@futdevpro/nts-dynamo 1.15.73 → 1.15.75

package/src/_modules/ai/_services/lmstudio-embedding.control-service.spec.ts

@@ -1,197 +1,197 @@
-import { CreateEmbeddingResponse } from 'openai/resources';
-import { DyFM_AI_Provider } from '@futdevpro/fsm-dynamo/ai';
-import { DyFM_Error } from '@futdevpro/fsm-dynamo';
-
-import { DyNTS_LMStudio_Embedding_ControlService } from './lmstudio-embedding.control-service';
-import { DyNTS_AI_CostEvent } from '../_models/interfaces/dynts-ai-cost-event.interface';
-
-describe('| DyNTS_LMStudio_Embedding_ControlService', () => {
-
-  const baseUrl: string = 'http://localhost:1234/v1';
-  const issuer: string = 'test-issuer';
-
-  /** Egy OpenAI-kompatibilis embeddings-válasz JSON-string-é csomagolva, fetch-mockhoz. */
-  function okResponse(embeddings: number[][], usage?: { prompt_tokens: number; total_tokens: number }): Response {
-    const body: string = JSON.stringify({
-      object: 'list',
-      data: embeddings.map((embedding, index) => ({ object: 'embedding', embedding: embedding, index: index })),
-      usage: usage,
-    });
-    return {
-      ok: true,
-      status: 200,
-      text: () => Promise.resolve(body),
-    } as Response;
-  }
-
-  /** Egy hibás (non-2xx) válasz fetch-mockhoz. */
-  function errorResponse(status: number, body: string): Response {
-    return {
-      ok: false,
-      status: status,
-      text: () => Promise.resolve(body),
-    } as Response;
-  }
-
-  let fetchSpy: jasmine.Spy;
-
-  beforeEach(() => {
-    fetchSpy = spyOn(globalThis, 'fetch');
-  });
-
-  describe('| constructor', () => {
-    it('| should throw a DyFM_Error if baseUrl is empty', () => {
-      expect(() => new DyNTS_LMStudio_Embedding_ControlService({ baseUrl: '' })).toThrowMatching(
-        (error: unknown) => error instanceof DyFM_Error,
-      );
-    });
-
-    it('| should set provider to LocalAI', () => {
-      const service: DyNTS_LMStudio_Embedding_ControlService = new DyNTS_LMStudio_Embedding_ControlService({ baseUrl: baseUrl });
-      expect(service.aiProvider).toBe(DyFM_AI_Provider.LocalAI);
-      expect(service.capabilities.embeddings).toBe(true);
-    });
-  });
-
-  describe('| createEmbeddings', () => {
-    it('| should POST to ${baseUrl}/embeddings and return vectors in order', async () => {
-      fetchSpy.and.returnValue(Promise.resolve(okResponse([[0.1, 0.2], [0.3, 0.4]])));
-
-      const service: DyNTS_LMStudio_Embedding_ControlService = new DyNTS_LMStudio_Embedding_ControlService({ baseUrl: baseUrl });
-      const result: number[][] | CreateEmbeddingResponse = await service.createEmbeddings({
-        texts: ['hello', 'world'],
-        model: 'nomic-embed-text-v1.5',
-        issuer: issuer,
-      });
-
-      expect(result).toEqual([[0.1, 0.2], [0.3, 0.4]]);
-      expect(fetchSpy).toHaveBeenCalledTimes(1);
-      const url: string = fetchSpy.calls.mostRecent().args[0] as string;
-      expect(url).toBe('http://localhost:1234/v1/embeddings');
-      const init: RequestInit = fetchSpy.calls.mostRecent().args[1] as RequestInit;
-      expect(init.method).toBe('POST');
-      const sentBody: { model: string; input: string[] } = JSON.parse(init.body as string);
-      expect(sentBody.model).toBe('nomic-embed-text-v1.5');
-      expect(sentBody.input).toEqual(['hello', 'world']);
-    });
-
-    it('| should trim trailing slashes from baseUrl', async () => {
-      fetchSpy.and.returnValue(Promise.resolve(okResponse([[0.5]])));
