import fs from 'fs';
import path from 'path';
import { OpenAIEmbeddings } from 'langchain/embeddings/openai';
import { MemoryVectorStore } from 'langchain/vectorstores/memory';
import { Document } from 'langchain/document';
export async function loadDocumentation(OPENAI_API_KEY: string): Promise<MemoryVectorStore> {
console.log('Loading and vectorizing documentation...');
const __dirname = path.dirname(new URL(import.meta.url).pathname);
const docsPath = path.resolve(__dirname, '..', '..', 'docs');
if (!fs.existsSync(docsPath)) {
throw new Error(`Documentation directory not found at ${docsPath}`);
}
const docFiles = fs.readdirSync(docsPath);
const documents: Document[] = [];
for (const file of docFiles) {
const filePath = path.join(docsPath, file);
const content = fs.readFileSync(filePath, 'utf-8');
documents.push(
new Document({
pageContent: content,
metadata: {
source: file,
type: path.extname(file).toLowerCase(),
lastUpdated: fs.statSync(filePath).mtime,
},
})
);
}
const embeddings = new OpenAIEmbeddings({
openAIApiKey: OPENAI_API_KEY,
});
const vectorStore = await MemoryVectorStore.fromDocuments(documents, embeddings);
console.log(`Successfully loaded ${documents.length} documentation files.`);
return vectorStore;
}
