From 7f8043003c7169f9df50c66a936e102a3f6b544f Mon Sep 17 00:00:00 2001 From: MT-Fire <798521692@qq.com> Date: Wed, 4 Mar 2026 14:17:39 +0800 Subject: [PATCH] =?UTF-8?q?feat:=20GLM=E6=99=AE=E9=80=9A=E5=AF=B9=E8=AF=9D?= =?UTF-8?q?=E5=8A=9F=E8=83=BD=E6=B5=8B=E8=AF=95=E6=88=90=E5=8A=9F?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- server/api/chat_routes.py | 25 ++- server/api/chat_routes_glm.py | 125 +++++++++++++ server/main.py | 158 ++++++++-------- server/utils/file_cache.py | 64 +++++++ server/utils/glm_adapter.py | 335 ++++++++++++++++++++++++++++++++++ 5 files changed, 631 insertions(+), 76 deletions(-) create mode 100644 server/api/chat_routes_glm.py create mode 100644 server/utils/file_cache.py create mode 100644 server/utils/glm_adapter.py diff --git a/server/api/chat_routes.py b/server/api/chat_routes.py index f27fb47..0899440 100644 --- a/server/api/chat_routes.py +++ b/server/api/chat_routes.py @@ -1,5 +1,6 @@ """ -API 路由定义 +API 路由定义（阿里云百炼 / DashScope 平台） +所有 DashScope 相关逻辑均集中在此文件，main.py 无感知任何平台细节。 """ import os import json @@ -12,6 +13,19 @@ from fastapi.responses import JSONResponse, StreamingResponse import dashscope from dashscope import Generation, MultiModalConversation + +def init(): + """ + 初始化百炼后端：设置 DashScope API Key。 + 由 main.py 在启动时调用（若 LLM_BACKEND=dashscope）。 + """ + api_key = os.getenv("ALIYUN_API_KEY") + if not api_key: + raise ValueError("dashscope 模式需要设置环境变量 ALIYUN_API_KEY") + dashscope.api_key = api_key + print(f"[DashScope] 初始化完成，ALIYUN_API_KEY 已配置") + + # 导入模型和工具函数（使用绝对路径） import sys from pathlib import Path @@ -936,4 +950,11 @@ def serve_upload_handler(filename: str): async def stop_generation_handler(message_id: str = None): """停止生成处理器""" message = f"已发出停止指令，消息ID: {message_id}" if message_id else "已发出停止指令" - return {"success": True, "message": message} \ No newline at end of file + return {"success": True, "message": message} + + +# ── 平台统一接口别名（供 main.py 的 _platform 动态调用）───────────── +# main.py 通过 _platform.chat_handler / _platform.models_handler 调用， +# 各平台模块需暴露相同名称的函数。 +chat_handler = chat_endpoint_handler # 聊天接口别名 +models_handler = get_models_handler # 模型列表别名 diff --git a/server/api/chat_routes_glm.py b/server/api/chat_routes_glm.py new file mode 100644 index 0000000..57cb1b9 --- /dev/null +++ b/server/api/chat_routes_glm.py @@ -0,0 +1,125 @@ +""" +GLM-4.6V 平台路由处理器（zai-sdk） +所有智谱 GLM 相关逻辑均集中在此文件，main.py 无感知任何平台细节。 +""" +import os +import sys +import json +from pathlib import Path +from fastapi import HTTPException +from fastapi.responses import JSONResponse, StreamingResponse +from utils.helpers import get_current_timestamp, generate_unique_id +from utils.logger import log_info + + +def init(): + """ + 初始化 GLM 后端：验证 API Key 是否配置。 + 由 main.py 在启动时调用（若 LLM_BACKEND=glm）。 + """ + api_key = os.getenv("ZHIPU_API_KEY") or os.getenv("GLM_API_KEY") + if not api_key: + raise ValueError("GLM 模式需要设置环境变量 ZHIPU_API_KEY（在 https://open.bigmodel.cn 申请）") + log_info(f"[GLM] 初始化完成，ZHIPU_API_KEY 已配置") + + +async def chat_handler(body: dict): + """ + GLM 聊天处理器（对外接口与百炼 chat_endpoint_handler 完全兼容）。 + 流式/非流式自动适配，支持图像、文档附件、联网搜索、深度思考。 + """ + from utils.glm_adapter import glm_stream_generator, glm_chat_sync + + if not isinstance(body, dict): + raise HTTPException(status_code=400, detail="请求体必须是 JSON 对象") + + messages = body.get("messages", []) + model = body.get("model", "glm-4.6v") + stream = body.get("stream", True) + temperature = body.get("temperature", 0.7) + max_tokens = body.get("max_tokens", body.get("maxTokens", 2000)) + web_search = body.get("webSearch", False) or body.get("deepSearch", False) + deep_think = body.get("deepThinking", False) + files = body.get("files", []) + + # 兼容前端简化格式（非 messages 结构） + if not messages: + msg_text = body.get("message", "") + sys_prompt = body.get("systemPrompt", "你是一个智能助手。") + user_content = msg_text if isinstance(msg_text, list) else [{"type": "text", "text": msg_text}] + messages = [ + {"role": "system", "content": sys_prompt}, + {"role": "user", "content": user_content}, + ] + + log_info(f"[GLM] model={model} stream={stream} web_search={web_search} " + f"thinking={deep_think} files={len(files)} msgs={len(messages)}") + + if stream: + return StreamingResponse( + glm_stream_generator( + messages=messages, model=model, temperature=temperature, + max_tokens=max_tokens, files=files or None, + web_search=web_search, deep_thinking=deep_think, + ), + media_type="text/event-stream", + ) + + result = glm_chat_sync( + messages=messages, model=model, temperature=temperature, + max_tokens=max_tokens, files=files or None, + web_search=web_search, deep_thinking=deep_think, + ) + resp = { + "id": f"chatcmpl-{generate_unique_id()}", + "object": "chat.completion", + "created": get_current_timestamp(), + "model": result["model"], + "choices": [ + { + "index": 0, + "message": {"role": "assistant", "content": result["content"]}, + "finish_reason": "stop", + } + ], + } + if result.get("usage"): + resp["usage"] = result["usage"] + return JSONResponse(content=resp) + + +def models_handler(): + """返回 GLM 可用模型列表""" + return { + "data": [ + { + "id": "glm-4.6v", + "name": "GLM-4.6V（推荐）", + "description": "最新旗舰模型，支持文本/图像/文档/深度思考", + "maxTokens": 128000, + "provider": "ZhipuAI", + }, + { + "id": "glm-4-flash", + "name": "GLM-4 Flash", + "description": "高性价比文本模型（0.2元/千token）", + "maxTokens": 128000, + "provider": "ZhipuAI", + }, + { + "id": "glm-4v-plus-0111", + "name": "GLM-4V Plus", + "description": "图像 + PDF/DOCX 原生多模态", + "maxTokens": 128000, + "provider": "ZhipuAI", + }, + { + "id": "glm-z1-flash", + "name": "GLM-Z1 Flash", + "description": "深度思考推理模型", + "maxTokens": 128000, + "provider": "ZhipuAI", + }, + ], + "object": "list", + } diff --git a/server/main.py b/server/main.py index f8c453d..f94379e 100644 --- a/server/main.py +++ b/server/main.py @@ -1,158 +1,168 @@ """ -改进版Python FastAPI服务器实现，使用DashScope Python SDK连接阿里云百炼平台API -拆分模块版本 +AI Chat API Server — 主入口（纯基础设施层） + +职责： + - 注入运行时依赖（venv site-packages） + - 读取 LLM_BACKEND 环境变量，动态加载对应平台模块 + - 注册 FastAPI 路由和中间件 + +平台代码位置（main.py 中不包含任何平台逻辑）： + - 百炼 DashScope → api/chat_routes.py + - 智谱 GLM-4.6V → api/chat_routes_glm.py + utils/glm_adapter.py """ import os -import json +import sys from datetime import datetime, timezone from pathlib import Path -import dashscope + +# ── 注入 venv site-packages（兼容 start.sh 用系统 python3 启动）──────── +# 必须在所有第三方 import 前执行 +_venv_lib = Path(__file__).parent / ".venv" / "lib" +if _venv_lib.exists(): + for _sp in sorted(_venv_lib.glob("python*/site-packages"), reverse=True): + if _sp.exists() and str(_sp) not in sys.path: + sys.path.insert(0, str(_sp)) + print(f"[启动] venv 注入：{_sp}") + break + +# ── 第三方导入 ──────────────────────────────────────────────────────── from dotenv import load_dotenv -from fastapi import FastAPI, HTTPException, File, UploadFile, Request +from fastapi import FastAPI, File, UploadFile, Request from fastapi.responses import JSONResponse -# 导入模块 -import sys sys.path.append('/home/mt/project/ai-chat-ui/server') +# ── 