diff --git a/audio_ai_chat/.env b/audio_ai_chat/.env new file mode 100644 index 0000000..6eda825 --- /dev/null +++ b/audio_ai_chat/.env @@ -0,0 +1,32 @@ +# 服务端口 +APP_PORT=8000 +# 日志级别（DEBUG/INFO/WARNING/ERROR） +LOG_LEVEL=INFO +# 日志文件路径 +LOG_FILE=logs/app.log + +# ASR服务配置 +ASR_SERVICE_URL=http://localhost:5000/asr +ASR_TIMEOUT=30 # 超时时间（秒） +ASR_RETRY_TIMES=2 # 重试次数 + +# LLM服务配置 +LLM_SERVICE_URL=http://localhost:6000/chat +LLM_TIMEOUT=60 +LLM_RETRY_TIMES=2 +LLM_MODEL=gpt-3.5-turbo # 可选：指定大模型版本 + +# TTS服务配置 +TTS_SERVICE_URL=http://localhost:7000/tts +TTS_TIMEOUT=30 +TTS_RETRY_TIMES=2 +TTS_VOICE=female # 可选：指定语音类型 + +# WebSocket配置 +WS_MAX_SIZE=10485760 # 最大消息大小（10MB） +WS_PING_INTERVAL=30 # 心跳检测间隔（秒） +WS_PING_TIMEOUT=10 # 心跳超时时间（秒） + +# 加密配置（ProtocolCodec依赖的密钥等） +ENCRYPT_KEY=your_secret_key_123 +ENCRYPT_IV=your_iv_456 \ No newline at end of file diff --git a/audio_ai_chat/.env.example b/audio_ai_chat/.env.example new file mode 100644 index 0000000..6eda825 --- /dev/null +++ b/audio_ai_chat/.env.example @@ -0,0 +1,32 @@ +# 服务端口 +APP_PORT=8000 +# 日志级别（DEBUG/INFO/WARNING/ERROR） +LOG_LEVEL=INFO +# 日志文件路径 +LOG_FILE=logs/app.log + +# ASR服务配置 +ASR_SERVICE_URL=http://localhost:5000/asr +ASR_TIMEOUT=30 # 超时时间（秒） +ASR_RETRY_TIMES=2 # 重试次数 + +# LLM服务配置 +LLM_SERVICE_URL=http://localhost:6000/chat +LLM_TIMEOUT=60 +LLM_RETRY_TIMES=2 +LLM_MODEL=gpt-3.5-turbo # 可选：指定大模型版本 + +# TTS服务配置 +TTS_SERVICE_URL=http://localhost:7000/tts +TTS_TIMEOUT=30 +TTS_RETRY_TIMES=2 +TTS_VOICE=female # 可选：指定语音类型 + +# WebSocket配置 +WS_MAX_SIZE=10485760 # 最大消息大小（10MB） +WS_PING_INTERVAL=30 # 心跳检测间隔（秒） +WS_PING_TIMEOUT=10 # 心跳超时时间（秒） + +# 加密配置（ProtocolCodec依赖的密钥等） +ENCRYPT_KEY=your_secret_key_123 +ENCRYPT_IV=your_iv_456 \ No newline at end of file diff --git a/audio_ai_chat/.idea/.gitignore b/audio_ai_chat/.idea/.gitignore new file mode 100644 index 0000000..10b731c --- /dev/null +++ b/audio_ai_chat/.idea/.gitignore @@ -0,0 +1,5 @@ +# 默认忽略的文件 +/shelf/ +/workspace.xml +# 基于编辑器的 HTTP 客户端请求 +/httpRequests/ diff --git a/audio_ai_chat/.idea/audio_ai_chat.iml b/audio_ai_chat/.idea/audio_ai_chat.iml new file mode 100644 index 0000000..4a03caf --- /dev/null +++ b/audio_ai_chat/.idea/audio_ai_chat.iml @@ -0,0 +1,10 @@ + + + + + + + + + + \ No newline at end of file diff --git a/audio_ai_chat/.idea/inspectionProfiles/Project_Default.xml b/audio_ai_chat/.idea/inspectionProfiles/Project_Default.xml new file mode 100644 index 0000000..7d133ac --- /dev/null +++ b/audio_ai_chat/.idea/inspectionProfiles/Project_Default.xml @@ -0,0 +1,12 @@ + + + + \ No newline at end of file diff --git a/audio_ai_chat/.idea/inspectionProfiles/profiles_settings.xml b/audio_ai_chat/.idea/inspectionProfiles/profiles_settings.xml new file mode 100644 index 0000000..105ce2d --- /dev/null +++ b/audio_ai_chat/.idea/inspectionProfiles/profiles_settings.xml @@ -0,0 +1,6 @@ + + + + \ No newline at end of file diff --git a/audio_ai_chat/.idea/misc.xml b/audio_ai_chat/.idea/misc.xml new file mode 100644 index 0000000..d6019ad --- /dev/null +++ b/audio_ai_chat/.idea/misc.xml @@ -0,0 +1,7 @@ + + + + + + \ No newline at end of file diff --git a/audio_ai_chat/.idea/modules.xml b/audio_ai_chat/.idea/modules.xml new file mode 100644 index 0000000..f7af7c8 --- /dev/null +++ b/audio_ai_chat/.idea/modules.xml @@ -0,0 +1,8 @@ + + + + + + + + \ No newline at end of file diff --git a/audio_ai_chat/.idea/vcs.xml b/audio_ai_chat/.idea/vcs.xml new file mode 100644 index 0000000..6c0b863 --- /dev/null +++ b/audio_ai_chat/.idea/vcs.xml @@ -0,0 +1,6 @@ + + + + + + \ No newline at end of file diff --git a/audio_ai_chat/README.md b/audio_ai_chat/README.md new file mode 100644 index 0000000..aeb6dc3 --- /dev/null +++ b/audio_ai_chat/README.md @@ -0,0 +1,71 @@ +# 语音AI对话系统（WebSocket版） +基于FastAPI实现，支持前端通过WebSocket推送语音数据，串联ASR（语音转文字）→ LLM（大模型对话）→ TTS（文字转语音）全流程，返回语音结果给前端。 + +## 快速启动 +1. 创建Conda虚拟环境：`conda create -n audio-ai-chat python=3.10 -y` +2. 激活环境：`conda activate audio-ai-chat` +3. 安装依赖：`pip install -r requirements.txt` +4. 配置环境变量：复制`.env.example`为`.env`，修改服务地址、密钥等配置 +5. 启动服务：`python run.py` +6. 接口文档：访问 `http://localhost:8000/docs`（FastAPI自动生成） + +## 目录说明 + audio_ai_chat/ # 项目根目录（包名，Python可导入） + ├── audio_ai_chat/ # 核心代码目录（与根目录同名，避免导入冲突） + │ ├── __init__.py # 包初始化文件（空文件即可） + │ ├── main.py # 项目启动类（FastAPI入口+WebSocket路由） + │ ├── config/ # 配置文件目录 + │ │ ├── __init__.py + │ │ ├── settings.py # 核心配置（ASR/LLM/TTS服务地址、WS参数等） + │ │ └── logger.py # 日志配置（格式、输出路径、级别） + │ ├── core/ # 核心业务逻辑（调整服务调用方式） + │ │ ├── __init__.py + │ │ ├── websocket_handler.py# 调整为：通过工厂类获取服务实例 + │ │ ├── asr/ # ASR服务目录（多版本实现） + │ │ │ ├── __init__.py + │ │ │ ├── base.py # ASR统一接口抽象类 + │ │ │ ├── version1.py # ASR版本1（如：本地离线版） + │ │ │ ├── version2.py # ASR版本2（如：百度云ASR） + │ │ │ └── factory.py # ASR工厂类（根据配置创建实例） + │ │ ├── llm/ # LLM服务目录（多版本实现） + │ │ │ ├── __init__.py + │ │ │ ├── base.py # LLM统一接口抽象类 + │ │ │ ├── openai_llm.py # LLM版本1（OpenAI） + │ │ │ ├── local_llm.py # LLM版本2（本地部署LLM） + │ │ │ └── factory.py # LLM工厂类 + │ │ └── tts/ # TTS服务目录（多版本实现） + │ │ ├── __init__.py + │ │ ├── base.py # TTS统一接口抽象类 + │ │ ├── ali_tts.py # TTS版本1（阿里云） + │ │ ├── pyttsx3_tts.py # TTS版本2（本地pyttsx3） + │ │ └── factory.py # TTS工厂类 + │ ├── codec/ # 数据编解码目录（存放加密/解密逻辑） + │ │ ├── __init__.py + │ │ └── ProtocolCodec.py # 已有：WS数据加密/解密、协议编解码 + │ ├── models/ # 数据模型目录（Pydantic/数据结构定义） + │ │ ├── __init__.py + │ │ └── ws_models.py # WebSocket消息结构（请求/响应模型、错误模型） + │ └── utils/ # 工具函数目录 + │ ├── __init__.py + │ ├── exceptions.py # 自定义异常（如服务调用失败、解码失败） + │ └── helpers.py # 通用工具（日志封装、异步重试、数据格式转换） + ├── logs/ # 日志输出目录（自动创建） + │ └── app.log # 应用日志文件（按配置滚动生成） + ├── .env # 环境变量文件（敏感配置，不提交Git） + ├── .env.example # 环境变量示例（提交Git，指导配置） + ├── requirements.txt # 依赖包清单 + ├── README.md # 项目说明（启动方式、配置说明、接口文档） + └── run.py # 项目启动脚本（简化启动命令） + +## 握手流程 + 前端 后端 + | | + |--- 建立 WebSocket 连接 ---->| + |<--- 连接接受（101状态码）---| + | | + |--- 发送身份信息（user_id+token）--->| + | | + |<--- 校验结果（成功/失败）---| + | | + |--- 发送业务数据（音频/文本）--->| + |<--- 推送业务结果（TTS/回复）---| \ No newline at end of file diff --git a/audio_ai_chat/audio_ai_chat/__init__.py b/audio_ai_chat/audio_ai_chat/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/audio_ai_chat/audio_ai_chat/__pycache__/__init__.cpython-310.pyc b/audio_ai_chat/audio_ai_chat/__pycache__/__init__.cpython-310.pyc new file mode 100644 index 0000000..2d4294d Binary files /dev/null and b/audio_ai_chat/audio_ai_chat/__pycache__/__init__.cpython-310.pyc differ diff --git a/audio_ai_chat/audio_ai_chat/__pycache__/main.cpython-310.pyc b/audio_ai_chat/audio_ai_chat/__pycache__/main.cpython-310.pyc new file mode 100644 index 0000000..7271d9f Binary files /dev/null and b/audio_ai_chat/audio_ai_chat/__pycache__/main.cpython-310.pyc differ diff --git a/python/WebSocketFrameHeader.py b/audio_ai_chat/audio_ai_chat/codec/ProtocolCodec.py similarity index 88% rename from python/WebSocketFrameHeader.py rename to audio_ai_chat/audio_ai_chat/codec/ProtocolCodec.py index 909f56c..de8b82e 100644 --- a/python/WebSocketFrameHeader.py +++ b/audio_ai_chat/audio_ai_chat/codec/ProtocolCodec.py @@ -5,7 +5,7 @@ import gzip from typing import Optional, Union, Dict, Any, List, Tuple -# -------------------------- 协议常量定义（与JS完全一致） -------------------------- +# -------------------------- 协议常量定义 -------------------------- class ProtocolConst: PROTOCOL_VERSION = 0b0001 # 协议版本（4位，字节0低4位，0~15） HEADER_SIZE = 8 # 头部固定字节数（字节0~7） @@ -13,13 +13,14 @@ class ProtocolConst: STRING_ENCODING = "utf-8" # 字符串默认编码 -# -------------------------- 枚举定义（与JS码值完全对齐） -------------------------- +# -------------------------- 枚举定义 -------------------------- class MessageType(IntEnum): """消息类型（4位，字节0高4位，0~15）""" PING = 0b0001 # 心跳（支持空包体） AUDIO_DATA = 0b0010 # 纯音频数据(pcm) TEXT_MESSAGE = 0b0011 # 纯文本消息 - CONTROL_CMD = 0b0100 # 控制指令（暂停/继续等） + CONTROL_CMD = 0b0100 # 控制指令 + IDENTITY = 0b0101 # 身份校验包 # 预留12种类型（0b0100 ~ 0b1111） @@ -161,11 +162,11 @@ class ProtocolCodec: return header + compressed_body @staticmethod - def unpack(packet: bytes) -> Tuple[MessageType, SerializationType, CompressionType, int, Any]: + def unpack(packet: bytes) -> Tuple[MessageType, int, Any]: """ 解析协议包（与JS unpack 方法完全兼容） :param packet: 完整协议包（头部 + 包体） - :return: (消息类型, 序列化方式, 压缩方式, 顺序号, 原始包体数据) + :return: (消息类型, 顺序号, 原始包体数据) # 仅返回核心必要信息 """ # 1. 