update the create embeddings end point to upload pdf files

api/controllers/embed.controller.ts

@@ -3,28 +3,28 @@ import { CreateDocumentEmbeddingHandler } from "../handlers/create-document-embe
 import { documentRequestSchema } from "../lib/validation-schemas";
 import { generateErrorResponse } from "../utils/utils";
 import { Result } from "../lib/result";
+import multer from "multer";
 export class EmbeddingController {
   path = "/embed";
   router = express.Router();
+  upload = multer();
   constructor() {
     this.initRoutes();
   }
 
   initRoutes() {
-    this.router.post(`${this.path}/documents`, this.createDocumentEmbeddings);
+    this.router.post(`${this.path}/documents`, this.upload.single("pdf"), this.createDocumentEmbeddings);
   }
 
-  async createDocumentEmbeddings(
-    req: express.Request,
-    res: express.Response,
-    next: express.NextFunction,
-  ) {
-    const embeddingHandler: CreateDocumentEmbeddingHandler =
-      new CreateDocumentEmbeddingHandler();
+  async createDocumentEmbeddings(req: express.Request, res: express.Response, next: express.NextFunction) {
+    if (!req.file) {
+      return res.json(Result.fail("No file uploaded", 400));
+    }
+    const file = req.file;
+    const { buffer } = file;
+    const embeddingHandler: CreateDocumentEmbeddingHandler = new CreateDocumentEmbeddingHandler(buffer);
     try {
-      const { title, documentType, domain } = documentRequestSchema.parse(
-        req.body,
-      );
+      const { title, documentType, domain } = documentRequestSchema.parse(req.body);
       const result = await embeddingHandler.handle({
         title,
         documentType,

api/handlers/create-document-embed.handler.ts

@@ -6,24 +6,17 @@ import { ICreateEmbeddingRequestDTO } from "../repositories/dtos/dtos";
 import { EmbeddingService } from "../services/embed.service";
 import { getValue } from "../utils";
 
-export class CreateDocumentEmbeddingHandler
-  implements IRequestHandler<ICreateEmbeddingRequestDTO, Result<boolean>>
-{
+export class CreateDocumentEmbeddingHandler implements IRequestHandler<ICreateEmbeddingRequestDTO, Result<boolean>> {
+  constructor(private readonly pdf: Buffer) {}
   private readonly apiKey: string = getValue("API_KEY");
-  embeddingService: EmbeddingService = new EmbeddingService(this.apiKey);
+
   async handle(request: ICreateEmbeddingRequestDTO): Promise<Result<boolean>> {
+    const embeddingService: EmbeddingService = new EmbeddingService(this.apiKey, this.pdf);
     try {
       const { title, documentType, domain } = request;
-      const result = await this.embeddingService.createDocumentsEmbeddings(
-        title,
-        documentType,
-        domain
-      );
+      const result = await embeddingService.createDocumentsEmbeddings(title, documentType, domain);
       if (!result) {
-        throw new HttpException(
-          HTTP_RESPONSE_CODE.BAD_REQUEST,
-          "An error occured, could not create embeddings"
-        );
+        throw new HttpException(HTTP_RESPONSE_CODE.BAD_REQUEST, "An error occured, could not create embeddings");
       }
       return result;
     } catch (error) {

api/interfaces/document-service.interface.ts

@@ -1,5 +1,5 @@
 export interface IDocumentService {
-  convertPDFToText(pdfFilePath: string): Promise<string>;
+  convertPDFToText(file: Buffer): Promise<string>;
   writeToFile(outputFilePath: string, text: string): void;
   breakTextIntoChunks(text: string, partSize: number): string[];
   adjustChunkToEndAtCharacter(chunk: string): string;

api/package.json

@@ -21,6 +21,7 @@
     "@google/generative-ai": "^0.1.3",
     "@prisma/client": "^5.11.0",
     "@types/express": "^4.17.21",
+    "@types/multer": "^1.4.11",
     "@types/pdf-parse": "^1.1.4",
     "@types/pg": "^8.11.0",
     "body-parser": "^1.20.2",
@@ -30,6 +31,7 @@
     "dotenv": "^16.4.1",
     "express": "^4.18.2",
     "langchain": "^0.1.9",
+    "multer": "^1.4.5-lts.1",
     "pdf-parse": "^1.1.1",
     "pg": "^8.11.3",
     "pgvector": "^0.1.7",

api/services/document.service.ts

@@ -3,10 +3,9 @@ import pdf from "pdf-parse";
 import { IDocumentService } from "../interfaces/document-service.interface";
 
 export class DocumentService implements IDocumentService {
-  async convertPDFToText(pdfFilePath: string): Promise<string> {
+  async convertPDFToText(file: Buffer): Promise<string> {
     try {
-      const dataBuffer = fs.readFileSync(pdfFilePath);
-      const data = await pdf(dataBuffer);
+      const data = await pdf(file);
       const text = this.formatText(data.text);
       return text;
     } catch (error) {

api/services/embed.service.ts

@@ -60,8 +60,10 @@ of information. For example, you could use this task type to embed articles, FAQ
 or product manuals to create a searchable knowledge base for customer support or information retrieval systems.*/
 
 export class EmbeddingService extends GenerativeAIService implements IEmbeddingService {
-  documentPath: string = getValue("PDF_ABSOLUTE_PATH");
-  constructor(apiKey: string) {
+  constructor(
+    apiKey: string,
+    private file: Buffer
+  ) {
     super(apiKey);
   }
   /**
@@ -194,10 +196,10 @@ export class EmbeddingService extends GenerativeAIService implements IEmbeddingS
   async createContentEmbeddings(): Promise<{ text: string; embeddings?: number[] }[]> {
     const documentService: IDocumentService = new DocumentService();
     let text: string;
-    if (!this.documentPath.length) {
+    if (!this.file.length) {
       throw new HttpException(HTTP_RESPONSE_CODE.BAD_REQUEST, "Could not read PDF file");
     }
-    text = await documentService.convertPDFToText(this.documentPath);
+    text = await documentService.convertPDFToText(this.file);
     const chunks: string[] = documentService.breakTextIntoChunks(text, 2000);
 
     const contentEmbed = chunks.map(