feat: Support blank nodes in attribute editor for fixed and default values (RDFA-321) by spah-soptim · Pull Request #22 · SOPTIM/RDFArchitect

spah-soptim · 2026-02-23T18:07:12Z

Description

Added RDF blank-node support for attribute cims:isFixed / cims:isDefault values, in addition to literals.

New AttributeValueNode base class (parent of CIMSIsFixed/CIMSIsDefault) carries value, dataType, and a blankNode flag.
Read: new ValueNodeParser replaces the previous literal/blank-node extraction in CIMQuerySolutionParser. The fetcher now passes the dataset Model through so blank-node properties can be enumerated.
Write: AttributeFixedDefaultResolver decides per-attribute whether the new value should be literal vs. blank-node - preserves existing shape, otherwise defaults to attributes.newValuesBlankNode (default false).
Replace/delete: CIMUpdates now emits a multi-statement UpdateRequest that first cleans orphan blank-node triples (since ?attribute SPO does not reach the inner triples of nested resources), then deletes the attribute, then inserts the new one.
AttributeMapper.buildXsdDatatype now goes through XSDDatatypeMapper (now for example xsd:string not xsd:String).
Frontend reactive mapper now passes through fixedValue/defaultValue.

Test Checklist

General Behavior

Components reload automatically when data changes
Editing features are disabled in readonly datasets
Dialogs pre-select current dataset/graph
Required fields are validated in dialogs
Discarding unsaved changes opens a discard cancel confirm dialog

Global MenuBar

Welcome Page

Navigation to Editor works
Tips are displayed
Security and data information displayed
Copyright and version information displayed

Editor - MenuBar

Search function works with all filters (All Datasets, Current Dataset, Current Graph, Current Package)
Search finds classes, attributes, associations, packages
"Enable Editing" button appears for readonly datasets

Editor - Navigation

Editor - Package View

Class diagram displays correctly
Loading animation shows while loading
Info cards show when no package or no classes available
Drag and zoom diagram works
"Reset View" button works
"Filter View" works
Click on class opens class editor

Editor - Class Editor

Display and edit class properties: UUID (readonly), Label, Namespace, Package, Derived from, Abstract, Stereotypes, Attributes, Associations, Comment
Delete class works
Save changes works
Discard changes works
Attribute Editor works
Association Editor works
attribute/association SHACL View works
Class SHACL View works

Prefixes Page

View, add, remove and edit namespaces works

Changelog Page

Select graph and display write operations works
Operations shown in reverse chronological order
Detailed view of changed triples works
Restoring graph to a version works

Compare Page

Compare two graphs works

Fixed and default attribute values can be persisted in two equivalent shapes in the source RDF: as a direct literal triple, or as a blank-node wrapper (subject -> predicate -> _:blank ; _:blank -> rdfs:Literal -> literal). Until now the model layer collapsed both shapes into a plain literal, so editing an attribute backed by a blank-node value silently dropped that blank node on write. Introduced a shared `AttributeValueNode` base for `CIMSIsFixed` and `CIMSIsDefault` carrying a `blankNode` flag, and a `ValueNodeParser` utility that turns an `RDFNode` into a `ParsedValue(value, dataType, blankNode)` while validating the blank-node shape (exactly one statement with `rdfs:Literal` predicate and a literal object). `CIMQuerySolutionParser` takes an optional `Model` so it can re-resolve blank-node properties from the source dataset; `CIMObjectFactory.createCIMAttributeList` and `CIMObjectFetcher.fetchCIMAttributeList` thread that model through. `CIMUpdates.appendValueNode` emits either a direct literal or a fresh blank-node wrapper depending on `blankNode`. Because removing an attribute via `?attr ?p ?o` does not reach the inner triples of an attached blank-node value, `replaceAttribute`/`replaceAttributes` now return an `UpdateRequest` composed of a separate blank-node cleanup, the SPO delete and the insert; `deleteClass` performs the same cleanup before deleting its attributes. `InMemorySparqlExecutor` gains an `executeSingleUpdate(UpdateRequest)` overload to run those compound requests in a single write transaction.

