♻️ add fallback poppler mechanism to prevent default-encrypted…

src/imageOperations/invoiceSplitterExtractor/invoiceSplitterExtractor.ts

@@ -3,6 +3,7 @@ import { MindeeError, MindeeMimeTypeError } from "../../errors";
 import { InvoiceSplitterV1 } from "../../product";
 import { LocalInputSource } from "../../input";
 import { ExtractedInvoiceSplitterImage } from "./extractedInvoiceSplitterImage";
+import { loadPdfWithFallback } from "../../pdf/pdfOperation";
 
 async function splitPdf(pdfDoc: PDFDocument, invoicePageGroups: number[][]): Promise<ExtractedInvoiceSplitterImage[]> {
   if (invoicePageGroups.length === 0) {
@@ -35,7 +36,7 @@ async function getPdfDoc(inputFile: LocalInputSource): Promise<PDFDocument> {
     throw new MindeeMimeTypeError("Invoice Splitter is only compatible with pdf documents.");
   }
 
-  const pdfDoc = await PDFDocument.load(inputFile.fileObject);
+  const pdfDoc = await loadPdfWithFallback(inputFile.fileObject);
   if (pdfDoc.getPageCount() < 2) {
     throw new MindeeError("Invoice Splitter is only compatible with multi-page-pdf documents.");
   }

src/imageOperations/multiReceiptsExtractor/multiReceiptsExtractor.ts

@@ -6,6 +6,7 @@ import { ExtractedMultiReceiptImage } from "./extractedMultiReceiptImage";
 import { LocalInputSource } from "../../input";
 import { extractFromPage } from "../common";
 import { PositionField } from "../../parsing/standard";
+import { loadPdfWithFallback } from "../../pdf/pdfOperation";
 
 /**
  * Given a page and a set of coordinates, extracts & assigns individual receipts to an ExtractedMultiReceiptImage
@@ -37,7 +38,7 @@ async function loadPdfDoc(inputFile: LocalInputSource) {
       '" Currently supported types are .png, .jpg and .pdf'
     );
   } else if (inputFile.isPdf()) {
-    pdfDoc = await PDFDocument.load(inputFile.fileObject);
+    pdfDoc = await loadPdfWithFallback(inputFile.fileObject);
   } else {
     pdfDoc = await PDFDocument.create();
     let image: PDFImage;

src/pdf/pdfCompressor.ts

@@ -5,6 +5,7 @@ import * as fs from "node:fs";
 import { Poppler } from "node-poppler";
 import { PDFDocument, PDFFont, PDFPage, rgb, StandardFonts } from "pdf-lib";
 import { compressImage } from "../imageOperations";
+import { loadPdfWithFallback } from "./pdfOperation";
 
 /**
  * Compresses each page of a provided PDF buffer.
@@ -128,7 +129,7 @@ async function compressPagesWithQuality(
   disableSourceText: boolean,
   extractedText: ExtractedPdfInfo | null
 ): Promise<Buffer[]> {
-  const pdfDoc = await PDFDocument.load(pdfData);
+  const pdfDoc = await loadPdfWithFallback(pdfData);
   const compressedPages: Buffer[] = [];
 
   for (let i = 0; i < extractedPdfInfo.pages.length; i++) {

src/pdf/pdfOperation.ts

@@ -3,12 +3,44 @@ import { PDFDocument } from "pdf-lib";
 import { PageOptions, PageOptionsOperation } from "../input";
 import { MindeeError } from "../errors";
 import { logger } from "../logger";
+import { Poppler } from "node-poppler";
+import { readFile } from "fs/promises";
+import tmp from "tmp";
+import fs from "node:fs";
 
 export interface SplitPdf {
   file: Buffer;
   totalPagesRemoved: number;
 }
 
+/**
+ * Attempts to load the file using pdf-lib, and falls back to node-poppler if unable to.
+ * @param file File buffer to be opened.
+ */
+export async function loadPdfWithFallback(file: string | Buffer) {
+  const document = await PDFDocument.load(file, { ignoreEncryption: true });
+  if (!document.isEncrypted) {
+    return document;
+  }
+  const poppler = new Poppler();
+
+  const tmpPdfOutput = tmp.fileSync();
+  const tmpPdfOutputPath = tmpPdfOutput.name;
+
+  try {
+    await poppler.pdfToCairo(file, tmpPdfOutputPath, {
+      pdfFile: true,
+      antialias: "default",
+    });
+
+    const convertedPdf = await readFile(tmpPdfOutputPath);
+    return await PDFDocument.load(convertedPdf, { ignoreEncryption: true });
+  } finally {
+    await fs.promises.unlink(tmpPdfOutputPath);
+  }
+}
+
+
 /**
  * Cut pages from a pdf file. If pages index are out of bound, it will throw an error.
  * @param file
@@ -19,9 +51,7 @@ export async function extractPages(
   file: Buffer,
   pageOptions: PageOptions
 ): Promise<SplitPdf> {
-  const currentPdf = await PDFDocument.load(file, {
-    ignoreEncryption: true,
-  });
+  const currentPdf = await loadPdfWithFallback(file);
 
   const newPdf = await PDFDocument.create();
 
@@ -84,8 +114,6 @@ export async function extractPages(
 }
 
 export async function countPages(file: Buffer): Promise<number> {
-  const currentPdf = await PDFDocument.load(file, {
-    ignoreEncryption: true,
-  });
+  const currentPdf = await loadPdfWithFallback(file);
   return currentPdf.getPageCount();
 }