-      const service: DyNTS_LMStudio_Embedding_ControlService = new DyNTS_LMStudio_Embedding_ControlService({
-        baseUrl: 'http://localhost:1234/v1///',
-      });
-      await service.createEmbeddings({ texts: ['x'], model: 'm', issuer: issuer });
-      expect(fetchSpy.calls.mostRecent().args[0]).toBe('http://localhost:1234/v1/embeddings');
-    });
-
-    it('| should add a Bearer header when apiKey is provided', async () => {
-      fetchSpy.and.returnValue(Promise.resolve(okResponse([[0.5]])));
-      const service: DyNTS_LMStudio_Embedding_ControlService = new DyNTS_LMStudio_Embedding_ControlService({
-        baseUrl: baseUrl,
-        apiKey: 'secret-token',
-      });
-      await service.createEmbeddings({ texts: ['x'], model: 'm', issuer: issuer });
-      const init: RequestInit = fetchSpy.calls.mostRecent().args[1] as RequestInit;
-      const headers: { [key: string]: string } = init.headers as { [key: string]: string };
-      expect(headers.Authorization).toBe('Bearer secret-token');
-    });
-
-    it('| should throw a DyFM_Error on a non-ok HTTP response', async () => {
-      fetchSpy.and.returnValue(Promise.resolve(errorResponse(500, 'boom')));
-      const service: DyNTS_LMStudio_Embedding_ControlService = new DyNTS_LMStudio_Embedding_ControlService({ baseUrl: baseUrl });
-
-      await expectAsync(
-        service.createEmbeddings({ texts: ['x'], model: 'm', issuer: issuer }),
-      ).toBeRejectedWith(jasmine.any(DyFM_Error));
-    });
-
-    it('| should throw a DyFM_Error if the vector count does not match the text count', async () => {
-      fetchSpy.and.returnValue(Promise.resolve(okResponse([[0.1]])));
-      const service: DyNTS_LMStudio_Embedding_ControlService = new DyNTS_LMStudio_Embedding_ControlService({ baseUrl: baseUrl });
-
-      await expectAsync(
-        service.createEmbeddings({ texts: ['a', 'b'], model: 'm', issuer: issuer }),
-      ).toBeRejectedWith(jasmine.any(DyFM_Error));
-    });
-
-    it('| should emit an embedding-batch cost-event with provider lm-studio', async () => {
-      fetchSpy.and.returnValue(Promise.resolve(okResponse([[0.1], [0.2]], { prompt_tokens: 7, total_tokens: 7 })));
-      const events: DyNTS_AI_CostEvent[] = [];
-      const service: DyNTS_LMStudio_Embedding_ControlService = new DyNTS_LMStudio_Embedding_ControlService({
-        baseUrl: baseUrl,
-        onCostEvent: (event: DyNTS_AI_CostEvent) => events.push(event),
-      });
-
-      await service.createEmbeddings({ texts: ['a', 'b'], model: 'm', issuer: issuer });
-
-      expect(events.length).toBe(1);
-      expect(events[0].callType).toBe('embedding-batch');
-      expect(events[0].provider).toBe('lm-studio');
-      expect(events[0].model).toBe('m');
-      expect(events[0].tokensUsed.input).toBe(7);
-      expect(events[0].issuer).toBe(issuer);
-    });
-
-    it('| should estimate tokens when the endpoint provides no usage', async () => {
-      fetchSpy.and.returnValue(Promise.resolve(okResponse([[0.1]])));
-      const events: DyNTS_AI_CostEvent[] = [];
-      const service: DyNTS_LMStudio_Embedding_ControlService = new DyNTS_LMStudio_Embedding_ControlService({
-        baseUrl: baseUrl,
-        onCostEvent: (event: DyNTS_AI_CostEvent) => events.push(event),
-      });
-
-      await service.createEmbeddings({ texts: ['abcdefgh'], model: 'm', issuer: issuer });
-
-      // 8 chars / 4 = 2 estimated tokens.
