Validate object-form semantic model dataset sources

CHANGELOG.md

@@ -165,6 +165,9 @@ request adding CHANGELOG notes for breaking (!) changes and possibly other secti
   `Unsupported access delegation mode: REMOTE_SIGNING`.
 - Iceberg REST: renaming a table or view with a missing `source` or `destination` now returns `400 Bad Request` instead of `500 Internal Server Error`.
 - Async file-cleanup tasks now bound how long they wait for object-store deletions via the new `polaris.tasks.file-deletion-timeout` (default 1h), so a stalled storage endpoint can no longer pin a task-executor thread indefinitely; a timeout is terminal for the current run rather than immediately retried, so it does not stack more deletions onto the stalled endpoint.
+- Semantic-model create and update requests now require `semantic_model` JSON to be an object and
+  validate every dataset source. Previously, invalid root shapes could bypass source validation or
+  be accepted under the obsolete array contract.
 - Python CLI `catalogs create --type external` now validates `--storage-type` and `--default-base-location` up front, matching the behavior for internal catalogs and the flags' documented "(Required)" status. Previously, omitting either produced an opaque pydantic `ValidationError` at request-build time.
 - Iceberg REST: server-side JSON processing failures (HTTP 500) now return the standard Iceberg
   error envelope (`{"error": {...}}`) instead of a flat `{"code", "message"}` body, so Iceberg

extensions/semantic-models/src/main/java/org/apache/polaris/service/catalog/semanticmodel/SemanticModelCatalog.java

@@ -310,25 +310,22 @@ private JsonNode validateDocument(SemanticModelDocument document) {
    * to the offending dataset.
    */
   private void resolveAndValidateSources(JsonNode semanticModel) {
-    if (!semanticModel.isArray()) {
-      return;
+    if (!semanticModel.isObject()) {
+      throw new BadRequestException("Field 'semantic_model' must be a JSON object");
     }
 
-    for (int modelIdx = 0; modelIdx < semanticModel.size(); modelIdx++) {
-      JsonNode datasets = semanticModel.get(modelIdx).get("datasets");
-      if (datasets == null || !datasets.isArray()) {
-        continue;
-      }
-      for (int datasetIdx = 0; datasetIdx < datasets.size(); datasetIdx++) {
-        String pointer =
-            String.format("/semantic_model/%d/datasets/%d/source", modelIdx, datasetIdx);
-        JsonNode source = datasets.get(datasetIdx).get("source");
-        if (source == null || !source.isTextual()) {
-          throw new BadRequestException(
-              "Semantic model dataset at %s must define a string 'source'", pointer);
-        }
-        resolveSourceOrThrow(source.asText(), pointer);
+    JsonNode datasets = semanticModel.get("datasets");
+    if (datasets == null || !datasets.isArray()) {
+      return;
+    }
+    for (int datasetIdx = 0; datasetIdx < datasets.size(); datasetIdx++) {
+      String pointer = String.format("/semantic_model/datasets/%d/source", datasetIdx);
+      JsonNode source = datasets.get(datasetIdx).get("source");
+      if (source == null || !source.isTextual()) {
+        throw new BadRequestException(
+            "Semantic model dataset at %s must define a string 'source'", pointer);
       }
+      resolveSourceOrThrow(source.asText(), pointer);
     }
   }
 

extensions/semantic-models/src/test/java/org/apache/polaris/service/catalog/semanticmodel/AbstractSemanticModelCatalogHandlerTest.java

@@ -116,11 +116,11 @@ protected static CreateSemanticModelRequest createRequest(String name, String mo
   }
 
   protected static SemanticModelDocument doc(String model) {
-    return SemanticModelDocument.builder().setVersion("0.1.1").setSemanticModel(model).build();
+    return SemanticModelDocument.builder().setVersion("0.2.0.dev0").setSemanticModel(model).build();
   }
 
   protected static String modelJson(String source) {
-    return "[{\"name\":\"m\",\"datasets\":[{\"name\":\"d\",\"source\":\"" + source + "\"}]}]";
+    return "{\"name\":\"m\",\"datasets\":[{\"name\":\"d\",\"source\":\"" + source + "\"}]}";
   }
 
   private void createCatalogNamespaceAndTable() {

extensions/semantic-models/src/test/java/org/apache/polaris/service/catalog/semanticmodel/SemanticModelCatalogTest.java

