chore Use pdf library to check file without extension (#115)

### Summary

Instead of manually checking filename with `.pdf` extension and return
is_pdf = false -> use currently pdf library to read file content to
decide if the file content is valid pdf

Author: yuming-long

src/hooks/custom/utils/pdf.ts

@@ -98,8 +98,7 @@ export async function splitPdf(
 }
 
 /**
- * Checks if the given file is a PDF. First it checks the `.pdf` file extension, then
- * it tries to load the file as a PDF using the `PDFDocument.load` method.
+ * Checks if the given file is a PDF by loading the file as a PDF using the `PDFDocument.load` method.
  * @param file - The file to check.
  * @returns A promise that resolves to three values, first is a boolean representing
  * whether there was an error during PDF load, second is a PDFDocument object or null
@@ -109,8 +108,8 @@ export async function splitPdf(
 export async function loadPdf(
   file: File | null
 ): Promise<[boolean, PDFDocument | null, number]> {
-  if (!file?.name.endsWith(".pdf")) {
-    console.info("Given file is not a PDF, so splitting is not enabled.");
+  if (!file) {
+    console.info("Given file is null, so splitting is not enabled.");
     return [true, null, 0];
   }
 
@@ -120,10 +119,6 @@ export async function loadPdf(
     const pagesCount = pdf.getPages().length;
     return [false, pdf, pagesCount];
   } catch (e) {
-    console.error(e);
-    console.warn(
-      "Attempted to interpret file as pdf, but error arose when splitting by pages. Reverting to non-split pdf handling path."
-    );
     return [true, null, 0];
   }
 }

test/unit/utils/pdf.test.ts

@@ -97,7 +97,7 @@ describe("Pdf utility functions", () => {
   });
 
   describe("loadPdf", () => {
-    it("should return true, null, and 0 if the file is not a PDF", async () => {
+    it("should return true, null, and 0 if the file is null", async () => {
       const result = await loadPdf(null);
 
       expect(result).toEqual([true, null, 0]);
@@ -115,6 +115,19 @@ describe("Pdf utility functions", () => {
       expect(file.content).not.toHaveBeenCalled();
     });
 
+    it("should return true, null, and 0 if the file is not a PDF without basing on file extension", async () => {
+      const file = {
+        name: "uuid1234",
+        content: jest.fn().mockResolvedValue(new ArrayBuffer(0)),
+      };
+
+      const result = await 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",
@@ -143,5 +156,24 @@ describe("Pdf utility functions", () => {
       expect(loadMock).toHaveBeenCalledTimes(1);
       expect(loadMock).toHaveBeenCalledWith(f.arrayBuffer());
     });
+
+    it("should return false, PDFDocument object, and the number of pages if the PDF is loaded successfully without basing on file extension", async () => {
+      const file = readFileSync("test/data/layout-parser-paper-fast.pdf");
+      const f = {
+        name: "uuid1234",
+        arrayBuffer: () => file.buffer,
+      };
+
+      jest.clearAllMocks(); // Reset Mocks Between Tests
+      const loadMock = jest.spyOn(PDFDocument, "load");
+
+      const [error, _, pages] = await loadPdf(f as any);
+
+      expect(error).toBeFalsy();
+      expect(pages).toEqual(2);
+      expect(loadMock).toHaveBeenCalledTimes(1);
+      expect(loadMock).toHaveBeenCalledWith(f.arrayBuffer());
+    });
+
   });
 });