校验包长度 if len(packet) < ProtocolConst.HEADER_SIZE: @@ -193,7 +194,7 @@ class ProtocolCodec: # 字节4~6：包体长度（24位大端序），字节7：保留位（忽略） body_len = (header[4] << 16) | (header[5] << 8) | header[6] - # 校验版本和包体长度 + # 校验版本和包体长度（校验逻辑保留，确保数据有效性） if version != ProtocolConst.PROTOCOL_VERSION: raise ValueError( f"协议版本不匹配：收到v{version}（0b{version:04b}），支持v{ProtocolConst.PROTOCOL_VERSION}（0b{ProtocolConst.PROTOCOL_VERSION:04b}）" @@ -236,16 +237,17 @@ class ProtocolCodec: else: raise ValueError(f"不支持的序列化方式：{serialization}") - return msg_type, serialization, compression, sequence, original_body + # 仅返回核心必要信息：消息类型、顺序号、原始包体数据 + return msg_type, sequence, original_body -# -------------------------- 使用示例（验证与JS兼容性） -------------------------- +# -------------------------- 使用示例（验证修改后功能正常） -------------------------- if __name__ == "__main__": # 示例1：PING消息（空包体，默认顺序号0） ping_packet = ProtocolCodec.pack(MessageType.PING) print(f"PING消息包长度：{len(ping_packet)}字节（仅头部）") - msg_type1, ser1, comp1, seq1, body1 = ProtocolCodec.unpack(ping_packet) - print(f"PING解析结果：类型={msg_type1.name}，序列化={ser1.name}，压缩={comp1.name}，顺序号={seq1}，包体={body1}\n") + msg_type1, seq1, body1 = ProtocolCodec.unpack(ping_packet) + print(f"PING解析结果：类型={msg_type1.name}，顺序号={seq1}，包体={body1}\n") # 示例2：纯文本消息（STRING序列化，指定顺序号） text_body = "Python与JS协议兼容测试（纯字符串）" @@ -256,8 +258,8 @@ if __name__ == "__main__": compression=CompressionType.NONE ) print(f"文本消息包长度：{len(text_packet)}字节") - msg_type2, ser2, comp2, seq2, body2 = ProtocolCodec.unpack(text_packet) - print(f"文本解析结果：类型={msg_type2.name}，序列化={ser2.name}，顺序号={seq2}，内容={body2}\n") + msg_type2, seq2, body2 = ProtocolCodec.unpack(text_packet) + print(f"文本解析结果：类型={msg_type2.name}，顺序号={seq2}，内容={body2}\n") # 示例3：控制指令（JSON序列化） control_body = {"cmd": ControlCommand.PAUSE.value, "reason": "用户主动暂停"} @@ -267,8 +269,8 @@ if __name__ == "__main__": sequence=1002 ) print(f"控制指令包长度：{len(control_packet)}字节") - msg_type3, ser3, comp3, seq3, body3 = ProtocolCodec.unpack(control_packet) - print(f"控制指令解析结果：类型={msg_type3.name}，序列化={ser3.name}，顺序号={seq3}，内容={body3}\n") + msg_type3, seq3, body3 = ProtocolCodec.unpack(control_packet) + print(f"控制指令解析结果：类型={msg_type3.name}，顺序号={seq3}，内容={body3}\n") # 示例4：音频数据（RAW序列化） audio_body = b"\x00\x01\x02\x03\x04\x05" * 100 # 模拟PCM数据 @@ -278,8 +280,8 @@ if __name__ == "__main__": sequence=1003 ) print(f"音频数据包长度：{len(audio_packet)}字节") - msg_type4, ser4, comp4, seq4, body4 = ProtocolCodec.unpack(audio_packet) - print(f"音频解析结果：类型={msg_type4.name}，序列化={ser4.name}，顺序号={seq4}，数据长度={len(body4)}字节\n") + msg_type4, seq4, body4 = ProtocolCodec.unpack(audio_packet) + print(f"音频解析结果：类型={msg_type4.name}，顺序号={seq4}，数据长度={len(body4)}字节\n") # 示例5：GZIP压缩测试（需JS端启用GZIP解压） long_text_body = "这是一段很长的文本，用于测试GZIP压缩效果" * 100 @@ -290,5 +292,5 @@ if __name__ == "__main__": compression=CompressionType.GZIP ) print(f"GZIP压缩后包长度：{len(gzip_packet)}字节（原始文本长度：{len(long_text_body.encode())}字节）") - msg_type5, ser5, comp5, seq5, body5 = ProtocolCodec.unpack(gzip_packet) - print(f"GZIP解析结果：类型={msg_type5.name}，压缩={comp5.name}，内容前50字：{body5[:50]}...") \ No newline at end of file + msg_type5, seq5, body5 = ProtocolCodec.unpack(gzip_packet) + print(f"GZIP解析结果：类型={msg_type5.name}，顺序号={seq5}，内容前50字：{body5[:50]}...") \ No newline at end of file diff --git a/audio_ai_chat/audio_ai_chat/codec/__init__.py b/audio_ai_chat/audio_ai_chat/codec/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/audio_ai_chat/audio_ai_chat/codec/__pycache__/ProtocolCodec.cpython-310.pyc b/audio_ai_chat/audio_ai_chat/codec/__pycache__/ProtocolCodec.cpython-310.pyc new file mode 100644 index 0000000..94f60fc Binary files /dev/null and b/audio_ai_chat/audio_ai_chat/codec/__pycache__/ProtocolCodec.cpython-310.pyc differ diff --git a/audio_ai_chat/audio_ai_chat/codec/__pycache__/__init__.cpython-310.pyc b/audio_ai_chat/audio_ai_chat/codec/__pycache__/__init__.cpython-310.pyc new file mode 100644 index 0000000..2e10ddf Binary files /dev/null and b/audio_ai_chat/audio_ai_chat/codec/__pycache__/__init__.cpython-310.pyc differ diff --git a/audio_ai_chat/audio_ai_chat/config/__init__.py b/audio_ai_chat/audio_ai_chat/config/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/audio_ai_chat/audio_ai_chat/config/__pycache__/__init__.cpython-310.pyc b/audio_ai_chat/audio_ai_chat/config/__pycache__/__init__.cpython-310.pyc new file mode 100644 index 0000000..4a3dac3 Binary files /dev/null and b/audio_ai_chat/audio_ai_chat/config/__pycache__/__init__.cpython-310.pyc differ diff --git a/audio_ai_chat/audio_ai_chat/config/__pycache__/logger.cpython-310.pyc b/audio_ai_chat/audio_ai_chat/config/__pycache__/logger.cpython-310.pyc new file mode 100644 index 0000000..3874682 Binary files /dev/null and b/audio_ai_chat/audio_ai_chat/config/__pycache__/logger.cpython-310.pyc differ diff --git a/audio_ai_chat/audio_ai_chat/config/__pycache__/settings.cpython-310.pyc b/audio_ai_chat/audio_ai_chat/config/__pycache__/settings.cpython-310.pyc new file mode 100644 index 0000000..caedb7c Binary files /dev/null and b/audio_ai_chat/audio_ai_chat/config/__pycache__/settings.cpython-310.pyc differ diff --git a/audio_ai_chat/audio_ai_chat/config/logger.py b/audio_ai_chat/audio_ai_chat/config/logger.py new file mode 100644 index 0000000..77f2653 --- /dev/null +++ b/audio_ai_chat/audio_ai_chat/config/logger.py @@ -0,0 +1,28 @@ +from loguru import logger +import sys +from audio_ai_chat.config.settings import settings + +# 移除默认日志输出 +logger.remove() + +# 添加控制台输出（开发环境） +logger.add( + sink=sys.stdout, + level=settings.LOG_LEVEL, + format="{time:YYYY-MM-DD HH:mm:ss.SSS} | {level: <8} | {name}:{function}:{line} - {message}", + enqueue=True, # 异步日志，提高性能 +) + +# 添加文件输出（按大小滚动，保留10个文件，每个文件最大50MB） +logger.add( + sink=settings.LOG_FILE, + level=settings.LOG_LEVEL, + format="{time:YYYY-MM-DD HH:mm:ss.SSS} | {level: <8} | {name}:{function}:{line} - {message}", + rotation="50 MB", + retention=10, + compression="zip", + enqueue=True, +) + +# 导出logger供其他模块使用 +__all__ = ["logger"] \ No newline at end of file diff --git a/audio_ai_chat/audio_ai_chat/config/settings.py b/audio_ai_chat/audio_ai_chat/config/settings.py new file mode 100644 index 0000000..b99fa65 --- /dev/null +++ b/audio_ai_chat/audio_ai_chat/config/settings.py @@ -0,0 +1,70 @@ +from pydantic_settings import BaseSettings, SettingsConfigDict +from pydantic import Field +from pathlib import Path + +ROOT_DIR = Path(__file__).parent.parent.parent + +class Settings(BaseSettings): + # 应用配置 + APP_PORT: int = Field(default=8000, description="服务端口") + LOG_LEVEL: str = Field(default="INFO", description="日志级别") + LOG_FILE: Path = Field(default=ROOT_DIR / "logs/app.log", description="日志文件路径") + + # ASR服务配置 + ASR_SERVICE_URL: str = Field(..., description="ASR服务地址") + ASR_TIMEOUT: int = Field(default=30, description="ASR超时时间（秒）") + ASR_RETRY_TIMES: int = Field(default=2, description="ASR重试次数") + + # LLM服务配置 + LLM_SERVICE_URL: str = Field(..., description="LLM服务地址") + LLM_TIMEOUT: int = Field(default=60, description="LLM超时时间（秒）") + LLM_RETRY_TIMES: int = Field(default=2, description="LLM重试次数") + LLM_MODEL: str = Field(default="default", description="LLM模型版本") + + # TTS服务配置 + TTS_SERVICE_URL: str = Field(..., description="TTS服务地址") + TTS_TIMEOUT: int = Field(default=30, description="TTS超时时间（秒）") + TTS_RETRY_TIMES: int = Field(default=2, description="TTS重试次数") + TTS_VOICE: str = Field(default="default", description="TTS语音类型") + + # WebSocket配置 + WS_MAX_SIZE: int = Field(default=10 * 1024 * 1024, description="WS最大消息大小（字节）") + WS_PING_INTERVAL: int = Field(default=30, description="WS心跳间隔（秒）") + WS_PING_TIMEOUT: int = Field(default=10, description="WS心跳超时（秒）") + + # 加密配置 + ENCRYPT_KEY: str = Field(..., description="数据加密密钥") + ENCRYPT_IV: str = Field(..., description="数据加密向量") + + # -------------------------- 新增：服务版本配置 -------------------------- + # ASR当前使用版本（对应ASR_REGISTRY中的key） + ASR_CURRENT_VERSION: str = Field(default="local_v1", description="ASR服务当前版本") + # 本地ASR专属配置（仅local_v1版本使用） + LOCAL_ASR_MODEL_PATH: Path = Field(default=ROOT_DIR / "models/asr/local_model", description="本地ASR模型路径") + # 百度云ASR专属配置（仅baidu_v2版本使用） + BAIDU_ASR_API_KEY: str = Field(default="", description="百度云ASR API Key") + BAIDU_ASR_SECRET_KEY: str = Field(default="", description="百度云ASR Secret Key") + + # LLM当前使用版本（对应LLM_REGISTRY中的key） + LLM_CURRENT_VERSION: str = Field(default="local", description="LLM服务当前版本") + # OpenAI LLM专属配置（仅openai版本使用） + OPENAI_API_KEY: str = Field(default="", description="OpenAI API Key") + OPENAI_BASE_URL: str = Field(default="https://api.openai.com/v1", description="OpenAI接口地址") + # 本地LLM专属配置（仅local版本使用） + LOCAL_LLM_MODEL_PATH: Path = Field(default=ROOT_DIR / "models/llm/local_model", description="本地LLM模型路径") + + # TTS当前使用版本（对应TTS_REGISTRY中的key） + TTS_CURRENT_VERSION: str = Field(default="pyttsx3", description="TTS服务当前版本") + # 阿里云TTS专属配置（仅aliyun版本使用） + ALI_TTS_ACCESS_KEY: str = Field(default="", description="阿里云TTS Access Key") + ALI_TTS_ACCESS_SECRET: str = Field(default="", description="阿里云TTS Access Secret") + ALI_TTS_REGION_ID: str = Field(default="cn-hangzhou", description="阿里云TTS区域ID") + + # 原有配置...（WS_MAX_SIZE、ENCRYPT_KEY等） + + model_config = SettingsConfigDict(env_file=".env", env_file_encoding="utf-8") + +settings = Settings() +# 确保模型目录存在（本地版本需要） +# settings.LOCAL_ASR_MODEL_PATH.parent.mkdir(parents=True, exist_ok=True) +# settings.LOCAL_LLM_MODEL_PATH.parent.mkdir(parents=True, exist_ok=True) \ No newline at end of file diff --git a/audio_ai_chat/audio_ai_chat/core/__init__.py b/audio_ai_chat/audio_ai_chat/core/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/audio_ai_chat/audio_ai_chat/core/__pycache__/__init__.cpython-310.pyc b/audio_ai_chat/audio_ai_chat/core/__pycache__/__init__.cpython-310.pyc new file mode 100644 index 0000000..f7ab12e Binary files /dev/null and b/audio_ai_chat/audio_ai_chat/core/__pycache__/__init__.cpython-310.pyc differ diff --git a/audio_ai_chat/audio_ai_chat/core/__pycache__/connection.cpython-310.pyc b/audio_ai_chat/audio_ai_chat/core/__pycache__/connection.cpython-310.pyc new file mode 100644 index 0000000..4a03bd7 Binary files /dev/null and b/audio_ai_chat/audio_ai_chat/core/__pycache__/connection.cpython-310.pyc differ diff --git a/audio_ai_chat/audio_ai_chat/core/__pycache__/websocket_handler.cpython-310.pyc b/audio_ai_chat/audio_ai_chat/core/__pycache__/websocket_handler.cpython-310.pyc new file mode 100644 index 0000000..94e62b8 Binary files /dev/null and b/audio_ai_chat/audio_ai_chat/core/__pycache__/websocket_handler.cpython-310.pyc differ diff --git a/audio_ai_chat/audio_ai_chat/core/asr/__init__.py b/audio_ai_chat/audio_ai_chat/core/asr/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/audio_ai_chat/audio_ai_chat/core/asr/__pycache__/__init__.cpython-310.pyc b/audio_ai_chat/audio_ai_chat/core/asr/__pycache__/__init__.cpython-310.pyc new file mode 100644 index 0000000..331d32e Binary files /dev/null and b/audio_ai_chat/audio_ai_chat/core/asr/__pycache__/__init__.cpython-310.pyc differ diff --git a/audio_ai_chat/audio_ai_chat/core/asr/__pycache__/base.cpython-310.pyc b/audio_ai_chat/audio_ai_chat/core/asr/__pycache__/base.cpython-310.pyc new file mode 100644 index 0000000..3dce723 Binary files /dev/null and b/audio_ai_chat/audio_ai_chat/core/asr/__pycache__/base.cpython-310.pyc differ diff --git a/audio_ai_chat/audio_ai_chat/core/asr/__pycache__/factory.cpython-310.pyc b/audio_ai_chat/audio_ai_chat/core/asr/__pycache__/factory.cpython-310.pyc new file mode 100644 index 0000000..cd9c8db Binary files /dev/null and b/audio_ai_chat/audio_ai_chat/core/asr/__pycache__/factory.cpython-310.pyc differ diff --git a/audio_ai_chat/audio_ai_chat/core/asr/base.py b/audio_ai_chat/audio_ai_chat/core/asr/base.py new file mode 100644 index 0000000..5f941cc --- /dev/null +++ b/audio_ai_chat/audio_ai_chat/core/asr/base.py @@ -0,0 +1,25 @@ +from abc import ABC, abstractmethod +from typing import Optional, Coroutine +from audio_ai_chat.config.settings import settings + +class ASRBase(ABC): + """ASR服务统一抽象接口""" + def __init__(self): + # 公共配置（所有ASR版本共享的超时、重试次数等） + self.timeout = settings.ASR_TIMEOUT + self.retry_times = settings.ASR_RETRY_TIMES + + @abstractmethod + async def recognize( + self, + voice_data: bytes, + user_id: Optional[str] = None, + **kwargs # 兼容不同版本的额外参数 + ) -> str: + """ + 语音识别核心方法（所有ASR版本必须实现） + :param voice_data: 语音二进制数据 + :param user_id: 用户ID（可选） + :return: 识别后的文本 + """ + pass \ No newline at end of file diff --git a/audio_ai_chat/audio_ai_chat/core/asr/factory.py b/audio_ai_chat/audio_ai_chat/core/asr/factory.py new file mode 100644 index 0000000..7e52290 --- /dev/null +++ b/audio_ai_chat/audio_ai_chat/core/asr/factory.py @@ -0,0 +1,26 @@ +from typing import Type +from audio_ai_chat.config.settings import settings +from audio_ai_chat.utils.exceptions import ServiceCallError +from .base import ASRBase +# from .version1 import LocalOfflineASR +# from .version2 import BaiduASR +# +# # 注册所有ASR版本：key=配置中的版本名，value=对应的类 +# ASR_REGISTRY: dict[str, Type[ASRBase]] = { +# "local_v1": LocalOfflineASR, +# "baidu_v2": BaiduASR, +# # 新增版本时，只需在这里注册："新版本名": 新类名 +# } + +# class ASRFactory: +# """ASR服务工厂类：根据配置创建对应版本的实例""" +# @staticmethod +# def get_asr_client() -> ASRBase: +# # 从配置中获取当前指定的ASR版本 +# current_version = settings.ASR_CURRENT_VERSION +# if current_version not in ASR_REGISTRY: +# raise ServiceCallError( +# f"不支持的ASR版本：{current_version}，可选版本：{list(ASR_REGISTRY.keys())}" +# ) +# # 创建并返回对应版本的实例 +# return ASR_REGISTRY[current_version]() \ No newline at end of file diff --git a/audio_ai_chat/audio_ai_chat/core/connection.py b/audio_ai_chat/audio_ai_chat/core/connection.py new file mode 100644 index 0000000..f2e0098 --- /dev/null +++ b/audio_ai_chat/audio_ai_chat/core/connection.py @@ -0,0 +1,145 @@ +from typing import Optional, Dict, Any, List +import asyncio +from audio_ai_chat.config.logger import logger +# from audio_ai_chat.core.llm.factory import LLMFactory +from audio_ai_chat.core.llm.base import LLMBase + + +class ConnectionContext: + """ + 单个WebSocket连接的上下文管理器：封装用户信息、大模型Session、消息队列等资源 + 每个WebSocket连接对应一个该类实例，确保资源隔离 + """ + + def __init__(self, client_id: str): + """ + 初始化连接上下文 + :param client_id: WebSocket连接唯一标识（如id(websocket)） + :param user_id: 用户唯一标识（从前端请求中获取） + """ + + self.client_id = client_id # 连接唯一ID + self.created_at = asyncio.get_event_loop().time() # 连接创建时间 + + # 1. 