Adds the resolver that decides, for every attribute being created or replaced, whether its fixed/default value should be persisted as a direct literal or as a blank-node wrapper: - if the attribute already exists in the graph, the existing shape wins (a blank-node value stays a blank-node value, a literal stays a literal), - otherwise the new `attributes.newValuesBlankNode` config flag decides; defaults to `false` so behaviour is unchanged for fresh installs. shape as `AssociationsService` — a single `resolve()` followed by a `AttributeFixedDefaultResolver` is a `@Service` with the config injected via `@Value` constructor parameter. It self-manages a READ transaction when none is active and reads inline when called from inside an existing WRITE transaction (e.g. `UpdateClassService.replaceClass`), so callers don't need transaction boilerplate. `AttributesService` keeps the same single `executeSingleUpdate()`. `AttributeMapper` now resolves the canonical XSD datatype URI (`xsd:string` instead of `xsd:String`) for fixed/default values via `XSDDatatypeMapper`; the resolver re-resolves it from the graph for non-primitive attribute datatypes. Frontend: `mapAttributeDtoToReactiveAttribute` passes `fixedValue` and `defaultValue` through so they are available on the reactive object. Signed-off-by: Jan-Hendrik Spahn <jan-hendrik.spahn@soptim.de>

…te insert Signed-off-by: Jan-Hendrik Spahn <jan-hendrik.spahn@soptim.de>

…TOCTOU - CIMUpdates.deleteAttributeValueNodes{,ForClass} now build their SPARQL via UpdateBuilder + ExprFactory.isBlank instead of raw string concat, so user-controlled classUUID and graphURI flow through Jena's node coercion and can no longer be used to inject SPARQL. - InMemorySparqlExecutor gains an executeSingleUpdate overload that takes a Function<Graph, UpdateRequest> and runs build + execute in a single WRITE transaction. AttributesService uses it so the fixed/default shape resolution and the resulting update happen atomically — the previous read-then-write split could see another writer change the existing value's shape between the two transactions. - AttributeFixedDefaultResolver no longer self-manages a READ transaction; the caller's open transaction is required (and is now always a WRITE). - ValueNodeParser.parseLiteral takes a Literal directly so the contract is enforced at the call site. - Add a comment explaining why insertAttribute uses BIND to force the WHERE clause to match exactly once. Signed-off-by: Jan-Hendrik Spahn <jan-hendrik.spahn@soptim.de>

Signed-off-by: Jan-Hendrik Spahn <jan-hendrik.spahn@soptim.de>

rema-soptim · 2026-04-30T14:27:49Z

            graph.begin(TxnType.WRITE);
            var cimClass = classMapper.toCIMObject(newClass);
+            // resolver inherits the open WRITE txn for its read-only attribute lookup
+            fixedDefaultResolver.resolve(graph, cimClass.getAttributes());


I still find this call questionable. It suggests that either classMapper.toCIMObject or CIMUpdates.replaceClass isn't working properly and needs an external fix.

rema-soptim · 2026-04-30T14:34:28Z

+    public void executeSingleUpdate(
+            GraphRewindableWithUUIDs graph, UpdateRequest update, String graphUri) {
+        executeSingleUpdate(graph, graphUri, g -> update);
+    }
+


Does this method make sense, if it's only used in unit tests?

rema-soptim · 2026-04-30T14:41:27Z

+    public void executeSingleUpdate(
+            GraphRewindableWithUUIDs graph,
+            String graphUri,
+            Function<GraphRewindableWithUUIDs, UpdateRequest> updateBuilder) {


I don't really see the need to use a lamba. Why not just prepare the argument beforehand and then pass it? And if you need to access the graph beforehand then just don't use the executeSingleUpdate, which makes this overly complicated.

rema-soptim · 2026-04-30T15:09:05Z

            String classUUID,
            List<CIMAttribute> attributes) {
-        var baseUpdate = CIMUpdates.deleteAttributes(prefixMapping, graphURI, classUUID);
+        var updateRequest = deleteAttributeValueNodesForClass(graphURI, classUUID);


Doesn't it make more sense to just enhance the delete Attribute Query instead of doing this separately, since value nodes are part of an attribute.

rema-soptim · 2026-04-30T15:12:13Z

            PrefixMapping prefixMapping, String graphURI, CIMAttribute attribute) {
-        var baseUpdate = deleteAttribute(prefixMapping, graphURI, attribute.getUuid());
-        return appendInsertAttribute(baseUpdate, attribute);
+        var updateRequest = deleteAttributeValueNodes(graphURI, attribute.getUuid());


Same here. this should not be its own method, but part of delete Attribute.

rema-soptim · 2026-04-30T15:26:39Z

+     * the properties of an attribute fixed/default value that was returned from the query as a
+     * blank node. May be {@code null} when blank-node values are not expected.
+     */
+    private final Model valueNodeModel;


