shuiyu67
dfc42d8583
- openai_engine: 请求体严格纯官方 OpenAI /chat/completions 格式（tool 消息去 name、 tool_calls 重建、tools 统一 type=function）；500(含 xunfei/Invalid Params)自动去掉 tools 降级重试；_friendly_error 友好提示；MODEL_CONTEXT_WINDOWS 补讯飞星火。 - ollama_engine/fc_tools/cli_core: tool 消息去 name。 - config.py resolve_input_value(<input> 语法)；ollama max_token_k(默认 <input:...>)； launcher 向导加 max_token_k 步骤。
538 lines
22 KiB
Python
538 lines
22 KiB
Python
"""
|
||
OpenAI 兼容格式引擎
|
||
支持所有兼容 OpenAI API 格式的服务（如 DeepSeek、Grok、硅基流动、本地 vLLM 等）
|
||
"""
|
||
|
||
import os
|
||
import sys
|
||
import json
|
||
import requests
|
||
from colorama import Fore, Style
|
||
import logging
|
||
|
||
# 设置日志
|
||
logger = logging.getLogger(__name__)
|
||
|
||
try:
|
||
from xcli_core.config import get_system_config
|
||
except Exception:
|
||
def get_system_config(key, default=None):
|
||
return default
|
||
|
||
# ── 常见模型的上下文窗口（token 数）──
|
||
# OpenAI /chat/completions 协议本身不返回模型窗口，需本地维护。
|
||
MODEL_CONTEXT_WINDOWS = {
|
||
# 讯飞星火（Spark，经 one-api/new-api 转发时模型名常为 spark-* / general*）
|
||
"spark-4.0": 128000, "spark-max": 8192, "spark-pro": 8192, "spark-lite": 8192,
|
||
"spark": 8192, "generalv3.5": 8192, "generalv3": 8192, "general": 4096,
|
||
"gpt-4o": 128000, "gpt-4o-mini": 128000, "gpt-4": 8192, "gpt-4-turbo": 128000,
|
||
"gpt-3.5-turbo": 16385,
|
||
"deepseek-chat": 64000, "deepseek-reasoner": 64000,
|
||
"qwen2.5": 32768, "qwen2": 32768, "qwen-max": 32768, "qwen-plus": 32768,
|
||
"llama3": 8192, "llama3.1": 128000, "llama3.2": 128000, "llama2": 4096,
|
||
"claude-3": 200000, "claude-3.5": 200000, "claude-3.7": 200000, "claude": 200000,
|
||
"grok": 131072, "grok-2": 131072, "grok-3": 131072,
|
||
"glm-4": 128000, "glm-4v": 128000,
|
||
"mistral": 32768, "mixtral": 32768, "mistral-large": 128000,
|
||
"gemini": 1000000, "gemini-1.5": 1000000, "gemini-2.0": 1000000, "gemini-2.5": 1000000,
|
||
"yi": 200000, "yi-large": 200000,
|
||
"kimi": 200000, "moonshot": 200000,
|
||
"abab": 200000, "minimax": 200000,
|
||
"baichuan": 192000, "chatglm": 32768,
|
||
}
|
||
|
||
# 添加项目根目录到sys.path
|
||
current_dir = os.path.dirname(os.path.abspath(__file__))
|
||
project_root = os.path.dirname(current_dir)
|
||
sys.path.insert(0, project_root)
|
||
|
||
|
||
class OpenaiAI:
|
||
"""OpenAI 兼容格式 AI 引擎 - 支持所有兼容 OpenAI API 的服务"""
|
||
requires_api_key = True
|
||
|
||
def get_help_info(self):
|
||
"""获取引擎帮助信息"""
|
||
help_text = """
|
||
OpenAI 兼容格式引擎帮助信息
|
||
============================
|
||
|
||
引擎名称: openai
|
||
描述: 通用 OpenAI API 兼容引擎，支持所有兼容 OpenAI 格式的服务
|
||
|
||
支持的服务:
|
||
- OpenAI 官方 API
|
||
- DeepSeek API
|
||
- Grok (xAI) API
|
||
- 硅基流动 (SiliconFlow) API
|
||
- 本地 vLLM / LiteLLM / OneAPI 等转发服务
|
||
- 任何兼容 /chat/completions 接口的服务
|
||
|
||
配置方式 (config.json):
|
||
{
|
||
"api": {
|
||
"engines": {
|
||
"openai": {
|
||
"api_key": "你的API密钥",
|
||
"base_url": "https://api.openai.com/v1",
|
||
"model": "gpt-4o"
|
||
}
|
||
}
|
||
}
|
||
}
|
||
|
||
可用命令:
|
||
