Fix semanticId matching: tolerate trailing eCl@ss/IRDI version suffix

core/src/main/java/it/extrared/extractor/strategies/AasExtractorStrategy.java

@@ -12,6 +12,7 @@
 import jakarta.enterprise.context.ApplicationScoped;
 import jakarta.json.*;
 import java.util.*;
+import java.util.regex.Pattern;
 import org.jboss.logging.Logger;
 
 /**
@@ -46,6 +47,7 @@ public class AasExtractorStrategy implements SearchKeyExtractorStrategy {
     private static final String VALUE = "value";
     private static final String SEMANTIC_ID = "semanticId";
     private static final String KEYS = "keys";
+    private static final Pattern IRDI_VERSION_SUFFIX = Pattern.compile("#\\d+$");
     private static final String LANGUAGE = "language";
     private static final String TEXT = "text";
 
@@ -160,7 +162,15 @@ private Object findBySemanticId(
 
     /**
      * Checks if {@code semanticId.keys[].value} of the element matches the configured semantic ID.
-     * Matching is suffix-insensitive to tolerate version suffixes (e.g. {@code #002}).
+     * Two tolerances let version-less configured targets (e.g. {@code 0173-1#02-AAO677}) match
+     * real-world AAS values:
+     *
+     * <ul>
+     *   <li>a leading IRI prefix on the AAS value is tolerated via {@code endsWith};
+     *   <li>a trailing eCl@ss/IRDI version suffix on the AAS value (e.g. {@code
+     *       0173-1#02-AAO677#002}) is tolerated by also comparing the value with that suffix
+     *       stripped — {@code endsWith} alone cannot see past it.
+     * </ul>
      */
     private boolean matchesSemanticId(JsonObject element, String targetSemanticId) {
         JsonObject semanticIdObj = getObject(element, SEMANTIC_ID);
@@ -171,17 +181,34 @@ private boolean matchesSemanticId(JsonObject element, String targetSemanticId) {
             if (key.getValueType() != JsonValue.ValueType.OBJECT) continue;
             JsonValue keyValue = key.asJsonObject().get(VALUE);
             if (keyValue instanceof JsonString js) {
-                // exact match OR the AAS value ends with our target (version-suffix tolerant)
                 String candidate = js.getString();
-                if (candidate.equalsIgnoreCase(targetSemanticId)
-                        || candidate.toUpperCase().endsWith(targetSemanticId.toUpperCase())) {
+                if (matchesTarget(candidate, targetSemanticId)
+                        || matchesTarget(withoutVersionSuffix(candidate), targetSemanticId)) {
                     return true;
                 }
             }
         }
         return false;
     }
 
+    /**
+     * Exact match (case-insensitive) OR the AAS value ends with the target (IRI-prefix tolerant).
+     */
+    private boolean matchesTarget(String candidate, String target) {
+        return candidate.equalsIgnoreCase(target)
+                || candidate.toUpperCase().endsWith(target.toUpperCase());
+    }
+
+    /**
+     * Strips a trailing IRDI version segment ({@code 0173-1#02-AAO677#002} → {@code
+     * 0173-1#02-AAO677}). Applied only when another {@code #} remains, so plain IRIs with a numeric
+     * fragment are left untouched.
+     */
+    private String withoutVersionSuffix(String semanticId) {
+        String stripped = IRDI_VERSION_SUFFIX.matcher(semanticId).replaceFirst("");
+        return stripped.contains("#") ? stripped : semanticId;
+    }
+
     // -------------------------------------------------------------------------
     // idShort path navigation
     // -------------------------------------------------------------------------

core/src/test/java/it/extrared/extractor/strategies/SearchKeyExtractorTest.java

@@ -60,6 +60,28 @@ public void extractFromAAS(UniAsserter uniAsserter) throws IOException {
         }
     }
 