工具/日志（与平台无关）─────────────────────────────────────────── +from utils.helpers import log_response +from utils.logger import get_logger + +logger = get_logger() + +# ── 加载环境变量 ────────────────────────────────────────────────────── +load_dotenv() + +LLM_BACKEND = os.getenv("LLM_BACKEND", "dashscope").lower().strip() +if LLM_BACKEND not in {"dashscope", "glm"}: + logger.warning(f"未知的 LLM_BACKEND='{LLM_BACKEND}'，回退到 dashscope") + LLM_BACKEND = "dashscope" + +# ── 动态加载平台模块 ────────────────────────────────────────────────── +if LLM_BACKEND == "glm": + import api.chat_routes_glm as _platform +else: + import api.chat_routes as _platform + +_platform.init() # 各平台自行完成初始化（API Key 校验等） + +# 通用路由处理器（文件上传、会话管理等，与平台无关，统一用百炼路由中的实现） from api.chat_routes import ( - chat_endpoint_handler, - get_models_handler, get_conversations_handler, get_conversation_handler, save_conversation_handler, delete_conversation_handler, upload_file_handler, serve_upload_handler, - stop_generation_handler + stop_generation_handler, ) -from models.chat_models import ChatRequest, ModelInfo -from utils.helpers import log_request, log_response -from utils.logger import get_logger -logger = get_logger() - - -# 加载环境变量 -load_dotenv() - -# 设置 DashScope API 密钥 -api_key = os.getenv("ALIYUN_API_KEY") -if not api_key: - raise ValueError("请在环境变量中设置 ALIYUN_API_KEY") - -dashscope.api_key = api_key - -# 创建 FastAPI 应用 -app = FastAPI(title="AI Chat API Server (Python)", version="2.0.0") +# ── FastAPI 应用 ────────────────────────────────────────────────────── +app = FastAPI( + title=f"AI Chat API（LLM_BACKEND={LLM_BACKEND}）", + version="3.0.0", +) @app.middleware("http") async def logging_middleware(request: Request, call_next): - """中间件：记录请求日志并美化输出""" start_time = datetime.now(timezone.utc) - client_ip = request.client.host if request.client else 'unknown' - - # 请求日志 + client_ip = request.client.host if request.client else "unknown" logger.info(f"→ {request.method} {request.url.path} | IP: {client_ip}") - response = await call_next(request) - - process_time = (datetime.now(timezone.utc) - start_time).total_seconds() * 1000 - status_emoji = "✅" if response.status_code < 400 else "❌" - - # 响应日志（包含端点、状态码、耗时） - logger.info(f"{status_emoji} {request.method} {request.url.path} | 状态: {response.status_code} | 耗时: {process_time:.0f}ms") - - # 记录结构化日志（写入日志文件） - log_response(response.status_code, process_time) - - response.headers["X-Process-Time"] = f"{process_time:.2f}ms" + ms = (datetime.now(timezone.utc) - start_time).total_seconds() * 1000 + icon = "✅" if response.status_code < 400 else "❌" + logger.info(f"{icon} {request.method} {request.url.path} | 状态: {response.status_code} | 耗时: {ms:.0f}ms") + log_response(response.status_code, ms) + response.headers["X-Process-Time"] = f"{ms:.2f}ms" return response +# ── 路由注册 ────────────────────────────────────────────────────────── + @app.get("/health") async def health_check(): - """健康检查端点""" - return {"status": "healthy", "timestamp": datetime.now(timezone.utc).isoformat()} + return { + "status": "healthy", + "backend": LLM_BACKEND, + "timestamp": datetime.now(timezone.utc).isoformat(), + } @app.post("/api/chat-ui/chat") async def chat_endpoint(request: Request): - """聊天接口 - 与阿里云百炼API兼容的接口""" - body = await request.json() - return await chat_endpoint_handler(body) + """聊天接口（自动路由到当前平台）""" + return await _platform.chat_handler(await request.json()) @app.get("/api/chat-ui/models") async def get_models(): - """获取模型列表""" - return