/engine.openai models - 列出可用模型（需要服务端支持）
|
||
/engine.openai set <模型名> - 切换模型
|
||
/engine.openai info - 显示当前配置信息
|
||
|
||
使用说明:
|
||
1. 在 config.json 中配置 api_key、base_url、model
|
||
2. 使用 /engine switch openai 切换到此引擎
|
||
3. 直接输入自然语言与 AI 对话
|
||
|
||
注意:
|
||
- base_url 填写 API 的基础地址，无需包含 /chat/completions
|
||
- 例如 OpenAI 官方填 https://api.openai.com/v1
|
||
- 本地服务填 http://localhost:8000/v1
|
||
"""
|
||
return help_text
|
||
|
||
def __init__(self):
|
||
self.name = "openai"
|
||
self.cli = None
|
||
|
||
# 从 config.json 读取配置
|
||
config_file = os.path.join(project_root, "config.json")
|
||
try:
|
||
with open(config_file, 'r', encoding='utf-8') as f:
|
||
config = json.load(f)
|
||
engine_config = config.get("api", {}).get("engines", {}).get("openai", {})
|
||
self.api_key = engine_config.get("api_key", "")
|
||
self.base_url = engine_config.get("base_url", "").rstrip('/')
|
||
self.model = engine_config.get("model", "")
|
||
self.max_history = config.get("system", {}).get("max_history", 10)
|
||
except Exception as e:
|
||
logger.warning(f"读取配置文件失败: {e}")
|
||
self.api_key = ""
|
||
self.base_url = ""
|
||
self.model = ""
|
||
self.max_history = 10
|
||
|
||
# 初始化对话历史
|
||
self.conversation_history = []
|
||
self.shared_conversation_history = None
|
||
|
||
# 最近一次请求的 token 用量（供 token 感知压缩使用）
|
||
self.last_prompt_tokens = None
|
||
|
||
# 思考标记（兼容深度思考模型）
|
||
self.thinking_start_marker = "<think>"
|
||
self.thinking_end_marker = "</think>"
|
||
|
||
# 检查配置
|
||
if not self.api_key:
|
||
print(f"{Fore.YELLOW}警告: openai 引擎的 API 密钥未设置{Style.RESET_ALL}")
|
||
if not self.base_url:
|
||
print(f"{Fore.YELLOW}警告: openai 引擎的 base_url 未设置{Style.RESET_ALL}")
|
||
else:
|
||
print(f"{Fore.GREEN}OpenAI 兼容引擎已启用 (base_url: {self.base_url}){Style.RESET_ALL}")
|
||
|
||
def _get_headers(self):
|
||
"""构建请求头"""
|
||
headers = {
|
||
'Content-Type': 'application/json'
|
||
}
|
||
if self.api_key:
|
||
headers['Authorization'] = f'Bearer {self.api_key}'
|
||
return headers
|
||
|
||
def _get_api_url(self):
|
||
"""获取完整的 API 地址"""
|
||
url = self.base_url
|
||
if not url:
|
||
return ""
|
||
if not url.endswith('/chat/completions'):
|
||
url = url.rstrip('/') + '/chat/completions'
|
||
return url
|
||
|
||
def generate_response(self, user_input, tool_results=None, system_prompt=None, tools=None):
|
||
"""
|
||
生成 AI 响应
|
||
|
||
Args:
|
||
user_input: 用户输入
|
||
tool_results: 工具执行结果
|
||
system_prompt: 系统提示词
|
||
tools: OpenAI FC 工具定义列表 (可选)
|
||
"""
|
||
try:
|
||
if not self.base_url:
|
||
return "错误: openai 引擎的 base_url 未配置，请在 config.json 中设置"
|
||
|
||
# 处理工具结果
|
||
if tool_results:
|
||
tool_result_text = f"工具执行结果: {tool_results.get('result', '无结果')}"
|
||
messages = self._build_messages_with_history(system_prompt, tool_result_text)
|
||
else:
|
||
messages = self._build_messages_with_history(system_prompt, user_input)
|
||
|
||
# 构建请求体
|
||
body = {
|
||
"model": self.model,
|
||
"messages": messages,
|
||
"stream": False
|
||
}
|
||
|
||
