dais-sdk 0.6.2__tar.gz → 0.6.4__tar.gz

{dais_sdk-0.6.2 → dais_sdk-0.6.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dais-sdk
-Version: 0.6.2
+Version: 0.6.4
 Summary: A wrapper of LiteLLM
 Author-email: BHznJNs <bhznjns@outlook.com>
 Requires-Python: >=3.14
@@ -11,11 +11,11 @@ Classifier: License :: OSI Approved :: MIT License
 Classifier: Programming Language :: Python :: 3 :: Only
 Classifier: Programming Language :: Python :: 3.14
 License-File: LICENSE
-Requires-Dist: litellm>=1.80.0
+Requires-Dist: litellm>=1.81.0
 Requires-Dist: pydantic>=2.0.0
 Requires-Dist: httpx==0.28.1
-Requires-Dist: mcp==1.25.0
-Requires-Dist: starlette==0.50.0
+Requires-Dist: mcp==1.26.0
+Requires-Dist: starlette==0.52.1
 Requires-Dist: uvicorn==0.40.0
 Project-URL: Source, https://github.com/Dais-Project/Dais-SDK
 Project-URL: Tracker, https://github.com/Dais-Project/Dais-SDK/issues

{dais_sdk-0.6.2 → dais_sdk-0.6.4}/pyproject.toml

@@ -15,16 +15,16 @@ classifiers = [
     "Programming Language :: Python :: 3.14",
 ]
 requires-python = ">=3.14"
-version = "0.6.2"
+version = "0.6.4"
 
 dependencies = [
-    "litellm>=1.80.0",
+    "litellm>=1.81.0",
     "pydantic>=2.0.0",
 
     # MCP client dependencies
     "httpx==0.28.1",
-    "mcp==1.25.0",
-    "starlette==0.50.0",
+    "mcp==1.26.0",
+    "starlette==0.52.1",
     "uvicorn==0.40.0",
 ]
 

{dais_sdk-0.6.2 → dais_sdk-0.6.4}/src/dais_sdk/mcp_client/local_mcp_client.py

@@ -4,7 +4,7 @@ from mcp import ClientSession, StdioServerParameters as StdioServerParams
 from mcp.client.stdio import stdio_client
 from .base_mcp_client import McpClient, Tool, ToolResult, McpSessionNotEstablishedError
 
-type LocalServerParams = StdioServerParams
+LocalServerParams = StdioServerParams
 
 class LocalMcpClient(McpClient):
     def __init__(self, name: str, params: LocalServerParams):

{dais_sdk-0.6.2 → dais_sdk-0.6.4}/src/dais_sdk/stream.py

@@ -74,6 +74,9 @@ class AssistantMessageCollector:
 
     def get_message(self) -> AssistantMessage:
         self.message_buf.tool_calls = self.tool_call_collector.get_tool_calls()
+        if self.message_buf.content is not None and self.message_buf.reasoning_content is None:
+            self.message_buf.content, self.message_buf.reasoning_content =\
+                AssistantMessage.extract_thinking_content(self.message_buf.content)
         return self.message_buf
 
     def clear(self):

{dais_sdk-0.6.2 → dais_sdk-0.6.4}/src/dais_sdk/types/message.py

@@ -1,5 +1,6 @@
 import json
 import dataclasses
+import re
 import uuid
 from abc import ABC, abstractmethod
 from typing import Any, Literal, cast
@@ -92,6 +93,22 @@ class AssistantMessage(ChatMessage):
     usage: LiteLlmUsage | None = None
     role: Literal["assistant"] = "assistant"
 
+    @staticmethod
+    def extract_thinking_content(text: str) -> tuple[str, str | None]:
+        """
+        Extract thinking content from the start of model responsed text.
+        Returns:
+            A tuple of (extracted_content, optional thinking_content)
+        """
+        pattern = r"<(think|thinking)>(.*?)</\1>"
+        match = re.match(pattern, text, re.DOTALL)
+        if match:
+            start, end = match.span()
+            thinking_content = match.group(2)
+            remaining_content = text[:start] + text[end:]
+            return remaining_content.strip(), thinking_content
+        return text, None
+
     @classmethod
     def from_litellm_message(cls, response: LiteLlmModelResponse) -> "AssistantMessage":
         choices = cast(list[LiteLlmModelResponseChoices], response.choices)
@@ -108,9 +125,16 @@ class AssistantMessage(ChatMessage):
                 type="function",
             ) for tool_call in cast(list[ChatCompletionMessageToolCall], message_tool_calls)]
 
+        content: str | None = message.get("content")
+        reasoning_content: str | None = message.get("reasoning_content")
+        if content is not None:
+            content, thinking_block_content = cls.extract_thinking_content(content)
+            if reasoning_content is None:
+                reasoning_content = thinking_block_content
+
         return cls.model_construct(
-            content=message.get("content"),
-            reasoning_content=message.get("reasoning_content"),
+            content=content,
+            reasoning_content=reasoning_content,
             tool_calls=tool_calls,
             audio=message.get("audio"),
             images=message.get("images"),