GenAi: switch to OpenAI official SDK and improve UI

Author: TyomaVader

build.gradle

@@ -288,9 +288,17 @@ configure(javaProjects) {
             dependency 'org.apache.bcel:bcel:6.6.0'
 
             // AI integration
-            dependency 'dev.langchain4j:langchain4j:1.7.1'
-            dependency 'dev.langchain4j:langchain4j-azure-open-ai-spring-boot-starter:1.7.1-beta14'
-            dependency 'dev.langchain4j:langchain4j-spring-boot-starter:1.7.1-beta14'
+            dependency 'dev.langchain4j:langchain4j:1.8.0'
+            dependency 'dev.langchain4j:langchain4j-spring-boot-starter:1.8.0-beta15'
+            dependency 'dev.langchain4j:langchain4j-open-ai-official:1.8.0-beta15'
+
+            // Conflict resolution
+            dependency 'com.google.errorprone:error_prone_annotations:2.33.0'
+            dependency 'org.jetbrains.kotlin:kotlin-stdlib:1.9.10'
+            dependency 'org.jetbrains.kotlin:kotlin-stdlib-jdk8:1.9.10'
+            dependency 'com.fasterxml:classmate:1.7.0'
+            dependency 'org.apache.httpcomponents.client5:httpclient5:5.5'
+            dependency 'org.apache.httpcomponents.core5:httpcore5:5.3.4'
 
             dependency 'org.reactivestreams:reactive-streams:1.0.4'
 

guide/configuration.md

@@ -167,15 +167,16 @@ services:
 
 ## Gen AI Configuration
 
-TimeBase Web Administrator can generate QQL queries with the help of an external Gen AI provider. 
+TimeBase Web Administrator can generate QQL queries with the help of an external Gen AI provider.
 The feature reads its settings from the `ai-api` section of `application.yaml`.
 