# ── Function Calling: 注入工具定义（统一纯官方格式）──
|
||
if tools:
|
||
pure_tools = []
|
||
for t in tools:
|
||
if not isinstance(t, dict):
|
||
continue
|
||
if t.get("type") == "function" and isinstance(t.get("function"), dict):
|
||
pure_tools.append(t)
|
||
elif isinstance(t.get("function"), dict):
|
||
pure_tools.append({"type": "function", "function": t["function"]})
|
||
if pure_tools:
|
||
body["tools"] = pure_tools
|
||
body["tool_choice"] = "auto"
|
||
|
||
# 发送请求
|
||
api_url = self._get_api_url()
|
||
response = requests.post(
|
||
url=api_url,
|
||
json=body,
|
||
headers=self._get_headers(),
|
||
timeout=None
|
||
)
|
||
|
||
if response.status_code == 200:
|
||
response_data = response.json()
|
||
# ── 记录 token 用量（供 token 感知压缩）──
|
||
usage = response_data.get("usage", {})
|
||
if isinstance(usage, dict) and usage.get("prompt_tokens"):
|
||
try:
|
||
self.last_prompt_tokens = int(usage["prompt_tokens"])
|
||
except (TypeError, ValueError):
|
||
self.last_prompt_tokens = None
|
||
if "choices" in response_data and len(response_data["choices"]) > 0:
|
||
message = response_data["choices"][0].get("message", {})
|
||
|
||
# ── Function Calling: 检查 tool_calls ──
|
||
tool_calls_raw = message.get("tool_calls")
|
||
if tool_calls_raw:
|
||
# 返回 FC 结果（由调用方处理工具执行）
|
||
from xcli_core.fc_tools import parse_fc_response
|
||
fc_calls = parse_fc_response(response_data)
|
||
if fc_calls:
|
||
# 返回一个特殊的 JSON 字符串，标记为 FC 调用
|
||
return json.dumps({
|
||
"_fc": True,
|
||
"tool_calls": fc_calls,
|
||
"raw_message": message
|
||
}, ensure_ascii=False)
|
||
|
||
# 普通文本响应
|
||
ai_response = message.get("content", "")
|
||
if ai_response:
|
||
# 清理无效的 UTF-8 代理字符
|
||
ai_response = ai_response.encode('utf-8', 'replace').decode('utf-8')
|
||
# 处理思考内容
|
||
ai_response = self._process_thinking_content(ai_response)
|
||
return ai_response or ""
|
||
else:
|
||
return f"API 返回格式错误: {response_data}"
|
||
elif response.status_code == 401:
|
||
return "API 认证失败，请检查 api_key 是否正确"
|
||
elif response.status_code == 404:
|
||
return f"API 端点不存在，请检查 base_url 是否正确: {api_url}"
|
||
elif response.status_code == 400:
|
||
# 工具定义可能不兼容，降级为无工具重试
|
||
if tools and "tools" in str(response.text).lower():
|
||
print(f"{Fore.YELLOW}FC 不支持，降级为普通模式{Style.RESET_ALL}")
|
||
body.pop("tools", None)
|
||
body.pop("tool_choice", None)
|
||
retry = requests.post(url=api_url, json=body, headers=self._get_headers(), timeout=None)
|
||
if retry.status_code == 200:
|
||
rd = retry.json()
|
||
if "choices" in rd and rd["choices"]:
|
||
ai_response = rd["choices"][0].get("message", {}).get("content", "")
|
||
return ai_response.encode('utf-8', 'replace').decode('utf-8') if ai_response else ""
|
||
return self._friendly_error(response.status_code, response)
|
||
else:
|
||
# 500 等错误：部分服务（如讯飞星火经 one-api 转发）不支持 FC 工具，
|
||
# 会直接报 500 Invalid Params —— 尝试去掉 tools 降级重试一次
|
||
lower = str(response.text).lower()
|
||
if tools and ("xunfei" in lower or "invalid params" in lower or "requestparamserror" in lower):
