Merge pull request #2097 from SheffeyG/fix-status-checking

Fix: add status checking for embedding model providers

astrbot/core/provider/manager.py

@@ -7,7 +7,7 @@ from astrbot.core.config.astrbot_config import AstrBotConfig
 from astrbot.core.db import BaseDatabase
 
 from .entities import ProviderType
-from .provider import Personality, Provider, STTProvider, TTSProvider
+from .provider import Personality, Provider, STTProvider, TTSProvider, EmbeddingProvider
 from .register import llm_tools, provider_cls_map
 
 
@@ -93,7 +93,7 @@ class ProviderManager:
         """加载的 Speech To Text Provider 的实例"""
         self.tts_provider_insts: List[TTSProvider] = []
         """加载的 Text To Speech Provider 的实例"""
-        self.embedding_provider_insts: List[Provider] = []
+        self.embedding_provider_insts: List[EmbeddingProvider] = []
         """加载的 Embedding Provider 的实例"""
         self.inst_map: dict[str, Provider] = {}
         """Provider 实例映射. key: provider_id, value: Provider 实例"""

astrbot/core/star/context.py

@@ -2,7 +2,7 @@ from asyncio import Queue
 from typing import List, Union
 
 from astrbot.core import sp
-from astrbot.core.provider.provider import Provider, TTSProvider, STTProvider
+from astrbot.core.provider.provider import Provider, TTSProvider, STTProvider, EmbeddingProvider
 from astrbot.core.provider.entities import ProviderType
 from astrbot.core.db import BaseDatabase
 from astrbot.core.config.astrbot_config import AstrBotConfig
@@ -141,6 +141,10 @@ class Context:
         """获取所有用于 STT 任务的 Provider。"""
         return self.provider_manager.stt_provider_insts
 
+    def get_all_embedding_providers(self) -> List[EmbeddingProvider]:
+        """获取所有用于 Embedding 任务的 Provider。"""
+        return self.provider_manager.embedding_provider_insts
+
     def get_using_provider(self, umo: str = None) -> Provider:
         """
         获取当前使用的用于文本生成任务的 LLM Provider(Chat_Completion 类型)。通过 /provider 指令切换。