 Example config:
 
 ```yaml
 ai-api:
   enabled: true
-  endpointUrl: "https://YOUR-RESOURCE-NAME.openai.azure.com" # endpoint must support Azure OpenAI API
+  provider: AZURE_LEGACY # OPENAI | AZURE | AZURE_LEGACY | GITHUB
+  endpointUrl: "https://YOUR-RESOURCE-NAME.openai.azure.com" # base url of the ai api, may be ommitted for OPENAI and GITHUB providers
   deploymentName: gpt-5-mini-2025-08-07 # deployment for QQL generation
   embeddingDeploymentName: text-embedding-3-small-1 # deployment for embeddings
   keys: # ai api keys per user
@@ -189,4 +190,17 @@ ai-api:
 If a user is not present in `ai-api.keys`, the system falls back to `security.oauth2.users[].aiApiKey` when it is defined.
 
 > [!IMPORTANT]
-> The Gen AI endpoint must be compatible with the Azure OpenAI API.
+> The Gen AI endpoint must be compatible with the selected provider's API.
+
+Supported AI providers:
+
+1. AZURE_LEGACY includes the deployment name in the request URL. Example:
+   `https://YOUR-RESOURCE-NAME.openai.azure.com/openai/deployments/gpt-5-mini-2025-08-07/chat/completions`
+   where `https://YOUR-RESOURCE-NAME.openai.azure.com` is the `endpointUrl` and
+   `gpt-5-mini-2025-08-07` is the `deploymentName`.
+2. AZURE uses the `endpointUrl` and provides the model name in the request body.
+3. OPENAI is the official OpenAI API. The endpoint defaults to
+   `https://api.openai.com/v1`, but may be any compatible endpoint.
+4. GITHUB uses the GitHub Models compatible API. The endpoint defaults to
+   `https://models.inference.ai.azure.com`. Use a GitHub personal access token
+   as the key — see GitHub docs [docs](https://docs.github.com/en/github-models/about-github-models) for details.

java/ws-server/build.gradle

@@ -82,8 +82,8 @@ dependencies {
 
     // AI integration
     implementation 'dev.langchain4j:langchain4j'
-    implementation 'dev.langchain4j:langchain4j-azure-open-ai-spring-boot-starter'
     implementation 'dev.langchain4j:langchain4j-spring-boot-starter'
+    implementation 'dev.langchain4j:langchain4j-open-ai-official'
 
     compileOnly 'com.webcohesion.enunciate:enunciate-core-annotations'
 

java/ws-server/src/main/java/com/epam/deltix/tbwg/webapp/config/GenAiConfig.java

@@ -54,27 +54,21 @@ public class GenAiConfig {
     private static final String FALLBACK_EMB_FILE = "qql-docs-embeddings.json";
 
     @Bean
-    public PerUserAzureChatModel perUserAzureChatModel(AiApiSettings settings,
-                                                       UserAiApiKeyProvider keyProvider) {
-        return new PerUserAzureChatModel(settings.getEndpointUrl(),
-                settings.getDeploymentName(),
-                keyProvider);
+    public ChatModel perUserChatModel(AiApiSettings settings,
+                                      UserAiApiKeyProvider keyProvider) {
+        return new PerUserOpenAiOfficialChatModel(settings, keyProvider);
     }
 
     @Bean
-    public PerUserAzureStreamingChatModel perUserAzureStreamingChatModel(AiApiSettings settings,
-                                                                         UserAiApiKeyProvider keyProvider) {
-        return new PerUserAzureStreamingChatModel(settings.getEndpointUrl(),
-                settings.getDeploymentName(),
-                keyProvider);
+    public StreamingChatModel perUserStreamingChatModel(AiApiSettings settings,
+                                                        UserAiApiKeyProvider keyProvider) {
+        return new PerUserOpenAiOfficialStreamingChatModel(settings, keyProvider);
     }
 
     @Bean
-    public PerUserAzureEmbeddingModel perUserAzureEmbeddingModel(AiApiSettings settings,
-                                                                 UserAiApiKeyProvider keyProvider) {
-        return new PerUserAzureEmbeddingModel(settings.getEndpointUrl(),
-                settings.getEmbeddingDeploymentName(),
-                keyProvider);
+    public EmbeddingModel perUserEmbeddingModel(AiApiSettings settings,
+                                                UserAiApiKeyProvider keyProvider) {
+        return new PerUserOpenAiOfficialEmbeddingModel(settings, keyProvider);
     }
 
     @Bean

java/ws-server/src/main/java/com/epam/deltix/tbwg/webapp/services/genai/GenAiService.java

@@ -30,6 +30,8 @@
 import com.epam.deltix.tbwg.webapp.settings.AiApiSettings;
 import com.epam.deltix.tbwg.webapp.websockets.subscription.SubscriptionChannel;
 import dev.langchain4j.model.chat.response.ChatResponse;
+import dev.langchain4j.model.chat.response.PartialResponse;
+import dev.langchain4j.model.chat.response.PartialResponseContext;
 import dev.langchain4j.service.Result;
 import dev.langchain4j.service.TokenStream;
 import org.springframework.boot.autoconfigure.condition.ConditionalOnBean;
@@ -158,8 +160,16 @@ private void iterativeGenerate(String username, String schema, String pinned,
 
             String wrappedPrompt = PerUserChatMaker.wrapUserInput(username, prompt);
             TokenStream ts = genAiHelperService.genQql(schema, pinned, wrappedPrompt)