|
||
print(f"{Fore.YELLOW}该接口可能不支持 FC 工具，降级为普通模式重试...{Style.RESET_ALL}")
|
||
body.pop("tools", None)
|
||
body.pop("tool_choice", None)
|
||
retry = requests.post(url=api_url, json=body, headers=self._get_headers(), timeout=None)
|
||
if retry.status_code == 200:
|
||
rd = retry.json()
|
||
retry_usage = rd.get("usage", {})
|
||
if isinstance(retry_usage, dict) and retry_usage.get("prompt_tokens"):
|
||
try:
|
||
self.last_prompt_tokens = int(retry_usage["prompt_tokens"])
|
||
except (TypeError, ValueError):
|
||
pass
|
||
if "choices" in rd and rd["choices"]:
|
||
ai_response = rd["choices"][0].get("message", {}).get("content", "")
|
||
ai_response = self._process_thinking_content(ai_response)
|
||
return ai_response.encode('utf-8', 'replace').decode('utf-8') if ai_response else ""
|
||
return self._friendly_error(response.status_code, response)
|
||
|
||
except requests.exceptions.ConnectionError:
|
||
return f"无法连接到 API 服务: {self.base_url}，请检查服务是否运行"
|
||
except Exception as e:
|
||
return f"API 调用失败: {str(e)}"
|
||
|
||
def _build_messages_with_history(self, system_prompt, user_input):
|
||
"""构建包含历史记录的消息列表（支持 FC tool 消息）"""
|
||
messages = []
|
||
|
||
# 添加系统提示词
|
||
if system_prompt:
|
||
clean_prompt = system_prompt.encode('utf-8', 'replace').decode('utf-8')
|
||
messages.append({
|
||
"role": "system",
|
||
"content": clean_prompt
|
||
})
|
||
|
||
# 添加历史对话记录（使用共享历史）
|
||
if self.shared_conversation_history is not None:
|
||
for msg in self.shared_conversation_history[-self.max_history:]:
|
||
if not isinstance(msg, dict) or "role" not in msg:
|
||
continue
|
||
|
||
role = msg["role"]
|
||
|
||
# tool 角色消息（FC 工具结果）—— 官方 OpenAI 格式不含 name 字段
|
||
if role == "tool":
|
||
tool_msg = {"role": "tool", "content": msg.get("content", "")}
|
||
if "tool_call_id" in msg:
|
||
tool_msg["tool_call_id"] = msg["tool_call_id"]
|
||
messages.append(tool_msg)
|
||
continue
|
||
|
||
# assistant 消息可能包含 tool_calls —— 重建成纯官方结构
|
||
if role == "assistant" and "tool_calls" in msg:
|
||
assistant_msg = {
|
||
"role": "assistant",
|
||
"content": msg.get("content") or "",
|
||
"tool_calls": self._pure_tool_calls(msg["tool_calls"])
|
||
}
|
||
messages.append(assistant_msg)
|
||
continue
|
||
|
||
# 普通 user/assistant 消息
|
||
if "content" in msg:
|
||
clean_content = msg["content"]
|
||
if isinstance(clean_content, str):
|
||
clean_content = clean_content.encode('utf-8', 'replace').decode('utf-8')
|
||
messages.append({"role": role, "content": clean_content})
|
||
|
||
# 添加当前用户输入
|
||
if user_input:
|
||
clean_input = user_input.encode('utf-8', 'replace').decode('utf-8')
|
||
messages.append({
|
||
"role": "user",
|
||
"content": clean_input
|
||
})
|
||
|
||
return messages
|
||
|
||
def _call_api_with_history(self, content, system_prompt=None):
|
||
"""使用对话历史调用 API"""
|
||
if not self.base_url:
|
||
return "错误: openai 引擎的 base_url 未配置"
|
||
|
||
messages = []
|
||
if system_prompt:
|
||
messages.append({"role": "system", "content": system_prompt})
