[feature/semantic-text] Add query and model compatibility tests (#107256

)

x-pack/plugin/inference/build.gradle

@@ -10,7 +10,7 @@ apply plugin: 'elasticsearch.internal-yaml-rest-test'
 
 restResources {
   restApi {
-    include '_common', 'bulk', 'indices', 'inference', 'index', 'get', 'update', 'reindex'
+    include '_common', 'bulk', 'indices', 'inference', 'index', 'get', 'update', 'reindex', 'search'
   }
 }
 

x-pack/plugin/inference/src/main/java/org/elasticsearch/xpack/inference/mapper/SemanticTextFieldMapper.java

@@ -217,7 +217,15 @@ protected void parseCreateField(DocumentParserContext context) throws IOExceptio
             try {
                 conflicts.check();
             } catch (Exception exc) {
-                throw new DocumentParsingException(xContentLocation, "Incompatible model_settings", exc);
+                throw new DocumentParsingException(
+                    xContentLocation,
+                    "Incompatible model settings for field ["
+                        + name()
+                        + "]. Check that the "
+                        + INFERENCE_ID_FIELD.getPreferredName()
+                        + " is not using different model settings",
+                    exc
+                );
             }
             mapper = this;
         }

x-pack/plugin/inference/src/yamlRestTest/resources/rest-api-spec/test/inference/20_semantic_text_field_mapper.yml

@@ -17,6 +17,7 @@ setup:
             "task_settings": {
             }
           }
+
   - do:
       inference.put_model:
         task_type: text_embedding
@@ -50,19 +51,109 @@ setup:
                 type: text
 
 ---
-"Sparse vector results format":
+"Dense vector results are indexed as nested chunks and searchable":
+    - do:
+        bulk:
+          index: test-index
+          refresh: true
+          body: |
+            {"index":{}}
+            {"dense_field": ["you know, for testing", "now with chunks"]}
+            {"index":{}}
+            {"dense_field": ["some more tests", "that include chunks"]}
+
     - do:
-        index:
+        search:
           index: test-index
-          id: doc_1
           body:
-            sparse_field: "you know, for testing"
+            query:
+              nested:
+                path: dense_field.inference.chunks
+                query:
+                  knn:
+                    field: dense_field.inference.chunks.embeddings
+                    query_vector_builder:
+                      text_embedding:
+                        model_id: dense-inference-id
+                        model_text: "you know, for testing"
+
+    - match: { hits.total.value: 2 }
+    - match: { hits.total.relation: eq }
+
+    # Search with inner hits
+    - do:
+        search:
+          _source: false
+          index: test-index
+          body:
+            query:
+              nested:
+                path: dense_field.inference.chunks
+                inner_hits:
+                  _source: false
+                  fields: [dense_field.inference.chunks.text]
+                query:
+                  knn:
+                    field: dense_field.inference.chunks.embeddings
+                    query_vector_builder:
+                      text_embedding:
+                        model_id: dense-inference-id
+                        model_text: "you know, for testing"
+
+    - match: { hits.total.value: 2 }
+    - match: { hits.total.relation: eq }
+    - match: { hits.hits.0.inner_hits.dense_field\.inference\.chunks.hits.total.value: 2 }
+    - exists:  hits.hits.0.inner_hits.dense_field\.inference\.chunks.hits.hits.0.fields.dense_field\.inference\.chunks.0.text
+    - exists:  hits.hits.0.inner_hits.dense_field\.inference\.chunks.hits.hits.1.fields.dense_field\.inference\.chunks.0.text
+
 
 ---
-"Dense vector results format":
+"Sparse vector results are indexed as nested chunks and searchable":
     - do:
-        index:
+        bulk:
+          index: test-index
+          refresh: true
+          body: |
+            {"index":{}}
+            {"sparse_field": ["you know, for testing", "now with chunks"]}
+            {"index":{}}
+            {"sparse_field": ["some more tests", "that include chunks"]}
+
+    - do:
+        search:
+          index: test-index
+          body:
+            query:
+              nested:
+                path: sparse_field.inference.chunks
+                query:
+                  text_expansion:
+                    sparse_field.inference.chunks.embeddings:
+                      model_id: sparse-inference-id
+                      model_text: "you know, for testing"
+
+    # Search with inner hits
+    - do:
+        search:
+          _source: false
           index: test-index
-          id: doc_1
           body:
-            dense_field: "you know, for testing"
+            query:
+              nested:
+                path: sparse_field.inference.chunks
+                inner_hits:
+                  _source: false
+                  fields: [sparse_field.inference.chunks.text]
+                query:
+                  text_expansion:
+                    sparse_field.inference.chunks.embeddings:
+                      model_id: sparse-inference-id
+                      model_text: "you know, for testing"
+
+    - match: { hits.total.value: 2 }
+    - match: { hits.total.relation: eq }
+    - match: { hits.hits.0.inner_hits.sparse_field\.inference\.chunks.hits.total.value: 2 }
+    - exists:  hits.hits.0.inner_hits.sparse_field\.inference\.chunks.hits.hits.0.fields.sparse_field\.inference\.chunks.0.text
+    - exists:  hits.hits.0.inner_hits.sparse_field\.inference\.chunks.hits.hits.1.fields.sparse_field\.inference\.chunks.0.text
+
+

x-pack/plugin/inference/src/yamlRestTest/resources/rest-api-spec/test/inference/30_semantic_text_inference_incompatible_models.yml

