge-graph#352.
+ */
+/**
+ * The distinct blank-node labels appearing in subject, object, or graph position
+ * across `quads`. Empty when the quads are blank-node-free.
+ */
+export function blankNodes(quads) {
+    const offenders = new Set();
+    for (const quad of quads) {
+        for (const term of [quad.subject, quad.object, quad.graph]) {
+            if (term.termType === 'BlankNode') {
+                offenders.add(term.value);
+            }
+        }
+    }
+    return [...offenders];
+}
+/**
+ * Throw if any quad carries a blank node. Use in producer tests to lock in the
+ * no-blank-nodes invariant (see module docs).
+ */
+export function assertNoBlankNodes(quads) {
+    const offenders = blankNodes(quads);
+    if (offenders.length > 0) {
+        throw new Error(`Output contains ${offenders.length} blank node(s), which fuse across ` +
+            `datasets when a file-based store cat-indexes per-dataset files. ` +
+            `Mint skolem IRIs instead (see skolemIri in @lde/dataset; ldelements/lde#474). ` +
+            `First: ${offenders.slice(0, 10).join(', ')}`);
+    }
+}
+/**
+ * A {@link QuadTransform} that passes quads through unchanged but throws on the
+ * first blank node it sees. Insert it just before the writer to turn the
+ * no-blank-nodes invariant into a hard pipeline failure (e.g. in a CI/staging
+ * run) rather than a per-test opt-in.
+ */
+export function failOnBlankNodes() {
+    return async function* (quads) {
+        for await (const quad of quads) {
+            for (const term of [quad.subject, quad.object, quad.graph]) {
+                if (term.termType === 'BlankNode') {
+                    throw new Error(`Blank node reached the writer (${term.value}); it would fuse ` +
+                        `across datasets in a cat-built index. Mint a skolem IRI instead ` +
+                        `(ldelements/lde#474): ${quad.subject.value} ${quad.predicate.value} …`);
+                }
+            }
+            yield quad;
+        }
+    };
+}

package/dist/index.d.ts

@@ -6,6 +6,7 @@ export * from './progressReporter.js';
 export * from './selector.js';
 export * from './stage.js';
 export * from './stageOutputResolver.js';
+export * from './guard/blankNodes.js';
 export * from './sparql/index.js';
 export * from './distribution/index.js';
 export * from './provenance/index.js';

package/dist/index.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAAA,cAAc,gBAAgB,CAAC;AAC/B,cAAc,iBAAiB,CAAC;AAChC,cAAc,YAAY,CAAC;AAC3B,cAAc,eAAe,CAAC;AAC9B,cAAc,uBAAuB,CAAC;AACtC,cAAc,eAAe,CAAC;AAC9B,cAAc,YAAY,CAAC;AAC3B,cAAc,0BAA0B,CAAC;AACzC,cAAc,mBAAmB,CAAC;AAClC,cAAc,yBAAyB,CAAC;AACxC,cAAc,uBAAuB,CAAC;AACtC,cAAc,mBAAmB,CAAC;AAClC,cAAc,oCAAoC,CAAC;AACnD,cAAc,wBAAwB,CAAC;AACvC,cAAc,oCAAoC,CAAC"}
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAAA,cAAc,gBAAgB,CAAC;AAC/B,cAAc,iBAAiB,CAAC;AAChC,cAAc,YAAY,CAAC;AAC3B,cAAc,eAAe,CAAC;AAC9B,cAAc,uBAAuB,CAAC;AACtC,cAAc,eAAe,CAAC;AAC9B,cAAc,YAAY,CAAC;AAC3B,cAAc,0BAA0B,CAAC;AACzC,cAAc,uBAAuB,CAAC;AACtC,cAAc,mBAAmB,CAAC;AAClC,cAAc,yBAAyB,CAAC;AACxC,cAAc,uBAAuB,CAAC;AACtC,cAAc,mBAAmB,CAAC;AAClC,cAAc,oCAAoC,CAAC;AACnD,cAAc,wBAAwB,CAAC;AACvC,cAAc,oCAAoC,CAAC"}

