"""
【通义千问 Qwen】API集成模块
用于意图理解和任务处理
"""
import json
import re
import logging
import dashscope
from dashscope import Generation
from database import config
from Progress.utils.logger_utils import log_time, log_step, log_var, log_call
from Progress.utils.logger_config import setup_logger
# --- 初始化日志器 ---
logger = logging.getLogger("ai_assistant")
DASHSCOPE_API_KEY = config.api_key
DASHSCOPE_MODEL = config.model
class QWENAssistant:
def __init__(self):
if not DASHSCOPE_API_KEY:
raise ValueError("缺少 DASHSCOPE_API_KEY，请检查配置文件")
dashscope.api_key = DASHSCOPE_API_KEY
self.model_name = DASHSCOPE_MODEL or 'qwen-max'
logger.info(f"✅ QWENAssistant 初始化完成，使用模型: {self.model_name}")
self.conversation_history = []
self.system_prompt = """
你是一个智能语音控制助手，能够理解用户的自然语言指令，并将其转化为可执行的任务计划。
你的职责是：
- 准确理解用户意图；
- 若涉及多个动作，需拆解为【执行计划】；
- 输出一个严格符合规范的 JSON 对象，供系统解析执行；
- 所有回复必须使用中文（仅限于 response_to_user 字段）；
🎯 输出格式要求（必须遵守）：
{
"intent": "system_control", // 意图类型："system_control" 或 "chat"
"task_type": "start_background_tasks",// 任务类型的简要描述（动态生成）
"execution_plan": [ // 执行步骤列表（每个步骤包含 operation, parameters, description）
{
"operation": "函数名", // 必须是已知操作之一
"parameters": { ... }, // 参数对象（按需提供）
"description": "该步骤的目的说明"
}
],
"response_to_user": "你要对用户说的话（用中文）",
"requires_confirmation": false, // 是否需要用户确认后再执行
"mode": "parallel" // 执行模式："parallel"（并行）或 "serial"（串行）
}
📌 已知 operation 列表（不可拼写错误）：
- play_music(music_path: str)
- stop_music()
- pause_music()
- resume_music()
- open_application(app_name: str)
- create_file(file_path: str, content?: str)
- read_file(file_path: str)
- write_file(file_path: str, content: str)
- set_reminder(reminder_time: str, message: str)
- speak_response(message: str)
📌 规则说明：
1. 只有当用户明确要求执行系统级任务时，才设置 intent="system_control"；
否则设为 intent="chat"（例如闲聊、问天气、讲笑话等）。
2. execution_plan 中的每一步都必须与用户需求直接相关；
❌ 禁止添加无关操作（如随便加 speak_response 或 play_music）！
3. mode 决定执行方式：
- 如果各步骤互不依赖 → "parallel"
- 如果有先后依赖（如先打开再写入）→ "serial"
4. response_to_user 是你对用户的自然语言反馈，必须简洁友好，使用中文。
5. requires_confirmation：
- 涉及删除、覆盖文件、长时间运行任务 → true
- 普通操作（打开应用、播放音乐）→ false
⚠️ 重要警告：
- 绝不允许照搬示例中的参数或路径！必须根据用户输入提取真实信息。
- 不得虚构不存在的 operation 或 parameter 名称。
- 不得省略任何字段，所有 key 都必须存在。
- 不得输出额外文本（如解释、注释、
```json
``` 包裹符），只输出纯 JSON 对象。
✅ 正确行为示例：
用户说：“帮我写一份自我介绍到 D:/intro.txt，并打开看看”
→ 应返回包含 write_file 和 read_file 的 serial 计划。
用户说：“播放 C:/Music/background.mp3 并告诉我准备好了”
→ 可以并行执行 play_music 和 speak_response。
用户说：“今天过得怎么样？”
→ intent="chat"，response_to_user="我很好，谢谢！"
🚫 错误行为：
- 把所有指令都变成和示例一样的操作组合；
- 在没有请求的情况下自动添加 speak_response；
- 使用未定义的操作如 run_script、send_email。
现在，请根据用户的最新指令生成对应的 JSON 响应。
"""
@log_time
@log_step("处理语音指令")
def process_voice_command(self, voice_text):
log_var("原始输入", voice_text)
if not voice_text.strip():
return self._create_fallback_response("我没有听清楚，请重新说话。")
self.conversation_history.append({"role": "user", "content": voice_text})
try:
messages = [{"role": "system", "content": self.system_prompt}]
messages.extend(self.conversation_history[-10:]) # 保留最近上下文
response = Generation.call(
model=self.model_name,
messages=messages,
temperature=0.5,
top_p=0.8,
max_tokens=1024
)
if response.status_code != 200:
logger.error(f"Qwen API 调用失败: {response.status_code}, {response.message}")
return self._create_fallback_response(f"服务暂时不可用: {response.message}")
ai_output = response.output['text'].strip()
log_var("模型输出", ai_output)
self.conversation_history.append({"role": "assistant", "content": ai_output})
# === 尝试解析 JSON ===
parsed = self._extract_and_validate_json(ai_output)
if parsed:
return parsed
else:
# 若无法解析为有效计划，则降级为普通对话
return self._create_fallback_response(ai_output)
except Exception as e:
logger.exception("处理语音指令时发生异常")
return self._create_fallback_response("抱歉，我遇到了一些技术问题，请稍后再试。")
def _extract_and_validate_json(self, text: str):
"""从文本中提取 JSON 并验证结构"""
try:
# 方法1：直接加载
data = json.loads(text)
return self._validate_plan_structure(data)
except json.JSONDecodeError:
pass
# 方法2：正则匹配第一个大括号包裹的内容
match = re.search(r'\{[\s\S]*\}', text)
if not match:
return None
try:
data = json.loads(match.group())
return self._validate_plan_structure(data)
except:
return None
def _validate_plan_structure(self, data: dict):
"""验证是否符合多任务计划格式"""
required_top_level = ["intent", "task_type", "execution_plan", "response_to_user", "requires_confirmation"]
for field in required_top_level:
if field not in data:
logger.warning(f"缺少必要字段: {field}")
return None
if not isinstance(data["execution_plan"], list) or len(data["execution_plan"]) == 0:
logger.warning("execution_plan 必须是非空数组")
return None
valid_operations = {
"play_music", "stop_music", "pause_music", "resume_music",
"open_application", "create_file", "read_file", "write_file",
