refactor: image model get_num_tokens override

apps/setting/models_provider/impl/aliyun_bai_lian_model_provider/model/image.py

@@ -2,12 +2,11 @@
 
 from typing import Dict
 
-from langchain_community.chat_models import ChatOpenAI
-
 from setting.models_provider.base_model_provider import MaxKBBaseModel
+from setting.models_provider.impl.base_chat_open_ai import BaseChatOpenAI
 
 
-class QwenVLChatModel(MaxKBBaseModel, ChatOpenAI):
+class QwenVLChatModel(MaxKBBaseModel, BaseChatOpenAI):
 
     @staticmethod
     def new_instance(model_type, model_name, model_credential: Dict[str, object], **model_kwargs):
@@ -18,6 +17,7 @@ class QwenVLChatModel(MaxKBBaseModel, ChatOpenAI):
             openai_api_base='https://dashscope.aliyuncs.com/compatible-mode/v1',
             # stream_options={"include_usage": True},
             streaming=True,
+            stream_usage=True,
             **optional_params,
         )
         return chat_tong_yi

apps/setting/models_provider/impl/azure_model_provider/model/image.py

@@ -1,7 +1,7 @@
-from typing import Dict
+from typing import Dict, List
 
+from langchain_core.messages import BaseMessage, get_buffer_string
 from langchain_openai import AzureChatOpenAI
-from langchain_openai.chat_models import ChatOpenAI
 
 from common.config.tokenizer_manage_config import TokenizerManage
 from setting.models_provider.base_model_provider import MaxKBBaseModel
@@ -26,3 +26,17 @@ class AzureOpenAIImage(MaxKBBaseModel, AzureChatOpenAI):
             streaming=True,
             **optional_params,
         )
+
+    def get_num_tokens_from_messages(self, messages: List[BaseMessage]) -> int:
+        try:
+            return super().get_num_tokens_from_messages(messages)
+        except Exception as e:
+            tokenizer = TokenizerManage.get_tokenizer()
+            return sum([len(tokenizer.encode(get_buffer_string([m]))) for m in messages])
+
+    def get_num_tokens(self, text: str) -> int:
+        try:
+            return super().get_num_tokens(text)
+        except Exception as e:
+            tokenizer = TokenizerManage.get_tokenizer()
+            return len(tokenizer.encode(text))

apps/setting/models_provider/impl/ollama_model_provider/model/image.py

@@ -1,15 +1,8 @@
 from typing import Dict
 from urllib.parse import urlparse, ParseResult
 
-from langchain_openai.chat_models import ChatOpenAI
-
-from common.config.tokenizer_manage_config import TokenizerManage
 from setting.models_provider.base_model_provider import MaxKBBaseModel
-
+from setting.models_provider.impl.base_chat_open_ai import BaseChatOpenAI
-
-def custom_get_token_ids(text: str):
-    tokenizer = TokenizerManage.get_tokenizer()
-    return tokenizer.encode(text)
 
 
 def get_base_url(url: str):
@@ -20,7 +13,7 @@ def get_base_url(url: str):
     return result_url[:-1] if result_url.endswith("/") else result_url
 
 
-class OllamaImage(MaxKBBaseModel, ChatOpenAI):
+class OllamaImage(MaxKBBaseModel, BaseChatOpenAI):
 
     @staticmethod
     def new_instance(model_type, model_name, model_credential: Dict[str, object], **model_kwargs):
@@ -34,5 +27,6 @@ class OllamaImage(MaxKBBaseModel, ChatOpenAI):
             openai_api_key=model_credential.get('api_key'),
             # stream_options={"include_usage": True},
             streaming=True,
+            stream_usage=True,
             **optional_params,
         )

apps/setting/models_provider/impl/openai_model_provider/model/image.py

@@ -1,17 +1,10 @@
 from typing import Dict
 
-from langchain_openai.chat_models import ChatOpenAI
-
-from common.config.tokenizer_manage_config import TokenizerManage
 from setting.models_provider.base_model_provider import MaxKBBaseModel
+from setting.models_provider.impl.base_chat_open_ai import BaseChatOpenAI
 
 
-def custom_get_token_ids(text: str):
+class OpenAIImage(MaxKBBaseModel, BaseChatOpenAI):
-    tokenizer = TokenizerManage.get_tokenizer()
-    return tokenizer.encode(text)
-
-
-class OpenAIImage(MaxKBBaseModel, ChatOpenAI):
 
     @staticmethod
     def new_instance(model_type, model_name, model_credential: Dict[str, object], **model_kwargs):
@@ -22,5 +15,6 @@ class OpenAIImage(MaxKBBaseModel, ChatOpenAI):
             openai_api_key=model_credential.get('api_key'),
             # stream_options={"include_usage": True},
             streaming=True,
+            stream_usage=True,
             **optional_params,
         )

apps/setting/models_provider/impl/qwen_model_provider/model/image.py

@@ -2,12 +2,11 @@
 
 from typing import Dict
 
-from langchain_community.chat_models import ChatOpenAI
-
 from setting.models_provider.base_model_provider import MaxKBBaseModel
+from setting.models_provider.impl.base_chat_open_ai import BaseChatOpenAI
 