package/dist/index.js

@@ -6,6 +6,7 @@ export * from './progressReporter.js';
 export * from './selector.js';
 export * from './stage.js';
 export * from './stageOutputResolver.js';
+export * from './guard/blankNodes.js';
 export * from './sparql/index.js';
 export * from './distribution/index.js';
 export * from './provenance/index.js';

package/dist/pipeline.d.ts

@@ -8,6 +8,18 @@ import type { ProvenanceStore } from './provenance/store.js';
 import type { StageOutputResolver } from './stageOutputResolver.js';
 import type { ProgressReporter } from './progressReporter.js';
 import { type TimeoutPolicy } from './sparql/timeoutPolicy.js';
+/**
+ * Context handed to a {@link PipelinePlugin.beforeStageWrite} transform: the
+ * `dataset` whose merged output is being written and the `stage` that produced
+ * it. The stage identity lets a transform mint stable IRIs keyed on
+ * `(dataset, stage)` instead of blank nodes, which would fuse across stages and
+ * datasets once the per-dataset outputs are merged into one graph (see issue
+ * #474).
+ */
+export interface BeforeStageWriteContext {
+    dataset: Dataset;
+    stage: string;
+}
 /** Plugin that hooks into pipeline lifecycle events. */
 export interface PipelinePlugin {
     name: string;
@@ -17,9 +29,7 @@ export interface PipelinePlugin {
      * – provenance, namespace normalisation – that apply regardless of which
      * executor produced a quad.
      */
-    beforeStageWrite?: QuadTransform<{
-        dataset: Dataset;
-    }>;
+    beforeStageWrite?: QuadTransform<BeforeStageWriteContext>;
 }
 export interface PipelineOptions {
     datasetSelector: DatasetSelector;
@@ -66,6 +76,7 @@ export declare class Pipeline {
     private readonly datasetSelector;
     private readonly stages;
     private readonly writer;
+    private readonly beforeStageWrite?;
     private readonly distributionResolver;
     private readonly chaining?;
     private readonly reporter?;
@@ -79,6 +90,13 @@ export declare class Pipeline {
     private recordOutcome;
     private reportValidators;
     private collectStages;
+    /**
+     * The writer a stage's merged output is written through: the user writer
+     * wrapped with the plugins' {@link PipelinePlugin.beforeStageWrite}
+     * transforms, carrying this `stage`'s identity so a transform can mint stable
+     * per-`(dataset, stage)` IRIs rather than blank nodes.
+     */
+    private stageWriter;
     /**
      * Run a stage with reporting and return whether it was supported.
      * Returns `true` if the stage produced results, `false` if NotSupported.

