refactor: Move asset preprocessing to its own worker out of the inference worker

1 files changed, 0 insertions, 44 deletions

diff --git a/apps/workers/utils.ts b/apps/workers/utils.ts
index 15634902..2f56d3f0 100644
--- a/apps/workers/utils.ts
+++ b/apps/workers/utils.ts
@@ -1,9 +1,3 @@
-import os from "os";
-import PDFParser from "pdf2json";
-import { createWorker } from "tesseract.js";
-
-import serverConfig from "@hoarder/shared/config";
-
 export function withTimeout<T, Ret>(
   func: (param: T) => Promise<Ret>,
   timeoutSec: number,
@@ -20,41 +14,3 @@ export function withTimeout<T, Ret>(
     ]);
   };
 }
-
-export async function readImageText(buffer: Buffer) {
-  if (serverConfig.ocr.langs.length == 1 && serverConfig.ocr.langs[0] == "") {
-    return null;
-  }
-  const worker = await createWorker(serverConfig.ocr.langs, undefined, {
-    cachePath: serverConfig.ocr.cacheDir ?? os.tmpdir(),
-  });
-  try {
-    const ret = await worker.recognize(buffer);
-    if (ret.data.confidence <= serverConfig.ocr.confidenceThreshold) {
-      return null;
-    }
-    return ret.data.text;
-  } finally {
-    await worker.terminate();
-  }
-}
-
-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) => {
-      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);
-  });
-}