fix: remove empty lines from text before text-to-speech conversion

Author: wxg0103

apps/application/serializers/application.py

@@ -34,7 +34,7 @@
 from common.db.search import native_search, native_page_search
 from common.exception.app_exception import AppApiException
 from common.field.common import UploadedFileField
-from common.utils.common import get_file_content, restricted_loads, generate_uuid
+from common.utils.common import get_file_content, restricted_loads, generate_uuid, _remove_empty_lines
 from knowledge.models import Knowledge, KnowledgeScope
 from knowledge.serializers.knowledge import KnowledgeSerializer, KnowledgeModelSerializer
 from maxkb.conf import PROJECT_DIR
@@ -931,8 +931,9 @@ def text_to_speech(self, instance, with_valid=True):
         if application.tts_model_enable:
             model = get_model_instance_by_model_workspace_id(application.tts_model_id, application.workspace_id,
                                                              **application.tts_model_params_setting)
+            content = _remove_empty_lines(instance.get('text', ''))
 
-            return model.text_to_speech(instance.get('text'))
+            return model.text_to_speech(content)
 
     def play_demo_text(self, instance, with_valid=True):
         text = '你好，这里是语音播放测试'

apps/common/utils/common.py

@@ -118,7 +118,7 @@ def markdown_to_plain_text(md: str) -> str:
     # 使用正则表达式去除所有 HTML 标签
     text = re.sub(r'<[^>]+>', '', text)
     # 先移除特定媒体标签（优先级高于通用HTML标签移除）
-    text = re.sub(r'<(audio|video)[^>]*>.*?</\1>', '', text, flags=re.DOTALL)  # 匹配音频/视频标签
+    text = re.sub(r'<(?:audio|video)(?:\s+[^>]*)?>[\s\S]*?(?:</(?:audio|video)>)?', '', text, flags=re.IGNORECASE)
     text = re.sub(r'<img[^>]*>', '', text)  # 匹配图片标签
     # 去除多余的空白字符（包括换行符、制表符等）
     text = re.sub(r'\s+', ' ', text)