package/dist/pipeline.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"pipeline.d.ts","sourceRoot":"","sources":["../src/pipeline.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,OAAO,EAAgB,MAAM,cAAc,CAAC;AAGrD,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,eAAe,CAAC;AACrD,OAAO,EAAE,KAAK,EAAE,MAAM,YAAY,CAAC;AACnC,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,YAAY,CAAC;AAChD,OAAO,KAAK,EAAE,MAAM,EAAE,MAAM,oBAAoB,CAAC;AAEjD,OAAO,EACL,KAAK,oBAAoB,EAG1B,MAAM,4BAA4B,CAAC;AAKpC,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,uBAAuB,CAAC;AAO7D,OAAO,KAAK,EAAE,mBAAmB,EAAE,MAAM,0BAA0B,CAAC;AACpE,OAAO,KAAK,EAEV,gBAAgB,EACjB,MAAM,uBAAuB,CAAC;AAE/B,OAAO,EAEL,KAAK,aAAa,EACnB,MAAM,2BAA2B,CAAC;AAEnC,wDAAwD;AACxD,MAAM,WAAW,cAAc;IAC7B,IAAI,EAAE,MAAM,CAAC;IACb;;;;;OAKG;IACH,gBAAgB,CAAC,EAAE,aAAa,CAAC;QAAE,OAAO,EAAE,OAAO,CAAA;KAAE,CAAC,CAAC;CACxD;AAED,MAAM,WAAW,eAAe;IAC9B,eAAe,EAAE,eAAe,CAAC;IACjC,MAAM,EAAE,KAAK,EAAE,CAAC;IAChB,OAAO,EAAE,MAAM,GAAG,MAAM,EAAE,CAAC;IAC3B,OAAO,CAAC,EAAE,cAAc,EAAE,CAAC;IAC3B,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,oBAAoB,CAAC,EAAE,oBAAoB,CAAC;IAC5C,QAAQ,CAAC,EAAE;QACT,mBAAmB,EAAE,mBAAmB,CAAC;QACzC,SAAS,EAAE,MAAM,CAAC;KACnB,CAAC;IACF,QAAQ,CAAC,EAAE,gBAAgB,CAAC;IAC5B;;;;;;OAMG;IACH,eAAe,CAAC,EAAE,eAAe,CAAC;IAClC;;;;;;OAMG;IACH,eAAe,CAAC,EAAE,MAAM,CAAC;IACzB;;;;;;;;;OASG;IACH,OAAO,CAAC,EAAE,MAAM,aAAa,CAAC;CAC/B;AAgFD,qBAAa,QAAQ;IACnB,OAAO,CAAC,QAAQ,CAAC,IAAI,CAAS;IAC9B,OAAO,CAAC,QAAQ,CAAC,eAAe,CAAkB;IAClD,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAU;IACjC,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAS;IAChC,OAAO,CAAC,QAAQ,CAAC,oBAAoB,CAAuB;IAC5D,OAAO,CAAC,QAAQ,CAAC,QAAQ,CAAC,CAA8B;IACxD,OAAO,CAAC,QAAQ,CAAC,QAAQ,CAAC,CAAmB;IAC7C,OAAO,CAAC,QAAQ,CAAC,cAAc,CAAsB;IACrD,OAAO,CAAC,QAAQ,CAAC,eAAe,CAAC,CAAkB;IACnD,OAAO,CAAC,QAAQ,CAAC,eAAe,CAAC,CAAS;gBAE9B,OAAO,EAAE,eAAe;IA0C9B,GAAG,IAAI,OAAO,CAAC,IAAI,CAAC;YAoBZ,cAAc;IA0I5B,+EAA+E;YACjE,aAAa;YAmBb,gBAAgB;IAW9B,OAAO,CAAE,aAAa;IAOtB;;;OAGG;YACW,QAAQ;IA0CtB,2EAA2E;YAC7D,eAAe;YAqBf,QAAQ;YA2DP,SAAS;CAczB"}
