Merge pull request #1323 from lblod/GN-5607-fix-old-template-literal-node-parsing

GN-5607: correctly parse older-style literal nodes

Author: abeforgit

.changeset/twelve-experts-pick.md

@@ -0,0 +1,5 @@
+---
+'@lblod/ember-rdfa-editor': patch
+---
+
+Add extra check to `postProcessTagAsRdfaNode` function to correctly parse older style literal nodes

packages/ember-rdfa-editor/src/utils/_private/rdfa-parser/post-process-as-rdfa-nodes.ts

@@ -5,6 +5,7 @@ export interface PostProcessArgs<N> {
   activeTag: IActiveTag<N>;
   attributes: Record<string, string>;
   isRootTag: boolean;
+  isExternalTriple: boolean;
   typedResource: true | ModelBlankNode<N> | ModelNamedNode<N> | null;
   markAsLiteralNode: (
     node: N,
@@ -76,11 +77,25 @@ export function postProcessTagAsRdfaNode<N>(args: PostProcessArgs<N>): void {
     activeTag,
     attributes,
     isRootTag,
+    isExternalTriple,
     typedResource,
     markAsLiteralNode,
     markAsResourceNode,
   } = args;
   const node = activeTag.node;
+  if (node && isExternalTriple) {
+    markAsResourceNode(
+      node,
+      unwrap(activeTag.subject),
+      activeTag,
+      activeTag.predicates?.find(
+        (pred) => pred.value === attributes['property'],
+      ),
+      activeTag.datatype,
+      activeTag.language,
+    );
+    return;
+  }
   if (!activeTag.skipElement && node) {
     // no rel or rev
     if (
@@ -120,7 +135,9 @@ export function postProcessTagAsRdfaNode<N>(args: PostProcessArgs<N>): void {
       } else {
         if (
           truthyAttribute(attributes, 'about') &&
-          !truthyAttribute(attributes, 'data-literal-node')
+          !truthyAttribute(attributes, 'data-literal-node') &&
+          // Is this correct, an alternative solution would be to check on the presence of a `resource` attr here?
+          !truthyAttribute(attributes, 'datatype')
         ) {
           // same exception as above, we always interpret (property +about -content) cases as literal nodes
           markAsResourceNode(

packages/ember-rdfa-editor/src/utils/_private/rdfa-parser/rdfa-parser.ts

@@ -1204,10 +1204,14 @@ export class RdfaParser<N> {
       }
     }
     const isRootTag: boolean = this.activeTagStack.length === 1;
+    const isExternalTriple = Boolean(
+      parentTag?.attributes['data-external-triple-container'],
+    );
     postProcessTagAsRdfaNode({
       activeTag,
       attributes: activeTag.attributes,
       isRootTag,
+      isExternalTriple,
       typedResource: activeTag.typedResource ?? null,
       markAsLiteralNode: this.markAsLiteralNode,
       markAsResourceNode: this.markAsResourceNode,

test-app/tests/unit/rdfa/rdfa-parsing-test.ts

@@ -733,4 +733,96 @@ module('rdfa | parsing', function () {
     const resultingDocument = controller.mainEditorState.doc;
     assert.propEqual(resultingDocument.toJSON(), initialDocument.toJSON());
   });
+  test('Older documents with older style literal nodes should be correctly parsed', function (assert) {
+    QUnit.dump.maxDepth = 10;
+    const html = `
+    <div
+    lang="nl-BE"
+    data-say-document="true"
+    data-literal-node="true"
+    >
+      <div
+        style="display: none"
+        class="say-hidden"
+        data-rdfa-container="true"
+      ></div>
+      <div data-content-container="true">
+        <div
+          class="say-editable say-block-rdfa"
+          about="http://data.lblod.info/id/besluiten/1"
+          data-say-id="30e2d215-a8b6-4bab-9d8b-aaab634589ea"
+        >
+          <div
+            style="display: none"
+            class="say-hidden"
+            data-rdfa-container="true"
+          >
+            <span
+              about="http://data.lblod.info/id/besluiten/1"
+              property="http://www.w3.org/1999/02/22-rdf-syntax-ns#type"
+              resource="http://data.vlaanderen.be/ns/besluit#Besluit"
+            ></span>
+          </div>
+          <div data-content-container="true">
+            <h4
+              property="http://data.europa.eu/eli/ontology#title"
+              about="http://data.lblod.info/id/besluiten/1"
+              datatype="http://www.w3.org/2001/XMLSchema#string"
+              data-say-id="literal-1"
+              lang=""
+            >Decision title</h4>
+          </div>
+        </div>
+      </div>
+    </div>
+    `;
+    const { doc, block_rdfa, heading } = testBuilders;
+    const df = new SayDataFactory();
+    const expectedDoc = doc(
+      {},
+      block_rdfa(
+        {
+          rdfaNodeType: 'resource',
+          subject: 'http://data.lblod.info/id/besluiten/1',
+          __rdfaId: '30e2d215-a8b6-4bab-9d8b-aaab634589ea',
+          properties: [
+            {
+              predicate: 'http://www.w3.org/1999/02/22-rdf-syntax-ns#type',
+              object: df.namedNode(
+                'http://data.vlaanderen.be/ns/besluit#Besluit',
+              ),
+            },
+            {
+              predicate: 'http://data.europa.eu/eli/ontology#title',
+              object: df.literalNode('literal-1'),
+            },
+          ] satisfies OutgoingTriple[],
+        },
+        heading(
+          {
+            rdfaNodeType: 'literal',
+            __rdfaId: 'literal-1',
+            backlinks: [
+              {
+                predicate: 'http://data.europa.eu/eli/ontology#title',
+                subject: sayDataFactory.resourceNode(
+                  'http://data.lblod.info/id/besluiten/1',
+                ),
+              },
+            ],
+            datatype: sayDataFactory.namedNode(
+              'http://www.w3.org/2001/XMLSchema#string',
+            ),
+            language: '',
+            level: 4,
+          },
+          'Decision title',
+        ),
+      ),
+    );
+    const { controller } = testEditor(schema, plugins);
+    controller.initialize(html);
+    const actualDoc = controller.mainEditorState.doc;
+    assert.propEqual(actualDoc.toJSON(), expectedDoc.toJSON());
+  });
 });