pdfPagesToBlob, splitPdf and load pdf unit tests

Author: mpolomdeepsense

test/unit/SplitPdfHook.test.ts

@@ -0,0 +1,126 @@
+import { PDFDocument } from "pdf-lib";
+import { readFileSync } from "node:fs";
+
+import { SplitPdfHook } from "../../src/hooks/custom/SplitPdfHook";
+
+describe("SplitPdfHook", () => {
+  let splitPdfHook: SplitPdfHook;
+  const filename = "test/data/layout-parser-paper.pdf";
+  let file: Buffer;
+  let pdf: PDFDocument;
+
+  beforeEach(async () => {
+    splitPdfHook = new SplitPdfHook();
+    file = readFileSync(filename);
+    pdf = await PDFDocument.load(file);
+  });
+
+  afterEach(() => {
+    jest.clearAllMocks();
+  });
+
+  describe("pdfPagesToBlob", () => {
+    it("should convert range of pages to a Blob object", async () => {
+      const copyMock = jest.spyOn(PDFDocument.prototype, "copyPages");
+      const saveMock = jest.spyOn(PDFDocument.prototype, "save");
+      const addMock = jest.spyOn(PDFDocument.prototype, "addPage");
+
+      // Call the method
+      const result = await splitPdfHook.pdfPagesToBlob(pdf, 4, 8);
+
+      // Verify the expected behavior
+      expect(copyMock).toHaveBeenCalledWith(pdf, [3, 4, 5, 6, 7]);
+      expect(saveMock).toHaveBeenCalled();
+      expect(addMock).toHaveBeenCalledTimes(5);
+      expect(result).toBeInstanceOf(Blob);
+      expect(result.type).toEqual("application/pdf");
+    });
+  });
+
+  describe("splitPdf", () => {
+    it("should split the PDF into one batch", async () => {
+      const result = await splitPdfHook.splitPdf(pdf, 1);
+
+      expect(result).toHaveLength(1);
+      expect(result[0]?.startPage).toBe(1);
+      expect(result[0]?.endPage).toBe(16);
+    });
+
+    it("should split the PDF into 3 batches", async () => {
+      const result = await splitPdfHook.splitPdf(pdf, 3);
+
+      // Verify the expected behavior
+      expect(result).toHaveLength(3);
+      expect(result[0]?.startPage).toBe(1);
+      expect(result[0]?.endPage).toBe(6);
+      expect(result[1]?.startPage).toBe(7);
+      expect(result[1]?.endPage).toBe(12);
+      expect(result[2]?.startPage).toBe(13);
+      expect(result[2]?.endPage).toBe(16);
+    });
+
+    it("should split the PDF into 4 batches", async () => {
+      const result = await splitPdfHook.splitPdf(pdf, 5);
+
+      // Verify the expected behavior
+      expect(result).toHaveLength(4);
+      expect(result[0]?.startPage).toBe(1);
+      expect(result[0]?.endPage).toBe(4);
+      expect(result[1]?.startPage).toBe(5);
+      expect(result[1]?.endPage).toBe(8);
+      expect(result[2]?.startPage).toBe(9);
+      expect(result[2]?.endPage).toBe(12);
+      expect(result[3]?.startPage).toBe(13);
+      expect(result[3]?.endPage).toBe(16);
+    });
+  });
+
+  describe("loadPdf", () => {
+    it("should return true, null, and 0 if the file is not a PDF", async () => {
+      const result = await splitPdfHook.loadPdf(null);
+
+      expect(result).toEqual([true, null, 0]);
+    });
+
+    it("should return true, null, and 0 if the file is not a PDF", async () => {
+      const file = {
+        name: "document.txt",
+        content: jest.fn().mockResolvedValue(new ArrayBuffer(0)),
+      };
+
+      const result = await splitPdfHook.loadPdf(file as any);
+
+      expect(result).toEqual([true, null, 0]);
+      expect(file.content).not.toHaveBeenCalled();
+    });
+
+    it("should return true, null, and 0 if there is an error while loading the PDF", async () => {
+      const file = {
+        name: "document.pdf",
+        arrayBuffer: jest.fn().mockRejectedValue(new ArrayBuffer(0)),
+      };
+
+      const result = await splitPdfHook.loadPdf(file as any);
+
+      expect(result).toEqual([true, null, 0]);
+      expect(file.arrayBuffer).toHaveBeenCalled();
+    });
+
+    it("should return false, PDFDocument object, and the number of pages if the PDF is loaded successfully", async () => {
+      const file = readFileSync("test/data/layout-parser-paper-fast.pdf");
+      const f = {
+        name: "document.pdf",
+        arrayBuffer: () => file.buffer,
+      };
+
+      const loadMock = jest.spyOn(PDFDocument, "load");
+
+      const [error, _, pages] = await splitPdfHook.loadPdf(f as any);
+
+      expect(error).toBeFalsy();
+      expect(pages).toEqual(2);
+      expect(loadMock).toHaveBeenCalledTimes(1);
+      expect(loadMock).toHaveBeenCalledWith(f.arrayBuffer());
+    });
+  });
+});