import { afterEach, beforeEach, describe, expect, it, Mock, vi } from 'vitest';
import { fetchWithCache } from '../../../src/cache';
import { GoogleImageProvider } from '../../../src/providers/google/image';
import { mockProcessEnv } from '../../util/utils';

vi.mock('../../../src/cache', async (importOriginal) => {
  return {
    ...(await importOriginal()),
    fetchWithCache: vi.fn(),
  };
});

vi.mock('../../../src/providers/google/util', async (importOriginal) => {
  return {
    ...(await importOriginal()),
    getGoogleClient: vi.fn(),
    loadCredentials: vi.fn(),
    resolveProjectId: vi.fn(),
    createAuthCacheDiscriminator: vi.fn().mockReturnValue(''),
  };
});

describe('GoogleImageProvider', async () => {
  const mockFetchWithCache = fetchWithCache as Mock;
  const utilMocks = await import('../../../src/providers/google/util');
  const mockGetGoogleClient = utilMocks.getGoogleClient as Mock;
  const mockLoadCredentials = utilMocks.loadCredentials as Mock;
  const mockResolveProjectId = utilMocks.resolveProjectId as Mock;

  beforeEach(() => {
    vi.clearAllMocks();
    mockProcessEnv({ GOOGLE_API_KEY: 'test-api-key' });
    mockProcessEnv({ GOOGLE_GENERATIVE_AI_API_KEY: undefined });
    mockProcessEnv({ GEMINI_API_KEY: undefined });
    mockProcessEnv({ GOOGLE_PROJECT_ID: undefined });
    mockProcessEnv({ GOOGLE_CLOUD_PROJECT: undefined });

    // Set up default mock behaviors
    mockLoadCredentials.mockImplementation(function (creds) {
      return creds;
    });
    mockResolveProjectId.mockResolvedValue('test-project');
  });

  afterEach(() => {
    mockProcessEnv({ GOOGLE_API_KEY: undefined });
    mockProcessEnv({ GOOGLE_PROJECT_ID: undefined });
    mockProcessEnv({ GOOGLE_CLOUD_PROJECT: undefined });
    mockProcessEnv({ GOOGLE_GENERATIVE_AI_API_KEY: undefined });
    mockProcessEnv({ GEMINI_API_KEY: undefined });
  });

  it('should construct with model name', () => {
    const provider = new GoogleImageProvider('imagen-3.0-generate-001');
    expect(provider.id()).toBe('google:image:imagen-3.0-generate-001');
    expect(provider.toString()).toBe('[Google Image Generation Provider imagen-3.0-generate-001]');
  });

  it('should use Google AI Studio when project ID is missing but API key is available', async () => {
    mockProcessEnv({ GOOGLE_PROJECT_ID: undefined });
    const provider = new GoogleImageProvider('imagen-3.0-generate-001');

    mockFetchWithCache.mockResolvedValueOnce({
      status: 200,
      data: {
        predictions: [
          {
            bytesBase64Encoded: 'base64data',
            mimeType: 'image/png',
          },
        ],
      },
      cached: false,
    });

    const result = await provider.callApi('Test prompt');

    expect(mockFetchWithCache).toHaveBeenCalledWith(
      'https://generativelanguage.googleapis.com/v1beta/models/imagen-3.0-generate-001:predict',
      expect.objectContaining({
        headers: expect.objectContaining({
          'x-goog-api-key': 'test-api-key',
        }),
      }),
      expect.any(Number),
      'json',
    );
    expect(result.output).toContain('data:image/png;base64,base64data');
    expect(result.images).toEqual([
      { data: 'data:image/png;base64,base64data', mimeType: 'image/png' },
    ]);
  });

  it('should return error when both project ID and API key are missing', async () => {
    mockProcessEnv({ GOOGLE_PROJECT_ID: undefined });
    mockProcessEnv({ GOOGLE_API_KEY: undefined });
    mockProcessEnv({ GOOGLE_GENERATIVE_AI_API_KEY: undefined });
    mockProcessEnv({ GEMINI_API_KEY: undefined });
    const provider = new GoogleImageProvider('imagen-3.0-generate-001');

    const result = await provider.callApi('Test prompt');

    expect(result.error).toContain('Imagen models require either:');
    expect(result.error).toContain('Google AI Studio');
    expect(result.error).toContain('Vertex AI');
  });

  describe('Vertex AI', () => {
    beforeEach(() => {
      mockProcessEnv({ GOOGLE_PROJECT_ID: 'test-project' });
    });

    it('should use OAuth authentication for Vertex AI', async () => {
      const provider = new GoogleImageProvider('imagen-3.0-generate-001', {
        config: {
          projectId: 'test-project',
        },
      });