-      expect(events[0].tokensUsed.input).toBe(2);
-    });
-  });
-
-  describe('| createEmbedding (single)', () => {
-    it('| should emit an embedding-single cost-event and return a single vector', async () => {
-      fetchSpy.and.returnValue(Promise.resolve(okResponse([[0.9, 0.8, 0.7]], { prompt_tokens: 3, total_tokens: 3 })));
-      const events: DyNTS_AI_CostEvent[] = [];
-      const service: DyNTS_LMStudio_Embedding_ControlService = new DyNTS_LMStudio_Embedding_ControlService({
-        baseUrl: baseUrl,
-        onCostEvent: (event: DyNTS_AI_CostEvent) => events.push(event),
-      });
-
-      const result: number[] | CreateEmbeddingResponse = await service.createEmbedding({
-        text: 'one',
-        model: 'm',
-        issuer: issuer,
-      });
-
-      expect(result).toEqual([0.9, 0.8, 0.7]);
-      expect(events[0].callType).toBe('embedding-single');
-    });
-
-    it('| should throw a DyFM_Error if text is empty', async () => {
-      const service: DyNTS_LMStudio_Embedding_ControlService = new DyNTS_LMStudio_Embedding_ControlService({ baseUrl: baseUrl });
-      await expectAsync(
-        service.createEmbedding({ text: '', model: 'm', issuer: issuer }),
-      ).toBeRejectedWith(jasmine.any(DyFM_Error));
-    });
-  });
-
-  describe('| testConnection', () => {
-    it('| should return true when the probe succeeds', async () => {
-      fetchSpy.and.returnValue(Promise.resolve(okResponse([[0.1, 0.2]])));
-      const service: DyNTS_LMStudio_Embedding_ControlService = new DyNTS_LMStudio_Embedding_ControlService({ baseUrl: baseUrl });
-      expect(await service.testConnection(issuer)).toBe(true);
-    });
-
-    it('| should return false (never throw) when the probe fails', async () => {
-      fetchSpy.and.returnValue(Promise.reject(new Error('connection refused')));
-      const service: DyNTS_LMStudio_Embedding_ControlService = new DyNTS_LMStudio_Embedding_ControlService({ baseUrl: baseUrl });
-      expect(await service.testConnection(issuer)).toBe(false);
-    });
-  });
-
-  describe('| getEmbeddingInfo', () => {
-    it('| should report the LocalAI provider and the given model', () => {
-      const service: DyNTS_LMStudio_Embedding_ControlService = new DyNTS_LMStudio_Embedding_ControlService({ baseUrl: baseUrl });
-      expect(service.getEmbeddingInfo('my-model')).toEqual({ provider: DyFM_AI_Provider.LocalAI, model: 'my-model' });
-    });
-  });
-});
+import { CreateEmbeddingResponse } from 'openai/resources';
+import { DyFM_AI_Provider } from '@futdevpro/fsm-dynamo/ai';
+import { DyFM_Error } from '@futdevpro/fsm-dynamo';
+
+import { DyNTS_LMStudio_Embedding_ControlService } from './lmstudio-embedding.control-service';
+import { DyNTS_AI_CostEvent } from '../_models/interfaces/dynts-ai-cost-event.interface';
+
+describe('| DyNTS_LMStudio_Embedding_ControlService', () => {
+
+  const baseUrl: string = 'http://localhost:1234/v1';
+  const issuer: string = 'test-issuer';
+
+  /** Egy OpenAI-kompatibilis embeddings-válasz JSON-string-é csomagolva, fetch-mockhoz. */
+  function okResponse(embeddings: number[][], usage?: { prompt_tokens: number; total_tokens: number }): Response {
+    const body: string = JSON.stringify({
+      object: 'list',
+      data: embeddings.map((embedding, index) => ({ object: 'embedding', embedding: embedding, index: index })),
+      usage: usage,
+    });
+    return {
+      ok: true,
+      status: 200,
+      text: () => Promise.resolve(body),
+    } as Response;
+  }
+
+  /** Egy hibás (non-2xx) válasz fetch-mockhoz. */
+  function errorResponse(status: number, body: string): Response {
+    return {
+      ok: false,
+      status: status,
+      text: () => Promise.resolve(body),
+    } as Response;
+  }
+
+  let fetchSpy: jasmine.Spy;
+
+  beforeEach(() => {
+    fetchSpy = spyOn(globalThis, 'fetch');
+  });
+