await get_models_handler() + """模型列表（由当前平台返回）""" + result = _platform.models_handler() + # 支持同步和异步两种返回 + if hasattr(result, "__await__"): + return await result + return result +# ── 通用路由（与平台无关）──────────────────────────────────────────── + @app.get("/api/chat-ui/conversations") async def get_conversations(): - """获取所有对话""" return await get_conversations_handler() @app.get("/api/chat-ui/conversations/{conversation_id}") async def get_conversation(conversation_id: str): - """获取特定对话""" return await get_conversation_handler(conversation_id) @app.post("/api/chat-ui/conversations") async def save_conversation(request: Request): - """保存或更新对话""" - data = await request.json() - return await save_conversation_handler(data) + return await save_conversation_handler(await request.json()) @app.delete("/api/chat-ui/conversations/{conversation_id}") async def delete_conversation(conversation_id: str): - """删除对话""" return await delete_conversation_handler(conversation_id) @app.post("/api/chat-ui/upload") async def upload_file(file: UploadFile = File(...)): - """文件上传接口""" return await upload_file_handler(file=file) @app.get("/uploads/{filename}") async def serve_upload(filename: str): - """提供上传文件的访问""" return serve_upload_handler(filename) @app.post("/api/chat-ui/stop") async def stop_generation(): - """停止生成接口""" return await stop_generation_handler() @app.post("/api/chat-ui/stop/{message_id}") async def stop_generation_by_id(message_id: str): - """根据消息ID停止生成""" return await stop_generation_handler(message_id) +# ── 程序入口 ────────────────────────────────────────────────────────── if __name__ == "__main__": import uvicorn - port = int(os.getenv("PORT", 8000)) - print("="*50) - print(f"Python AI Chat Server 启动中...") - print(f"监听端口: {port}") - print(f"API Key 状态: {'已配置' if api_key else '未配置'}") - print("="*50) - - if not api_key: - print("警告: 未在环境变量中检测到 ALIYUN_API_KEY!") - print("请在 .env 文件中添加您的百炼 API Key。") - else: - print("API Key 已检测到。") - + print("=" * 55) + print(f" AI Chat Server v3.0 启动中...") + print(f" 后端平台 : {LLM_BACKEND.upper()} [LLM_BACKEND={LLM_BACKEND}]") + print(f" 监听端口 : {port}") + print(f" 切换平台 : 修改 .env 中 LLM_BACKEND=glm|dashscope，重启") + print("=" * 55) uvicorn.run(app, host="0.0.0.0", port=port) \ No newline at end of file diff --git a/server/utils/file_cache.py b/server/utils/file_cache.py new file mode 100644 index 0000000..619035e --- /dev/null +++ b/server/utils/file_cache.py @@ -0,0 +1,64 @@ +""" +GLM 文件 ID 缓存（基于磁盘的简单 KV，sha256 → file_id，3天有效期） +""" +import hashlib +import json +import time +import threading +from pathlib import Path + +_CACHE_FILE = Path(__file__).parent.parent / "uploads" / ".glm_file_cache.json" +_lock = threading.Lock() +_TTL = 3 * 24 * 3600 # 3天 + + +def _load() -> dict: + try: + if _CACHE_FILE.exists(): + return json.loads(_CACHE_FILE.read_text(encoding="utf-8")) + except Exception: + pass + return {} + + +def _save(data: dict) -> None: + try: + _CACHE_FILE.parent.mkdir(parents=True, exist_ok=True) + _CACHE_FILE.write_text(json.dumps(data, ensure_ascii=False, indent=2), encoding="utf-8") + except Exception as e: + print(f"[file_cache] 写入失败：{e}") + + +def sha256_of_file(file_path: Path) -> str: + h = hashlib.sha256() + with open(file_path, "rb") as f: + for chunk in iter(lambda: f.read(65536), b""): + h.update(chunk) + return h.hexdigest() + + +def get(file_hash: str) -> dict | None: + with _lock: + data = _load() + entry = data.get(file_hash) + if not entry: + return None + if entry.get("expires_at", 