|
||
if self.shared_conversation_history is not None:
|
||
for msg in self.shared_conversation_history[-self.max_history:]:
|
||
if isinstance(msg, dict) and "role" in msg and "content" in msg:
|
||
messages.append(msg)
|
||
messages.append({"role": "user", "content": content})
|
||
|
||
body = {
|
||
"model": self.model,
|
||
"messages": messages,
|
||
"stream": False
|
||
}
|
||
|
||
try:
|
||
response = requests.post(
|
||
url=self._get_api_url(),
|
||
json=body,
|
||
headers=self._get_headers(),
|
||
timeout=None
|
||
)
|
||
|
||
if response.status_code == 200:
|
||
response_data = response.json()
|
||
if "choices" in response_data and len(response_data["choices"]) > 0:
|
||
ai_response = response_data["choices"][0]["message"]["content"]
|
||
ai_response = ai_response.encode('utf-8', 'replace').decode('utf-8')
|
||
return self._process_thinking_content(ai_response)
|
||
return f"API 调用失败: {response.status_code} - {response.text}"
|
||
except Exception as e:
|
||
return f"API 调用失败: {str(e)}"
|
||
|
||
def _process_thinking_content(self, response):
|
||
"""处理思考内容，将思考部分以灰色文本显示"""
|
||
import re
|
||
|
||
if not self.thinking_start_marker or not self.thinking_end_marker:
|
||
return response
|
||
|
||
if self.thinking_start_marker in response and self.thinking_end_marker in response:
|
||
parts = response.split(self.thinking_start_marker)
|
||
if len(parts) > 1:
|
||
remaining = parts[1]
|
||
if self.thinking_end_marker in remaining:
|
||
thinking_part, reply_part = remaining.split(self.thinking_end_marker, 1)
|
||
thinking_content = thinking_part.strip()
|
||
reply_content = reply_part.strip()
|
||
if thinking_content:
|
||
print(f"{Fore.LIGHTBLACK_EX}[思考: {thinking_content}]{Style.RESET_ALL}")
|
||
return reply_content if reply_content else ""
|
||
|
||
return response
|
||
|
||
@staticmethod
|
||
def _friendly_error(status_code, response):
|
||
"""把错误响应里的可读信息提取出来，附带常见原因提示"""
|
||
try:
|
||
data = response.json()
|
||
err = data.get("error", {})
|
||
msg = err.get("message", "") if isinstance(err, dict) else ""
|
||
except Exception:
|
||
msg = ""
|
||
if not msg:
|
||
return f"API 调用失败: {status_code} - {response.text}"
|
||
|
||
hint = ""
|
||
low = str(msg).lower()
|
||
if "invalid params" in low or "requestparamserror" in low or "xunfei" in low:
|
||
hint = ("\n提示: 讯飞星火接口报'参数错误'，常见原因:"
|
||
"\n 1) model 名与 one-api/new-api 渠道里的模型映射不一致（如 spark-4.0-ultra / spark-max / generalv3.5）"
|
||
"\n 2) 该模型/渠道不支持 function calling（引擎已尝试去掉 tools 重试）"
|
||
"\n 3) base_url 需为 OpenAI 兼容转发地址（如 one-api 的 https://域名/v1）")
|
||
return f"API 调用失败: {status_code} - {msg}{hint}"
|
||
|
||
@staticmethod
|
||
def _pure_tool_calls(tool_calls):
|
||
"""把历史里的 tool_calls 重建成官方 OpenAI 格式:
|
||
[{"id": ..., "type": "function", "function": {"name": ..., "arguments": "json串"}}]
|
||
剔除内部字段（action/tool/args 等），arguments 统一为字符串。
|
||
"""
|
||
pure = []
|
||
for tc in (tool_calls or []):
|
||
if not isinstance(tc, dict):
|
||
continue
|
||