+  describe('| constructor', () => {
+    it('| should throw a DyFM_Error if baseUrl is empty', () => {
+      expect(() => new DyNTS_LMStudio_Embedding_ControlService({ baseUrl: '' })).toThrowMatching(
+        (error: unknown) => error instanceof DyFM_Error,
+      );
+    });
+
+    it('| should set provider to LocalAI', () => {
+      const service: DyNTS_LMStudio_Embedding_ControlService = new DyNTS_LMStudio_Embedding_ControlService({ baseUrl: baseUrl });
+      expect(service.aiProvider).toBe(DyFM_AI_Provider.LocalAI);
+      expect(service.capabilities.embeddings).toBe(true);
+    });
+  });
+
+  describe('| createEmbeddings', () => {
+    it('| should POST to ${baseUrl}/embeddings and return vectors in order', async () => {
+      fetchSpy.and.returnValue(Promise.resolve(okResponse([[0.1, 0.2], [0.3, 0.4]])));
+
+      const service: DyNTS_LMStudio_Embedding_ControlService = new DyNTS_LMStudio_Embedding_ControlService({ baseUrl: baseUrl });
+      const result: number[][] | CreateEmbeddingResponse = await service.createEmbeddings({
+        texts: ['hello', 'world'],
+        model: 'nomic-embed-text-v1.5',
+        issuer: issuer,
+      });
+
+      expect(result).toEqual([[0.1, 0.2], [0.3, 0.4]]);
+      expect(fetchSpy).toHaveBeenCalledTimes(1);
+      const url: string = fetchSpy.calls.mostRecent().args[0] as string;
+      expect(url).toBe('http://localhost:1234/v1/embeddings');
+      const init: RequestInit = fetchSpy.calls.mostRecent().args[1] as RequestInit;
+      expect(init.method).toBe('POST');
+      const sentBody: { model: string; input: string[] } = JSON.parse(init.body as string);
+      expect(sentBody.model).toBe('nomic-embed-text-v1.5');
+      expect(sentBody.input).toEqual(['hello', 'world']);
+    });
+
+    it('| should trim trailing slashes from baseUrl', async () => {
+      fetchSpy.and.returnValue(Promise.resolve(okResponse([[0.5]])));
+      const service: DyNTS_LMStudio_Embedding_ControlService = new DyNTS_LMStudio_Embedding_ControlService({
+        baseUrl: 'http://localhost:1234/v1///',
+      });
+      await service.createEmbeddings({ texts: ['x'], model: 'm', issuer: issuer });
+      expect(fetchSpy.calls.mostRecent().args[0]).toBe('http://localhost:1234/v1/embeddings');
+    });
+
+    it('| should add a Bearer header when apiKey is provided', async () => {
+      fetchSpy.and.returnValue(Promise.resolve(okResponse([[0.5]])));
+      const service: DyNTS_LMStudio_Embedding_ControlService = new DyNTS_LMStudio_Embedding_ControlService({
+        baseUrl: baseUrl,
+        apiKey: 'secret-token',
+      });
+      await service.createEmbeddings({ texts: ['x'], model: 'm', issuer: issuer });
+      const init: RequestInit = fetchSpy.calls.mostRecent().args[1] as RequestInit;
+      const headers: { [key: string]: string } = init.headers as { [key: string]: string };
+      expect(headers.Authorization).toBe('Bearer secret-token');
+    });
+
+    it('| should throw a DyFM_Error on a non-ok HTTP response', async () => {
+      fetchSpy.and.returnValue(Promise.resolve(errorResponse(500, 'boom')));
+      const service: DyNTS_LMStudio_Embedding_ControlService = new DyNTS_LMStudio_Embedding_ControlService({ baseUrl: baseUrl });
+
+      await expectAsync(
+        service.createEmbeddings({ texts: ['x'], model: 'm', issuer: issuer }),
+      ).toBeRejectedWith(jasmine.any(DyFM_Error));
+    });
+
+    it('| should throw a DyFM_Error if the vector count does not match the text count', async () => {
+      fetchSpy.and.returnValue(Promise.resolve(okResponse([[0.1]])));
+      const service: DyNTS_LMStudio_Embedding_ControlService = new DyNTS_LMStudio_Embedding_ControlService({ baseUrl: baseUrl });
+
+      await expectAsync(
+        service.createEmbeddings({ texts: ['a', 'b'], model: 'm', issuer: issuer }),
+      ).toBeRejectedWith(jasmine.any(DyFM_Error));
+    });
+