0) <= time.time(): + data.pop(file_hash, None) + _save(data) + return None + return entry + + +def set(file_hash: str, file_id: str) -> None: + with _lock: + data = _load() + data[file_hash] = {"file_id": file_id, "expires_at": time.time() + _TTL} + _save(data) + + +def delete(file_hash: str) -> None: + with _lock: + data = _load() + data.pop(file_hash, None) + _save(data) diff --git a/server/utils/glm_adapter.py b/server/utils/glm_adapter.py new file mode 100644 index 0000000..22612bb --- /dev/null +++ b/server/utils/glm_adapter.py @@ -0,0 +1,335 @@ +""" +GLM-4.6V 适配层（基于 zai-sdk） +SDK：pip install zai-sdk +模型：glm-4.6v（支持文本/图像/文档/深度思考） +""" +import os +import sys +import json +import base64 +import threading +from pathlib import Path +from typing import AsyncGenerator + + +# ── 自动注入 venv site-packages ─────────────────────────────────────── +def _ensure_venv(): + server_dir = Path(__file__).parent.parent + for sp in sorted((server_dir / ".venv" / "lib").glob("python*/site-packages"), reverse=True): + if sp.exists() and str(sp) not in sys.path: + sys.path.insert(0, str(sp)) + print(f"[GLM] venv 注入：{sp}") + break + + +# ── 客户端单例 ──────────────────────────────────────────────────────── +_client = None + + +def get_client(): + global _client + if _client is None: + _ensure_venv() + try: + from zai import ZhipuAiClient + except ImportError: + raise ImportError("GLM 模式需要安装 zai-sdk：.venv/bin/pip install zai-sdk") + api_key = os.getenv("ZHIPU_API_KEY") or os.getenv("GLM_API_KEY") + if not api_key: + raise ValueError("GLM 模式需要设置环境变量 ZHIPU_API_KEY") + _client = ZhipuAiClient(api_key=api_key) + print("[GLM] ZhipuAiClient 初始化完成（zai-sdk）") + return _client + + +# ── 模型映射 ────────────────────────────────────────────────────────── +DEFAULT_TEXT_MODEL = "glm-4.6v" # glm-4.6v 文本+视觉统一模型 +DEFAULT_VISION_MODEL = "glm-4.6v" + +MODEL_MAP = { + "qwen-max": "glm-4.6v", + "qwen-plus": "glm-4.6v", + "qwen-turbo": "glm-4.6v", + "qwen-vl-max": "glm-4.6v", + "qwen-vl-plus": "glm-4.6v", +} + + +def resolve_model(model: str, has_vision: bool = False) -> str: + if model.startswith("glm"): + return model + return MODEL_MAP.get(model, DEFAULT_TEXT_MODEL) + + +# ── 文件上传（含 file_id 缓存）─────────────────────────────────────── +def upload_file_for_extract(local_path: Path) -> str: + from utils.file_cache import sha256_of_file, get as cache_get, set as cache_set + + file_hash = sha256_of_file(local_path) + cached = cache_get(file_hash) + if cached: + print(f"[GLM] file_id 缓存命中：{local_path.name} → {cached['file_id']}") + return cached["file_id"] + + client = get_client() + mime_map = { + ".pdf": "application/pdf", + ".docx": "application/vnd.openxmlformats-officedocument.wordprocessingml.document", + ".doc": "application/msword", + ".xlsx": "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet", + ".xls": "application/vnd.ms-excel", + ".pptx": "application/vnd.openxmlformats-officedocument.presentationml.presentation", + ".ppt": "application/vnd.ms-powerpoint", + } + mime = mime_map.get(local_path.suffix.lower(), "application/octet-stream") + print(f"[GLM] 上传文件：{local_path.name}（{mime}）") + with open(local_path, "rb") as f: + file_obj = client.files.create(file=(local_path.name, f, mime), purpose="file-extract") + file_id = file_obj.id + cache_set(file_hash, file_id) + print(f"[GLM] 上传成功：file_id={file_id}") + return file_id + + +# ── 图像编码 ───────────────────────────────────────────────────────── +def encode_image(image_source: str) -> dict: + """将图像来源统一转为 OpenAI image_url 格式""" + if image_source.startswith("data:image") or image_source.startswith(("http://", "https://")): + return {"type": "image_url", "image_url": {"url": image_source}} + # 本地路径 → base64 + local = Path(image_source.replace("file://", "").lstrip("/")) + if not local.exists(): + local = Path.cwd() / local + ext = local.suffix.lstrip(".") + with open(local, "rb") as f: + b64 = base64.b64encode(f.read()).decode() + return {"type": "image_url", "image_url": {"url": f"data:image/{ext};base64,{b64}"}} + + +# ── 消息格式转换 ────────────────────────────────────────────────────── +def build_glm_messages(messages: list, files: list | None = None) -> tuple[list, bool]: + """ + 将 OpenAI 格式的 messages + files 转换为 zai-sdk 所需格式。 + 返回 (glm_messages, has_vision)。 + """ + from urllib.parse import urlparse + + glm_messages = [] + has_vision = False + + for msg in messages: + if not isinstance(msg, dict): + glm_messages.append({"role": "user", "content": str(msg)}) + continue + role = msg.get("role", "user") + content = msg.get("content", "") + + if isinstance(content, str): + glm_messages.append({"role": role, "content": content}) + elif isinstance(content, list): + new_content = [] + for item in content: + if not isinstance(item, dict): + new_content.append({"type": "text", "text": str(item)}) + continue + t = item.get("type") + if t == "text": + new_content.append({"type": "text", "text": item.get("text", "")}) + elif t == "image_url": + has_vision = True + img_val = item.get("image_url", "") + img_src = img_val.get("url", "") if isinstance(img_val, dict) else img_val + new_content.append(encode_image(img_src)) + else: + new_content.append({"type": "text", "text": str(item)}) + glm_messages.append({"role": role, "content": new_content}) + else: + glm_messages.append({"role": role, "content": str(content)}) + + # 处理独立附件列表 + if files: + doc_exts = {".pdf", ".doc", ".docx", ".xlsx", ".xls", ".pptx", ".ppt"} + img_exts = {".jpg", ".jpeg", ".png", ".gif", ".bmp", ".webp"} + inserts = [] + + for file_url in files: + parsed = urlparse(file_url) + filename = parsed.path.split("/")[-1] + suffix = Path(filename).suffix.lower() + rel = parsed.path.lstrip("/") + local = Path(rel) + + if suffix in img_exts: + has_vision = True + try: + inserts.append(encode_image(f"file://{rel}")) + except Exception as e: + print(f"[GLM] 图像编码失败：{e}") + elif suffix in doc_exts: + has_vision = True + if local.exists(): + try: + fid = upload_file_for_extract(local) + inserts.append({"type": "file", "file": {"file_id": fid}}) + except Exception as e: + inserts.append({"type": "text", "text": f"[文件上传失败：{filename}，{e}]"}) + else: + inserts.append({"type": "text", "text": f"[附件：{filename}，类型：{suffix}]"}) + + if inserts: + for i in range(len(glm_messages) - 1, -1, -1): + if glm_messages[i].get("role") == "user": + old = glm_messages[i]["content"] + if isinstance(old, str): + new_content = inserts + [{"type": "text", "text": old}] + elif isinstance(old, list): + new_content = inserts + old + else: + new_content = inserts + glm_messages[i] = {"role": "user", "content": new_content} + break + + return glm_messages, has_vision + + +# ── 哨兵对象 ───────────────────────────────────────────────────────── +_SENTINEL = object() + + +# ── 流式 SSE 生成器 ─────────────────────────────────────────────────── +async def glm_stream_generator( + messages: list, + model: str, + temperature: float, + max_tokens: int, + files: list | None = None, + web_search: bool = False, + deep_thinking: bool = False, +) -> AsyncGenerator[str, None]: + """ + GLM 流式 SSE 生成器。 + 使用 queue.Queue + 专用线程（生产者）+ asyncio 消费者模式， + 让 zai-sdk 同步迭代器在单一线程内安全运行。 + """ + import asyncio + import queue + + from utils.helpers import get_current_timestamp, generate_unique_id + + glm_msgs, has_vision = build_glm_messages(messages, files) + actual_model = resolve_model(model, has_vision) + + extra_kwargs: dict = {} + if web_search: + extra_kwargs["tools"] = [ + {"type": "web_search", "web_search": {"search_result": True}} + ] + if deep_thinking: + extra_kwargs["thinking"] = {"type": "enabled"} + + print(f"[GLM] 流式请求：model={actual_model} vision={has_vision} " + f"web_search={web_search} thinking={deep_thinking}") + + chunk_queue: queue.Queue = queue.Queue(maxsize=128) + + def _producer(): + try: + client = get_client() + resp = client.chat.completions.create( + model=actual_model, + messages=glm_msgs, + stream=True, + temperature=temperature, + max_tokens=max_tokens, + **extra_kwargs, + ) + for chunk in resp: + chunk_queue.put(chunk) + except Exception as exc: + chunk_queue.put(exc) + finally: + chunk_queue.put(_SENTINEL) + + t = threading.Thread(target=_producer, daemon=True) + t.start() + + loop = asyncio.get_running_loop() + + while True: + item = await loop.run_in_executor(None, chunk_queue.get) + + if item is _SENTINEL: + break + + if isinstance(item, Exception): + print(f"[GLM] 生产者异常：{item}") + yield f"data: {json.dumps({'error': {'message': str(item), 'type': 'glm_error'}}, ensure_ascii=False)}\n\n" + break + + try: + delta = item.choices[0].delta + text = getattr(delta, "content", "") or "" + if not text: + continue + data = { + "id": f"chatcmpl-{generate_unique_id()}", + "object": "chat.completion.chunk", + "created": get_current_timestamp(), + "model": actual_model, + "choices": [{"index": 0, "delta": {"content": text}, "finish_reason": None}], + } + yield f"data: {json.dumps(data, ensure_ascii=False)}\n\n" + except Exception as e: + print(f"[GLM] chunk 解析异常：{e}") + + finish = { + "id": f"chatcmpl-{generate_unique_id()}", + "object": "chat.completion.chunk", + "created": get_current_timestamp(), + "model": actual_model, + "choices": [{"index": 0, "delta": {}, "finish_reason": "stop"}], + } + yield f"data: {json.dumps(finish, ensure_ascii=False)}\n\n" + yield "data: [DONE]\n\n" + + +# ── 非流式调用 ──────────────────────────────────────────────────────── +def glm_chat_sync( + messages: list, + model: str, + temperature: float, + max_tokens: int, + files: list | None = None, + web_search: bool = False, + deep_thinking: bool = False, +) -> dict: + glm_msgs, has_vision = build_glm_messages(messages, files) + actual_model = resolve_model(model, has_vision) + + extra_kwargs: dict = {} + if web_search: + extra_kwargs["tools"] = [ + {"type": "web_search", "web_search": {"search_result": True}} + ] + if deep_thinking: + extra_kwargs["thinking"] = {"type": "enabled"} + + client = get_client() + print(f"[GLM] 非流式请求：model={actual_model}") + resp = client.chat.completions.create( + model=actual_model, + messages=glm_msgs, + stream=False, + temperature=temperature, + max_tokens=max_tokens, + **extra_kwargs, + ) + content = resp.choices[0].message.content or "" + usage = None + if hasattr(resp, "usage") and resp.usage: + usage = { + "promptTokens": resp.usage.prompt_tokens, + "completionTokens": resp.usage.completion_tokens, + "totalTokens": resp.usage.total_tokens, + } + return {"content": content, "model": actual_model, "usage": usage}