大模型独立Session（每个连接创建一个新的LLM客户端实例） + # self.llm_session: LLMBase = LLMFactory.get_llm_client() # 独立Session + self.chat_history: List[Dict[str, str]] = [] # 该连接的对话历史（[(user: "...", assistant: "..."), ...]） + + # 2. 异步消息队列（用于缓存TTS结果，有序推送给前端） + self.message_queue: asyncio.Queue[bytes] = asyncio.Queue() + + # 3. 连接状态（可选：如是否正在处理请求、是否断开等） + self.is_active: bool = True + self.is_processing: bool = False + self.name = None + self.user_id = None + self.token = None + + def set_user_info(self, token: str, user_id: str, name: str = "匿名用户"): + """ + 二次设置用户信息（身份校验通过后调用） + :param token: + :param user_id: 用户唯一标识（必填） + :param name: 用户名（可选，默认匿名） + """ + if not user_id: + raise ValueError("user_id不能为空") + self.name = name + self.user_id = user_id + self.token = token + logger.debug(f"客户端 {self.client_id} 设置用户信息：user_id={user_id}, name={name}") + + async def add_message_to_queue(self, message: bytes): + """将TTS结果添加到消息队列（异步安全）""" + if not self.is_active: + raise ValueError(f"连接已断开（client_id={self.client_id}），无法添加消息") + await self.message_queue.put(message) + logger.debug(f"消息队列添加数据：client_id={self.client_id}，队列长度={self.message_queue.qsize()}") + + async def get_message_from_queue(self) -> Optional[bytes]: + """从消息队列获取消息（异步阻塞，直到有消息或连接断开）""" + try: + # 超时时间：30秒（避免无限阻塞） + return await asyncio.wait_for(self.message_queue.get(), timeout=30.0) + except asyncio.TimeoutError: + logger.debug(f"消息队列超时：client_id={self.client_id}，无新消息") + return None + + def update_chat_history(self, user_text: str, assistant_text: str): + """更新该连接的对话历史""" + self.chat_history.append({ + "user": user_text, + "assistant": assistant_text + }) + # 可选：限制历史长度（避免内存溢出） + if len(self.chat_history) > 50: + self.chat_history.pop(0) # 删除最早的历史 + + def close(self): + """关闭连接上下文，释放资源""" + self.is_active = False + self.is_processing = False + # 清空消息队列（可选） + while not self.message_queue.empty(): + try: + self.message_queue.get_nowait() + except asyncio.QueueEmpty: + break + logger.info(f"连接上下文已关闭：client_id={self.client_id}，user_id={self.user_id}") + + def __del__(self): + """析构函数：确保资源释放""" + self.close() + + +class ConnectionManager: + """ + WebSocket连接全局管理器：维护所有活跃连接的上下文 + 提供创建、查询、删除连接上下文的接口（线程/异步安全） + """ + + def __init__(self): + # 存储所有活跃连接：key=client_id（int），value=ConnectionContext实例 + self.connections: Dict[int, ConnectionContext] = {} + # 异步锁：确保多连接并发操作时的数据安全 + self._lock = asyncio.Lock() + + async def create_connection(self, client_id: int, user_id: Optional[str] = None) -> ConnectionContext: + """创建新的连接上下文（线程安全）""" + async with self._lock: + # 避免重复创建（同一client_id不会重复连接） + if client_id in self.connections: + logger.warning(f"连接已存在：client_id={client_id}，将覆盖旧连接") + self.connections[client_id].close() + + # 创建新的连接上下文（包含独立LLM Session和消息队列） + context = ConnectionContext(client_id=client_id, user_id=user_id) + self.connections[client_id] = context + logger.info( + f"创建新连接上下文：client_id={client_id}，user_id={user_id}，当前活跃连接数={len(self.connections)}") + return context + + async def get_connection(self, client_id: int) -> Optional[ConnectionContext]: + """获取指定client_id的连接上下文（线程安全）""" + async with self._lock: + context = self.connections.get(client_id) + if context and not context.is_active: + # 清理已断开的连接 + del self.connections[client_id] + return None + return context + + async def remove_connection(self, client_id: int): + """删除连接上下文（线程安全）""" + async with self._lock: + context = self.connections.pop(client_id, None) + if context: + context.close() + logger.info(f"移除连接上下文：client_id={client_id}，当前活跃连接数={len(self.connections)}") + + async def get_active_connections_count(self) -> int: + """获取当前活跃连接数（线程安全）""" + async with self._lock: + # 过滤已断开的连接 + self.connections = {k: v for k, v in self.connections.items() if v.is_active} + return len(self.connections) \ No newline at end of file diff --git a/audio_ai_chat/audio_ai_chat/core/llm/__init__.py b/audio_ai_chat/audio_ai_chat/core/llm/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/audio_ai_chat/audio_ai_chat/core/llm/__pycache__/__init__.cpython-310.pyc b/audio_ai_chat/audio_ai_chat/core/llm/__pycache__/__init__.cpython-310.pyc new file mode 100644 index 0000000..c2daf54 Binary files /dev/null and b/audio_ai_chat/audio_ai_chat/core/llm/__pycache__/__init__.cpython-310.pyc differ diff --git a/audio_ai_chat/audio_ai_chat/core/llm/__pycache__/base.cpython-310.pyc b/audio_ai_chat/audio_ai_chat/core/llm/__pycache__/base.cpython-310.pyc new file mode 100644 index 0000000..ab938f1 Binary files /dev/null and b/audio_ai_chat/audio_ai_chat/core/llm/__pycache__/base.cpython-310.pyc differ diff --git a/audio_ai_chat/audio_ai_chat/core/llm/__pycache__/factory.cpython-310.pyc b/audio_ai_chat/audio_ai_chat/core/llm/__pycache__/factory.cpython-310.pyc new file mode 100644 index 0000000..6f0a486 Binary files /dev/null and b/audio_ai_chat/audio_ai_chat/core/llm/__pycache__/factory.cpython-310.pyc differ diff --git a/audio_ai_chat/audio_ai_chat/core/llm/base.py b/audio_ai_chat/audio_ai_chat/core/llm/base.py new file mode 100644 index 0000000..48087c4 --- /dev/null +++ b/audio_ai_chat/audio_ai_chat/core/llm/base.py @@ -0,0 +1,26 @@ +from abc import ABC, abstractmethod +from typing import Optional, Coroutine + +class LLMBase(ABC): + """LLM服务统一抽象接口""" + def __init__(self): + self.timeout = settings.LLM_TIMEOUT + self.retry_times = settings.LLM_RETRY_TIMES + self.model = settings.LLM_MODEL # 模型版本（不同LLM可能支持不同模型） + + @abstractmethod + async def chat( + self, + text: str, + user_id: Optional[str] = None, + history: Optional[list] = None, # 对话历史（部分LLM支持） + **kwargs + ) -> str: + """ + 大模型对话核心方法 + :param text: 用户输入文本（ASR识别结果） + :param user_id: 用户ID（可选） + :param history: 对话历史（可选，格式：[(用户输入, 模型回答), ...]） + :return: 模型生成的回答文本 + """ + pass \ No newline at end of file diff --git a/audio_ai_chat/audio_ai_chat/core/llm/factory.py b/audio_ai_chat/audio_ai_chat/core/llm/factory.py new file mode 100644 index 0000000..548e2c3 --- /dev/null +++ b/audio_ai_chat/audio_ai_chat/core/llm/factory.py @@ -0,0 +1,22 @@ +from typing import Type +from audio_ai_chat.config.settings import settings +from audio_ai_chat.utils.exceptions import ServiceCallError +from .base import LLMBase +# +# from .openai_llm import OpenAILLM +# from .local_llm import LocalLLM +# +# LLM_REGISTRY: dict[str, Type[LLMBase]] = { +# "openai": OpenAILLM, +# "local": LocalLLM, +# } +# +# class LLMFactory: +# @staticmethod +# def get_llm_client() -> LLMBase: +# current_version = settings.LLM_CURRENT_VERSION +# if current_version not in LLM_REGISTRY: +# raise ServiceCallError( +# f"不支持的LLM版本：{current_version}，可选版本：{list(LLM_REGISTRY.keys())}" +# ) +# return LLM_REGISTRY[current_version]() \ No newline at end of file diff --git a/audio_ai_chat/audio_ai_chat/core/tts/__init__.py b/audio_ai_chat/audio_ai_chat/core/tts/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/audio_ai_chat/audio_ai_chat/core/tts/base.py b/audio_ai_chat/audio_ai_chat/core/tts/base.py new file mode 100644 index 0000000..9721b67 --- /dev/null +++ b/audio_ai_chat/audio_ai_chat/core/tts/base.py @@ -0,0 +1,26 @@ +from abc import ABC, abstractmethod +from typing import Optional, Coroutine + +class TTSBase(ABC): + """TTS服务统一抽象接口""" + def __init__(self): + self.timeout = settings.TTS_TIMEOUT + self.retry_times = settings.TTS_RETRY_TIMES + self.voice = settings.TTS_VOICE # 语音类型 + + @abstractmethod + async def synthesize( + self, + text: str, + user_id: Optional[str] = None, + speed: float = 1.0, # 语速（默认1.0） + **kwargs + ) -> bytes: + """ + 文本转语音核心方法 + :param text: 待合成文本（LLM回答结果） + :param user_id: 用户ID（可选） + :param speed: 语速（0.5~2.0） + :return: 语音二进制数据 + """ + pass \ No newline at end