+    @Test
+    @RunOnVertxContext
+    public void extractFromAASWithVersionedSemanticIds(UniAsserter uniAsserter) throws IOException {
+        // Same data as dpp-aas.json, but with vendor-specific idShorts and versioned eCl@ss
+        // semanticIds (e.g. 0173-1#02-AAW338#001) — the form real IDTA submodels carry. The
+        // idShort-path fallback cannot match here, so extraction must succeed via the
+        // semanticId scan alone; 'granularity' (idShortPath-only) keeps working as before.
+        try (InputStream is =
+                getClass().getResourceAsStream("/example-dpp/dpp-aas-versioned.json")) {
+            Uni<Map<String, Object>> result = searchKeyExtractor.extractSearchKeys(is);
+            uniAsserter.assertThat(
+                    () -> result,
+                    r -> {
+                        assertEquals(4, r.size());
+                        assertEquals("product", r.get("granularity"));
+                        assertEquals(45.8, r.get("carbonFootprint"));
+                        assertEquals(189.5, r.get("weight"));
+                        assertEquals("EcoPhone X Pro AAS", r.get("productName"));
+                    });
+        }
+    }
+
     @Test
     @RunOnVertxContext
     public void extractFromJsonLd(UniAsserter uniAsserter) throws IOException {

core/src/test/resources/example-dpp/dpp-aas-versioned.json

@@ -0,0 +1,183 @@
+{
+  "modelType": "AssetAdministrationShellEnvironment",
+
+  "assetAdministrationShells": [
+    {
+      "modelType": "AssetAdministrationShell",
+      "id": "urn:example:aas:smartphone-002",
+      "idShort": "SmartphoneEcoPhoneXProVendor",
+      "assetInformation": {
+        "assetKind": "Instance",
+        "globalAssetId": "urn:epc:id:sgtin:0614141.112345.002",
+        "specificAssetIds": [
+          { "name": "GTIN",          "value": "8712345678901" },
+          { "name": "commodityCode", "value": "8517120000"   }
+        ]
+      },
+      "submodels": [
+        { "type": "ExternalReference", "keys": [{ "type": "Submodel", "value": "urn:example:sm:nameplate:smartphone-002"       }] },
+        { "type": "ExternalReference", "keys": [{ "type": "Submodel", "value": "urn:example:sm:technicaldata:smartphone-002"   }] },
+        { "type": "ExternalReference", "keys": [{ "type": "Submodel", "value": "urn:example:sm:carbonfootprint:smartphone-002" }] },
+        { "type": "ExternalReference", "keys": [{ "type": "Submodel", "value": "urn:example:sm:dppadmin:smartphone-002"        }] }
+      ]
+    }
+  ],
+
+  "_comment": "Same data as dpp-aas.json, but with vendor-specific (non-standard) idShorts and versioned eCl@ss semanticIds (the form real IDTA submodels carry, e.g. 0173-1#02-AAW338#001). Extraction of productName/weight/carbonFootprint must succeed via the semanticId scan alone: the idShort-path fallback cannot match these names. Only 'granularity' (configured by idShortPath alone) keeps its standard idShorts.",
+
+  "submodels": [
+
+    {
+      "modelType": "Submodel",
+      "id": "urn:example:sm:nameplate:smartphone-002",
+      "idShort": "VendorNameplate",
+      "semanticId": {
+        "type": "ExternalReference",
+        "keys": [{ "type": "GlobalReference", "value": "https://admin-shell.io/idta/nameplate/3/0/Nameplate" }]
+      },
+      "submodelElements": [
+        {
+          "modelType": "MultiLanguageProperty",
+          "idShort": "BrandName",
+          "semanticId": {
+            "type": "ExternalReference",
+            "keys": [{ "type": "GlobalReference", "value": "0173-1#02-AAO677#002" }]
+          },
+          "value": [
+            { "language": "en", "text": "Alpha Electronics Inc." }
+          ]
+        },
+        {
+          "modelType": "MultiLanguageProperty",
+          "idShort": "ProductDesignation",
+          "semanticId": {
+            "type": "ExternalReference",
+            "keys": [{ "type": "GlobalReference", "value": "0173-1#02-AAW338#001" }]
+          },
+          "value": [
+            { "language": "en", "text": "EcoPhone X Pro AAS" }
+          ]
+        }
+      ]
+    },