"set_reminder", "speak_response"
}
for step in data["execution_plan"]:
op = step.get("operation")
params = step.get("parameters", {})
if not op or op not in valid_operations:
logger.warning(f"无效操作: {op}")
return None
if not isinstance(params, dict):
logger.warning(f"parameters 必须是对象: {params}")
return None
# 补全默认值
if "mode" not in data:
data["mode"] = "parallel"
return data
def _create_fallback_response(self, message: str):
"""降级响应：用于非结构化输出"""
return {
"intent": "chat",
"task_type": "reply",
"response_to_user": message,
"requires_confirmation": False,
"execution_plan": [],
"mode": "serial"
}
def _create_response(self, intent, action, parameters, response, needs_confirmation):
resp = {"intent": intent, "action": action, "parameters": parameters, "response": response, "needs_confirmation": needs_confirmation}
log_var("返回响应", resp)
return resp
@log_time
def generate_text(self, prompt, task_type="general"):
log_var("任务类型", task_type)
log_var("提示词长度", len(prompt))
try:
system_prompt = f"""
你是一个专业的文本生成助手。根据用户的要求生成高质量的文本内容。
任务类型：{task_type}
要求：{prompt}
请生成相应的文本内容，确保内容准确、有逻辑、语言流畅。
"""
response = Generation.call(
model=self.model_name,
messages=[
{"role": "system", "content": system_prompt},
{"role": "user", "content": prompt}
],
temperature=0.8,
max_tokens=2000
)
if response.status_code == 200:
result = response.output['text']
log_var("生成结果长度", len(result))
return result
else:
error_msg = f"文本生成失败: {response.message}"
logger.error(error_msg)
return error_msg
except Exception as e:
logger.exception("文本生成出错")
return f"抱歉，生成文本时遇到错误：{str(e)}"
@log_time
def summarize_text(self, text):
log_var("待总结文本长度", len(text))
try:
prompt = f"请总结以下文本的主要内容：\n\n{text}"
response = Generation.call(
model=self.model_name,
messages=[{"role": "user", "content": prompt}],
temperature=0.3,
max_tokens=500
)
if response.status_code == 200:
result = response.output['text']
log_var("总结结果长度", len(result))
return result
else:
error_msg = f"总结失败: {response.message}"
logger.error(error_msg)
return error_msg
except Exception as e:
logger.exception("文本总结出错")
return f"抱歉，总结文本时遇到错误：{str(e)}"
@log_time
def translate_text(self, text, target_language="英文"):
log_var("目标语言", target_language)
log_var("原文长度", len(text))
try:
prompt = f"请将以下文本翻译成{target_language}：\n\n{text}"
response = Generation.call(
model=self.model_name,
messages=[{"role": "user", "content": prompt}],
temperature=0.3,
max_tokens=1000
)
if response.status_code == 200:
result = response.output['text']
log_var("翻译结果长度", len(result))
return result
else:
error_msg = f"翻译失败: {response.message}"
logger.error(error_msg)
return error_msg
except Exception as e:
logger.exception("文本翻译出错")
return f"抱歉，翻译文本时遇到错误：{str(e)}"
"""
【系统控制模块】System Controller
提供音乐播放、文件操作、应用启动、定时提醒等本地系统级功能
"""
import inspect
import os
import subprocess
import platform
import threading
import time
from typing import Any, Dict
import psutil
import pygame
from datetime import datetime
import logging
import schedule
from Progress.app.text_to_speech import TTSEngine
from Progress.utils.ai_tools import call_llm_to_choose_function
from database import config
from Progress.utils.ai_tools import FUNCTION_SCHEMA, ai_callable
from Progress.utils.logger_utils import log_time, log_step, log_var, log_call
from Progress.utils.logger_config import setup_logger
RESOURCE_PATH = config.resource_path
DEFAULT_MUSIC_PATH = os.path.join(RESOURCE_PATH, config.music_path)
DEFAULT_DOCUMENT_PATH = os.path.join(RESOURCE_PATH, config.doc_path)
logger = logging.getLogger("ai_assistant")
class SystemController:
def __init__(self):
self.system = platform.system()
self.tts_engine = TTSEngine()
self.music_player = None
self._init_music_player()
self.task_counter = 0
self.scheduled_tasks = {}
@ai_callable(
description="使用语音合成技术播报一段文本回复用户",
params={
"message": "要朗读的文本内容"
},
intent="response",
action="speak"
)
@log_step("语音回复用户")
@log_time
def _speak_response(self, message: str):
"""
AI 回复用户的语音播报接口
"""
if not self.tts_engine.is_available():
logger.warning("🔊 TTS 引擎不可用")
return False, "TTS 引擎未就绪"
try:
logger.info(f"📢 播报: {message}")
success = self.tts_engine.speak(message, interrupt=True)
return success, "语音已播放" if success else "播放失败"
except Exception as e:
logger.exception("💥 播报异常")
return False, str(e)
@log_step("初始化音乐播放器")
@log_time
def _init_music_player(self):
try:
pygame.mixer.init()
self.music_player = pygame.mixer.music
logger.info("✅ 音乐播放器初始化成功")
except Exception as e:
logger.exception("❌ 音乐播放器初始化失败")
self.music_player = None
@log_step("播放音乐")
@log_time
@ai_callable(
description="播放音乐文件或指定歌手的歌曲",
params={"path": "音乐文件路径", "artist": "歌手名称"},