of file diff --git a/audio_ai_chat/audio_ai_chat/core/tts/factory.py b/audio_ai_chat/audio_ai_chat/core/tts/factory.py new file mode 100644 index 0000000..b381fdb --- /dev/null +++ b/audio_ai_chat/audio_ai_chat/core/tts/factory.py @@ -0,0 +1,22 @@ +from typing import Type +from audio_ai_chat.config.settings import settings +from audio_ai_chat.utils.exceptions import ServiceCallError +from .base import TTSBase +# +# from .ali_tts import AliTTS +# from .pyttsx3_tts import Pyttsx3TTS +# +# TTS_REGISTRY: dict[str, Type[TTSBase]] = { +# "aliyun": AliTTS, +# "pyttsx3": Pyttsx3TTS, +# } +# +# class TTSFactory: +# @staticmethod +# def get_tts_client() -> TTSBase: +# current_version = settings.TTS_CURRENT_VERSION +# if current_version not in TTS_REGISTRY: +# raise ServiceCallError( +# f"不支持的TTS版本：{current_version}，可选版本：{list(TTS_REGISTRY.keys())}" +# ) +# return TTS_REGISTRY[current_version]() \ No newline at end of file diff --git a/audio_ai_chat/audio_ai_chat/core/websocket_handler.py b/audio_ai_chat/audio_ai_chat/core/websocket_handler.py new file mode 100644 index 0000000..cd8b4a4 --- /dev/null +++ b/audio_ai_chat/audio_ai_chat/core/websocket_handler.py @@ -0,0 +1,302 @@ +from fastapi import WebSocket +from typing import Dict, List, Optional + +from pyexpat.errors import messages + +from audio_ai_chat.config.logger import logger +# from audio_ai_chat.core.asr.factory import ASRFactory # 导入ASR工厂 +# from audio_ai_chat.core.llm.factory import LLMFactory # 导入LLM工厂 +# from audio_ai_chat.core.tts.factory import TTSFactory # 导入TTS工厂 +from audio_ai_chat.config.settings import settings +from audio_ai_chat.utils.exceptions import ServiceCallError +from fastapi import WebSocket, WebSocketDisconnect +from audio_ai_chat.codec.ProtocolCodec import ProtocolCodec +from typing import Optional, Dict, Callable, Awaitable, List, Any, Coroutine +from dataclasses import dataclass, field +import sys +import json +import asyncio +import websockets +import uuid +import logging +from audio_ai_chat.core.connection import ConnectionContext +from audio_ai_chat.codec.ProtocolCodec import ProtocolCodec, MessageType + + +class WebSocketConnectionManager: + """WebSocket连接管理器""" + + def __init__(self): + # 活跃连接列表 + self.active_connections: List[WebSocket] = [] + # 关键映射：client_id -> ConnectionContext（快速获取用户专属上下文） + self.client_context_map: Dict[str, ConnectionContext] = {} + # self.asr_client = ASRFactory.get_asr_client() + # self.tts_client = TTSFactory.get_tts_client() + # 用户LLM会话存储 + # self.user_llm_conversations: Dict[str, LLMConversation] = {} + # 全局唤醒事件 + self.consume_wakeup = asyncio.Event() + + async def connect(self, websocket: WebSocket) -> ConnectionContext: + """ + 建立连接+身份校验（前端主动发送身份信息） + 超时逻辑：5秒内未收到前端身份信息，自动关闭连接 + """ + # 1. 接受连接 + await websocket.accept() + self.active_connections.append(websocket) + # 生成唯一标识 + client_id = str(id(websocket)) + context = ConnectionContext( + client_id=client_id + ) + logger.info( + f"连接 {client_id} 已接受，等待前端发送身份信息（5秒超时）...，当前连接数: {len(self.active_connections)}") + + try: + # 2. 超时控制：5秒内未收到身份信息，关闭连接 + try: + # 关键：用 receive_bytes 接收二进制包（而非 receive_json） + ping_packet = await asyncio.wait_for( + websocket.receive_bytes(), + timeout=5.0 # 身份校验超时时间 + ) + except asyncio.TimeoutError: + # 发送超时错误响应（二进制格式） + # todo 告知前端失败 + # await websocket.close(code=1008) # todo 错误码值待定 + # self.active_connections.remove(websocket) + raise TimeoutError(f"连接 {client_id} 身份包超时") + + # 3. 解包二进制包 + msg_type, _, body = ProtocolCodec.unpack(ping_packet) + if msg_type != MessageType.IDENTITY: + raise ValueError(f"连接 {client_id} 首个包类型错误（期望1，实际{MessageType.IDENTITY}）") + + logger.debug(f"解包成功：, body={body}") + # 4. 提取并校验身份信息（核心：按需扩展校验逻辑） + identity_data = json.loads(body.decode("utf-8")) + user_id = identity_data.get("user_id") + token = identity_data.get("token") + name = identity_data.get("name") + # todo 校验身份信息 raise PermissionError(f"用户 {user_id} 身份校验失败") + context.set_user_info(token, user_id, name) + self.client_context_map[client_id] = context + # 5. 响应前端：校验成功 + await context.add_message_to_queue( + message=ProtocolCodec.pack(MessageType.IDENTITY, {'messages': '身份校验成功，连接已就绪'})) + + logger.info(f"用户 {user_id}（{name}）身份校验通过，连接就绪（client_id: {client_id}）") + return context + except: + pass + + # 初始化LLM会话 + # self._init_llm_conversation(user_id) + + # return conn_id, user_id, conn_id # conn_id 同时作为 tts_session_id + + # def _init_llm_conversation(self, user_id: str): + # """初始化用户LLM会话""" + # if user_id not in self.user_llm_conversations: + # self.user_llm_conversations[user_id] = LLMConversation( + # user_id=user_id, + # scene_description="语音识别对话场景" + # ) + + def disconnect(self, websocket: WebSocket, conn_id: str): + """断开连接并清理资源""" + if websocket in self.active_connections: + self.active_connections.remove(websocket) + logger.info(f"连接 {conn_id} 已断开，当前连接数: {len(self.active_connections)}") + + # async def setup_tts_manager(self, result_queue: asyncio.Queue) -> TTSManager: + # """初始化TTS管理器""" + # + # def handle_tts_result(req_id: str, result: Dict[str, Any]): + # """TTS结果回调处理""" + # try: + # status = result.get("status") + # if status == "completed": + # audio_data = result.get("audio_data") + # if audio_data is not None and len(audio_data) > 0: + # # 转换为PCM格式 + # pcm_data = (audio_data.astype(np.float32) * 32767).astype(np.int16) + # pcm_bytes = pcm_data.tobytes() + # result_queue.put_nowait(pcm_bytes) + # except Exception as e: + # logger.error(f"TTS结果处理失败: {str(e)}") + # + # self.tts_client = TTSFactory.get_tts_client() + # tts_manager = TTSManager() + # tts_manager.set_result_callback(handle_tts_result) + # tts_manager.set_playback_enabled(False) + # await tts_manager.initialize() + # return tts_manager + + async def asr_result_callback(self, result: dict, websocket: WebSocket, + user_id: str, result_queue: asyncio.Queue): + """ASR结果回调处理""" + try: + logger.info(f"ASR识别结果: {result}") + final_asr_text = result.get("text", "").strip() + + # 转发ASR结果到前端队列 + if final_asr_text: + print(f"插入ASR结果时队列大小: {result_queue.qsize()}") + self.consume_wakeup.set() # 唤醒消费协程 + + # 异步调用大模型 + llm_conversation = self.user_llm_conversations.get(user_id) + if llm_conversation: + asyncio.create_task( + self.call_llm_and_send( + query=final_asr_text, + conversation=llm_conversation, + websocket=websocket + ) + ) + except Exception as e: + logger.error(f"ASR回调执行失败: {str(e)}") + + # async def llm_stream_callback(self, chunk: str, tts_manager: TTSManager): + # """大模型流式回调处理""" + # if not chunk: + # return + # try: + # # 提交TTS合成请求 + # await tts_manager.synthesize(chunk) + # await asyncio.sleep(0) # 让出调度权 + # except Exception as e: + # logger.error(f"LLM流式回调处理失败: {str(e)}") + + # async def call_llm_and_send(self, query: str, conversation: LLMConversation, websocket: WebSocket): + # """调用大模型并处理结果""" + # logger.info(f"调用大模型 - 用户({conversation.user_id}): {query}") + # try: + # conv_id, full_reply = await llm_client.send_message( + # query=query, + # conversation=conversation, + # stream_callback=self.llm_stream_callback, + # response_mode="streaming" + # ) + # logger.info(f"大模型回复完成 - 会话ID: {conv_id}, 完整回复: {full_reply}") + # except Exception as e: + # logger.error(f"大模型调用失败: {str(e)}") + # if not websocket.client_state.disconnected: + # await websocket.send_json({ + # "type": "llm_error", + # "data": {"error": str(e)} + # }) + + async def recv_frontend_data(self, websocket: WebSocket, asr_conn): + """接收前端音频数据并推送到ASR""" + while not asr_conn.stop_event.is_set(): + try: + raw_bytes = await websocket.receive_bytes() + + # success = await