astrbot/dashboard/routes/config.py

@@ -1,3 +1,4 @@
+import numbers
 import typing
 import traceback
 from .route import Route, Response, RouteContext
@@ -187,6 +188,7 @@ class ConfigRoute(Route):
         """辅助函数：测试单个 provider 的可用性"""
         meta = provider.meta()
         provider_name = provider.provider_config.get("id", "Unknown Provider")
+        provider_type = provider.provider_config.get("provider_type", "Unknown Type")
         logger.debug(f"Got provider meta: {meta}")
         if not provider_name and meta:
             provider_name = meta.id
@@ -197,6 +199,7 @@ class ConfigRoute(Route):
             "model": getattr(meta, "model", "Unknown Model"),
             "type": getattr(meta, "type", "Unknown Type"),
             "name": provider_name,
+            "provider_type": provider_type,
             "status": "unavailable",  # 默认为不可用
             "error": None,
         }
@@ -204,40 +207,60 @@ class ConfigRoute(Route):
             f"Attempting to check provider: {status_info['name']} (ID: {status_info['id']}, Type: {status_info['type']}, Model: {status_info['model']})"
         )
         try:
-            logger.debug(f"Sending 'Ping' to provider: {status_info['name']}")
-            response = await asyncio.wait_for(
-                provider.text_chat(prompt="REPLY `PONG` ONLY"), timeout=45.0
-            )
-            logger.debug(f"Received response from {status_info['name']}: {response}")
-            # 只要 text_chat 调用成功返回一个 LLMResponse 对象 (即 response 不为 None)，就认为可用
-            if response is not None:
-                status_info["status"] = "available"
-                response_text_snippet = ""
-                if hasattr(response, "completion_text") and response.completion_text:
-                    response_text_snippet = (
-                        response.completion_text[:70] + "..."
-                        if len(response.completion_text) > 70
-                        else response.completion_text
-                    )
-                elif hasattr(response, "result_chain") and response.result_chain:
-                    try:
-                        response_text_snippet = (
-                            response.result_chain.get_plain_text()[:70] + "..."
-                            if len(response.result_chain.get_plain_text()) > 70
-                            else response.result_chain.get_plain_text()
-                        )
-                    except Exception as _:
-                        pass
-                logger.info(
-                    f"Provider {status_info['name']} (ID: {status_info['id']}) is available. Response snippet: '{response_text_snippet}'"
+            if status_info["provider_type"] == "chat_completion":
+                logger.debug(f"Sending 'Ping' to provider: {status_info['name']}")
+                response = await asyncio.wait_for(
+                    provider.text_chat(prompt="REPLY `PONG` ONLY"), timeout=45.0
                 )
+                logger.debug(f"Received response from {status_info['name']}: {response}")
+                # 只要 text_chat 调用成功返回一个 LLMResponse 对象 (即 response 不为 None)，就认为可用
+                if response is not None:
+                    status_info["status"] = "available"
+                    response_text_snippet = ""
+                    if hasattr(response, "completion_text") and response.completion_text:
+                        response_text_snippet = (
+                            response.completion_text[:70] + "..."
+                            if len(response.completion_text) > 70
+                            else response.completion_text
+                        )
+                    elif hasattr(response, "result_chain") and response.result_chain:
+                        try:
+                            response_text_snippet = (
+                                response.result_chain.get_plain_text()[:70] + "..."
+                                if len(response.result_chain.get_plain_text()) > 70
+                                else response.result_chain.get_plain_text()
+                            )
+                        except Exception as _:
+                            pass
+                    logger.info(
+                        f"Provider {status_info['name']} (ID: {status_info['id']}) is available. Response snippet: '{response_text_snippet}'"
+                    )
+                else:
+                    # 这个分支理论上不应该被走到，除非 text_chat 实现可能返回 None
+                    status_info["error"] = (
+                        "Test call returned None, but expected an LLMResponse object."
+                    )
+                    logger.warning(
+                        f"Provider {status_info['name']} (ID: {status_info['id']}) test call returned None."
+                    )
+            elif status_info["provider_type"] == "embedding":
+                logger.debug(f"Sending 'astrbot' to embedding provider: {status_info['name']}")
+                response = await asyncio.wait_for(
+                    provider.get_embedding("astrbot"), timeout=45.0
+                )
+                logger.debug(f"Received response from {status_info['name']}: {response}")
+                # 若返回向量则认为该嵌入模型可用
+                if response and isinstance(response, typing.Iterable) and all(isinstance(x, numbers.Number) for x in response):
+                    status_info["status"] = "available"
+                    logger.info(
+                        f"Provider {status_info['name']} (ID: {status_info['id']}) is available. Response snippet: '{str(response)[:10]}...'"
+                    )
             else:
-                # 这个分支理论上不应该被走到，除非 text_chat 实现可能返回 None
                 status_info["error"] = (
-                    "Test call returned None, but expected an LLMResponse object."
+                    f"Status checking for provider type '{status_info['type']}' not implemented."
                 )
                 logger.warning(
-                    f"Provider {status_info['name']} (ID: {status_info['id']}) test call returned None."
+                    f"Provider {status_info['name']}'s status checking not implemented yet"
                 )
 
         except asyncio.TimeoutError:
@@ -263,7 +286,7 @@ class ConfigRoute(Route):
         # 记录更详细的traceback信息，但只在是严重错误时
         if status_code == 500:
             log_fn(traceback.format_exc())
-        return Response().error(message, status_code=status_code).__dict__
+        return Response().error(message).__dict__
 
     async def check_one_provider_status(self):
         """API: check a single LLM Provider's status by id"""
@@ -274,11 +297,9 @@ class ConfigRoute(Route):
         logger.info(f"API call: /config/provider/check_one id={provider_id}")
         try:
             all_providers = self.core_lifecycle.star_context.get_all_providers()
+            all_providers += self.core_lifecycle.star_context.get_all_embedding_providers()
             # replace manual loop with next(filter(...))
-            target = next(
-                (p for p in all_providers if p.provider_config.get("id") == provider_id),
-                None
-            )
+            target = next(filter(lambda p: p.provider_config.get("id") == provider_id, all_providers), None)
             if not target:
                 return self._error_response(f"Provider with id '{provider_id}' not found", 404, logger.warning)