The CIMQuerySolutionParser should not have a model. Its job is to parse the result of a query, not query the model itself. Instead the query should be extended. For attributes that would be the query created in CIMQueries.getAttributesQuery.

rema-soptim · 2026-04-30T15:29:13Z

@@ -0,0 +1,128 @@
+/*


If you just parse a query result in CIMQuerySolutionParser this file is either obsolete or heavily refactored to parse a query result instead of searching the model itself

rema-soptim · 2026-04-30T15:31:59Z

+                graph,
+                graphUri,
+                g -> {
+                    fixedDefaultResolver.resolve(g, cimAttribute);


should either inside CIMUpdates.insertAttribute or attributeMapper.toCIMObject

rema-soptim · 2026-04-30T15:32:19Z

+                graph,
+                graphUri,
+                g -> {
+                    fixedDefaultResolver.resolve(g, cimAttribute);


rema-soptim · 2026-04-30T15:32:26Z

+                graph,
+                graphUri,
+                g -> {
+                    fixedDefaultResolver.resolve(g, attributeCIMObjects);


spah-soptim self-assigned this Feb 23, 2026

spah-soptim added the bug Something isn't working label Feb 23, 2026

spah-soptim changed the title ~~RDFA321: Support Blank Nodes in Attribute Editor (Fixed/Default values)~~ RDFA-321: Support Blank Nodes in Attribute Editor (Fixed/Default values) Feb 23, 2026

spah-soptim force-pushed the feature/RDFA-321-attribute-editor-fix branch 2 times, most recently from e463e38 to 95ced91 Compare February 25, 2026 15:12

spah-soptim requested a review from rema-soptim February 25, 2026 15:54

spah-soptim marked this pull request as draft March 10, 2026 16:51

spah-soptim changed the title ~~RDFA-321: Support Blank Nodes in Attribute Editor (Fixed/Default values)~~ feat: Support blank nodes in attribute editor for fixed and default values (RDFA-321) Mar 11, 2026

spah-soptim force-pushed the feature/RDFA-321-attribute-editor-fix branch from d3a4163 to f5b701b Compare April 27, 2026 06:14

github-code-quality Bot found potential problems Apr 27, 2026

View reviewed changes

Comment thread backend/src/main/java/org/rdfarchitect/models/cim/data/dto/relations/CIMSIsFixed.java Dismissed

Comment thread backend/src/main/java/org/rdfarchitect/models/cim/data/dto/relations/CIMSIsDefault.java Dismissed

spah-soptim added 4 commits April 30, 2026 01:26

fix(cim): stop duplicating blank-node fixed/default values on attribu…

001f61e

…te insert Signed-off-by: Jan-Hendrik Spahn <jan-hendrik.spahn@soptim.de>

spah-soptim force-pushed the feature/RDFA-321-attribute-editor-fix branch from f5b701b to d8ac3d3 Compare April 30, 2026 06:30

chore: ran backend linting

943dc18

Signed-off-by: Jan-Hendrik Spahn <jan-hendrik.spahn@soptim.de>

spah-soptim marked this pull request as ready for review April 30, 2026 08:13

spah-soptim removed the request for review from rema-soptim April 30, 2026 08:13

Merge branch 'main' into feature/RDFA-321-attribute-editor-fix

877e2fa

rema-soptim requested changes Apr 30, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: Support blank nodes in attribute editor for fixed and default values (RDFA-321)#22

feat: Support blank nodes in attribute editor for fixed and default values (RDFA-321)#22
spah-soptim wants to merge 6 commits intomainfrom
feature/RDFA-321-attribute-editor-fix

spah-soptim commented Feb 23, 2026 •

edited

Loading

Uh oh!

Uh oh!

Uh oh!

rema-soptim Apr 30, 2026

Uh oh!

rema-soptim Apr 30, 2026

Uh oh!

rema-soptim Apr 30, 2026

Uh oh!

rema-soptim Apr 30, 2026

Uh oh!

rema-soptim Apr 30, 2026

Uh oh!

rema-soptim Apr 30, 2026

Uh oh!

rema-soptim Apr 30, 2026

Uh oh!

rema-soptim Apr 30, 2026

Uh oh!

rema-soptim Apr 30, 2026

Uh oh!

rema-soptim Apr 30, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

spah-soptim commented Feb 23, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Test Checklist

General Behavior

Global MenuBar

Welcome Page

Editor - MenuBar

Editor - Navigation

Editor - Package View

Editor - Class Editor

Prefixes Page

Changelog Page

Compare Page

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

spah-soptim commented Feb 23, 2026 •

edited

Loading