feat: pdf parser changed

README.md

@@ -35,7 +35,7 @@
 - [**ai**](https://sdk.vercel.ai/docs/introduction) - AI framework for TypeScript.
 - [**Vercel**](https://vercel.com/) - Deploy your app anywhere.
 - [**Vercel Analytics**](https://vercel.com/analytics) - Analytics for Vercel.
-- [**pdf2json**](https://github.com/modesty/pdf2json) - Convert PDF to JSON.
+- [**pdf-parse**](https://gitlab.com/autokent/pdf-parse) - Convert PDF to Text.
 
 <!-- Help me to describe features of this project -->
 

next.config.mjs

@@ -1,7 +1,11 @@
 import withBundleAnalyzer from '@next/bundle-analyzer';
 
 /** @type {import('next').NextConfig} */
-const nextConfig = {};
+const nextConfig = {
+  experimental: {
+    serverComponentsExternalPackages: ["pdf-parse"],
+  },
+};
 
 const bundleAnalyzer = withBundleAnalyzer({
   enabled: process.env.ANALYZE === 'true',

package.json

@@ -25,7 +25,7 @@
     "cmdk": "^1.0.0",
     "next": "14.2.3",
     "next-themes": "^0.3.0",
-    "pdf2json": "^3.1.3",
+    "pdf-parse": "^1.1.1",
     "react": "^18",
     "react-dom": "^18",
     "react-dropzone": "^14.2.3",
@@ -37,6 +37,7 @@
   "devDependencies": {
     "@next/bundle-analyzer": "^14.2.8",
     "@types/node": "^20",
+    "@types/pdf-parse": "^1.1.4",
     "@types/react": "^18",
     "@types/react-dom": "^18",
     "eslint": "^8",

src/app/api/translate-document/route.ts

@@ -1,21 +1,14 @@
 import { z } from "zod";
 import { createOpenAI } from "@ai-sdk/openai";
 import { streamText } from "ai";
-import PDFParser, { Output } from "pdf2json";
+import pdf from "pdf-parse";
+
+import { validPrefixes } from "@/lib/constants";
 
-import { dataURLtoFile } from "@/lib/utils";
 
 // Allow streaming responses up to 30 seconds
 export const maxDuration = 30;
 
-const pdfParser = new PDFParser();
-
-const waitForPdfData = (): Promise<Output> => {
-  return new Promise((resolve, reject) => {
-    pdfParser.on("pdfParser_dataError", reject);
-    pdfParser.on("pdfParser_dataReady", resolve);
-  });
-};
 
 const RequestSchema = z.object({
   fromLanguage: z.string(),
@@ -43,18 +36,40 @@ export async function POST(req: Request) {
   // Controller for the translation
   const { fromLanguage, toLanguage, document, apiKey } = data;
 
-  const file = dataURLtoFile(document);
-  const fileArrayBuffer = await file.arrayBuffer();
-  pdfParser.parseBuffer(fileArrayBuffer as Buffer, 9);
+  const matchedPrefix = validPrefixes.find((prefix) =>
+    document.startsWith(prefix)
+  );
+
+  if (!matchedPrefix) {
+    return Response.json(
+      {
+        success: false,
+        message: "The Data URI format is invalid",
+      },
+      { status: 401 }
+    );
+  }
+
+  const base64Data = document.slice(matchedPrefix.length);
 
-  const pdfData = await waitForPdfData();
+  const pdfBuffer = Buffer.from(base64Data, "base64");
 
-  const textToTranslate = pdfData.Pages.map((page) => {
-    return page.Texts.map((text) => {
-      return text.R.map(({ T }) => decodeURIComponent(T).trim()).join("");
-    }).join("");
-  });
+  let textToTranslate = "";
+
+  try {
+    const data = await pdf(pdfBuffer);
+    textToTranslate = data.text;
+  } catch (error) {    
+    return Response.json(
+      {
+        success: false,
+        message: "Error parsing the PDF document",
+      },
+      { status: 500 }
+    );
+  }
 
+  // Service for the translation
   const openai = createOpenAI({
     compatibility: "strict",
     apiKey,

src/app/layout.tsx

@@ -27,7 +27,7 @@ export const metadata: Metadata = {
   description:
     "Translate your text, images, and documents in multiple languages",
   keywords:
-    "translate, translator, translation, text, image, document, pdf, ai, openai, gpt, chatgpt, vercel, nextjs, tailwindcss, shadcn, radix, react, typescript, bun, pdf2json",
+    "translate, translator, translation, text, image, document, pdf, ai, openai, gpt, chatgpt, vercel, nextjs, tailwindcss, shadcn, radix, react, typescript, bun, pdf-parse",
   metadataBase: new URL("https://talk-translate.vercel.app"),
 };
 

src/lib/constants.ts

@@ -82,3 +82,8 @@ export const enum SearchParams {
   OPTION = "op",
   TEXT = "text",
 }
+
+export const validPrefixes = [
+  "data:application/pdf;base64,",
+  "data:@file/pdf;base64,",
+];