Merge pull request #3196 from xinnan-tech/py-device-calling

Py device calling

main/xiaozhi-server/core/connection.py

@@ -190,6 +190,9 @@ class ConnectionHandler:
         # 初始化提示词管理器
         self.prompt_manager = PromptManager(self.config, self.logger)
 
+        # 初始化通话状态
+        self.calling = False
+
     async def handle_connection(self, ws: websockets.ServerConnection):
         try:
             # 获取运行中的事件循环（必须在异步上下文中）

main/xiaozhi-server/core/handle/sendAudioHandle.py

@@ -47,7 +47,8 @@ async def sendAudioMessage(conn: "ConnectionHandler", sentenceType, audios, text
         conn.logger.bind(tag=TAG).info(f"发送音频消息: {sentenceType}, {text}")
 
     # 发送结束消息（如果是最后一个文本）
-    if sentenceType == SentenceType.LAST:
+    # 通话需要维持speaking状态
+    if not conn.calling and sentenceType == SentenceType.LAST:
         await send_tts_message(conn, "stop", None)
         if conn.close_after_chat:
             await conn.close()

main/xiaozhi-server/core/handle/textHandler/listenMessageHandler.py

@@ -1,17 +1,21 @@
 import time
+import uuid
 import asyncio
 from typing import Dict, Any, TYPE_CHECKING
 
 if TYPE_CHECKING:
     from core.connection import ConnectionHandler
 
+from core.utils.dialogue import Message
+from core.providers.asr.dto.dto import InterfaceType
 from core.handle.receiveAudioHandle import startToChat
 from core.handle.reportHandle import enqueue_asr_report
 from core.handle.sendAudioHandle import send_stt_message, send_tts_message
 from core.handle.textMessageHandler import TextMessageHandler
 from core.handle.textMessageType import TextMessageType
 from core.utils.util import remove_punctuation_and_length
-from core.providers.asr.dto.dto import InterfaceType
+from core.providers.tts.dto.dto import ContentType, TTSMessageDTO, SentenceType
+
 
 TAG = __name__
 
@@ -54,6 +58,25 @@ class ListenTextMessageHandler(TextMessageHandler):
                     original_text
                 )
 
+                # 检查是否是设备呼叫指令 [device_call]
+                if original_text.startswith("[device_call]"):
+                    # 提取 tag 后的文本
+                    call_text = original_text[len("[device_call]"):].strip()
+                    conn.logger.bind(tag=TAG).info(f"收到设备呼叫指令: {call_text}")
+
+                    # 准备开始新会话
+                    conn.sentence_id = uuid.uuid4().hex
+
+                    await send_stt_message(conn, call_text)
+                    conn.tts.store_tts_text(conn.sentence_id, call_text)
+                    conn.tts.tts_text_queue.put(TTSMessageDTO(sentence_id=conn.sentence_id, sentence_type=SentenceType.FIRST, content_type=ContentType.ACTION))
+                    conn.tts.tts_one_sentence(conn, ContentType.TEXT, content_detail=call_text)
+                    conn.tts.tts_text_queue.put(TTSMessageDTO(sentence_id=conn.sentence_id, sentence_type=SentenceType.LAST, content_type=ContentType.ACTION))
+
+                    # 添加到对话历史，让模型理解上下文
+                    conn.dialogue.put(Message(role="assistant", content=call_text))
+                    return
+
                 # 识别是否是唤醒词
                 is_wakeup_words = filtered_text in conn.config.get("wakeup_words")
                 # 是否开启唤醒词回复