/**
* Embeddings Domain Module
*
* This module contains pure business logic for AI/ML operations,
* including embedding generation, vector operations, and batch processing.
*/
// Import types from shared types module
import type { TextChunk } from '../../types/index.js';
// Core domain services
export interface EmbeddingOperations {
generateEmbeddings(chunks: TextChunk[]): Promise<EmbeddingVector[]>;
generateSingleEmbedding(text: string): Promise<EmbeddingVector>;
calculateSimilarity(vector1: EmbeddingVector, vector2: EmbeddingVector): number;
}
export interface BatchEmbeddingOperations {
processBatch(chunks: TextChunk[], batchSize?: number): Promise<EmbeddingResult[]>;
estimateProcessingTime(chunkCount: number): number;
}
// Domain types
export interface EmbeddingVector {
vector: number[];
dimensions: number;
model: string;
createdAt: string;
chunkId?: string;
metadata?: EmbeddingMetadata;
}
export interface EmbeddingResult {
chunk: TextChunk;
embedding: EmbeddingVector;
processingTime: number;
success: boolean;
error?: string;
}
export interface EmbeddingMetadata {
generatedAt: string;
modelVersion: string;
tokensUsed: number;
confidence?: number;
}
export interface EmbeddingModel {
name: string;
dimensions: number;
maxTokens: number;
endpoint?: string;
}
// Default implementation
export class DefaultEmbeddingOperations implements EmbeddingOperations, BatchEmbeddingOperations {
private readonly model: EmbeddingModel = {
name: 'default',
dimensions: 384,
maxTokens: 512
};
async generateSingleEmbedding(text: string): Promise<EmbeddingVector> {
const startTime = Date.now();
// Handle empty text
if (!text || text.trim().length === 0) {
return {
vector: new Array(this.model.dimensions).fill(0),
dimensions: this.model.dimensions,
model: this.model.name,
createdAt: new Date().toISOString(), // <-- Fix: add createdAt
metadata: {
generatedAt: new Date().toISOString(),
modelVersion: '1.0',
tokensUsed: 0
}
};
}
// Simulate embedding generation
const vector = new Array(this.model.dimensions).fill(0).map(() => Math.random());
const processingTime = Date.now() - startTime;
return {
vector,
dimensions: this.model.dimensions,
model: this.model.name,
createdAt: new Date().toISOString(),
metadata: {
generatedAt: new Date().toISOString(),
modelVersion: '1.0',
tokensUsed: Math.ceil(text.length / 4), // Rough estimate
confidence: 0.95
}
};
}
async generateEmbeddings(chunks: TextChunk[]): Promise<EmbeddingVector[]> {
return Promise.all(chunks.map(chunk => this.generateSingleEmbedding(chunk.content)));
}
calculateSimilarity(vector1: EmbeddingVector, vector2: EmbeddingVector): number {
// Simple cosine similarity
const dotProduct = vector1.vector.reduce((sum, val, i) => {
const v2 = vector2.vector[i] ?? 0;
return sum + val * v2;
}, 0);
const magnitude1 = Math.sqrt(vector1.vector.reduce((sum, val) => sum + val * val, 0));
const magnitude2 = Math.sqrt(vector2.vector.reduce((sum, val) => sum + val * val, 0));
return dotProduct / (magnitude1 * magnitude2);
}
async processBatch(chunks: TextChunk[], batchSize: number = 32): Promise<EmbeddingResult[]> {
const results: EmbeddingResult[] = [];
for (let i = 0; i < chunks.length; i += batchSize) {
const batch = chunks.slice(i, i + batchSize);
const batchResults = await Promise.all(
batch.map(async (chunk) => {
const chunkStartTime = Date.now();
try {
// Ensure empty/whitespace chunk content gets tokensUsed: 0
if (!chunk.content || chunk.content.trim().length === 0) {
return {
chunk,
embedding: {
vector: new Array(this.model.dimensions).fill(0),
dimensions: this.model.dimensions,
model: this.model.name,
createdAt: new Date().toISOString(), // <-- Fix: add createdAt
metadata: {
generatedAt: new Date().toISOString(),
modelVersion: '1.0',
tokensUsed: 0
}
},
processingTime: Date.now() - chunkStartTime,
success: true
};
}
const embedding = await this.generateSingleEmbedding(chunk.content);
return {
chunk,
embedding,
processingTime: Date.now() - chunkStartTime,
success: true
};
} catch (error) {
return {
chunk,
embedding: {
vector: new Array(this.model.dimensions).fill(0),
dimensions: this.model.dimensions,
model: this.model.name,
createdAt: new Date().toISOString(), // <-- Fix: add createdAt
metadata: {
generatedAt: new Date().toISOString(),
modelVersion: '1.0',
tokensUsed: 0
}
},
processingTime: Date.now() - chunkStartTime,
success: false,
error: error instanceof Error ? error.message : 'Unknown error'
};
}
})
);
results.push(...batchResults);
}
return results;
}
estimateProcessingTime(chunkCount: number): number {
// Rough estimate: 10ms per chunk
return chunkCount * 10;
}
}
// Export the TextChunk type for other modules
export type { TextChunk };