+{"version":3,"file":"pipeline.d.ts","sourceRoot":"","sources":["../src/pipeline.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,OAAO,EAAgB,MAAM,cAAc,CAAC;AAGrD,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,eAAe,CAAC;AACrD,OAAO,EAAE,KAAK,EAAE,MAAM,YAAY,CAAC;AACnC,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,YAAY,CAAC;AAChD,OAAO,KAAK,EAAE,MAAM,EAAE,MAAM,oBAAoB,CAAC;AAEjD,OAAO,EACL,KAAK,oBAAoB,EAG1B,MAAM,4BAA4B,CAAC;AAKpC,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,uBAAuB,CAAC;AAO7D,OAAO,KAAK,EAAE,mBAAmB,EAAE,MAAM,0BAA0B,CAAC;AACpE,OAAO,KAAK,EAEV,gBAAgB,EACjB,MAAM,uBAAuB,CAAC;AAE/B,OAAO,EAEL,KAAK,aAAa,EACnB,MAAM,2BAA2B,CAAC;AAEnC;;;;;;;GAOG;AACH,MAAM,WAAW,uBAAuB;IACtC,OAAO,EAAE,OAAO,CAAC;IACjB,KAAK,EAAE,MAAM,CAAC;CACf;AAED,wDAAwD;AACxD,MAAM,WAAW,cAAc;IAC7B,IAAI,EAAE,MAAM,CAAC;IACb;;;;;OAKG;IACH,gBAAgB,CAAC,EAAE,aAAa,CAAC,uBAAuB,CAAC,CAAC;CAC3D;AAED,MAAM,WAAW,eAAe;IAC9B,eAAe,EAAE,eAAe,CAAC;IACjC,MAAM,EAAE,KAAK,EAAE,CAAC;IAChB,OAAO,EAAE,MAAM,GAAG,MAAM,EAAE,CAAC;IAC3B,OAAO,CAAC,EAAE,cAAc,EAAE,CAAC;IAC3B,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,oBAAoB,CAAC,EAAE,oBAAoB,CAAC;IAC5C,QAAQ,CAAC,EAAE;QACT,mBAAmB,EAAE,mBAAmB,CAAC;QACzC,SAAS,EAAE,MAAM,CAAC;KACnB,CAAC;IACF,QAAQ,CAAC,EAAE,gBAAgB,CAAC;IAC5B;;;;;;OAMG;IACH,eAAe,CAAC,EAAE,eAAe,CAAC;IAClC;;;;;;OAMG;IACH,eAAe,CAAC,EAAE,MAAM,CAAC;IACzB;;;;;;;;;OASG;IACH,OAAO,CAAC,EAAE,MAAM,aAAa,CAAC;CAC/B;AAkFD,qBAAa,QAAQ;IACnB,OAAO,CAAC,QAAQ,CAAC,IAAI,CAAS;IAC9B,OAAO,CAAC,QAAQ,CAAC,eAAe,CAAkB;IAClD,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAU;IACjC,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAS;IAChC,OAAO,CAAC,QAAQ,CAAC,gBAAgB,CAAC,CAAyC;IAC3E,OAAO,CAAC,QAAQ,CAAC,oBAAoB,CAAuB;IAC5D,OAAO,CAAC,QAAQ,CAAC,QAAQ,CAAC,CAA8B;IACxD,OAAO,CAAC,QAAQ,CAAC,QAAQ,CAAC,CAAmB;IAC7C,OAAO,CAAC,QAAQ,CAAC,cAAc,CAAsB;IACrD,OAAO,CAAC,QAAQ,CAAC,eAAe,CAAC,CAAkB;IACnD,OAAO,CAAC,QAAQ,CAAC,eAAe,CAAC,CAAS;gBAE9B,OAAO,EAAE,eAAe;IA2C9B,GAAG,IAAI,OAAO,CAAC,IAAI,CAAC;YAoBZ,cAAc;IA0I5B,+EAA+E;YACjE,aAAa;YAmBb,gBAAgB;IAW9B,OAAO,CAAE,aAAa;IAOtB;;;;;OAKG;IACH,OAAO,CAAC,WAAW;IAMnB;;;OAGG;YACW,QAAQ;IA0CtB,2EAA2E;YAC7D,eAAe;YAqBf,QAAQ;YA+DP,SAAS;CAczB"}