push_audio_data(asr_conn, raw_bytes) + # if not success: + # logger.warning("音频数据插入ASR失败（队列满/连接失效）") + except WebSocketDisconnect: + logger.info("前端主动断开连接") + asr_conn.stop_event.set() + break + except Exception as e: + logger.error(f"接收前端数据失败: {str(e)}") + # asr_conn.stop_event.set() + # if not websocket.client_state.disconnected: + # await websocket.send_json({"error": f"接收数据失败: {str(e)}"}) + # break + + async def send_results(self, websocket: WebSocket, result_queue: asyncio.Queue, asr_conn): + """从结果队列发送数据到前端""" + while True: + try: + # 等待队列数据或超时 + result = await asyncio.wait_for(result_queue.get(), timeout=0.05) + if not websocket.client_state.disconnected: + await websocket.send_bytes(result) + except asyncio.TimeoutError: + if asr_conn.stop_event.is_set(): + break + continue + except Exception as e: + logger.error(f"发送结果到前端失败: {str(e)}") + asr_conn.stop_event.set() + break + + async def handle_connection(self, websocket: WebSocket): + """处理单个WebSocket连接的完整生命周期""" + + tts_manager = None + asr_conn = None + asr_task = None + result_queue = asyncio.Queue(maxsize=10000) + + context = await self.connect(websocket) + print(context) + + try: + pass + # 1. 建立连接 + + # 2. 初始化TTS + # tts_manager = await self.setup_tts_manager(result_queue) + + # 3. 获取ASR连接 + # asr_conn = await get_idle_asr_connection() + # if not asr_conn: + # await websocket.send_json({"error": "ASR服务暂时不可用", "text": ""}) + # return + + # 4. 启动ASR通信协程 + # asr_callback = lambda res: self.asr_result_callback(res, websocket, user_id, result_queue) + # asr_task = asyncio.create_task(handle_asr_communication(asr_conn, asr_callback)) + # + # # 5. 启动数据接收和发送协程 + # task_recv = asyncio.create_task(self.recv_frontend_data(websocket, asr_conn)) + # task_send = asyncio.create_task(self.send_results(websocket, result_queue, asr_conn)) + # + # # 6. 等待任一任务完成 + # done, pending = await asyncio.wait( + # [task_recv, task_send], + # return_when=asyncio.FIRST_COMPLETED + # ) + + except Exception as e: + logger.error(f"WebSocket连接处理异常: {str(e)}") + if asr_conn: + asr_conn.stop_event.set() + if not websocket.client_state.disconnected: + await websocket.send_json({"error": str(e)}) + finally: + pass + # 7. 资源清理 + # logger.info(f"开始清理连接 {conn_id} 的资源") + # # 停止ASR + # if asr_conn: + # asr_conn.stop_event.set() + # + # # 取消任务 + # if asr_task and not asr_task.done(): + # asr_task.cancel() + # try: + # await asr_task + # except asyncio.CancelledError: + # pass + # + # # 清理TTS + # if tts_manager: + # await tts_manager.cleanup() # 假设TTSManager有cleanup方法，无则忽略 + # + # # 断开连接 + # if websocket: + # self.disconnect(websocket, conn_id) + # try: + # await websocket.close() + # except Exception: + # pass + # + # logger.info(f"连接 {conn_id} 资源清理完成") diff --git a/audio_ai_chat/audio_ai_chat/main.py b/audio_ai_chat/audio_ai_chat/main.py new file mode 100644 index 0000000..eb9b63f --- /dev/null +++ b/audio_ai_chat/audio_ai_chat/main.py @@ -0,0 +1,56 @@ +from fastapi import FastAPI, WebSocket, WebSocketDisconnect +from fastapi.middleware.cors import CORSMiddleware +from audio_ai_chat.config.settings import settings +from audio_ai_chat.config.logger import logger +from audio_ai_chat.core.websocket_handler import WebSocketConnectionManager +# from audio_ai_chat.models.ws_models import WSRequest, WSResponse, WSError +from audio_ai_chat.codec.ProtocolCodec import ProtocolCodec # 已有加密类 +from audio_ai_chat.utils.exceptions import CodecError, ServiceCallError +from contextlib import asynccontextmanager +from frontend_ws import frontend_websocket_handler + +# FastAPI 启动时初始化 ASR 连接池 +@asynccontextmanager +async def lifespan(app: FastAPI): + # 启动时执行（原 startup 逻辑） + print(' FastAPI 启动时初始化 ASR 连接池') + # await init_asr_pool() + yield # 应用运行中 + # 关闭时执行（可选，比如清理连接池） + print("应用关闭，开始清理 ASR 连接池...") + # await close_asr_pool() + + +app = FastAPI( + title="语音AI对话系统", + description="整合前端/ASR/大模型/TTS的异步网关服务", + version="1.0", + lifespan=lifespan # 绑定生命周期 +) + +# 跨域配置（允许前端访问） +app.add_middleware( + CORSMiddleware, + allow_origins=["*"], # 生产环境替换为具体前端域名 + allow_credentials=True, + allow_methods=["*"], + allow_headers=["*"], +) + +# 初始化WebSocket连接管理器 +ws_manager = WebSocketConnectionManager() +# 初始化编解码器（从配置读取密钥） +codec = ProtocolCodec() + + +@app.websocket("/ws/audio") +async def websocket_audio(websocket: WebSocket): + # 完全委托给frontend_websocket_handler处理 + await ws_manager.handle_connection(websocket) + + + +@app.get("/health") +async def health_check(): + """服务健康检查接口""" + return {"status": "healthy", "service": "audio-ai-chat"} \ No newline at end of file diff --git a/audio_ai_chat/audio_ai_chat/utils/__init__.py b/audio_ai_chat/audio_ai_chat/utils/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/audio_ai_chat/audio_ai_chat/utils/__pycache__/__init__.cpython-310.pyc b/audio_ai_chat/audio_ai_chat/utils/__pycache__/__init__.cpython-310.pyc new file mode 100644 index 0000000..ccec488 Binary files /dev/null and b/audio_ai_chat/audio_ai_chat/utils/__pycache__/__init__.cpython-310.pyc differ diff --git a/audio_ai_chat/audio_ai_chat/utils/__pycache__/exceptions.cpython-310.pyc b/audio_ai_chat/audio_ai_chat/utils/__pycache__/exceptions.cpython-310.pyc new file mode 100644 index 0000000..6a7230c Binary files /dev/null and b/audio_ai_chat/audio_ai_chat/utils/__pycache__/exceptions.cpython-310.pyc differ diff --git a/audio_ai_chat/audio_ai_chat/utils/exceptions.py b/audio_ai_chat/audio_ai_chat/utils/exceptions.py new file mode 100644 index 0000000..d298a7b --- /dev/null +++ b/audio_ai_chat/audio_ai_chat/utils/exceptions.py @@ -0,0 +1,15 @@ +class AudioAIChatBaseException(Exception): + """项目基础异常""" + pass + +class CodecError(AudioAIChatBaseException): + """编解码异常（加密/解密失败）""" + pass + +class ServiceCallError(AudioAIChatBaseException): + """服务调用异常（ASR/LLM/TTS调用失败）""" + pass + +class ValidationError(AudioAIChatBaseException): + """数据验证异常（请求参数不符合模型）""" + pass \ No newline at end of file diff --git a/audio_ai_chat/audio_ai_chat/utils/helpers.py b/audio_ai_chat/audio_ai_chat/utils/helpers.py new file mode 100644 index 0000000..78042a5 --- /dev/null +++ b/audio_ai_chat/audio_ai_chat/utils/helpers.py @@ -0,0 +1,26 @@ +from typing import Callable, Awaitable, TypeVar +from tenacity import retry, stop_after_attempt, wait_exponential +from audio_ai_chat.config.logger import logger + +T = TypeVar("T") + +def async_retry( + times: int = 3, + delay: int = 1, + max_delay: int = 5 +) -> Callable[[Callable[..., Awaitable[T]]], Callable[..., Awaitable[T]]]: + """异步重试装饰器（通用）""" + def decorator(func: Callable[..., Awaitable[T]]) -> Callable[..., Awaitable[T]]: + @retry( + stop=stop_after_attempt(times), + wait=wait_exponential(multiplier=delay, min=delay, max=max_delay), + reraise=True + ) + async def wrapper(*args, **kwargs) -> T: + try: + return await func(*args, **kwargs) + except Exception as e: + logger.warning(f"函数 {func.