intent="music",
action="play",
concurrent=True
)
def play_music(self, music_path=None):
target_path = music_path or DEFAULT_MUSIC_PATH
if not os.path.exists(target_path):
msg = f"📁 路径不存在: {target_path}"
logger.warning(msg)
return False, msg
music_files = self._find_music_files(target_path)
if not music_files:
msg = "🎵 未找到支持的音乐文件"
logger.info(msg)
return False, msg
try:
self.stop_music()
self.music_player.load(music_files[0])
self.music_player.play(-1)
success_msg = f"🎶 正在播放: {os.path.basename(music_files[0])}"
logger.info(success_msg)
return True, success_msg
except Exception as e:
logger.exception("💥 播放音乐失败")
return False, f"播放失败: {str(e)}"
@ai_callable(
description="停止当前播放的音乐",
params={},
intent="music",
action="stop"
)
def stop_music(self):
try:
if self.music_player and pygame.mixer.get_init():
self.music_player.stop()
logger.info("⏹️ 音乐已停止")
return True, "音乐已停止"
except Exception as e:
logger.exception("❌ 停止音乐失败")
return False, f"停止失败: {str(e)}"
@ai_callable(
description="暂停当前正在播放的音乐。",
params={},
intent="muxic",
action="pause"
)
def pause_music(self):
"""暂停音乐"""
try:
self.music_player.pause()
return True, "音乐已暂停"
except Exception as e:
return False, f"暂停音乐失败: {str(e)}"
@ai_callable(
description="恢复播放当前正在播放的音乐。",
params={},
intent="music",
action="resume"
)
def resume_music(self):
"""恢复音乐"""
try:
self.music_player.unpause()
return True, "音乐已恢复"
except Exception as e:
return False, f"恢复音乐失败: {str(e)}"
@ai_callable(
description="打开应用程序或浏览器访问网址",
params={"app_name": "应用名称（如 记事本、浏览器）", "url": "网页地址"},
intent="system",
action="open_app",
concurrent=True
)
def open_application(self, app_name: str, url: str = None):
def _run():
"""
AI 调用入口：打开指定应用程序
参数由 AI 解析后传入
"""
# === 别名映射表 ===
alias_map = {
# 浏览器相关
"浏览器": "browser", "browser": "browser",
"chrome": "browser", "google chrome": "browser", "谷歌浏览器": "browser",
"edge": "browser", "firefox": "browser", "safari": "browser",
# 文本编辑器
"记事本": "text_editor", "notepad": "text_editor", "text_editer": "text_editor", "文本编辑器": "text_editor",
# 文件管理器
"文件管理器": "explorer", "explorer": "explorer", "finder": "explorer",
# 计算器
"计算器": "calc", "calc": "calc", "calculator": "calc",
# 终端
"终端": "terminal", "terminal": "terminal", "cmd": "terminal", "powershell": "terminal",
"shell": "terminal", "命令行": "terminal"
}
app_key = alias_map.get(app_name.strip())
if not app_key:
error_msg = f"🚫 不支持的应用: {app_name}。支持的应用有：浏览器、记事本、计算器、终端、文件管理器等。"
logger.warning(error_msg)
return False, error_msg
try:
if app_key == "browser":
target_url = url or "https://www.baidu.com"
success, msg = self._get_browser_command(target_url)
logger.info(f"🌐 {msg}")
return success, msg
else:
# 获取对应命令生成函数
cmd_func_name = f"_get_{app_key}_command"
cmd_func = getattr(self, cmd_func_name, None)
if not cmd_func:
return False, f"❌ 缺少命令生成函数: {cmd_func_name}"
cmd = cmd_func()
subprocess.Popen(cmd, shell=True)
success_msg = f"🚀 已发送指令打开 {app_name}"
logger.info(success_msg)
return True, success_msg
except Exception as e:
logger.exception(f"💥 启动应用失败: {app_name}")
return False, f"启动失败: {str(e)}"
thread = threading.Thread(target=_run,daemon=True)
thread.start()
return True,f"正在尝试打开{app_name}..."
@ai_callable(
description="创建一个新文本文件并写入内容",
params={"file_path": "文件路径", "content": "要写入的内容"},
intent="file",
action="create",
concurrent=True
)
def create_file(self, file_path, content=""):
def _run():
try:
os.makedirs(os.path.dirname(file_path), exist_ok=True)
with open(file_path, 'w', encoding='utf-8') as f:
f.write(content)
return True, f"文件已创建: {file_path}"
except Exception as e:
logger.exception("❌ 创建文件失败")
return False, f"创建失败: {str(e)}"
thread = threading.Thread(target=_run, daemon=True)
thread.start()
return True, f"正在尝试创建文件并写入文本..."
@ai_callable(
description="读取文本文件内容",
params={"file_path": "文件路径"},
intent="file",
action="read",
concurrent=True
)
def read_file(self, file_path):
def _run():
"""读取文件"""
try:
with open(DEFAULT_DOCUMENT_PATH+file_path, 'r', encoding='utf-8') as f:
content = f.read()
return True, content
except Exception as e:
return False, f"读取文件失败: {str(e)}"
thread = threading.Thread(target=_run,daemon=True)
thread.start()
return True,f"正在尝试读取文件..."
@ai_callable(
description="读取文本文件内容",
params={"file_path": "文件路径","content":"写入的内容"},
intent="file",
action="write",
concurrent=True
)
def write_file(self, file_path, content):
def _run():
"""写入文件"""
try:
with open(DEFAULT_DOCUMENT_PATH+file_path, 'w', encoding='utf-8') as f:
f.write(content)
return True, f"文件已保存: {file_path}"
except Exception as e:
return False, f"写入文件失败: {str(e)}"
thread = threading.Thread(target=_run,daemon=True)
thread.start()
return True,f"正在尝试向{file_path}写入文本..."