package/dist/pipeline.js

@@ -63,15 +63,14 @@ class FanOutWriter {
 class TransformWriter {
     inner;
     transform;
-    constructor(inner, transform) {
+    stage;
+    constructor(inner, transform, stage) {
         this.inner = inner;
         this.transform = transform;
+        this.stage = stage;
     }
     async write(dataset, quads) {
-        await this.inner.write(dataset, this.transform(quads, { dataset }));
-    }
-    async flush(dataset) {
-        await this.inner.flush?.(dataset);
+        await this.inner.write(dataset, this.transform(quads, { dataset, stage: this.stage }));
     }
 }
 export class Pipeline {
@@ -79,6 +78,7 @@ export class Pipeline {
     datasetSelector;
     stages;
     writer;
+    beforeStageWrite;
     distributionResolver;
     chaining;
     reporter;
@@ -96,17 +96,18 @@ export class Pipeline {
         this.name = options.name ?? '';
         this.datasetSelector = options.datasetSelector;
         this.stages = options.stages;
-        let writer = Array.isArray(options.writers)
+        // The user writer is the post-merge target; the plugins' beforeStageWrite
+        // transforms wrap it per stage (see stageWriter) so each carries the stage
+        // identity it needs to mint stable, non-fusing IRIs.
+        this.writer = Array.isArray(options.writers)
             ? new FanOutWriter(options.writers)
             : options.writers;
         const transforms = options.plugins
             ?.map((p) => p.beforeStageWrite)
             .filter((t) => t !== undefined);
-        if (transforms?.length) {
-            const composed = (quads, context) => transforms.reduce((q, fn) => fn(q, context), quads);
-            writer = new TransformWriter(writer, composed);
-        }
-        this.writer = writer;
+        this.beforeStageWrite = transforms?.length
+            ? (quads, context) => transforms.reduce((q, fn) => fn(q, context), quads)
+            : undefined;
         this.distributionResolver =
             options.distributionResolver ?? new SparqlDistributionResolver();
         this.chaining = options.chaining;
@@ -210,7 +211,7 @@ export class Pipeline {
                         await this.runChain(dataset, resolved.distribution, stage, timeout);
                     }
                     else {
-                        await this.runStage(dataset, resolved.distribution, stage, this.writer, timeout);
+                        await this.runStage(dataset, resolved.distribution, stage, this.stageWriter(stage.name), timeout);
                     }
                 }
                 catch (error) {
@@ -269,6 +270,17 @@ export class Pipeline {
                 yield* this.collectStages(stage.stages);
         }
     }
+    /**
+     * The writer a stage's merged output is written through: the user writer
+     * wrapped with the plugins' {@link PipelinePlugin.beforeStageWrite}
+     * transforms, carrying this `stage`'s identity so a transform can mint stable
+     * per-`(dataset, stage)` IRIs rather than blank nodes.
+     */
+    stageWriter(stage) {
+        return this.beforeStageWrite
+            ? new TransformWriter(this.writer, this.beforeStageWrite, stage)
+            : this.writer;
+    }
     /**
      * Run a stage with reporting and return whether it was supported.
      * Returns `true` if the stage produced results, `false` if NotSupported.
@@ -335,8 +347,9 @@ export class Pipeline {
                     currentDistribution = await stageOutputResolver.resolve(childWriter.getOutputPath(dataset));
                 }
             }
-            // 3. Concatenate all output files → user writer.
-            await this.writer.write(dataset, this.readFiles(outputFiles));
+            // 3. Concatenate all output files → user writer, applying the plugins'
+            // beforeStageWrite transforms once for the chain under the parent stage.
+            await this.stageWriter(stage.name).write(dataset, this.readFiles(outputFiles));
         }
         finally {
             await stageOutputResolver.cleanup();

package/dist/plugin/provenance.d.ts

@@ -1,10 +1,7 @@
 import type { QuadTransform } from '../stage.js';
-import type { PipelinePlugin } from '../pipeline.js';
-import type { Dataset } from '@lde/dataset';
+import type { BeforeStageWriteContext, PipelinePlugin } from '../pipeline.js';
 /** QuadTransform that appends PROV-O provenance quads. */
-export declare const provenanceTransform: QuadTransform<{
-    dataset: Dataset;
-}>;
+export declare const provenanceTransform: QuadTransform<BeforeStageWriteContext>;
 /** Pipeline plugin that appends PROV-O provenance to every stage's output. */
 export declare function provenancePlugin(): PipelinePlugin;
 //# sourceMappingURL=provenance.d.ts.map

package/dist/plugin/provenance.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"provenance.d.ts","sourceRoot":"","sources":["../../src/plugin/provenance.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,aAAa,CAAC;AACjD,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,gBAAgB,CAAC;AACrD,OAAO,KAAK,EAAE,OAAO,EAAE,MAAM,cAAc,CAAC;AAkB5C,0DAA0D;AAC1D,eAAO,MAAM,mBAAmB,EAAE,aAAa,CAAC;IAAE,OAAO,EAAE,OAAO,CAAA;CAAE,CAGC,CAAC;AAEtE,8EAA8E;AAC9E,wBAAgB,gBAAgB,IAAI,cAAc,CAKjD"}
+{"version":3,"file":"provenance.d.ts","sourceRoot":"","sources":["../../src/plugin/provenance.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,aAAa,CAAC;AACjD,OAAO,KAAK,EAAE,uBAAuB,EAAE,cAAc,EAAE,MAAM,gBAAgB,CAAC;AAmB9E,0DAA0D;AAC1D,eAAO,MAAM,mBAAmB,EAAE,aAAa,CAAC,uBAAuB,CAGK,CAAC;AAE7E,8EAA8E;AAC9E,wBAAgB,gBAAgB,IAAI,cAAc,CAKjD"}

