From 9516ce1aa6b0489c5105be9415507277efa4058d Mon Sep 17 00:00:00 2001
From: Sha Zhou <sha.zhou@jina.ai>
Date: Tue, 19 Aug 2025 17:01:04 +0800
Subject: [PATCH] fix regex issue

---
 src/tools/embeddings.ts | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/src/tools/embeddings.ts b/src/tools/embeddings.ts
index 8ac8c6c..d9891ed 100644
--- a/src/tools/embeddings.ts
+++ b/src/tools/embeddings.ts
@@ -251,5 +251,6 @@ function truncateInputString(input: string | Record<string, string>): string {
 }
 
 function trimLeadingSymbols(str: string): string {
-  return str.replace(/^(?:[\u{1F000}-\u{1F9FF}]|[\u{2600}-\u{27BF}]|[\u{FE00}-\u{FE0F}]|[\u{DFE5}]|\s)+/gu, '');
+  const regex = /^[\p{S}\p{P}\p{Z}\p{C}\p{Emoji}]+/u;
+  return str.replace(regex, '');
 }
\ No newline at end of file