@ai_callable(
description="获取当前系统信息，包括操作系统、CPU、内存等。",
params={},
intent="system",
action="get_system_info",
concurrent=True
)
def get_system_info(self):
def _run():
"""获取系统信息"""
try:
info = {
"操作系统": platform.system(),
"系统版本": platform.version(),
"处理器": platform.processor(),
"内存使用率": f"{psutil.virtual_memory().percent}%",
"CPU使用率": f"{psutil.cpu_percent()}%",
"磁盘使用率": f"{psutil.disk_usage('/').percent}%"
}
return True, info
except Exception as e:
return False, f"获取系统信息失败: {str(e)}"
thread = threading.Thread(target=_run,daemon=True)
thread.start()
return True,f"正在尝试获取系统信息..."
@ai_callable(
description="设置一个定时提醒",
params={"message": "提醒内容", "delay_minutes": "延迟分钟数"},
intent="system",
action="set_reminder",
concurrent=True
)
def set_reminder(self, message, delay_minutes):
def _run():
"""设置提醒"""
try:
self.task_counter += 1
task_id = f"reminder_{self.task_counter}"
def reminder_job():
print(f"提醒: {message}")
# 这里可以添加通知功能
schedule.every(delay_minutes).minutes.do(reminder_job)
self.scheduled_tasks[task_id] = {
"message": message,
"delay": delay_minutes,
"created": datetime.now()
}
return True, f"提醒已设置: {delay_minutes}分钟后提醒 - {message}"
except Exception as e:
return False, f"设置提醒失败: {str(e)}"
thread = threading.Thread(target=_run,daemon=True)
thread.start()
return True,f"正在设置提醒..."
@ai_callable(
description="并发执行多个任务",
params={"tasks": "任务列表，每个包含operation和arguments"},
intent="system",
action="execute_concurrent",
concurrent=True
)
def _run_parallel_tasks(self, tasks: list):
def _run_single(task):
op = task.get("operation")
args = task.get("arguments",{})
func = getattr(self,op,None)
if func and callable(func):
try:
func(**args)
except Exception as e:
logger.error(f"执行任务{op}失败：{e}")
for task in tasks:
thread = threading.Thread(target=_run_single,args=(task,),daemon=True)
thread.start()
return True,f"已并发执行{len(tasks)}个任务"
def run_scheduled_tasks(self):
"""运行定时任务"""
schedule.run_pending()
def _find_music_files(self, directory):
"""查找音乐文件"""
music_extensions = ['.mp3', '.wav', '.flac', '.m4a', '.ogg']
music_files = []
try:
for root, dirs, files in os.walk(directory):
for file in files:
if any(file.lower().endswith(ext) for ext in music_extensions):
music_files.append(os.path.join(root, file))
except Exception as e:
print(f"搜索音乐文件失败: {e}")
return music_files
def _get_text_editor_command(self):
"""获取文本编辑器启动命令"""
if self.system == "Windows":
return "notepad"
elif self.system == "Darwin": # macOS
return "open -a TextEdit"
else: # Linux
return "gedit"
def _get_explorer_command(self):
"""获取文件管理器启动命令"""
if self.system == "Windows":
return "explorer"
elif self.system == "Darwin": # macOS
return "open -a Finder"
else: # Linux
return "nautilus"
def _get_calc_command(self):
"""获取计算器启动命令"""
if self.system == "Windows":
return "calc"
elif self.system == "Darwin": # macOS
return "open -a Calculator"
else: # Linux
return "gnome-calculator"
def _get_terminal_command(self):
"""获取终端启动命令"""
if self.system == "Windows":
return "cmd"
elif self.system == "Darwin": # macOS
return "open -a Terminal"
else: # Linux
return "gnome-terminal"
def _get_browser_command(self, url="https://www.baidu.com"):
try:
import webbrowser
if webbrowser.open(url):
logger.info(f"🌐 已使用默认浏览器打开: {url}")
return True, f"正在打开浏览器访问: {url}"
else:
return False, "无法打开浏览器"
except Exception as e:
logger.error(f"❌ 浏览器打开异常: {e}")
return False, str(e)
class TaskOrchestrator:
def __init__(self, system_controller):
self.system_controller = system_controller
self.function_map = self._build_function_map()
self.running_scheduled_tasks = False
logger.info(f"🔧 任务编排器已加载 {len(self.function_map)} 个可调用函数")
def _build_function_map(self) -> Dict[str, callable]:
"""构建函数名 → 方法对象的映射"""
mapping = {}
for item in FUNCTION_SCHEMA:
func_name = item["name"]
func = getattr(self.system_controller, func_name, None)
if func and callable(func):
mapping[func_name] = func
else:
logger.warning(f"⚠️ 未找到或不可调用: {func_name}")
return mapping
def _convert_arg_types(self, func: callable, args: dict) -> dict:
"""
尝试将参数转为函数期望的类型（简单启发式）
注意：Python 没有原生参数类型签名，这里做基础转换
"""
converted = {}
sig = inspect.signature(func)
for name, param in sig.parameters.items():
value = args.get(name)
if value is None:
continue
# 简单类型推断（基于默认值）
ann = param.annotation
if isinstance(ann, type):
try:
if ann == int and not isinstance(value, int):
converted[name] = int(value)
elif ann == float and not isinstance(value, float):
converted[name] = float(value)
else:
converted[name] = value
except (ValueError, TypeError):
converted[name] = value # 保持原始值，让函数自己处理
else:
converted[name] = value
return converted
def _start_scheduled_task_loop(self):
"""后台线程运行定时任务"""
def run_loop():
while self.running_scheduled_tasks:
schedule.run_pending()
time.sleep(1)
if not self.running_scheduled_tasks:
self.running_scheduled_tasks = True
thread = threading.Thread(target=run_loop, daemon=True)
thread.start()
logger.info("⏰ 已启动定时任务监听循环")
@log_time
def execute_from_ai_decision(self, user_input: str) -> Dict[str, Any]:
"""
主入口：接收用户输入 → AI 决策 → 执行函数 → 返回结构化结果
支持两种模式：
1. 传统单任务：{"function": "...", "arguments": {...}}
2. 多步骤任务：{"execution_plan": [...], "mode": "parallel|serial"}
"""