      const mockClient = {
        request: vi.fn().mockResolvedValue({
          data: {
            predictions: [
              {
                image: {
                  mimeType: 'image/png',
                  bytesBase64Encoded: 'base64data',
                },
              },
            ],
          },
        }),
      };

      mockGetGoogleClient.mockResolvedValue({
        client: mockClient,
        projectId: 'test-project',
      });

      const result = await provider.callApi('Test prompt');

      expect(mockGetGoogleClient).toHaveBeenCalled();

      expect(mockClient.request).toHaveBeenCalledWith({
        url: expect.stringContaining('aiplatform.googleapis.com'),
        method: 'POST',
        headers: expect.objectContaining({
          'Content-Type': 'application/json',
        }),
        data: expect.objectContaining({
          instances: [{ prompt: 'Test prompt' }],
        }),
        timeout: 300000,
      });

      expect(result.output).toContain('data:image/png;base64,base64data');
      expect(result.images).toEqual([
        { data: 'data:image/png;base64,base64data', mimeType: 'image/png' },
      ]);
    });

    it('should handle OAuth errors', async () => {
      const provider = new GoogleImageProvider('imagen-3.0-generate-001', {
        config: {
          projectId: 'test-project',
        },
      });

      mockGetGoogleClient.mockRejectedValue(new Error('Google auth library not found'));

      const result = await provider.callApi('Test prompt');

      expect(result.error).toContain('Failed to call Vertex AI');
      expect(result.error).toContain('Google auth library not found');
    });
  });

  it('should support different model name formats', () => {
    const provider1 = new GoogleImageProvider('imagen-3.0-generate-001');
    expect(provider1.id()).toBe('google:image:imagen-3.0-generate-001');

    // When model name already includes 'imagen', it should be preserved
    const provider2 = new GoogleImageProvider('gemini/imagen-3.0-generate-001');
    expect(provider2.id()).toBe('google:image:gemini/imagen-3.0-generate-001');

    // When model name doesn't include 'imagen', ID still includes full model name
    const provider3 = new GoogleImageProvider('3.0-generate-001');
    expect(provider3.id()).toBe('google:image:3.0-generate-001');
  });

  it('should handle model path prefixing correctly', async () => {
    const testCases = [
      { input: 'imagen-3.0-generate-001', expected: 'imagen-3.0-generate-001' },
      { input: '3.0-generate-001', expected: 'imagen-3.0-generate-001' },
      { input: 'custom-imagen-model', expected: 'imagen-custom-imagen-model' }, // Should be prefixed
      { input: 'imagen-4.0-ultra', expected: 'imagen-4.0-ultra' },
    ];

    for (const { input, expected } of testCases) {
      const provider = new GoogleImageProvider(input, {
        config: {
          projectId: 'test-project', // Ensure Vertex AI is used
        },
      });

      // Mock the API response to extract the model path from the request
      const mockClient = {
        request: vi.fn().mockResolvedValue({
          data: {
            predictions: [{ bytesBase64Encoded: 'test', mimeType: 'image/png' }],
          },
        }),
      };

      mockGetGoogleClient.mockResolvedValue({
        client: mockClient,
        projectId: 'test-project',
      });

      await provider.callApi('test prompt');

      expect(mockClient.request).toHaveBeenCalledWith(
        expect.objectContaining({
          url: expect.stringContaining(`/models/${expected}:predict`),
        }),
      );
    }
  });

  it('should return correct cost for different models', async () => {
    // Test costs through actual API responses
    const testCases = [
      // Imagen 4 GA names
      { model: 'imagen-4.0-ultra-generate-001', expectedCost: 0.06 },
      { model: 'imagen-4.0-generate-001', expectedCost: 0.04 },
      { model: 'imagen-4.0-fast-generate-001', expectedCost: 0.02 },
      // Imagen 4 preview aliases
      { model: 'imagen-4.0-ultra-generate-preview-06-06', expectedCost: 0.06 },
      { model: 'imagen-4.0-generate-preview-06-06', expectedCost: 0.04 },
      { model: 'imagen-4.0-fast-generate-preview-06-06', expectedCost: 0.02 },
      { model: '3.0-generate-001', expectedCost: 0.04 }, // Without prefix
      { model: 'unknown-model', expectedCost: 0.04 }, // Default cost
    ];

    for (const { model, expectedCost } of testCases) {
      const provider = new GoogleImageProvider(model);
      mockFetchWithCache.mockResolvedValueOnce({
        status: 200,
        data: {
          predictions: [{ bytesBase64Encoded: 'base64data', mimeType: 'image/png' }],
        },
        cached: false,
      });