+
+    {
+      "modelType": "Submodel",
+      "id": "urn:example:sm:technicaldata:smartphone-002",
+      "idShort": "TechData",
+      "semanticId": {
+        "type": "ExternalReference",
+        "keys": [{ "type": "GlobalReference", "value": "https://admin-shell.io/ZVEI/TechnicalData/Submodel/1/2" }]
+      },
+      "submodelElements": [
+        {
+          "modelType": "SubmodelElementCollection",
+          "idShort": "GeneralProperties",
+          "semanticId": {
+            "type": "ExternalReference",
+            "keys": [{ "type": "GlobalReference", "value": "https://admin-shell.io/ZVEI/TechnicalData/MainSection/1/1" }]
+          },
+          "value": [
+            {
+              "modelType": "Property",
+              "idShort": "ProductNetWeight",
+              "semanticId": {
+                "type": "ExternalReference",
+                "keys": [{ "type": "GlobalReference", "value": "0173-1#02-AAB419#012" }]
+              },
+              "valueType": "xs:decimal",
+              "value": "189.5"
+            },
+            {
+              "modelType": "Property",
+              "idShort": "ProductNetWeightUnit",
+              "valueType": "xs:string",
+              "value": "g"
+            }
+          ]
+        }
+      ]
+    },
+
+    {
+      "modelType": "Submodel",
+      "id": "urn:example:sm:carbonfootprint:smartphone-002",
+      "idShort": "PCF",
+      "semanticId": {
+        "type": "ExternalReference",
+        "keys": [{ "type": "GlobalReference", "value": "https://admin-shell.io/idta/CarbonFootprint/CarbonFootprint/1/0" }]
+      },
+      "submodelElements": [
+        {
+          "modelType": "SubmodelElementList",
+          "idShort": "Footprints",
+          "semanticId": {
+            "type": "ExternalReference",
+            "keys": [{ "type": "GlobalReference", "value": "https://admin-shell.io/idta/CarbonFootprint/ProductCarbonFootprints/1/0" }]
+          },
+          "typeValueListElement": "SubmodelElementCollection",
+          "value": [
+            {
+              "modelType": "SubmodelElementCollection",
+              "idShort": "Footprint00",
+              "semanticId": {
+                "type": "ExternalReference",
+                "keys": [{ "type": "GlobalReference", "value": "https://admin-shell.io/idta/CarbonFootprint/ProductCarbonFootprint/1/0" }]
+              },
+              "value": [
+                {
+                  "modelType": "Property",
+                  "idShort": "CO2eq",
+                  "semanticId": {
+                    "type": "ExternalReference",
+                    "keys": [{ "type": "GlobalReference", "value": "0173-1#02-ABG855#003" }]
+                  },
+                  "valueType": "xs:decimal",
+                  "value": "45.8"
+                },
+                {
+                  "modelType": "Property",
+                  "idShort": "CO2eqUnit",
+                  "semanticId": {
+                    "type": "ExternalReference",
+                    "keys": [{ "type": "GlobalReference", "value": "0173-1#02-ABG856#003" }]
+                  },
+                  "valueType": "xs:string",
+                  "value": "piece"
+                }
+              ]
+            }
+          ]
+        }
+      ]
+    },
+
+    {
+      "modelType": "Submodel",
+      "id": "urn:example:sm:dppadmin:smartphone-002",
+      "idShort": "DPPAdministration",
+      "displayName": [{ "language": "en", "text": "DPP Administration Metadata" }],
+      "submodelElements": [
+        {
+          "modelType": "Property",
+          "idShort": "UniqueDPPID",
+          "valueType": "xs:string",
+          "value": "urn:uuid:b2c3d4e5-f6a7-4b5c-9d0e-1f2a3b4c5d6e"
+        },
+        {
+          "modelType": "Property",
+          "idShort": "Granularity",
+          "valueType": "xs:string",
+          "value": "product"
+        },
+        {
+          "modelType": "Property",
+          "idShort": "Status",
+          "valueType": "xs:string",
+          "value": "Active"
+        }
+      ]
+    }
+
+  ]
+}