# Step 1: 调用 LLM 获取决策
decision = call_llm_to_choose_function(user_input)
if not decision:
return {
"success": False,
"message": "抱歉，我没有理解您的请求。",
"data": None
}
# === 判断是否为多任务计划 ===
if "execution_plan" in decision and isinstance(decision["execution_plan"], list):
return self._execute_task_plan(decision)
# === 否则走传统单任务流程 ===
func_name = decision.get("function")
args = decision.get("arguments", {})
if not func_name:
return {
"success": False,
"message": "AI 返回的函数名为空。",
"data": None
}
func = self.function_map.get(func_name)
if not func:
logger.warning(f"❌ 函数不存在: {func_name}")
return {
"success": False,
"message": f"系统不支持操作：{func_name}",
"data": None
}
try:
safe_args = self._convert_arg_types(func, args)
except Exception as e:
logger.warning(f"参数转换失败: {e}")
safe_args = args
try:
logger.info(f"🚀 正在执行: {func_name}({safe_args})")
result = func(**safe_args)
if isinstance(result, tuple):
success, msg = result
elif isinstance(result, dict):
success = result.get("success", False)
msg = result.get("message", str(result))
else:
success = True
msg = str(result)
# 特殊逻辑：设置提醒后启动调度循环
if func_name == "set_reminder" and success:
self._start_scheduled_task_loop()
return {
"success": success,
"message": msg,
"data": None,
"operation": func_name,
"input": args
}
except TypeError as e:
error_msg = f"参数错误，请检查输入格式: {str(e)}"
logger.error(error_msg)
return {
"success": False,
"message": error_msg,
"data": None
}
except Exception as e:
logger.exception(f"💥 执行 {func_name} 时发生异常")
return {
"success": False,
"message": f"执行失败：{str(e)}",
"data": None
}
@log_step("执行多任务计划")
@log_time
def _execute_task_plan(self, plan: dict) -> Dict[str, Any]:
"""
执行由多个 operation 组成的任务计划
支持 serial / parallel 模式
"""
execution_plan = plan.get("execution_plan", [])
mode = plan.get("mode", "parallel").lower()
response_to_user = plan.get("response_to_user", "任务已提交。")
if not execution_plan:
return {"success": False, "message": "执行计划为空"}
def run_single_step(step: dict):
op = step.get("operation")
params = step.get("parameters", {})
func = self.function_map.get(op)
if not func:
logger.warning(f"⚠️ 未知操作: {op}")
return False, f"不支持的操作: {op}"
try:
safe_params = self._convert_arg_types(func, params)
result = func(**safe_params)
if isinstance(result, tuple):
return result
return True, str(result)
except Exception as e:
logger.exception(f"执行 {op} 失败")
return False, str(e)
threads = []
results = []
if mode == "parallel":
for step in execution_plan:
t = threading.Thread(target=lambda s=step: results.append(run_single_step(s)), daemon=True)
t.start()
threads.append(t)
for t in threads:
t.join()
else: # serial
for step in execution_plan:
res = run_single_step(step)
results.append(res)
if not res[0]: # 失败则中断
break
# 汇总结果
all_success = all(r[0] for r in results)
messages = [r[1] for r in results]
final_message = response_to_user + " " + " | ".join(messages) if messages else response_to_user
return {
"success": all_success,
"message": final_message.strip(),
"data": {"step_results": results},
"operation": "task_plan",
"input": plan
}
import pyttsx3
import threading
import queue
import time
import logging
logger = logging.getLogger("ai_assistant")
class TTSEngine:
def __init__(self):
self.engine = None
self.is_speaking = False
self.speech_queue = queue.Queue()
self._lock = threading.Lock()
self._worker_thread = None
self._stop_event = threading.Event()
self._initialize_engine()
def _initialize_engine(self):
try:
self.engine = pyttsx3.init()
# 设置语速和音量
self.engine.setProperty('rate', 180)
self.engine.setProperty('volume', 0.9)
# 尝试设置中文语音
voices = self.engine.getProperty('voices')
for v in voices:
if 'chinese' in v.name.lower() or 'zh' in v.name or '普通话' in v.name:
self.engine.setProperty('voice', v.id)
break
# 绑定回调
self.engine.connect('started-utterance', self._on_start)
self.engine.connect('finished-utterance', self._on_finish)
logger.info("✅ TTS 引擎初始化成功")
except Exception as e:
logger.error(f"❌ 初始化 TTS 失败: {e}")
self.engine = None
def _on_start(self, name):
logger.info(f"▶️ 开始播报: {name}")
with self._lock:
self.is_speaking = True
def _on_finish(self, name, completed):
status = "完成" if completed else "中断"
logger.info(f"⏹️ 结束播报: {name} ({status})")
with self._lock:
self.is_speaking = False
# 自动处理下一条
self._process_next()
def _process_next(self):
"""尝试从队列取出下一条语音播报"""
try:
text = self.speech_queue.get_nowait()
logger.debug(f"📥 准备播放队列中的语音: {text}")
self.engine.say(text, name=text[:50])
self.speech_queue.task_done()
except queue.Empty:
pass # 队列空了，无需处理
def _event_loop_worker(self):
"""TTS 专用事件循环线程"""
logger.debug("🔊 TTS 工作线程已启动")
try:
# 🔁 启动非阻塞事件循环
self.engine.startLoop(False)
while not self._stop_event.is_set():
if not self.is_speaking and not self.speech_queue.empty():
try:
text = self.speech_queue.get(timeout=0.5)
self.engine.say(text, name=text[:50]) # 触发 started-utterance
self.speech_queue.task_done()
except queue.Empty:
continue
else:
time.sleep(0.1) # 避免 CPU 占用过高
# 显式关闭事件循环
self.engine.endLoop()
logger.debug("🔚 TTS 事件循环已退出")
except Exception as e:
logger.exception(f"💥 TTS 线程异常: {e}")
self.engine.endLoop()
def speak(self, text: str, interrupt: bool = True):
"""添加语音到播报队列"""
if not self.engine or not text.strip():
return False