+    it('| should emit an embedding-batch cost-event with provider lm-studio', async () => {
+      fetchSpy.and.returnValue(Promise.resolve(okResponse([[0.1], [0.2]], { prompt_tokens: 7, total_tokens: 7 })));
+      const events: DyNTS_AI_CostEvent[] = [];
+      const service: DyNTS_LMStudio_Embedding_ControlService = new DyNTS_LMStudio_Embedding_ControlService({
+        baseUrl: baseUrl,
+        onCostEvent: (event: DyNTS_AI_CostEvent) => events.push(event),
+      });
+
+      await service.createEmbeddings({ texts: ['a', 'b'], model: 'm', issuer: issuer });
+
+      expect(events.length).toBe(1);
+      expect(events[0].callType).toBe('embedding-batch');
+      expect(events[0].provider).toBe('lm-studio');
+      expect(events[0].model).toBe('m');
+      expect(events[0].tokensUsed.input).toBe(7);
+      expect(events[0].issuer).toBe(issuer);
+    });
+
+    it('| should estimate tokens when the endpoint provides no usage', async () => {
+      fetchSpy.and.returnValue(Promise.resolve(okResponse([[0.1]])));
+      const events: DyNTS_AI_CostEvent[] = [];
+      const service: DyNTS_LMStudio_Embedding_ControlService = new DyNTS_LMStudio_Embedding_ControlService({
+        baseUrl: baseUrl,
+        onCostEvent: (event: DyNTS_AI_CostEvent) => events.push(event),
+      });
+
+      await service.createEmbeddings({ texts: ['abcdefgh'], model: 'm', issuer: issuer });
+
+      // 8 chars / 4 = 2 estimated tokens.
+      expect(events[0].tokensUsed.input).toBe(2);
+    });
+  });
+
+  describe('| createEmbedding (single)', () => {
+    it('| should emit an embedding-single cost-event and return a single vector', async () => {
+      fetchSpy.and.returnValue(Promise.resolve(okResponse([[0.9, 0.8, 0.7]], { prompt_tokens: 3, total_tokens: 3 })));
+      const events: DyNTS_AI_CostEvent[] = [];
+      const service: DyNTS_LMStudio_Embedding_ControlService = new DyNTS_LMStudio_Embedding_ControlService({
+        baseUrl: baseUrl,
+        onCostEvent: (event: DyNTS_AI_CostEvent) => events.push(event),
+      });
+
+      const result: number[] | CreateEmbeddingResponse = await service.createEmbedding({
+        text: 'one',
+        model: 'm',
+        issuer: issuer,
+      });
+
+      expect(result).toEqual([0.9, 0.8, 0.7]);
+      expect(events[0].callType).toBe('embedding-single');
+    });
+
+    it('| should throw a DyFM_Error if text is empty', async () => {
+      const service: DyNTS_LMStudio_Embedding_ControlService = new DyNTS_LMStudio_Embedding_ControlService({ baseUrl: baseUrl });
+      await expectAsync(
+        service.createEmbedding({ text: '', model: 'm', issuer: issuer }),
+      ).toBeRejectedWith(jasmine.any(DyFM_Error));
+    });
+  });
+
+  describe('| testConnection', () => {
+    it('| should return true when the probe succeeds', async () => {
+      fetchSpy.and.returnValue(Promise.resolve(okResponse([[0.1, 0.2]])));
+      const service: DyNTS_LMStudio_Embedding_ControlService = new DyNTS_LMStudio_Embedding_ControlService({ baseUrl: baseUrl });
+      expect(await service.testConnection(issuer)).toBe(true);
+    });
+
+    it('| should return false (never throw) when the probe fails', async () => {
+      fetchSpy.and.returnValue(Promise.reject(new Error('connection refused')));
+      const service: DyNTS_LMStudio_Embedding_ControlService = new DyNTS_LMStudio_Embedding_ControlService({ baseUrl: baseUrl });
+      expect(await service.testConnection(issuer)).toBe(false);
+    });
+  });
+
+  describe('| getEmbeddingInfo', () => {
+    it('| should report the LocalAI provider and the given model', () => {
+      const service: DyNTS_LMStudio_Embedding_ControlService = new DyNTS_LMStudio_Embedding_ControlService({ baseUrl: baseUrl });
+      expect(service.getEmbeddingInfo('my-model')).toEqual({ provider: DyFM_AI_Provider.LocalAI, model: 'my-model' });
+    });
+  });
+});