@@ -0,0 +1,190 @@
+setup:
+  - skip:
+      version: " - 8.12.99"
+      reason: semantic_text introduced in 8.13.0 # TODO change when 8.13.0 is released
+
+  - do:
+      inference.put_model:
+        task_type: sparse_embedding
+        inference_id: sparse-inference-id
+        body: >
+          {
+            "service": "test_service",
+            "service_settings": {
+              "model": "my_model",
+              "api_key": "abc64"
+            },
+            "task_settings": {
+            }
+          }
+  - do:
+      inference.put_model:
+        task_type: text_embedding
+        inference_id: dense-inference-id
+        body: >
+          {
+            "service": "text_embedding_test_service",
+            "service_settings": {
+              "model": "my_model",
+              "dimensions": 10,
+              "similarity": "cosine",
+              "api_key": "abc64"
+            },
+            "task_settings": {
+            }
+          }
+
+  - do:
+      indices.create:
+        index: test-index
+        body:
+          mappings:
+            properties:
+              sparse_field:
+                type: semantic_text
+                inference_id: sparse-inference-id
+              dense_field:
+                type: semantic_text
+                inference_id: dense-inference-id
+
+  # Index a doc to set mappings internally
+  - do:
+      index:
+        index: test-index
+        id: doc_1
+        body:
+          dense_field: "inference test"
+          sparse_field: "another inference test"
+
+---
+"Fails for non-compatible dimensions":
+
+  - do:
+      inference.delete_model:
+        task_type: text_embedding
+        inference_id: dense-inference-id
+
+  - do:
+      inference.put_model:
+        task_type: text_embedding
+        inference_id: dense-inference-id
+        body: >
+          {
+            "service": "text_embedding_test_service",
+            "service_settings": {
+              "model": "my_model",
+              "dimensions": 20,
+              "similarity": "cosine",
+              "api_key": "abc64"
+            },
+            "task_settings": {
+            }
+          }
+
+  - do:
+      catch: /Incompatible model settings for field \[dense_field\].+/
+      index:
+        index: test-index
+        id: doc_2
+        body:
+          dense_field: "some other test"
+
+---
+"Fails for non-compatible similarity":
+
+  - do:
+      inference.delete_model:
+        task_type: text_embedding
+        inference_id: dense-inference-id
+
+  - do:
+      inference.put_model:
+        task_type: text_embedding
+        inference_id: dense-inference-id
+        body: >
+          {
+            "service": "text_embedding_test_service",
+            "service_settings": {
+              "model": "my_model",
+              "dimensions": 10,
+              "similarity": "dot_product",
+              "api_key": "abc64"
+            },
+            "task_settings": {
+            }
+          }
+
+  - do:
+      catch: /Incompatible model settings for field \[dense_field\].+/
+      index:
+        index: test-index
+        id: doc_2
+        body:
+          dense_field: "some other test"
+
+---
+"Fails for non-compatible task type for dense vectors":
+
+  - do:
+      inference.delete_model:
+        task_type: text_embedding
+        inference_id: dense-inference-id
+
+
+  - do:
+      inference.put_model:
+        task_type: sparse_embedding
+        inference_id: dense-inference-id
+        body: >
+          {
+            "service": "test_service",
+            "service_settings": {
+              "model": "my_model",
+              "api_key": "abc64"
+            },
+            "task_settings": {
+            }
+          }
+
+  - do:
+      catch: /Incompatible model settings for field \[dense_field\].+/
+      index:
+        index: test-index
+        id: doc_2
+        body:
+          dense_field: "some other test"
+
+---
+"Fails for non-compatible task type for sparse vectors":
+
+  - do:
+      inference.delete_model:
+        task_type: sparse_embedding
+        inference_id: sparse-inference-id
+
+  - do:
+      inference.put_model:
+        task_type: text_embedding
+        inference_id: sparse-inference-id
+        body: >
+          {
+            "service": "text_embedding_test_service",
+            "service_settings": {
+              "model": "my_model",
+              "dimensions": 10,
+              "similarity": "cosine",
+              "api_key": "abc64"
+            },
+            "task_settings": {
+            }
+          }
+
+  - do:
+      catch: /Incompatible model settings for field \[sparse_field\].+/
+      index:
+        index: test-index
+        id: doc_2
+        body:
+          sparse_field: "some other test"
+
+