🐛 fix default-encrypted & XFA pdfs not opening

package.json

@@ -56,14 +56,15 @@
     "typescript": "^5.6.3"
   },
   "dependencies": {
+    "@cantoo/pdf-lib": "^2.3.2",
     "commander": "~9.4.1",
     "file-type": "~16.5.4",
     "form-data": "~3.0.1",
     "node-poppler": "^7.2.2",
-    "pdf-lib": "^1.17.1",
     "pdf.js-extract": "^0.2.1",
     "sharp": "^0.33.5",
-    "tmp": "^0.2.3"
+    "tmp": "^0.2.3",
+    "tslib": "^2.8.1"
   },
   "keywords": [
     "typescript",

src/imageOperations/common/imageExtractor.ts

@@ -1,4 +1,4 @@
-import { PDFDocument, PDFPage } from "pdf-lib";
+import { PDFDocument, PDFPage } from "@cantoo/pdf-lib";
 import { getMinMaxX, getMinMaxY, Polygon } from "../../geometry";
 
 /**

src/imageOperations/invoiceSplitterExtractor/invoiceSplitterExtractor.ts

@@ -1,4 +1,4 @@
-import { PDFDocument } from "pdf-lib";
+import { PDFDocument } from "@cantoo/pdf-lib";
 import { MindeeError, MindeeMimeTypeError } from "../../errors";
 import { InvoiceSplitterV1 } from "../../product";
 import { LocalInputSource } from "../../input";
@@ -35,7 +35,10 @@ 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 PDFDocument.load(inputFile.fileObject, {
+    ignoreEncryption: true,
+    password: ""
+  });
   if (pdfDoc.getPageCount() < 2) {
     throw new MindeeError("Invoice Splitter is only compatible with multi-page-pdf documents.");
   }

src/imageOperations/multiReceiptsExtractor/multiReceiptsExtractor.ts

@@ -1,4 +1,4 @@
-import { PDFDocument, PDFImage, PDFPage } from "pdf-lib";
+import { PDFDocument, PDFImage, PDFPage } from "@cantoo/pdf-lib";
 import { MindeeError, MindeeMimeTypeError } from "../../errors";
 import { Polygon } from "../../geometry";
 import { MultiReceiptsDetectorV1 } from "../../product";
@@ -37,7 +37,10 @@ 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 PDFDocument.load(inputFile.fileObject, {
+      ignoreEncryption: true,
+      password: ""
+    });
   } else {
     pdfDoc = await PDFDocument.create();
     let image: PDFImage;

src/pdf/pdfCompressor.ts

@@ -3,7 +3,7 @@ import tmp from "tmp";
 import { ExtractedPdfInfo, extractTextFromPdf, hasSourceText } from "./pdfUtils";
 import * as fs from "node:fs";
 import { Poppler } from "node-poppler";
-import { PDFDocument, PDFFont, PDFPage, rgb, StandardFonts } from "pdf-lib";
+import { PDFDocument, PDFFont, PDFPage, rgb, StandardFonts } from "@cantoo/pdf-lib";
 import { compressImage } from "../imageOperations";
 
 /**
@@ -128,7 +128,10 @@ async function compressPagesWithQuality(
   disableSourceText: boolean,
   extractedText: ExtractedPdfInfo | null
 ): Promise<Buffer[]> {
-  const pdfDoc = await PDFDocument.load(pdfData);
+  const pdfDoc = await PDFDocument.load(pdfData, {
+    ignoreEncryption: true,
+    password: ""
+  });
   const compressedPages: Buffer[] = [];
 
   for (let i = 0; i < extractedPdfInfo.pages.length; i++) {

src/pdf/pdfOperation.ts

@@ -1,5 +1,5 @@
 import { errorHandler } from "../errors/handler";
-import { PDFDocument } from "pdf-lib";
+import { PDFDocument } from "@cantoo/pdf-lib";
 import { PageOptions, PageOptionsOperation } from "../input";
 import { MindeeError } from "../errors";
 import { logger } from "../logger";
@@ -21,6 +21,7 @@ export async function extractPages(
 ): Promise<SplitPdf> {
   const currentPdf = await PDFDocument.load(file, {
     ignoreEncryption: true,
+    password: ""
   });
 
   const newPdf = await PDFDocument.create();
@@ -86,6 +87,7 @@ export async function extractPages(
 export async function countPages(file: Buffer): Promise<number> {
   const currentPdf = await PDFDocument.load(file, {
     ignoreEncryption: true,
+    password: ""
   });
   return currentPdf.getPageCount();
 }

tests/pdf/pdfTypes.spec.ts

@@ -0,0 +1,43 @@
+import * as mindee from "../../src";
+import path from "path";
+import { expect } from "chai";
+import * as pdf from "../../src/pdf";
+import { PageOptions } from "../../src/input";
+import { PageOptionsOperation } from "../../src";
+import * as fs from "node:fs";
+
+describe("Test pdf lib", () => {
+  let client: mindee.Client;
+  beforeEach(async () => {
+    client = new mindee.Client();
+  });
+  it("should open a simple XFA form PDF.", async () => {
+    const inputDoc = client.docFromPath(path.join(__dirname, "../data/file_types/pdf/XfaForm.pdf"));
+
+    await inputDoc.init();
+    expect(await pdf.countPages(inputDoc.fileObject)).to.eq(1);
+  });
+
+  it("should open an encrypted XFA form PDF.", async () => {
+    const inputDoc = client.docFromPath(path.join(__dirname, "../data/file_types/pdf/XfaForm_15p_encrypted.pdf"));
+
+    await inputDoc.init();
+    expect(await pdf.countPages(inputDoc.fileObject)).to.eq(15);
+  });
+
+
+  it("should be able to perform page operations on an encrypted XFA form PDF.", async () => {
+    const inputDoc = client.docFromPath(path.join(__dirname, "../data/file_types/pdf/XfaForm_15p_encrypted.pdf"));
+
+    await inputDoc.init();
+
+    const pageOptions: PageOptions = {
+      pageIndexes: [0, 1],
+      operation: PageOptionsOperation.KeepOnly,
+      onMinPages: 1,
+    };
+    const splitPdf = await pdf.extractPages(inputDoc.fileObject, pageOptions);
+    expect(splitPdf.totalPagesRemoved).to.eq(13);
+    expect(await pdf.countPages(splitPdf.file)).to.eq(2);
+  });
+});