 
-class QwenVLChatModel(MaxKBBaseModel, ChatOpenAI):
+class QwenVLChatModel(MaxKBBaseModel, BaseChatOpenAI):
 
     @staticmethod
     def new_instance(model_type, model_name, model_credential: Dict[str, object], **model_kwargs):
@@ -18,6 +17,7 @@ class QwenVLChatModel(MaxKBBaseModel, ChatOpenAI):
             openai_api_base='https://dashscope.aliyuncs.com/compatible-mode/v1',
             # stream_options={"include_usage": True},
             streaming=True,
+            stream_usage=True,
             **optional_params,
         )
         return chat_tong_yi

apps/setting/models_provider/impl/tencent_model_provider/model/image.py

@@ -1,17 +1,10 @@
 from typing import Dict
 
-from langchain_openai.chat_models import ChatOpenAI
-
-from common.config.tokenizer_manage_config import TokenizerManage
 from setting.models_provider.base_model_provider import MaxKBBaseModel
+from setting.models_provider.impl.base_chat_open_ai import BaseChatOpenAI
 
 
-def custom_get_token_ids(text: str):
+class TencentVision(MaxKBBaseModel, BaseChatOpenAI):
-    tokenizer = TokenizerManage.get_tokenizer()
-    return tokenizer.encode(text)
-
-
-class TencentVision(MaxKBBaseModel, ChatOpenAI):
 
     @staticmethod
     def new_instance(model_type, model_name, model_credential: Dict[str, object], **model_kwargs):
@@ -22,5 +15,6 @@ class TencentVision(MaxKBBaseModel, ChatOpenAI):
             openai_api_key=model_credential.get('api_key'),
             # stream_options={"include_usage": True},
             streaming=True,
+            stream_usage=True,
             **optional_params,
         )

apps/setting/models_provider/impl/volcanic_engine_model_provider/model/image.py

@@ -1,17 +1,10 @@
 from typing import Dict
 
-from langchain_openai.chat_models import ChatOpenAI
-
-from common.config.tokenizer_manage_config import TokenizerManage
 from setting.models_provider.base_model_provider import MaxKBBaseModel
+from setting.models_provider.impl.base_chat_open_ai import BaseChatOpenAI
 
 
-def custom_get_token_ids(text: str):
+class VolcanicEngineImage(MaxKBBaseModel, BaseChatOpenAI):
-    tokenizer = TokenizerManage.get_tokenizer()
-    return tokenizer.encode(text)
-
-
-class VolcanicEngineImage(MaxKBBaseModel, ChatOpenAI):
 
     @staticmethod
     def new_instance(model_type, model_name, model_credential: Dict[str, object], **model_kwargs):
@@ -22,5 +15,6 @@ class VolcanicEngineImage(MaxKBBaseModel, ChatOpenAI):
             openai_api_base=model_credential.get('api_base'),
             # stream_options={"include_usage": True},
             streaming=True,
+            stream_usage=True,
             **optional_params,
         )

apps/setting/models_provider/impl/xinference_model_provider/model/image.py

@@ -1,17 +1,10 @@
 from typing import Dict
 
-from langchain_openai.chat_models import ChatOpenAI
-
-from common.config.tokenizer_manage_config import TokenizerManage
 from setting.models_provider.base_model_provider import MaxKBBaseModel
+from setting.models_provider.impl.base_chat_open_ai import BaseChatOpenAI
 
 
-def custom_get_token_ids(text: str):
+class XinferenceImage(MaxKBBaseModel, BaseChatOpenAI):
-    tokenizer = TokenizerManage.get_tokenizer()
-    return tokenizer.encode(text)
-
-
-class XinferenceImage(MaxKBBaseModel, ChatOpenAI):
 
     @staticmethod
     def new_instance(model_type, model_name, model_credential: Dict[str, object], **model_kwargs):
@@ -22,5 +15,6 @@ class XinferenceImage(MaxKBBaseModel, ChatOpenAI):
             openai_api_key=model_credential.get('api_key'),
             # stream_options={"include_usage": True},
             streaming=True,
+            stream_usage=True,
             **optional_params,
         )

apps/setting/models_provider/impl/zhipu_model_provider/model/image.py

@@ -1,17 +1,10 @@
 from typing import Dict
 
-from langchain_openai.chat_models import ChatOpenAI
-
-from common.config.tokenizer_manage_config import TokenizerManage
 from setting.models_provider.base_model_provider import MaxKBBaseModel
+from setting.models_provider.impl.base_chat_open_ai import BaseChatOpenAI
 
 
-def custom_get_token_ids(text: str):
+class ZhiPuImage(MaxKBBaseModel, BaseChatOpenAI):
-    tokenizer = TokenizerManage.get_tokenizer()
-    return tokenizer.encode(text)
-
-
-class ZhiPuImage(MaxKBBaseModel, ChatOpenAI):
 
     @staticmethod
     def new_instance(model_type, model_name, model_credential: Dict[str, object], **model_kwargs):
@@ -22,5 +15,6 @@ class ZhiPuImage(MaxKBBaseModel, ChatOpenAI):
             openai_api_base='https://open.bigmodel.cn/api/paas/v4',
             # stream_options={"include_usage": True},
             streaming=True,
+            stream_usage=True,
             **optional_params,
         )