vectorvein 0.1.49__tar.gz → 0.1.51__tar.gz

{vectorvein-0.1.49 → vectorvein-0.1.51}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: vectorvein
-Version: 0.1.49
+Version: 0.1.51
 Summary: Default template for PDM package
 Author-Email: Anderson <andersonby@163.com>
 License: MIT

{vectorvein-0.1.49 → vectorvein-0.1.51}/pyproject.toml

@@ -17,7 +17,7 @@ description = "Default template for PDM package"
 name = "vectorvein"
 readme = "README.md"
 requires-python = ">=3.10"
-version = "0.1.49"
+version = "0.1.51"
 
 [project.license]
 text = "MIT"

{vectorvein-0.1.49 → vectorvein-0.1.51}/src/vectorvein/chat_clients/minimax_client.py

@@ -193,11 +193,19 @@ class MiniMaxChatClient(BaseChatClient):
         if max_tokens is None:
             max_output_tokens = self.model_setting.max_output_tokens
             if max_output_tokens is not None:
-                token_counts = get_token_counts({"messages": messages, "tools_params": tools_params})
+                token_counts = get_token_counts(
+                    text={"messages": messages, "tools_params": tools_params},
+                    model=self.model,
+                    use_token_server_first=True,
+                )
                 max_tokens = self.model_setting.context_length - token_counts
                 max_tokens = min(max(max_tokens, 1), max_output_tokens)
             else:
-                token_counts = get_token_counts({"messages": messages, "tools_params": tools_params})
+                token_counts = get_token_counts(
+                    text={"messages": messages, "tools_params": tools_params},
+                    model=self.model,
+                    use_token_server_first=True,
+                )
                 max_tokens = self.model_setting.context_length - token_counts
 
         self.url = self.endpoint.api_base
@@ -418,11 +426,19 @@ class AsyncMiniMaxChatClient(BaseAsyncChatClient):
         if max_tokens is None:
             max_output_tokens = self.model_setting.max_output_tokens
             if max_output_tokens is not None:
-                token_counts = get_token_counts({"messages": messages, "tools_params": tools_params})
+                token_counts = get_token_counts(
+                    text={"messages": messages, "tools_params": tools_params},
+                    model=self.model,
+                    use_token_server_first=True,
+                )
                 max_tokens = self.model_setting.context_length - token_counts
                 max_tokens = min(max(max_tokens, 1), max_output_tokens)
             else:
-                token_counts = get_token_counts({"messages": messages, "tools_params": tools_params})
+                token_counts = get_token_counts(
+                    text={"messages": messages, "tools_params": tools_params},
+                    model=self.model,
+                    use_token_server_first=True,
+                )
                 max_tokens = self.model_setting.context_length - token_counts
 
         self.url = self.endpoint.api_base

{vectorvein-0.1.49 → vectorvein-0.1.51}/src/vectorvein/chat_clients/utils.py

@@ -114,7 +114,7 @@ def convert_type(value, value_type):
         return value  # 如果类型未知，返回原始值
 
 
-def get_token_counts(text: str | dict, model: str = "", use_token_server_first: bool = False) -> int:
+def get_token_counts(text: str | dict, model: str = "", use_token_server_first: bool = True) -> int:
     if use_token_server_first and settings.token_server is not None:
         base_url = (
             settings.token_server.url
@@ -226,7 +226,11 @@ def get_token_counts(text: str | dict, model: str = "", use_token_server_first:
         result = response.json()
         return result["totalTokens"]
     elif model.startswith("claude"):
-        return Anthropic().count_tokens(text)
+        return (
+            Anthropic()
+            .beta.messages.count_tokens(messages=[{"role": "user", "content": text}], model=model)
+            .input_tokens
+        )
     elif model.startswith("deepseek"):
         from deepseek_tokenizer import deepseek_tokenizer