fix: use dynamic imports for pdfjs-dist to support both ESM and CommonJS environments

Author: beshkenadze

examples/nike-10k.ts

@@ -16,13 +16,12 @@ async function main() {
   }
 
   // Path to Nike's 10-K report
-  const pdfPath = path.join(__dirname, "../example_data/nke-10k-2023.pdf");
+  const pdfPath = path.join(__dirname, "../../example_data/nke-10k-2023.pdf");
 
   // Initialize the loader with specific settings for large documents
   const loader = new MistralOcrLoader(pdfPath, {
     apiKey,
     splitPages: true,
-    forceSingleMode: true, // Process each page individually for reliability
     modelName: "mistral-ocr-latest",
   });
 

examples/simple.ts

@@ -18,7 +18,10 @@ async function main() {
   }
 
   // Path to your PDF file - using the sample file
-  const pdfPath = path.join(__dirname, "../example_data/file-sample_150kB.pdf");
+  const pdfPath = path.join(
+    __dirname,
+    "../../example_data/file-sample_150kB.pdf"
+  );
 
   // Initialize the loader
   const loader = new MistralOcrLoader(pdfPath, {

src/pdf-utils/pdf-parser/pdf-parser.ts

@@ -1,5 +1,9 @@
 import path from "path";
-import { getDocument, version } from "pdfjs-dist/legacy/build/pdf.mjs";
+// Create a factory function to dynamically import pdfjs-dist
+async function getPdfLib() {
+  return import("pdfjs-dist/legacy/build/pdf.mjs");
+}
+
 // CMap and font configurations
 const CMAP_URL = path.join(process.cwd(), "node_modules/pdfjs-dist/cmaps/");
 const CMAP_PACKED = true;
@@ -106,14 +110,17 @@ export async function parsePDF(
     ...options,
   };
 
+  // Dynamically import pdfjs-dist
+  const { getDocument, version } = await getPdfLib();
+
   // Initialize result object
   const result: PDFParseResult = {
     numpages: 0,
     numrender: 0,
     info: null,
     metadata: null,
     text: "",
-    version: version, // Hard-coded version as we're not importing the full pdfjsLib
+    version: version,
   };
 
   try {

src/pdf-utils/pdf-to-image.ts

@@ -1,7 +1,10 @@
 import fs from "fs";
 import path from "path";
-import { getDocument } from "pdfjs-dist/legacy/build/pdf.mjs";
 import sharp from "sharp";
+// Create a factory function to dynamically import pdfjs-dist
+async function getPdfLib() {
+  return import("pdfjs-dist/legacy/build/pdf.mjs");
+}
 
 // CMap and font configurations
 const CMAP_URL = path.join(process.cwd(), "node_modules/pdfjs-dist/cmaps/");
@@ -41,6 +44,9 @@ export async function convertPdfToImage(
     const quality = options.quality || 100;
     const pageNumber = options.pageNumber || 1;
 
+    // Dynamically import pdfjs-dist
+    const { getDocument } = await getPdfLib();
+
     // Load the PDF document
     let pdfData: Uint8Array;
     if (typeof input === "string") {
@@ -144,6 +150,9 @@ export async function convertPdfToImages(
       pdfData = new Uint8Array(input);
     }
 
+    // Dynamically import pdfjs-dist
+    const { getDocument } = await getPdfLib();
+
     const loadingTask = getDocument({
       data: pdfData,
       cMapUrl: CMAP_URL,