__name__} 调用失败，将重试（剩余次数：{retry.statistics.get('attempt_number', 0)}）：{str(e)}") + raise + return wrapper + return decorator \ No newline at end of file diff --git a/audio_ai_chat/logs/app.log b/audio_ai_chat/logs/app.log new file mode 100644 index 0000000..e69de29 diff --git a/audio_ai_chat/requirements.txt b/audio_ai_chat/requirements.txt new file mode 100644 index 0000000..d336d12 --- /dev/null +++ b/audio_ai_chat/requirements.txt @@ -0,0 +1,9 @@ +fastapi>=0.104.1 +uvicorn>=0.24.0 # FastAPI运行服务器 +websockets>=12.0 # WebSocket支持 +pydantic>=2.4.2 # 数据模型校验 +python-dotenv>=1.0.0 # 读取.env文件 +loguru>=0.7.2 # 日志工具（比原生logging更简洁） +requests>=2.31.0 # 同步HTTP请求（若ASR/LLM/TTS用HTTP接口） +aiohttp>=3.9.1 # 异步HTTP请求（推荐，契合FastAPI异步特性） +tenacity>=8.2.3 # 重试机制（服务调用失败自动重试） diff --git a/audio_ai_chat/run.py b/audio_ai_chat/run.py new file mode 100644 index 0000000..bd05115 --- /dev/null +++ b/audio_ai_chat/run.py @@ -0,0 +1,16 @@ +# run.py +import uvicorn +from dotenv import load_dotenv + +# 加载环境变量 +load_dotenv() + +if __name__ == "__main__": + uvicorn.run( + app="audio_ai_chat.main:app", + host="0.0.0.0", + port=8000, + reload=True, + log_level="info", # 关键：开启详细日志 + access_log=True, + ) \ No newline at end of file diff --git a/audio_ai_chat/ws_server.log b/audio_ai_chat/ws_server.log new file mode 100644 index 0000000..e69de29 diff --git a/python/ProtocolCodec.py b/python/ProtocolCodec.py new file mode 100644 index 0000000..10e63fe --- /dev/null +++ b/python/ProtocolCodec.py @@ -0,0 +1,295 @@ +from enum import IntEnum +import struct +import json +import gzip +from typing import Optional, Union, Dict, Any, List, Tuple + + +# -------------------------- 协议常量定义 -------------------------- +class ProtocolConst: + PROTOCOL_VERSION = 0b0001 # 协议版本（4位，字节0低4位，0~15） + HEADER_SIZE = 8 # 头部固定字节数（字节0~7） + MAX_BODY_SIZE = 0xFFFFFF # 最大包体大小（24位，≈16MB，字节4~6存储） + STRING_ENCODING = "utf-8" # 字符串默认编码 + + +# -------------------------- 枚举定义 -------------------------- +class MessageType(IntEnum): + """消息类型（4位，字节0高4位，0~15）""" + PING = 0b0001 # 心跳（支持空包体） + AUDIO_DATA = 0b0010 # 纯音频数据(pcm) + TEXT_MESSAGE = 0b0011 # 纯文本消息 + CONTROL_CMD = 0b0100 # 控制指令 + # 预留12种类型（0b0100 ~ 0b1111） + + +class SerializationType(IntEnum): + """序列化方式（3位，字节1高3位，1~8）""" + RAW = 0b001 # 原始二进制（1） + JSON = 0b010 # JSON格式（2） + STRING = 0b011 # 直接字符串（3） + # 预留5种方式（0b100 ~ 0b111） + + +class CompressionType(IntEnum): + """压缩方式（3位，字节1中3位，1~8）""" + NONE = 0b001 # 无压缩（默认值1） + GZIP = 0b010 # gzip压缩（2） + # 预留6种方式（0b011 ~ 0b111） + + +class ControlCommand(IntEnum): + """控制指令类型（配合MessageType.CONTROL_CMD使用）""" + HEARTBEAT = 0b0001 # 心跳响应 + PAUSE = 0b0010 # 暂停 + RESUME = 0b0011 # 继续 + STOP = 0b0100 # 停止 + + +# -------------------------- 协议工具类（与JS协议结构一致） -------------------------- +class ProtocolCodec: + @staticmethod + def pack( + msg_type: MessageType, + body: Union[bytes, str, Dict[str, Any], List[Any], None] = None, + sequence: int = 0, + serialization: Optional[SerializationType] = None, + compression: CompressionType = CompressionType.NONE + ) -> bytes: + """ + 封装协议包（与JS pack 方法完全兼容） + :param msg_type: 消息类型 + :param body: 包体数据（PING消息可传None/空，其他类型必填） + :param sequence: 消息顺序号（0~65535，默认0） + :param serialization: 序列化方式（None时自动推导） + :param compression: 压缩方式（默认无压缩） + :return: 完整协议包（bytes） + """ + # 校验顺序号范围（0~65535） + if not isinstance(sequence, int) or not (0 <= sequence <= 0xFFFF): + raise ValueError(f"消息顺序号必须是0~65535的整数，当前传入：{sequence}") + + # 特殊处理：PING消息支持空包体 + if msg_type == MessageType.PING: + # PING消息强制RAW序列化 + serialization = SerializationType.RAW + # 空包体转为空bytes + if body is None: + body = b"" + if not isinstance(body, bytes): + raise TypeError(f"PING消息仅支持空包体或bytes类型，当前传入：{type(body)}") + else: + # 非PING消息包体不能为空 + if body is None: + raise ValueError(f"非PING消息（{msg_type.name}）包体不能为空") + + # 1. 自动推导序列化方式（非PING消息） + if serialization is None and msg_type != MessageType.PING: + if msg_type == MessageType.AUDIO_DATA: + serialization = SerializationType.RAW + elif msg_type == MessageType.TEXT_MESSAGE: + serialization = SerializationType.STRING + elif msg_type == MessageType.CONTROL_CMD: + serialization = SerializationType.JSON + else: + raise ValueError(f"不支持的消息类型：{msg_type}") + + # 2. 校验枚举值范围 + if not (0b001 <= serialization.value <= 0b111): + raise ValueError(f"序列化方式必须在1~8（0b001~0b111）范围内，当前传入：{serialization.value}") + if not (0b001 <= compression.value <= 0b111): + raise ValueError(f"压缩方式必须在1~8（0b001~0b111）范围内，当前传入：{compression.value}") + + # 3. 序列化包体 + serialized_body: bytes + if serialization == SerializationType.RAW: + if not isinstance(body, bytes): + raise TypeError("RAW序列化要求body必须是bytes类型") + serialized_body = body + elif serialization == SerializationType.STRING: + if not isinstance(body, str): + raise TypeError("STRING序列化要求body必须是str类型") + serialized_body = body.encode(ProtocolConst.STRING_ENCODING) + elif serialization == SerializationType.JSON: + if isinstance(body, str): + serialized_body = body.encode(ProtocolConst.STRING_ENCODING) + elif isinstance(body, (dict, list)): + serialized_body = json.dumps(body, ensure_ascii=False, separators=(',', ':')).encode(ProtocolConst.STRING_ENCODING) + else: + raise TypeError("JSON序列化要求body必须是str/dict/list类型") + else: + raise ValueError(f"不支持的序列化方式：{serialization}") + + # 4. 压缩包体 + compressed_body: bytes + if compression == CompressionType.GZIP: + compressed_body = gzip.compress(serialized_body) + elif compression == CompressionType.NONE: + compressed_body = serialized_body + else: + raise ValueError(f"不支持的压缩方式：{compression}") + + # 5. 校验包体大小（24位最大支持0xFFFFFF字节） + body_len = len(compressed_body) + if body_len > ProtocolConst.MAX_BODY_SIZE: + raise OverflowError( + f"包体过大（{body_len}字节），最大支持{ProtocolConst.MAX_BODY_SIZE}字节（≈16MB）" + ) + + # 6. 构造头部（与JS头部结构完全一致） + # 字节0：消息类型(高4位) + 协议版本(低4位) + byte0 = ((msg_type.value & 0x0F) << 4) | (ProtocolConst.PROTOCOL_VERSION & 0x0F) + # 字节1：序列化方式(高3位) + 压缩方式(中3位) + 保留位(低2位) + byte1 = ((serialization.value & 0x07) << 5) | ((compression.value & 0x07) << 2) | 0x00 + # 字节2~3：消息顺序号（16位大端序） + byte2_3 = struct.pack(">H", sequence) + # 字节4~6：包体长度（24位大端序） + byte4 = (body_len >> 16) & 0xFF + byte5 = (body_len >> 8) & 0xFF + byte6 = body_len & 0xFF + # 字节7：保留位（固定0x00） + byte7 = 0x00 + + # 拼接头部 + header = ( + bytes([byte0, byte1]) + + byte2_3 + + bytes([byte4, byte5, byte6, byte7]) + ) + assert len(header) == ProtocolConst.HEADER_SIZE, f"头部长度错误：实际{len(header)}字节，预期{ProtocolConst.HEADER_SIZE}字节" + + return header + compressed_body + + @staticmethod + def unpack(packet: bytes) -> Tuple[MessageType, int, Any]: + """ + 解析协议包（与JS unpack 方法完全兼容） + :param packet: 完整协议包（头部 + 包体） + :return: (消息类型, 顺序号, 原始包体数据) # 仅返回核心必要信息 + """ + # 1. 校验包长度 + if len(packet) < ProtocolConst.HEADER_SIZE: + raise ValueError( + f"包长度过短（{len(packet)}字节），至少需要{ProtocolConst.HEADER_SIZE}字节头部" + ) + + # 2. 解析头部 + header = packet[:ProtocolConst.HEADER_SIZE] + body = packet[ProtocolConst.HEADER_SIZE:] + + # 字节0：消息类型(高4位) + 协议版本(低4位) + byte0 = header[0] + msg_type = MessageType((byte0 >> 4) & 0x0F) + version = byte0 & 0x0F + + # 字节1：序列化方式(高3位) + 压缩方式(中3位) + 保留位(低2位) + byte1 = header[1] + serialization = SerializationType((byte1 >> 5) & 0x07) + compression = CompressionType((byte1 >> 2) & 0x07) + + # 字节2~3：消息顺序号（16位大端序） + sequence = struct.unpack(">H", header[2:4])[0] + + # 字节4~6：包体长度（24位大端序），字节7：保留位（忽略） + body_len = (header[4] << 16) | (header[5] << 8) | header[6] + + # 校验版本和包体长度（校验逻辑保留，确保数据有效性） + if version != ProtocolConst.PROTOCOL_VERSION: + raise ValueError( + f"协议版本不匹配：收到v{version}（0b{version:04b}），支持v{ProtocolConst.PROTOCOL_VERSION}（0b{ProtocolConst.PROTOCOL_VERSION:04b}）" + ) + if len(body) != body_len: + raise ValueError( + f"包体长度不匹配：头部声明{body_len}字节，实际{len(body)}字节" + ) + + # 3. 解压包体 + decompressed_body: bytes + if compression == CompressionType.GZIP: + try: + decompressed_body = gzip.decompress(body) + except Exception as e: + raise ValueError(f"GZIP解压失败：{str(e)}") + elif compression == CompressionType.NONE: + decompressed_body = body + else: + raise ValueError(f"不支持的压缩方式：{compression}") + + # 4. 