cleaned = text.strip()
with self._lock:
if interrupt and self.is_speaking:
self.stop_current_speech()
self.speech_queue.put(cleaned)
if not self._worker_thread or not self._worker_thread.is_alive():
self._stop_event.clear()
self._worker_thread = threading.Thread(target=self._event_loop_worker, daemon=True)
self._worker_thread.start()
return True
def stop_current_speech(self):
"""停止当前语音播报"""
with self._lock:
if self.is_speaking and self.engine:
self.engine.stop() # 会触发 finished-utterance 回调
# 清空队列
while not self.speech_queue.empty():
try:
self.speech_queue.get_nowait()
self.speech_queue.task_done()
except queue.Empty:
break
def is_available(self):
return self.engine is not None and not self._stop_event.is_set()
"""
【语音识别模块】Speech Recognition (Offline)
使用麦克风进行实时语音识别，基于 Vosk 离线模型
支持单次识别 & 持续监听模式
音量可视化、模型路径检查、资源安全释放
"""
import threading
import time
import logging
import json
import numpy as np
import os
from vosk import Model, KaldiRecognizer
import pyaudio
from database import config
from Progress.utils.logger_utils import log_time, log_step, log_var, log_call
from Progress.utils.logger_config import setup_logger
# --- 配置参数 ---
VOICE_TIMEOUT = config.timeout # 最大等待语音输入时间（秒）
VOICE_PHRASE_TIMEOUT = config.phrase_timeout # 单句话最长录音时间
VOSK_MODEL_PATH = "./vosk-model-small-cn-0.22" # 注意标准命名是 zh-cn
# --- 初始化日志器 ---
logger = logging.getLogger("ai_assistant")
class SpeechRecognizer:
def __init__(self):
self.model = None
self.recognizer = None
self.audio = None
self.is_listening = False
self.callback = None # 用户注册的回调函数：callback(text)
self._last_text = ""
self._listen_thread = None
self.sample_rate = 16000 # Vosk 要求采样率 16kHz
self.chunk_size = 1600 # 推荐帧大小（对应 ~100ms）
self._load_model()
self._init_audio_system()
@log_step("加载 Vosk 离线模型")
@log_time
def _load_model(self):
"""加载本地 Vosk 模型"""
if not os.path.exists(VOSK_MODEL_PATH):
raise FileNotFoundError(f"❌ Vosk 模型路径不存在: {VOSK_MODEL_PATH}\n"
"请从 https://alphacephei.com/vosk/models 下载中文小模型并解压至此路径")
try:
logger.info(f"📦 正在加载模型: {VOSK_MODEL_PATH}")
self.model = Model(VOSK_MODEL_PATH)
log_call("✅ 模型加载成功")
except Exception as e:
logger.critical(f"🔴 加载 Vosk 模型失败: {e}")
raise RuntimeError("Failed to load Vosk model") from e
@log_step("初始化音频系统")
@log_time
def _init_audio_system(self):
"""初始化 PyAudio 并创建全局 recognizer"""
try:
self.audio = pyaudio.PyAudio()
# 创建默认识别器（可在每次识别前 Reset）
self.recognizer = KaldiRecognizer(self.model, self.sample_rate)
logger.debug("✅ 音频系统初始化完成")
except Exception as e:
logger.exception("❌ 初始化音频系统失败")
raise
@property
def last_text(self) -> str:
return self._last_text
def is_available(self) -> bool:
"""检查麦克风是否可用"""
if not self.audio:
return False
try:
stream = self.audio.open(
format=pyaudio.paInt16,
channels=1,
rate=self.sample_rate,
input=True,
frames_per_buffer=self.chunk_size
)
stream.close()
return True
except Exception as e:
logger.error(f"🔴 麦克风不可用或无权限: {e}")
return False
@log_step("执行单次语音识别")
@log_time
def listen_and_recognize(self, timeout=None) -> str:
"""
单次语音识别：阻塞直到识别完成或超时
"""
timeout = timeout or VOICE_TIMEOUT
start_time = time.time()
in_speech = False
result_text = ""
logger.debug(f"🎙️ 开始单次语音识别 (timeout={timeout:.1f}s)...")
logger.info("🔊 请说话...")
stream = None
try:
# 打开音频流
stream = self.audio.open(
format=pyaudio.paInt16,
channels=1,
rate=self.sample_rate,
input=True,
frames_per_buffer=self.chunk_size
)
# 重置识别器状态
self.recognizer.Reset()
while (time.time() - start_time) < timeout:
data = stream.read(self.chunk_size, exception_on_overflow=False)
# 分析音量（可视化反馈）
audio_np = np.frombuffer(data, dtype=np.int16)
volume = np.abs(audio_np).mean()
bar = "█" * int(volume // 20)
logger.debug(f"📊 音量: {volume:5.1f} |{bar:10}|")
# 将音频送入 Vosk
if self.recognizer.AcceptWaveform(data):
final_result = json.loads(self.recognizer.Result())
text = final_result.get("text", "").strip()
if text:
result_text = text
break
else:
partial = json.loads(self.recognizer.PartialResult())
partial_text = partial.get("partial", "")
if partial_text.strip():
in_speech = True # 标记已经开始说话
# 如果还没开始说话，则允许超时；否则继续等待说完
if not in_speech and (time.time() - start_time) > timeout:
logger.info("💤 超时未检测到语音输入")
break
if result_text:
self._last_text = result_text
logger.info(f"🎯 识别结果: '{result_text}'")
return result_text
else:
logger.info("❓ 未识别到有效内容")
self._last_text = ""
return ""
except Exception as e:
logger.exception("🔴 执行单次语音识别时发生异常")
self._last_text = ""
return ""
finally:
# 确保资源释放
if stream:
try:
stream.stop_stream()
stream.close()
except Exception as e:
logger.warning(f"⚠️ 关闭音频流失败: {e}")
@log_step("启动持续语音监听")
def start_listening(self, callback=None, language=None):
"""
启动后台线程持续监听语音输入
:param callback: 回调函数，接受一个字符串参数 text
:param language: 语言代码（忽略，由模型决定）
"""
if self.is_listening:
logger.warning("⚠️ 已在监听中，忽略重复启动")
return
if not callable(callback):
logger.error("🔴 回调函数无效，请传入可调用对象")
return
self.callback = callback
self.is_listening = True
self._listen_thread = threading.Thread(target=self._background_listen, args=(language,), daemon=True)
self._listen_thread.start()
logger.info("🟢 已启动后台语音监听")
@log_step("停止语音监听")
def stop_listening(self):
"""安全停止后台监听"""
if not self.is_listening:
return
self.is_listening = False
logger.info("🛑 正在停止语音监听...")