-                    .onPartialResponse((String part) -> {
-                        if (!ctx.stopped().get() && part != null && !part.isEmpty()) {
+                    .onPartialResponseWithContext((PartialResponse partResp, PartialResponseContext partCtx) -> {
+                        if (ctx.stopped().get()) {
+                            partCtx.streamingHandle().cancel();
+                            latch.countDown();
+                            LOG.info("Generation cancelled by user.");
+                            return;
+                        }
+
+                        String part = partResp.text();
+                        if (part != null && !part.isEmpty()) {
                             streamed.append(part);
                             send(ch, QqlGenMessage.builder("PART")
                                     .data(part));
@@ -175,7 +185,6 @@ private void iterativeGenerate(String username, String schema, String pinned,
                                     .error(e.getMessage()).finalEvent(true));
                         latch.countDown();
                     });
-
             ts.start();
             try {
                 if (!latch.await(STREAM_TIMEOUT.toMillis(), TimeUnit.MILLISECONDS)) {
@@ -223,7 +232,7 @@ private void send(SubscriptionChannel ch, QqlGenMessage.Builder b) {
         ch.sendMessage(b.build());
     }
 
-    private String loadOverview() {
+    private static String loadOverview() {
         try {
             return Resources.toString(
                     Resources.getResource("qql_gen/overview.md"),
@@ -273,19 +282,19 @@ private static void describeBasic(DXTickStream stream, StringBuilder b) {
         b.append("\n\n");
     }
 
-    private String repairPrompt(String intent, String prev, String err) {
+    private static String repairPrompt(String intent, String prev, String err) {
         return "User intent:\n" + intent +
                 "\n\nPrevious QQL (fix minimally):\n" + prev +
                 "\n\nCompiler error:\n" + err +
                 "\n\nOutput ONLY corrected QQL.";
     }
 
-    private String summaryPinned(String pinned) {
+    private static String summaryPinned(String pinned) {
         if (pinned == null) return "No pinned docs.";
         return "Pinned size=" + pinned.length();
     }
 
-    private Set<String> parseStreamKeys(String raw) {
+    private static Set<String> parseStreamKeys(String raw) {
         if (raw == null || raw.isBlank()) return Collections.emptySet();
         return Stream.of(raw.split(","))
                 .map(String::trim)

java/ws-server/src/main/java/com/epam/deltix/tbwg/webapp/services/genai/models/PerUserAzureStreamingChatModel.java

@@ -20,6 +20,7 @@
 import dev.langchain4j.data.message.UserMessage;
 import dev.langchain4j.model.chat.request.ChatRequest;
 import dev.langchain4j.model.chat.request.ChatRequestParameters;
+import dev.langchain4j.model.openaiofficial.OpenAiOfficialChatRequestParameters;
 
 import java.util.ArrayList;
 import java.util.List;
@@ -105,4 +106,8 @@ public static ProcessResult prepare(ChatRequest chatRequest) {
     public static ChatRequestParameters defaultParams(String modelName) {
         return ChatRequestParameters.builder().modelName(modelName).build();
     }
+
+    public static OpenAiOfficialChatRequestParameters defaultOpenAiOfficialParams(String modelName) {
+        return OpenAiOfficialChatRequestParameters.builder().modelName(modelName).build();
+    }
 }

java/ws-server/src/main/java/com/epam/deltix/tbwg/webapp/services/genai/models/PerUserChatMaker.java

@@ -16,26 +16,24 @@
  */
 package com.epam.deltix.tbwg.webapp.services.genai.models;
 
-import dev.langchain4j.model.azure.AzureOpenAiChatModel;
+import com.epam.deltix.tbwg.webapp.settings.AiApiSettings;
 import dev.langchain4j.model.chat.ChatModel;
 import dev.langchain4j.model.chat.request.ChatRequest;
 import dev.langchain4j.model.chat.request.ChatRequestParameters;
 import dev.langchain4j.model.chat.response.ChatResponse;
+import dev.langchain4j.model.openaiofficial.OpenAiOfficialChatModel;
 
-public class PerUserAzureChatModel implements ChatModel {
+public class PerUserOpenAiOfficialChatModel implements ChatModel {
 
-    private final String endpoint;
-    private final String deploymentName;
+    private final AiApiSettings settings;
     private final UserAiApiKeyProvider keyProvider;
     private final ChatRequestParameters defaultParams;
 
-    public PerUserAzureChatModel(String endpoint,
-                                 String deploymentName,
-                                 UserAiApiKeyProvider keyProvider) {
-        this.endpoint = endpoint;
-        this.deploymentName = deploymentName;
+    public PerUserOpenAiOfficialChatModel(AiApiSettings settings,
+                                          UserAiApiKeyProvider keyProvider) {
+        this.settings = settings;
         this.keyProvider = keyProvider;
-        this.defaultParams = PerUserChatMaker.defaultParams(deploymentName);
+        this.defaultParams = PerUserChatMaker.defaultOpenAiOfficialParams(settings.chatModelNameForParams());
     }
 
     @Override
@@ -49,13 +47,33 @@ public ChatResponse doChat(ChatRequest chatRequest) {
         String username = pr.username();
         ChatRequest effectiveRequest = pr.request();
 
-        ChatModel model = AzureOpenAiChatModel.builder()
-                .endpoint(endpoint)
-                .apiKey(keyProvider.resolve(username))
-                .deploymentName(deploymentName)
-                .build();
+        ChatModel model = createDelegate(username);
 
         return model.doChat(effectiveRequest);
     }
 
+    private ChatModel createDelegate(String username) {
+        OpenAiOfficialChatModel.Builder builder = OpenAiOfficialChatModel.builder()
+                .apiKey(keyProvider.resolve(username));
+
+        String endpoint = settings.getEndpointUrl();
+        if (endpoint != null && !endpoint.isBlank()) {
+            builder.baseUrl(endpoint);
+        }
+
+        String deployment = settings.requireDeploymentName();
+        switch (settings.providerOrDefault()) {
+            case OPENAI -> builder.modelName(deployment);
+            case AZURE -> builder.isAzure(true)
+                    .modelName(deployment)
+                    .azureDeploymentName(deployment);
+            case AZURE_LEGACY -> builder.isAzure(true)
+                    .azureDeploymentName(deployment)
+                    .modelName("");
+            case GITHUB -> builder.isGitHubModels(true)
+                    .modelName(deployment);
+        }
+
+        return builder.build();
+    }
 }

…/genai/models/PerUserAzureChatModel.java …dels/PerUserOpenAiOfficialChatModel.javajava/ws-server/src/main/java/com/epam/deltix/tbwg/webapp/services/genai/models/PerUserAzureChatModel.java renamed to java/ws-server/src/main/java/com/epam/deltix/tbwg/webapp/services/genai/models/PerUserOpenAiOfficialChatModel.java java/ws-server/src/main/java/com/epam/deltix/tbwg/webapp/services/genai/models/PerUserAzureChatModel.java renamed to java/ws-server/src/main/java/com/epam/deltix/tbwg/webapp/services/genai/models/PerUserOpenAiOfficialChatModel.java

@@ -16,26 +16,24 @@
  */
 package com.epam.deltix.tbwg.webapp.services.genai.models;
 
+import com.epam.deltix.tbwg.webapp.settings.AiApiSettings;
 import dev.langchain4j.data.embedding.Embedding;
 import dev.langchain4j.data.segment.TextSegment;
-import dev.langchain4j.model.azure.AzureOpenAiEmbeddingModel;
 import dev.langchain4j.model.embedding.EmbeddingModel;
+import dev.langchain4j.model.openaiofficial.OpenAiOfficialEmbeddingModel;
 import dev.langchain4j.model.output.Response;
 
 import java.util.ArrayList;
 import java.util.List;
 
-public class PerUserAzureEmbeddingModel implements EmbeddingModel {
+public class PerUserOpenAiOfficialEmbeddingModel implements EmbeddingModel {
 
-    private final String endpoint;
-    private final String deploymentName;
+    private final AiApiSettings settings;
     private final UserAiApiKeyProvider keyProvider;
 
-    public PerUserAzureEmbeddingModel(String endpoint,
-                                      String deploymentName,
-                                      UserAiApiKeyProvider keyProvider) {
-        this.endpoint = endpoint;
-        this.deploymentName = deploymentName;
+    public PerUserOpenAiOfficialEmbeddingModel(AiApiSettings settings,
+                                               UserAiApiKeyProvider keyProvider) {
+        this.settings = settings;
         this.keyProvider = keyProvider;
     }
 
@@ -60,11 +58,32 @@ public Response<List<Embedding>> embedAll(List<TextSegment> textSegments) {
             effectiveSegments.add(TextSegment.from(unwrappedText, segment.metadata()));
         }
 
-        EmbeddingModel model = AzureOpenAiEmbeddingModel.builder()
-                .endpoint(endpoint)
-                .deploymentName(deploymentName)
-                .apiKey(keyProvider.resolve(username))
-                .build();
+        EmbeddingModel model = createDelegate(username);
         return model.embedAll(effectiveSegments);
     }
+
+    private EmbeddingModel createDelegate(String username) {
+        OpenAiOfficialEmbeddingModel.Builder builder = OpenAiOfficialEmbeddingModel.builder()
+                .apiKey(keyProvider.resolve(username));
+
+        String endpoint = settings.getEndpointUrl();
+        if (endpoint != null && !endpoint.isBlank()) {
+            builder.baseUrl(endpoint);
+        }
+
+        String deployment = settings.requireEmbeddingDeploymentName();
+        switch (settings.providerOrDefault()) {
+            case OPENAI -> builder.modelName(deployment);
+            case AZURE -> builder.isAzure(true)
+                    .modelName(deployment)
+                    .azureDeploymentName(deployment);
+            case AZURE_LEGACY -> builder.isAzure(true)
+                    .azureDeploymentName(deployment)
+                    .modelName("");
+            case GITHUB -> builder.isGitHubModels(true)
+                    .modelName(deployment);
+        }
+
+        return builder.build();
+    }
 }