From 35760617a5f1e7f3cacbf9413e085b013329d0d6 Mon Sep 17 00:00:00 2001
From: Harry Yu <hyu05@wesleyan.edu>
Date: Thu, 24 Oct 2024 22:54:13 -0400
Subject: [PATCH] Updated the input clean up (issue #595)

---
 src/background/analysis/interactDB/ml/jsrun.js | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/src/background/analysis/interactDB/ml/jsrun.js b/src/background/analysis/interactDB/ml/jsrun.js
index ad86cfcb..2899b5ff 100644
--- a/src/background/analysis/interactDB/ml/jsrun.js
+++ b/src/background/analysis/interactDB/ml/jsrun.js
@@ -46,6 +46,10 @@ export async function loadModel() {
  * @returns {Promise<boolean>} boolean
  */
 export async function useModel(input, test_model) {
+  // Replace multiple backslashes followed by a double quote with
+  // a single escaped backslash and double quote to normalize the input
+  const input_cleaned = input.replace(/\\+\"/g, '\\"');
+  
   if (test_model != undefined) {
     model = test_model;
   }
@@ -55,7 +59,7 @@ export async function useModel(input, test_model) {
   if (!tokenizer) {
     tokenizer = new BertTokenizer(true);
   }
-  var tokens = await tokenizer.tokenize(input);
+  var tokens = await tokenizer.tokenize(input_cleaned);
   const tLen = tokens.length;
   var attArr = [];
   if (tLen < 384) {