if self._listen_thread and self._listen_thread != threading.current_thread():
self._listen_thread.join(timeout=3)
if self._listen_thread.is_alive():
logger.warning("🟡 监听线程未能及时退出（可能阻塞）")
elif self._listen_thread == threading.current_thread():
logger.error("❌ 无法在当前线程中 join 自己！请检查调用栈")
else:
logger.debug("No thread to join")
logger.info("✅ 语音监听已停止")
@log_time
def _background_listen(self, language=None):
"""后台循环监听线程"""
logger.debug("🎧 后台监听线程已启动")
stream = None
try:
stream = self.audio.open(
format=pyaudio.paInt16,
channels=1,
rate=self.sample_rate,
input=True,
frames_per_buffer=self.chunk_size
)
except Exception as e:
logger.error(f"🔴 无法打开音频流: {e}")
return
try:
while self.is_listening:
try:
data = stream.read(self.chunk_size, exception_on_overflow=False)
# 关键：先送入数据
if self.recognizer.AcceptWaveform(data):
# 已完成一句话识别
result_json = self.recognizer.Result()
result_dict = json.loads(result_json)
text = result_dict.get("text", "").strip()
if text and self.callback:
logger.info(f"🔔 回调触发: '{text}'")
self.callback(text)
# 🟢 完成后立即重置识别器状态
self.recognizer.Reset()
else:
# 获取部分结果（可用于实时显示）
partial = json.loads(self.recognizer.PartialResult())
partial_text = partial.get("partial", "")
if partial_text.strip():
logger.debug(f"🗣️ 当前语音片段: '{partial_text}'")
except Exception as e:
logger.exception("Background listening error")
time.sleep(0.05) # 小延迟降低 CPU 占用
finally:
if stream:
stream.stop_stream()
stream.close()
logger.debug("🔚 后台监听线程退出")
from functools import wraps
import inspect
import logging
from Progress.app.qwen_assistant import QWENAssistant
# 全局注册表
REGISTERED_FUNCTIONS = {}
FUNCTION_SCHEMA = []
FUNCTION_MAP = {} # (intent, action) -> method_name
_qwen_assistant = None
logger = logging.getLogger("ai_assistant")
def ai_callable(
*,
description: str,
params: dict,
intent: str = None,
action: str = None,
concurrent: bool = False # 新增：是否允许并发执行
):
def decorator(func):
func_name = func.__name__
metadata = {
"func": func,
"description": description,
"params": params,
"intent": intent,
"action": action,
"signature": str(inspect.signature(func)),
"concurrent": concurrent # 记录是否可并发
}
REGISTERED_FUNCTIONS[func_name] = metadata
FUNCTION_SCHEMA.append({
"name": func_name,
"description": description,
"parameters": params
})
if intent and action:
key = (intent, action)
if key in FUNCTION_MAP:
raise ValueError(f"冲突：语义 ({intent}, {action}) 已被函数 {FUNCTION_MAP[key]} 占用")
FUNCTION_MAP[key] = func_name
@wraps(func)
def wrapper(*args, **kwargs):
return func(*args, **kwargs)
wrapper._ai_metadata = metadata
return wrapper
return decorator
def get_qwen_assistant():
global _qwen_assistant
if _qwen_assistant is None:
_qwen_assistant = QWENAssistant()
return _qwen_assistant
def call_llm_to_choose_function(user_query: str) -> dict:
"""
使用 Qwen 助手理解用户意图，并返回完整的执行计划或单任务指令。
返回可能是：
- 单任务: {"function": "...", "arguments": {...}}
- 多任务: {"execution_plan": [...], "mode": "...", "response_to_user": "..."}
"""
assistant = get_qwen_assistant()
try:
ai_response = assistant.process_voice_command(user_query)
# 如果是聊天类，直接回复
if ai_response.get("intent") == "chat":
return {
"function": "_speak_response",
"arguments": {"message": ai_response.get("response_to_user", "好的")}
}
# 如果是文本生成任务
if ai_response.get("intent") == "text":
return {
"function": "_generate_text_task",
"arguments": {
"task_type": ai_response.get("action", "general"),
"prompt": user_query
}
}
# 如果是多任务计划
if "execution_plan" in ai_response and ai_response["execution_plan"]:
return ai_response # 直接透传给 orchestrator 处理
# 尝试从 FUNCTION_MAP 匹配单任务
intent = ai_response.get("intent")
action = ai_response.get("action")
params = ai_response.get("parameters", {})
func_name = FUNCTION_MAP.get((intent, action))
if func_name and func_name in REGISTERED_FUNCTIONS:
return {
"function": func_name,
"arguments": params
}
# 默认兜底
return {
"function": "_speak_response",
"arguments": {"message": ai_response.get("response_to_user", "抱歉，我无法完成这项操作。")}
}
except Exception as e:
logger.exception("❌ AI 决策出错")
return {}
"""
【AI语音助手】主程序入口
集成语音识别、Qwen 意图理解、TTS 与动作执行
✅ 已修复：不再访问 _last_text 私有字段
✅ 增强：异常防护、类型提示、唤醒词预留接口
"""
import sys
import time
import logging
# --- 导入日志工具 ---
from Progress.utils.logger_config import setup_logger
