From 994fceece379bec3571ace6e53398330159f76ee Mon Sep 17 00:00:00 2001
From: crazywoola <100913391+crazywoola@users.noreply.github.com>
Date: Mon, 11 Dec 2023 15:53:37 +0800
Subject: [PATCH] fix: qa regex (#1738)

---
 api/core/indexing_runner.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/api/core/indexing_runner.py b/api/core/indexing_runner.py
index 56c45bfd85..2baecedfd7 100644
--- a/api/core/indexing_runner.py
+++ b/api/core/indexing_runner.py
@@ -632,8 +632,8 @@ class IndexingRunner:
         return text
 
     def format_split_text(self, text):
-        regex = r"Q\d+:\s*(.*?)\s*A\d+:\s*([\s\S]*?)(?=Q|$)"
-        matches = re.findall(regex, text, re.MULTILINE)
+        regex = r"Q\d+:\s*(.*?)\s*A\d+:\s*([\s\S]*?)(?=Q\d+:|$)" 
+        matches = re.findall(regex, text, re.UNICODE)
 
         return [
             {