      const result = await provider.callApi('Test prompt');
      expect(result.cost).toBe(expectedCost);
    }
  });

  it('should not report cost for cached responses', async () => {
    const provider = new GoogleImageProvider('imagen-4.0-fast-generate-001');
    mockFetchWithCache.mockResolvedValueOnce({
      status: 200,
      data: {
        predictions: [{ bytesBase64Encoded: 'base64data', mimeType: 'image/png' }],
      },
      cached: true,
    });

    const result = await provider.callApi('Test prompt');

    expect(result.cached).toBe(true);
    expect(result.cost).toBeUndefined();
  });

  describe('Google AI Studio', () => {
    beforeEach(() => {
      mockProcessEnv({ GOOGLE_PROJECT_ID: undefined });
      mockProcessEnv({ GOOGLE_API_KEY: 'test-api-key' });
    });

    it('should make correct API request to Google AI Studio', async () => {
      const provider = new GoogleImageProvider('imagen-4.0-generate-preview-06-06', {
        config: {
          n: 2,
          aspectRatio: '16:9',
          safetyFilterLevel: 'block_few',
          personGeneration: 'allow_adult',
        },
      });

      mockFetchWithCache.mockResolvedValueOnce({
        status: 200,
        data: {
          predictions: [
            {
              bytesBase64Encoded: 'base64data1',
              mimeType: 'image/png',
            },
            {
              bytesBase64Encoded: 'base64data2',
              mimeType: 'image/png',
            },
          ],
        },
        cached: false,
      });

      const result = await provider.callApi('Test prompt');

      expect(mockFetchWithCache).toHaveBeenCalledWith(
        'https://generativelanguage.googleapis.com/v1beta/models/imagen-4.0-generate-preview-06-06:predict',
        expect.objectContaining({
          method: 'POST',
          headers: expect.objectContaining({
            'Content-Type': 'application/json',
            'x-goog-api-key': 'test-api-key',
          }),
          body: JSON.stringify({
            instances: [
              {
                prompt: 'Test prompt',
              },
            ],
            parameters: {
              sampleCount: 2,
              aspectRatio: '16:9',
              personGeneration: 'allow_adult',
              safetySetting: 'block_low_and_above',
            },
          }),
        }),
        expect.any(Number),
        'json',
      );

      // First image as output for blob externalization
      expect(result.output).toBe('data:image/png;base64,base64data1');
      // All images in structured field
      expect(result.images).toEqual([
        { data: 'data:image/png;base64,base64data1', mimeType: 'image/png' },
        { data: 'data:image/png;base64,base64data2', mimeType: 'image/png' },
      ]);
      expect(result.cached).toBe(false);
      expect(result.cost).toBe(0.08); // 2 images * 0.04
    });

    it('should handle API errors from Google AI Studio', async () => {
      const provider = new GoogleImageProvider('imagen-3.0-generate-001');

      mockFetchWithCache.mockResolvedValueOnce({
        status: 400,
        data: {
          error: {
            message: 'Invalid request',
          },
        },
      });

      const result = await provider.callApi('Test prompt');

      expect(result.error).toBe('Invalid request');
    });

    it('should handle missing API key for Google AI Studio', async () => {
      mockProcessEnv({ GOOGLE_API_KEY: undefined });
      mockProcessEnv({ GOOGLE_GENERATIVE_AI_API_KEY: undefined });
      mockProcessEnv({ GEMINI_API_KEY: undefined });
      const provider = new GoogleImageProvider('imagen-3.0-generate-001');

      const result = await provider.callApi('Test prompt');

      expect(result.error).toContain('Imagen models require either:');
    });

    it('should support different API key environment variables', async () => {
      mockProcessEnv({ GOOGLE_API_KEY: undefined });
      mockProcessEnv({ GEMINI_API_KEY: 'gemini-key' });

      const provider = new GoogleImageProvider('imagen-3.0-generate-001');

      mockFetchWithCache.mockResolvedValueOnce({
        status: 200,
        data: {
          predictions: [
            {
              bytesBase64Encoded: 'base64data',
              mimeType: 'image/png',
            },
          ],
        },
        cached: false,
      });

      const result = await provider.callApi('Test prompt');

      expect(mockFetchWithCache).toHaveBeenCalledWith(
        expect.any(String),
        expect.objectContaining({
          headers: expect.objectContaining({
            'x-goog-api-key': 'gemini-key',
          }),
        }),
        expect.any(Number),
        'json',
      );
      expect(result.output).toContain('data:image/png;base64,base64data');
    });
  });
});