@@ -81,7 +81,7 @@ class SemanticModelCatalogTest {
   private static final SemanticModelIdentifier IDENTIFIER =
       SemanticModelIdentifier.builder().setNamespace(List.of("sales")).setName(MODEL).build();
   private static final String VALID_MODEL_JSON =
-      "[{\"name\":\"m\",\"datasets\":[{\"name\":\"d\",\"source\":\"sales.store_sales\"}]}]";
+      "{\"name\":\"m\",\"datasets\":[{\"name\":\"d\",\"source\":\"sales.store_sales\"}]}";
 
   private PolarisResolutionManifestCatalogView view;
   private PolarisMetaStoreManager metaStoreManager;
@@ -119,7 +119,7 @@ void setUp() {
 
   private SemanticModelDocument doc(String semanticModelJson) {
     return SemanticModelDocument.builder()
-        .setVersion("0.1.1")
+        .setVersion("0.2.0.dev0")
         .setSemanticModel(semanticModelJson)
         .build();
   }
@@ -143,7 +143,7 @@ private void stubResolvableSource() {
   private void stubExistingModel(int entityVersion) {
     SemanticModelEntity stored =
         new SemanticModelEntity.Builder(NS, MODEL)
-            .setSpecVersion("0.1.1")
+            .setSpecVersion("0.2.0.dev0")
             .setContent(VALID_MODEL_JSON)
             .setId(10L)
             .setCatalogId(CATALOG_ID)
@@ -167,7 +167,7 @@ void createResolvesSourcesAndPersists() {
         catalog.createSemanticModel(IDENTIFIER, doc(VALID_MODEL_JSON));
 
     assertThat(response.getDocument().getSemanticModel()).isEqualTo(VALID_MODEL_JSON);
-    assertThat(response.getDocument().getVersion()).isEqualTo("0.1.1");
+    assertThat(response.getDocument().getVersion()).isEqualTo("0.2.0.dev0");
     assertThat(response.getEntityVersion()).isEqualTo("1");
   }
 
@@ -178,12 +178,20 @@ void createRejectsEmptyDocument() {
         .hasMessageContaining("must not be empty");
   }
 
+  @Test
+  void createRejectsArrayDocument() {
+    assertThatThrownBy(
+            () -> catalog.createSemanticModel(IDENTIFIER, doc("[" + VALID_MODEL_JSON + "]")))
+        .isInstanceOf(BadRequestException.class)
+        .hasMessageContaining("must be a JSON object");
+  }
+
   @Test
   void createRejectsDatasetWithoutSource() {
-    String noSource = "[{\"name\":\"m\",\"datasets\":[{\"name\":\"d\"}]}]";
+    String noSource = "{\"name\":\"m\",\"datasets\":[{\"name\":\"d\"}]}";
     assertThatThrownBy(() -> catalog.createSemanticModel(IDENTIFIER, doc(noSource)))
         .isInstanceOf(BadRequestException.class)
-        .hasMessageContaining("/semantic_model/0/datasets/0/source")
+        .hasMessageContaining("/semantic_model/datasets/0/source")
         .hasMessageContaining("must define a string 'source'");
   }
 
@@ -192,7 +200,7 @@ void createRejectsUnresolvedSource() {
     // No stub for the source table -> passthrough resolution returns null.
     assertThatThrownBy(() -> catalog.createSemanticModel(IDENTIFIER, doc(VALID_MODEL_JSON)))
         .isInstanceOf(BadRequestException.class)
-        .hasMessageContaining("/semantic_model/0/datasets/0/source")
+        .hasMessageContaining("/semantic_model/datasets/0/source")
         .hasMessageContaining("store_sales");
   }
 

polaris-core/src/main/java/org/apache/polaris/core/semantic/SemanticModelEntity.java

@@ -39,7 +39,7 @@
  */
 public class SemanticModelEntity extends PolarisEntity {
 
-  /** The declared Ossie spec version of the stored document (e.g. {@code 0.1.1}). */
+  /** The declared Ossie spec version of the stored document (e.g. {@code 0.2.0.dev0}). */
   public static final String SPEC_VERSION_KEY = "semantic-model.spec-version";
 
   /** The Ossie document serialized as a JSON string. */

spec/polaris-catalog-apis/semantic-models-api.yaml

@@ -284,12 +284,12 @@ components:
         version:
           type: string
           description: The Apache Ossie spec version.
-          example: "0.1.1"
+          example: "0.2.0.dev0"
         semantic_model:
           type: string
-          description: The Apache Ossie semantic model serialized as a JSON string.
+          description: The Apache Ossie semantic model serialized as a JSON object string.
       example:
-        version: "0.1.1"
+        version: "0.2.0.dev0"
         semantic_model: '{"name":"tpcds_retail_model","description":"retail analytics","datasets":[{"name":"store_sales","source":"public.store_sales","primary_key":["ss_item_sk","ss_ticket_number"]}],"metrics":[{"name":"total_sales","expression":{"dialects":[{"dialect":"ANSI_SQL","expression":"SUM(ss_ext_sales_price)"}]}}]}'
 
     SemanticModelIdentifier: