@dataclouder/nest-vertex 0.0.57 → 0.0.58

package/controllers/stt/adapter-stt.controller.d.ts

@@ -0,0 +1,10 @@
+import { type MulterFile } from '@webundsoehne/nest-fastify-file-upload';
+import { GroqService } from '../../services/whisper/groq.service';
+import { LocalSttService } from '../../services/whisper/local-stt.service';
+export declare class AdapterSttController {
+    private readonly groqService;
+    private readonly localSttService;
+    private readonly logger;
+    constructor(groqService: GroqService, localSttService: LocalSttService);
+    processAudio(file: MulterFile, provider?: 'local' | 'groq'): Promise<any>;
+}

package/controllers/stt/adapter-stt.controller.js

@@ -0,0 +1,79 @@
+"use strict";
+var __decorate = (this && this.__decorate) || function (decorators, target, key, desc) {
+    var c = arguments.length, r = c < 3 ? target : desc === null ? desc = Object.getOwnPropertyDescriptor(target, key) : desc, d;
+    if (typeof Reflect === "object" && typeof Reflect.decorate === "function") r = Reflect.decorate(decorators, target, key, desc);
+    else for (var i = decorators.length - 1; i >= 0; i--) if (d = decorators[i]) r = (c < 3 ? d(r) : c > 3 ? d(target, key, r) : d(target, key)) || r;
+    return c > 3 && r && Object.defineProperty(target, key, r), r;
+};
+var __metadata = (this && this.__metadata) || function (k, v) {
+    if (typeof Reflect === "object" && typeof Reflect.metadata === "function") return Reflect.metadata(k, v);
+};
+var __param = (this && this.__param) || function (paramIndex, decorator) {
+    return function (target, key) { decorator(target, key, paramIndex); }
+};
+var AdapterSttController_1;
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.AdapterSttController = void 0;
+const common_1 = require("@nestjs/common");
+const nest_fastify_file_upload_1 = require("@webundsoehne/nest-fastify-file-upload");
+const swagger_1 = require("@nestjs/swagger");
+const nest_core_1 = require("@dataclouder/nest-core");
+const groq_service_1 = require("../../services/whisper/groq.service");
+const local_stt_service_1 = require("../../services/whisper/local-stt.service");
+let AdapterSttController = AdapterSttController_1 = class AdapterSttController {
+    groqService;
+    localSttService;
+    logger = new common_1.Logger(AdapterSttController_1.name);
+    constructor(groqService, localSttService) {
+        this.groqService = groqService;
+        this.localSttService = localSttService;
+    }
+    async processAudio(file, provider = 'local') {
+        this.logger.log(`Receive request with provider: ${provider}`);
+        if (!file || !file.buffer) {
+            this.logger.error('No file buffer received.');
+            return { error: 'No file uploaded or file buffer is missing.' };
+        }
+        this.logger.log(`Received file: ${file.originalname}, mimetype: ${file.mimetype}, size: ${file.size}`);
+        try {
+            let result;
+            if (provider === 'groq') {
+                result = await this.groqService.transcribeAudio(file.buffer, file.originalname, file.mimetype);
+            }
+            else {
+                result = await this.localSttService.transcribeAudio(file.buffer, file.originalname, file.mimetype);
+            }
+            return result;
+        }
+        catch (error) {
+            this.logger.error('Error during transcription process:', error);
+            return { error: 'Failed to transcribe audio.', details: error.message };
+        }
+    }
+};
+exports.AdapterSttController = AdapterSttController;
+__decorate([
+    (0, common_1.Post)('transcribe-bytes'),
+    (0, common_1.UseInterceptors)((0, nest_fastify_file_upload_1.FileInterceptor)('file')),
+    (0, swagger_1.ApiConsumes)('multipart/form-data'),
+    (0, nest_fastify_file_upload_1.ApiFileBody)('file'),
+    (0, swagger_1.ApiQuery)({
+        name: 'provider',
+        enum: ['local', 'groq'],
+        required: false,
+        description: 'The provider to use for transcription. Defaults to "local".',
+    }),
+    __param(0, (0, common_1.UploadedFile)('file')),
+    __param(1, (0, common_1.Query)('provider')),
+    __metadata("design:type", Function),
+    __metadata("design:paramtypes", [Object, String]),
+    __metadata("design:returntype", Promise)
+], AdapterSttController.prototype, "processAudio", null);
+exports.AdapterSttController = AdapterSttController = AdapterSttController_1 = __decorate([
+    (0, swagger_1.ApiTags)('Speech-to-Text'),
+    (0, common_1.Controller)('api/ai-services/adapter/stt'),
+    (0, common_1.UseFilters)(nest_core_1.AllExceptionsHandler),
+    __metadata("design:paramtypes", [groq_service_1.GroqService,
+        local_stt_service_1.LocalSttService])
+], AdapterSttController);
+//# sourceMappingURL=adapter-stt.controller.js.map

package/controllers/stt/groq-stt.controller.js

@@ -55,7 +55,7 @@ __decorate([
 ], GroqSttController.prototype, "processAudio", null);
 exports.GroqSttController = GroqSttController = GroqSttController_1 = __decorate([
     (0, swagger_1.ApiTags)('Speech-to-Text'),
-    (0, common_1.Controller)('api/ai-services/groq-stt'),
+    (0, common_1.Controller)('api/ai-services/groq/stt'),
     (0, common_1.UseFilters)(nest_core_1.AllExceptionsHandler),
     __metadata("design:paramtypes", [groq_service_1.GroqService])
 ], GroqSttController);

package/controllers/stt/local-stt.controller.js

@@ -55,7 +55,7 @@ __decorate([
 ], LocalSttController.prototype, "processAudio", null);
 exports.LocalSttController = LocalSttController = LocalSttController_1 = __decorate([
     (0, swagger_1.ApiTags)('Speech-to-Text'),
-    (0, common_1.Controller)('api/ai-services/local-stt'),
+    (0, common_1.Controller)('api/ai-services/local/stt'),
     (0, common_1.UseFilters)(nest_core_1.AllExceptionsHandler),
     __metadata("design:paramtypes", [local_stt_service_1.LocalSttService])
 ], LocalSttController);

package/controllers/tts/vertex-gemini-tts.controller.js

@@ -79,7 +79,7 @@ __decorate([
     __metadata("design:returntype", Promise)
 ], VertexGeminiTtsController.prototype, "synthesizeSpeech", null);
 exports.VertexGeminiTtsController = VertexGeminiTtsController = __decorate([
-    (0, swagger_1.ApiTags)('Vertex Gemini TTS'),
+    (0, swagger_1.ApiTags)('Text To Speech Gemini'),
     (0, common_1.Controller)('api/vertex-gemini/tts'),
     (0, common_1.UseFilters)(nest_core_1.AllExceptionsHandler),
     __metadata("design:paramtypes", [vertex_gemini_tts_service_1.VertexGeminiTtsService])

package/controllers/tts/vertex-tts-adapter.controller.js

@@ -104,7 +104,7 @@ __decorate([
     __metadata("design:returntype", Promise)
 ], VertexTtsAdapterController.prototype, "listVoices", null);
 exports.VertexTtsAdapterController = VertexTtsAdapterController = __decorate([
-    (0, swagger_1.ApiTags)('Vertex TTS Adapter'),
+    (0, swagger_1.ApiTags)('Text To Speech Adapter'),
     (0, common_1.Controller)('api/ai-services/adapter/tts'),
     (0, common_2.UseFilters)(nest_core_1.AllExceptionsHandler),
     __metadata("design:paramtypes", [vertex_tts_service_1.VertextTtsService])

package/models/adapter.models.d.ts

@@ -21,7 +21,7 @@ export interface ChatJsonResponse extends MessageLLM {
 export interface ChatLLMRequestAdapter {
     messages: MessageLLM[];
     provider?: string;
-    model?: IAIModel;
+    model?: Partial<IAIModel>;
     returnJson?: boolean;
     keyType?: TierType;
 }

package/nest-vertex.module.js

@@ -45,6 +45,7 @@ const nest_auth_1 = require("@dataclouder/nest-auth");
 const groq_llm_service_1 = require("./services/llm/groq-llm.service");
 const groq_llm_controller_1 = require("./controllers/llm/groq-llm.controller");
 const video_gen_adapter_controller_1 = require("./controllers/video/video-gen-adapter.controller");
+const adapter_stt_controller_1 = require("./controllers/stt/adapter-stt.controller");
 let NestVertexModule = class NestVertexModule {
 };
 exports.NestVertexModule = NestVertexModule;
@@ -112,6 +113,7 @@ exports.NestVertexModule = NestVertexModule = __decorate([
             groq_stt_controller_1.GroqSttController,
             local_stt_controller_1.LocalSttController,
             vertex_gemini_tts_controller_1.VertexGeminiTtsController,
+            adapter_stt_controller_1.AdapterSttController,
         ],
     })
 ], NestVertexModule);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@dataclouder/nest-vertex",
-  "version": "0.0.57",
+  "version": "0.0.58",
   "description": "NestJS Vertex AI library for Dataclouder",
   "author": "dataclouder",
   "license": "MIT",

package/services/vertex-gemini-chat.service.d.ts

@@ -15,6 +15,7 @@ export declare class GeminiChatService {
     chatStream(messages: MessageLLM[], model?: string, keyType?: TierType): Promise<AsyncIterable<ChatMessageDict>>;
     listModels(): Promise<Record<string, string>[]>;
     getDefaultQualityModel(quality: EModelQuality): string;
+    private getBestModel;
     private _extractJsonWithRecovery;
     chatAndExtractJson(messages: MessageLLM[], model?: string, keyType?: TierType): Promise<ChatJsonResponse>;
     describeImageByUrl(dto: DescribeImageRequestAdapter): Promise<any>;

package/services/vertex-gemini-chat.service.js

@@ -93,8 +93,8 @@ let GeminiChatService = GeminiChatService_1 = class GeminiChatService {
             this.logger.debug(`Received Gemini response text. ${response?.text.slice(0, 50).replace(/\n/g, '')} ...`);
             const responseText = response?.text ?? '';
             const tokens = {
-                input: response?.usageMetadata?.candidatesTokenCount,
-                output: response?.usageMetadata?.promptTokenCount,
+                input: response?.usageMetadata?.promptTokenCount,
+                output: response?.usageMetadata?.candidatesTokenCount,
                 total: response?.usageMetadata?.totalTokenCount,
             };
             return {
@@ -201,6 +201,19 @@ let GeminiChatService = GeminiChatService_1 = class GeminiChatService {
                 return gemini_models_1.GeminiModels.Gemini2_5Pro;
         }
     }
+    getBestModel(conversation) {
+        if (!conversation.model) {
+            return { provider: 'google', modelName: gemini_models_1.GeminiModels.Gemini2_5Lite };
+        }
+        const { quality, modelName, provider } = conversation.model;
+        if (quality) {
+            return { provider: 'google', modelName: this.getDefaultQualityModel(quality) };
+        }
+        if (modelName && provider) {
+            return { provider, modelName };
+        }
+        return { provider: 'google', modelName: gemini_models_1.GeminiModels.Gemini2_5Lite };
+    }
     async _extractJsonWithRecovery(responseText, model, keyType) {
         try {
             const json = (0, llm_models_1.extractJsonFromResponse)(responseText);
@@ -286,36 +299,30 @@ let GeminiChatService = GeminiChatService_1 = class GeminiChatService {
     }
     async chatWithConversation(conversation) {
         const startTime = Date.now();
-        if (!conversation?.model) {
-            conversation.model = { provider: 'google', modelName: gemini_models_1.GeminiModels.Gemini2_5Lite, id: 'no-id' };
-        }
+        const { provider, modelName } = this.getBestModel(conversation);
         const tierType = conversation.tierType || key_balancer_models_1.TierType.TIER_1;
-        if (conversation?.model?.quality) {
-            conversation.model.provider = 'google';
-            conversation.model.modelName = this.getDefaultQualityModel(conversation.model.quality);
-        }
         const returnJson = conversation.returnJson;
         if (returnJson) {
-            const obj = await this.chatAndExtractJson(conversation.messages, conversation.model.modelName, tierType);
+            const obj = await this.chatAndExtractJson(conversation.messages, modelName, tierType);
             const endTime = Date.now();
             const processTime = (endTime - startTime) / 1000;
             const metadata = {
                 type: 'json',
-                provider: conversation.model.provider,
-                model: conversation.model.modelName,
+                provider,
+                model: modelName,
                 processTime,
                 ...obj?.metadata,
             };
             return { content: obj.json, role: adapter_models_1.ChatRole.Assistant, metadata };
         }
         else {
-            const response = await this.chat(conversation.messages, conversation.model.modelName, tierType);
+            const response = await this.chat(conversation.messages, modelName, tierType);
             const endTime = Date.now();
             const processTime = (endTime - startTime) / 1000;
             return {
                 content: response.content,
                 role: response.role,
-                metadata: { provider: conversation.model.provider, model: conversation.model.modelName, processTime, tokens: response.metadata.tokens },
+                metadata: { provider, model: modelName, processTime, tokens: response.metadata.tokens },
             };
         }
     }

package/services/whisper/local-stt.service.js

@@ -51,8 +51,10 @@ let LocalSttService = LocalSttService_1 = class LocalSttService {
     logger = new common_1.Logger(LocalSttService_1.name);
     openai;
     constructor() {
+        const aiServerHost = process.env.AI_LAB_HOST;
+        console.log('Connecting to host AI_LAB_HOST', aiServerHost);
         this.openai = new openai_1.default({
-            baseURL: 'http://192.168.2.224:8000/v1',
+            baseURL: `${aiServerHost}:3171/v1`,
             apiKey: '',
         });
     }
@@ -103,13 +105,22 @@ let LocalSttService = LocalSttService_1 = class LocalSttService {
             this.logger.warn(`Could not determine valid extension for mime type ${mimeType}. Using original filename: ${originalFileName}. Transcription may fail if the filename lacks a supported extension.`);
         }
         try {
+            console.log(' -> Request to...', this.openai.baseURL);
             const result = await this.openai.audio.transcriptions.create({
                 model: 'rtlingo/mobiuslabsgmbh-faster-whisper-large-v3-turbo',
                 file: await (0, openai_1.toFile)(fileBuffer, effectiveFileName, { type: mimeType }),
                 response_format: 'verbose_json',
                 timestamp_granularities: ['word'],
             });
-            this.logger.log(`Transcription successful for file: ${effectiveFileName}`);
+            if (result?.segments) {
+                delete result.segments;
+            }
+            if (result?.words?.length) {
+                result.words.forEach((word) => {
+                    delete word?.probability;
+                });
+            }
+            this.logger.log(`Transcription successful for file: ${result.text}, lang: ${result.language}, duration: ${result.duration}`);
             return result;
         }
         catch (error) {