1 files changed, 32 insertions, 0 deletions
diff --git a/apps/workers/utils.ts b/apps/workers/utils.ts
index 2f56d3f0..f8c48408 100644
--- a/apps/workers/utils.ts
+++ b/apps/workers/utils.ts
@@ -1,3 +1,5 @@
+import PDFParser from "pdf2json";
+
 export function withTimeout<T, Ret>(
   func: (param: T) => Promise<Ret>,
   timeoutSec: number,
@@ -14,3 +16,33 @@ export function withTimeout<T, Ret>(
     ]);
   };
 }
+
+export async function readPDFText(buffer: Buffer): Promise<{
+  text: string;
+  metadata: Record<string, string>;
+}> {
+  return new Promise((resolve, reject) => {
+    // Need raw text flag represents as number (1), reference : https://github.com/modesty/pdf2json/issues/76#issuecomment-236569265
+    const pdfParser = new PDFParser(null, 1);
+    pdfParser.on("pdfParser_dataError", reject);
+    pdfParser.on("pdfParser_dataReady", (pdfData) => {
+      // eslint-disable-next-line
+      resolve({
+        // The type isn't set correctly, reference : https://github.com/modesty/pdf2json/issues/327
+        // eslint-disable-next-line
+        text: (pdfParser as any).getRawTextContent(),
+        metadata: pdfData.Meta,
+      });
+    });
+    pdfParser.parseBuffer(buffer);
+  });
+}
+
+export function truncateContent(content: string, length = 1500) {
+  let words = content.split(" ");
+  if (words.length > length) {
+    words = words.slice(length);
+    content = words.join(" ");
+  }
+  return content;
+}