反序列化包体（空包体返回None） + original_body: Any + if len(decompressed_body) == 0: + original_body = None + elif serialization == SerializationType.RAW: + original_body = decompressed_body + elif serialization == SerializationType.STRING: + try: + original_body = decompressed_body.decode(ProtocolConst.STRING_ENCODING) + except UnicodeDecodeError: + raise ValueError(f"STRING反序列化失败：{ProtocolConst.STRING_ENCODING}解码错误") + elif serialization == SerializationType.JSON: + try: + original_body = json.loads(decompressed_body.decode(ProtocolConst.STRING_ENCODING)) + except UnicodeDecodeError: + raise ValueError(f"JSON反序列化失败：{ProtocolConst.STRING_ENCODING}解码错误") + except json.JSONDecodeError as e: + raise ValueError(f"JSON反序列化失败：格式错误（{str(e)}）") + else: + raise ValueError(f"不支持的序列化方式：{serialization}") + + # 仅返回核心必要信息：消息类型、顺序号、原始包体数据 + return msg_type, sequence, original_body + + +# -------------------------- 使用示例（验证修改后功能正常） -------------------------- +if __name__ == "__main__": + # 示例1：PING消息（空包体，默认顺序号0） + ping_packet = ProtocolCodec.pack(MessageType.PING) + print(f"PING消息包长度：{len(ping_packet)}字节（仅头部）") + msg_type1, seq1, body1 = ProtocolCodec.unpack(ping_packet) + print(f"PING解析结果：类型={msg_type1.name}，顺序号={seq1}，包体={body1}\n") + + # 示例2：纯文本消息（STRING序列化，指定顺序号） + text_body = "Python与JS协议兼容测试（纯字符串）" + text_packet = ProtocolCodec.pack( + msg_type=MessageType.TEXT_MESSAGE, + body=text_body, + sequence=1001, + compression=CompressionType.NONE + ) + print(f"文本消息包长度：{len(text_packet)}字节") + msg_type2, seq2, body2 = ProtocolCodec.unpack(text_packet) + print(f"文本解析结果：类型={msg_type2.name}，顺序号={seq2}，内容={body2}\n") + + # 示例3：控制指令（JSON序列化） + control_body = {"cmd": ControlCommand.PAUSE.value, "reason": "用户主动暂停"} + control_packet = ProtocolCodec.pack( + msg_type=MessageType.CONTROL_CMD, + body=control_body, + sequence=1002 + ) + print(f"控制指令包长度：{len(control_packet)}字节") + msg_type3, seq3, body3 = ProtocolCodec.unpack(control_packet) + print(f"控制指令解析结果：类型={msg_type3.name}，顺序号={seq3}，内容={body3}\n") + + # 示例4：音频数据（RAW序列化） + audio_body = b"\x00\x01\x02\x03\x04\x05" * 100 # 模拟PCM数据 + audio_packet = ProtocolCodec.pack( + msg_type=MessageType.AUDIO_DATA, + body=audio_body, + sequence=1003 + ) + print(f"音频数据包长度：{len(audio_packet)}字节") + msg_type4, seq4, body4 = ProtocolCodec.unpack(audio_packet) + print(f"音频解析结果：类型={msg_type4.name}，顺序号={seq4}，数据长度={len(body4)}字节\n") + + # 示例5：GZIP压缩测试（需JS端启用GZIP解压） + long_text_body = "这是一段很长的文本，用于测试GZIP压缩效果" * 100 + gzip_packet = ProtocolCodec.pack( + msg_type=MessageType.TEXT_MESSAGE, + body=long_text_body, + sequence=1004, + compression=CompressionType.GZIP + ) + print(f"GZIP压缩后包长度：{len(gzip_packet)}字节（原始文本长度：{len(long_text_body.encode())}字节）") + msg_type5, seq5, body5 = ProtocolCodec.unpack(gzip_packet) + print(f"GZIP解析结果：类型={msg_type5.name}，顺序号={seq5}，内容前50字：{body5[:50]}...") \ No newline at end of file diff --git a/python/__pycache__/frontend_ws.cpython-310.pyc b/python/__pycache__/frontend_ws.cpython-310.pyc index c7646e6..cf95f18 100644 Binary files a/python/__pycache__/frontend_ws.cpython-310.pyc and b/python/__pycache__/frontend_ws.cpython-310.pyc differ diff --git a/python/tts_output_33cfc4e1.pcm b/python/tts_output_33cfc4e1.pcm deleted file mode 100644 index 294df1a..0000000 Binary files a/python/tts_output_33cfc4e1.pcm and /dev/null differ diff --git a/python/tts_output_46c5a16f.pcm b/python/tts_output_46c5a16f.pcm deleted file mode 100644 index 1e42a4c..0000000 Binary files a/python/tts_output_46c5a16f.pcm and /dev/null differ diff --git a/python/tts_output_4cafe62e.mp3 b/python/tts_output_4cafe62e.mp3 deleted file mode 100644 index 5f6e332..0000000 Binary files a/python/tts_output_4cafe62e.mp3 and /dev/null differ diff --git a/python/tts_output_83430d06.pcm b/python/tts_output_83430d06.pcm deleted file mode 100644 index 7f9a3e0..0000000 Binary files a/python/tts_output_83430d06.pcm and /dev/null differ diff --git a/python/tts_output_ab2c61fb.pcm b/python/tts_output_ab2c61fb.pcm deleted file mode 100644 index 7a805ff..0000000 Binary files a/python/tts_output_ab2c61fb.pcm and /dev/null differ diff --git a/python/tts_output_c5f4106a.mp3 b/python/tts_output_c5f4106a.mp3 deleted file mode 100644 index 9174284..0000000 Binary files a/python/tts_output_c5f4106a.mp3 and /dev/null differ diff --git a/python/tts_output_cb0e0ad4.pcm b/python/tts_output_cb0e0ad4.pcm deleted file mode 100644 index d88283b..0000000 Binary files a/python/tts_output_cb0e0ad4.pcm and /dev/null differ diff --git a/测试流式传输uniapp/pages/index/index - 副本 (2).vue b/测试流式传输uniapp/pages/index/index - 副本 (2).vue new file mode 100644 index 0000000..f4c2824 --- /dev/null +++ b/测试流式传输uniapp/pages/index/index - 副本 (2).vue @@ -0,0 +1,285 @@ + + + + + \ No newline at end of file diff --git a/测试流式传输uniapp/pages/index/index - 副本.vue b/测试流式传输uniapp/pages/index/index - 副本.vue index da453ab..8d3870c 100644 --- a/测试流式传输uniapp/pages/index/index - 副本.vue +++ b/测试流式传输uniapp/pages/index/index - 副本.vue @@ -79,6 +79,7 @@ success: () => { console.log('web'); this.ws.onOpen((res) => { + console.log('WebSocket连接已打开', res); this.ws.onMessage((res) => { let messageData = res.data; diff --git a/测试流式传输uniapp/pages/index/index.vue b/测试流式传输uniapp/pages/index/index.vue index 4dc9bdd..3d8ad99 100644 --- a/测试流式传输uniapp/pages/index/index.vue +++ b/测试流式传输uniapp/pages/index/index.vue @@ -4,225 +4,40 @@ {{ status }} 当前分贝：{{ currentDecibels }} - - + xx - + - \ No newline at end of file + diff --git a/测试流式传输uniapp/pages/index/useWebSocket.js b/测试流式传输uniapp/pages/index/useWebSocket.js new file mode 100644 index 0000000..e7efe35 --- /dev/null +++ b/测试流式传输uniapp/pages/index/useWebSocket.js @@ -0,0 +1,77 @@ +import { ref } from 'vue'; + +// 在组件 setup 函数中使用 +export default function useWebSocket() { + // 存储 WebSocket 实例 + const ws = ref(null); + const wsUrl = 'ws://127.0.0.1:8000/ws/audio'; + // 初始化 WebSocket 连接 + const initWebSocket = () => { + ws.value = uni.connectSocket({ + url: wsUrl, + success: () => { + console.log('web'); + + // 监听连接打开 + ws.value.onOpen((res) => { + console.log('WebSocket连接已打开', res); + + // 监听消息接收 + ws.value.onMessage((res) => { + let messageData = res.data; + + // 处理不同类型的数据 + if (typeof messageData === 'string') { + // 文本数据 + try { + const parsedData = JSON.parse(messageData); + handleMessage(parsedData); + } catch (e) { + handleMessage(messageData); + } + } else if (messageData instanceof ArrayBuffer) { + // 二进制数据 + try { + // 注意：组合式 API 中需通过 ref 获取子组件实例 + const aaaRef = ref(null); // 需在组件中声明 + aaaRef.value?.appendBuffer(messageData); + } catch (e) { + console.log('二进制数据解析失败:', e); + } + } + }); + }); + }, + fail: () => { + console.log('fail'); + }, + }); + }; + + // 消息处理函数（根据实际业务逻辑修改） + const handleMessage = (data) => { + // 原组件中的 handleMessage 逻辑迁移到这里 + console.log('收到消息:', data); + }; + + // 辅助函数：ArrayBuffer 转字符串（如果后续需要使用） + const arrayBufferToStringCompat = (buffer) => { + const decoder = new TextDecoder('utf-8'); + return decoder.decode(buffer); + }; + + // 关闭连接函数（可选，按需暴露） + const closeWebSocket = () => { + if (ws.value) { + ws.value.close(); + console.log('WebSocket连接已关闭'); + } + }; + + return { + ws, + initWebSocket, + closeWebSocket, + handleMessage + }; +}; \ No newline at end of file diff --git a/测试流式传输uniapp/unpackage/dist/cache/.vite/deps/_metadata.json b/测试流式传输uniapp/unpackage/dist/cache/.vite/deps/_metadata.json index 568a6e3..11f7af3 100644 --- a/测试流式传输uniapp/unpackage/dist/cache/.vite/deps/_metadata.json +++ b/测试流式传输uniapp/unpackage/dist/cache/.vite/deps/_metadata.json @@ -1,8 +1,8 @@ { - "hash": "030e727a", - "configHash": "c22f3258", - "lockfileHash": "c10b225f", - "browserHash": "77dd6173", + "hash": "56a2a6b1", + "configHash": "0d2436b6", + "lockfileHash": "1c743963", + "browserHash": "42eed156", "optimized": {}, "chunks": {} } \ No newline at end of file