package/dist/plugin/provenance.js

@@ -1,5 +1,6 @@
+import { hashSuffix, skolemIri } from '@lde/dataset';
 import { DataFactory } from 'n3';
-const { namedNode, literal, blankNode, quad } = DataFactory;
+const { namedNode, literal, quad } = DataFactory;
 const RDF_TYPE = namedNode('http://www.w3.org/1999/02/22-rdf-syntax-ns#type');
 const PROV_ENTITY = namedNode('http://www.w3.org/ns/prov#Entity');
 const PROV_ACTIVITY = namedNode('http://www.w3.org/ns/prov#Activity');
@@ -8,7 +9,7 @@ const PROV_STARTED_AT_TIME = namedNode('http://www.w3.org/ns/prov#startedAtTime'
 const PROV_ENDED_AT_TIME = namedNode('http://www.w3.org/ns/prov#endedAtTime');
 const XSD_DATE_TIME = namedNode('http://www.w3.org/2001/XMLSchema#dateTime');
 /** QuadTransform that appends PROV-O provenance quads. */
-export const provenanceTransform = (quads, { dataset }) => appendProvenanceQuads(quads, dataset.iri.toString(), new Date());
+export const provenanceTransform = (quads, { dataset, stage }) => appendProvenanceQuads(quads, dataset.iri.toString(), stage, new Date());
 /** Pipeline plugin that appends PROV-O provenance to every stage's output. */
 export function provenancePlugin() {
     return {
@@ -16,13 +17,20 @@ export function provenancePlugin() {
         beforeStageWrite: provenanceTransform,
     };
 }
-async function* appendProvenanceQuads(quads, iri, startedAt) {
+async function* appendProvenanceQuads(quads, iri, stage, startedAt) {
     for await (const q of quads) {
         yield q;
     }
     const endedAt = new Date();
     const subject = namedNode(iri);
-    const activity = blankNode();
+    // Skolemise the activity to a stable IRI keyed on (dataset, stage) instead of
+    // a blank node. Per-dataset outputs are merged into one graph (the DKG index
+    // cats every dataset's file together), where blank-node labels are not unique
+    // across documents and would fuse unrelated activities into one node — one
+    // prov:Activity wrongly wasGeneratedBy several datasets (see issue #474).
+    // The IRI also makes a re-run idempotent: same (dataset, stage) → same node.
+    // The `.well-known/prov#activity-<hash>` shape mirrors the linkset skolem.
+    const activity = namedNode(skolemIri(`${iri}/.well-known/prov#activity`, hashSuffix(stage)));
     yield quad(subject, RDF_TYPE, PROV_ENTITY);
     yield quad(subject, PROV_WAS_GENERATED_BY, activity);
     yield quad(activity, RDF_TYPE, PROV_ACTIVITY);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@lde/pipeline",
-  "version": "0.30.13",
+  "version": "0.30.15",
   "repository": {
     "url": "git+https://github.com/ldelements/lde.git",
     "directory": "packages/pipeline"
@@ -24,10 +24,10 @@
     "!**/*.tsbuildinfo"
   ],
   "dependencies": {
-    "@lde/dataset": "0.7.6",
-    "@lde/dataset-registry-client": "0.8.2",
-    "@lde/distribution-probe": "0.1.9",
-    "@lde/sparql-importer": "0.6.4",
+    "@lde/dataset": "0.7.7",
+    "@lde/dataset-registry-client": "0.8.3",
+    "@lde/distribution-probe": "0.1.10",
+    "@lde/sparql-importer": "0.6.5",
     "@lde/sparql-server": "0.4.11",
     "@rdfjs/types": "^2.0.1",
     "@traqula/generator-sparql-1-1": "^1.1.4",