func = tc.get("function")
|
||
if isinstance(func, dict):
|
||
name = func.get("name", tc.get("name", ""))
|
||
args = func.get("arguments", "{}")
|
||
else:
|
||
name = tc.get("name", "")
|
||
args = tc.get("arguments", "{}")
|
||
if isinstance(args, str):
|
||
args_str = args
|
||
else:
|
||
try:
|
||
args_str = json.dumps(args, ensure_ascii=False)
|
||
except (TypeError, ValueError):
|
||
args_str = "{}"
|
||
pure.append({
|
||
"id": tc.get("id", f"call_{name}"),
|
||
"type": "function",
|
||
"function": {"name": name, "arguments": args_str}
|
||
})
|
||
return pure
|
||
|
||
def set_model(self, model_name):
|
||
"""设置模型"""
|
||
self.model = model_name
|
||
print(f"{Fore.GREEN}已切换模型为: {model_name}{Style.RESET_ALL}")
|
||
|
||
def context_window(self):
|
||
"""
|
||
返回当前模型的上下文窗口（token 数）。
|
||
优先级: config.system.context_window > 模型名查表 > 默认 128000。
|
||
注: OpenAI /chat/completions 协议本身不返回模型窗口，需本地维护。
|
||
"""
|
||
cfg = get_system_config("context_window", None)
|
||
if cfg:
|
||
try:
|
||
return int(cfg)
|
||
except (TypeError, ValueError):
|
||
pass
|
||
model = (self.model or "").lower()
|
||
for key, win in MODEL_CONTEXT_WINDOWS.items():
|
||
if key and key in model:
|
||
return win
|
||
return 128000
|
||
|
||
def list_models(self):
|
||
"""列出可用模型（调用 API 的 /models 端点）"""
|
||
if not self.base_url:
|
||
return ["错误: base_url 未配置"]
|
||
|
||
try:
|
||
models_url = self.base_url.rstrip('/')
|
||
if models_url.endswith('/chat/completions'):
|
||
models_url = models_url[:-len('/chat/completions')]
|
||
models_url = models_url.rstrip('/') + '/models'
|
||
|
||
response = requests.get(
|
||
url=models_url,
|
||
headers=self._get_headers(),
|
||
timeout=10
|
||
)
|
||
|
||
if response.status_code == 200:
|
||
data = response.json()
|
||
if "data" in data:
|
||
models = [m.get("id", "unknown") for m in data["data"]]
|
||
return models
|
||
else:
|
||
return ["API 返回格式不标准，无法解析模型列表"]
|
||
else:
|
||
return [f"获取模型列表失败: {response.status_code}"]
|
||
except Exception as e:
|
||
return [f"获取模型列表失败: {str(e)}"]
|
||
|
||
def handle_command(self, command):
|
||
"""处理引擎特定命令"""
|
||
if command == "models":
|
||
models = self.list_models()
|
||
print(f"{Fore.GREEN}可用模型列表:{Style.RESET_ALL}")
|
||
for model in models:
|
||
status = " (当前使用)" if model == self.model else ""
|
||
print(f"{Fore.GREEN} - {model}{status}{Style.RESET_ALL}")
|
||
return True
|
||
|
||
elif command.startswith("set "):
|
||
model_name = command[4:].strip()
|
||
if model_name:
|
||
self.set_model(model_name)
|
||
return True
|
||
else:
|
||
print(f"{Fore.RED}请提供模型名称. 用法: /engine.openai set <模型名>{Style.RESET_ALL}")
|
||
return False
|
||
|
||
elif command == "info":
|
||
print(f"{Fore.GREEN}OpenAI 兼容引擎配置:{Style.RESET_ALL}")
|
||
print(f" base_url: {self.base_url}")
|
||
print(f" model: {self.model}")
|
||
print(f" api_key: {'已设置' if self.api_key else '未设置'}")
|
||
return True
|
||
|
||
else:
|
||
print(f"{Fore.RED}未知命令. 可用命令: models, set, info{Style.RESET_ALL}")
|
||
return False
|