from Progress.utils.logger_utils import log_time, log_step, log_var, log_call
# --- 显式导入各模块核心类 ---
from Progress.app.voice_recognizer import SpeechRecognizer
from Progress.app.qwen_assistant import QWENAssistant
from Progress.app.text_to_speech import TTSEngine
from Progress.app.system_controller import SystemController, TaskOrchestrator
from database import config
# --- 初始化全局日志器 ---
logger = logging.getLogger("ai_assistant")
@log_step("初始化语音识别模块")
@log_time
def initialize_speech_recognizer() -> SpeechRecognizer:
try:
recognizer = SpeechRecognizer()
if not recognizer.is_available():
raise RuntimeError("麦克风不可用，请检查设备连接和权限")
log_call("✅ 语音识别器初始化完成")
return recognizer
except Exception as e:
logger.critical(f"🔴 初始化语音识别失败: {e}")
raise
@log_step("初始化 AI 助手模块")
@log_time
def initialize_qwen_assistant() -> QWENAssistant:
try:
assistant = QWENAssistant()
log_call("✅ Qwen 助手初始化完成")
return assistant
except Exception as e:
logger.critical(f"🔴 初始化 Qwen 助手失败: {e}")
raise
@log_step("初始化文本转语音模块")
@log_time
def initialize_tts_engine() -> TTSEngine:
try:
tts_engine = TTSEngine()
if not tts_engine.is_available():
raise RuntimeError("TTS引擎初始化失败")
log_call("✅ TTS 引擎初始化完成")
return tts_engine
except Exception as e:
logger.critical(f"🔴 初始化 TTS 失败: {e}")
raise
@log_step("初始化动作执行器")
@log_time
def initialize_action_executor() -> TaskOrchestrator:
system_controller = SystemController()
task_orchestrator = TaskOrchestrator(system_controller=system_controller)
log_call("✅ 动作执行器初始化完成")
return task_orchestrator
@log_step("安全执行单次交互")
@log_time
def handle_single_interaction_safe(
recognizer: SpeechRecognizer,
assistant: QWENAssistant,
tts_engine: TTSEngine,
executor: TaskOrchestrator
):
try:
handle_single_interaction(recognizer, assistant, tts_engine, executor)
except Exception as e:
logger.exception("⚠️ 单次交互过程中发生异常，已降级处理")
error_msg = "抱歉，我在处理刚才的操作时遇到了一点问题。"
logger.info(f"🗣️ 回复: {error_msg}")
tts_engine.speak(error_msg)
@log_step("处理一次语音交互")
@log_time
def handle_single_interaction(
recognizer: SpeechRecognizer,
assistant: QWENAssistant,
tts_engine: TTSEngine,
executor: TaskOrchestrator
):
"""
单次完整交互：听 -> 识别 -> AI 决策 -> 执行 -> 回复
"""
# 1. 听
text = recognizer.listen_and_recognize(timeout=5)
if not text:
logger.info("🔇 未检测到有效语音")
return
logger.info(f"🗣️ 用户说: '{text}'")
# 2. AI 决策 + 执行（一体化由 TaskOrchestrator 完成）
# 注意：这里不再需要单独调用 assistant.think_and_decide()
# 因为现在的 execute_from_ai_decision 内部已经集成了 LLM 调用
result = executor.execute_from_ai_decision(text)
# 3. 构造回复语句
if result["success"]:
ai_reply = str(result["message"])
logger.info(f"✅ 操作成功: {result['operation']} -> {ai_reply}")
else:
error_msg = result["message"]
ai_reply = f"抱歉，{error_msg if '抱歉' not in error_msg else error_msg[3:]}"
logger.warning(f"❌ 执行失败: {error_msg}")
# 4. 说
logger.info(f"🤖 回复: {ai_reply}")
tts_engine.speak(ai_reply)
@log_step("启动 AI 语音助手")
@log_time
def main():
logger.info("🚀 正在启动 AI 语音助手系统...")
try:
recognizer = initialize_speech_recognizer()
assistant = initialize_qwen_assistant()
tts_engine = initialize_tts_engine()
executor = initialize_action_executor()
log_call("✅ 所有模块初始化完成，进入监听循环")
log_call("\n" + "—" * 50)
log_call("🎙️ 语音助手已就绪")
log_call("💡 说出你的命令，例如：'打开浏览器'、'写一篇春天的文章'")
log_call("🛑 说出‘退出’、‘关闭’、‘停止’或‘拜拜’来结束程序")
log_call("—" * 50 + "\n")
while True:
try:
handle_single_interaction(recognizer, assistant, tts_engine, executor)
except KeyboardInterrupt:
logger.info("🛑 用户主动中断 (Ctrl+C)，准备退出...")
raise # 让 main 捕获并退出
except Exception as e:
logger.exception("⚠️ 单次交互过程中发生异常，已降级处理")
error_msg = "抱歉，我在处理刚才的操作时遇到了一点问题。"
logger.info(f"🗣️ 回复: {error_msg}")
tts_engine.speak(error_msg)
last_text = recognizer.last_text.lower()
exit_keywords = ['退出', '关闭', '停止', '拜拜', '再见']
if any(word in last_text for word in exit_keywords):
logger.info("🎯 用户请求退出，程序即将终止")
break
time.sleep(0.5)
logger.info("👋 语音助手已安全退出")
except KeyboardInterrupt:
logger.info("🛑 用户通过 Ctrl+C 中断程序")
print("\n👋 再见！")
except Exception as e:
logger.exception("❌ 主程序运行时发生未预期异常")
print(f"\n🚨 程序异常终止：{e}")
sys.exit(1)
if __name__ == "__main__":
if not logging.getLogger().handlers:
setup_logger(name="ai_assistant", log_dir="logs", level=logging.INFO)
main()
这里有一个严重问题，播放音乐并不会生成多任务，我需要在生成任何一个任务时自动附加一个语音播报
扫一扫
629
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
