unique_toolkit 1.19.1__tar.gz → 1.19.2__tar.gz

{unique_toolkit-1.19.1 → unique_toolkit-1.19.2}/CHANGELOG.md

@@ -5,11 +5,14 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/), 
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [1.19.2] - 2025-10-29
+- Removing unused tool specific `get_tool_call_result_for_loop_history` function
+- Removing unused experimental config `full_sources_serialize_dump` in `history_manager`
+
 ## [1.19.1] - 2025-10-29
 - Make api operations more flexible
 - Make verification button text adaptable
 
-
 ## [1.19.0] - 2025-10-28
 - Enable additional headers on openai and langchain client
 

{unique_toolkit-1.19.1 → unique_toolkit-1.19.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: unique_toolkit
-Version: 1.19.1
+Version: 1.19.2
 Summary: 
 License: Proprietary
 Author: Cedric Klinkert
@@ -118,11 +118,14 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/), 
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [1.19.2] - 2025-10-29
+- Removing unused tool specific `get_tool_call_result_for_loop_history` function
+- Removing unused experimental config `full_sources_serialize_dump` in `history_manager`
+
 ## [1.19.1] - 2025-10-29
 - Make api operations more flexible
 - Make verification button text adaptable
 
-
 ## [1.19.0] - 2025-10-28
 - Enable additional headers on openai and langchain client
 

{unique_toolkit-1.19.1 → unique_toolkit-1.19.2}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "unique_toolkit"
-version = "1.19.1"
+version = "1.19.2"
 description = ""
 authors = [
     "Cedric Klinkert <cedric.klinkert@unique.ch>",

unique_toolkit-1.19.2/unique_toolkit/_common/feature_flags/schema.py

@@ -0,0 +1,9 @@
+from pydantic import BaseModel
+
+from unique_toolkit.agentic.tools.config import get_configuration_dict
+
+
+class FeatureExtendedSourceSerialization(BaseModel):
+    """Mixin for experimental feature in Source serialization"""
+
+    model_config = get_configuration_dict()

{unique_toolkit-1.19.1 → unique_toolkit-1.19.2}/unique_toolkit/agentic/history_manager/history_manager.py

@@ -3,6 +3,9 @@ from typing import Annotated, Awaitable, Callable
 
 from pydantic import BaseModel, Field
 
+from unique_toolkit._common.feature_flags.schema import (
+    FeatureExtendedSourceSerialization,
+)
 from unique_toolkit._common.validators import LMI
 from unique_toolkit.agentic.history_manager.loop_token_reducer import LoopTokenReducer
 from unique_toolkit.agentic.history_manager.utils import transform_chunks_to_string
@@ -42,11 +45,7 @@ class UploadedContentConfig(BaseModel):
     )
 
 
-class ExperimentalFeatures(BaseModel):
-    full_sources_serialize_dump: bool = Field(
-        default=False,
-        description="If True, the sources will be serialized in full, otherwise only the content will be serialized.",
-    )
+class ExperimentalFeatures(FeatureExtendedSourceSerialization): ...
 
 
 class HistoryManagerConfig(BaseModel):
@@ -172,8 +171,6 @@ class HistoryManager:
         stringified_sources, sources = transform_chunks_to_string(
             content_chunks,
             self._source_enumerator,
-            None,  # Use None for SourceFormatConfig
-            self._config.experimental_features.full_sources_serialize_dump,
         )
 
         self._source_enumerator += len(

unique_toolkit-1.19.2/unique_toolkit/agentic/history_manager/utils.py

@@ -0,0 +1,96 @@
+import json
+import logging
+from copy import deepcopy
+from typing import Any
+
+from unique_toolkit.agentic.tools.schemas import Source
+from unique_toolkit.content.schemas import ContentChunk
+from unique_toolkit.language_model.schemas import (
+    LanguageModelAssistantMessage,
+    LanguageModelMessage,
+    LanguageModelToolMessage,
+)
+
+logger = logging.getLogger(__name__)
+
+
+def convert_tool_interactions_to_content_messages(
+    loop_history: list[LanguageModelMessage],
+) -> list[LanguageModelMessage]:
+    new_loop_history = []
+    copy_loop_history = deepcopy(loop_history)
+
+    for message in copy_loop_history:
+        if isinstance(message, LanguageModelAssistantMessage) and message.tool_calls:
+            new_loop_history.append(_convert_tool_call_to_content(message))
+
+        elif isinstance(message, LanguageModelToolMessage):
+            new_loop_history.append(_convert_tool_call_response_to_content(message))
+        else:
+            new_loop_history.append(message)
+
+    return new_loop_history
+
+
+def _convert_tool_call_to_content(
+    assistant_message: LanguageModelAssistantMessage,
+) -> LanguageModelAssistantMessage:
+    assert assistant_message.tool_calls is not None
+    new_content = "The assistant requested the following tool_call:"
+    for tool_call in assistant_message.tool_calls:
+        new_content += (
+            f"\n\n- {tool_call.function.name}: {tool_call.function.arguments}"
+        )
+    assistant_message.tool_calls = None
+    assistant_message.content = new_content
+
+    return assistant_message
+
+
+def _convert_tool_call_response_to_content(
+    tool_message: LanguageModelToolMessage,
+) -> LanguageModelAssistantMessage:
+    new_content = f"The assistant received the following tool_call_response: {tool_message.name}, {tool_message.content}"
+    assistant_message = LanguageModelAssistantMessage(
+        content=new_content,
+    )
+    return assistant_message
+
+
+def transform_chunks_to_string(
+    content_chunks: list[ContentChunk],
+    max_source_number: int,
+) -> tuple[str, list[dict[str, Any]]]:
+    """Transform content chunks into a string of sources.
+
+    Args:
+        content_chunks (list[ContentChunk]): The content chunks to transform
+        max_source_number (int): The maximum source number to use
+
+    Returns:
+        str: String for the tool call response
+    """
+    if not content_chunks:
+        return "No relevant sources found.", []
+    sources: list[dict[str, Any]] = [
+        {
+            "source_number": max_source_number + i,
+            "content": chunk.text,
+        }
+        for i, chunk in enumerate(content_chunks)
+    ]
+    return json.dumps(sources), sources
+
+
+def load_sources_from_string(
+    source_string: str,
+) -> list[Source] | None:
+    """Transform JSON string from language model tool message in the tool call response into Source objects"""
+
+    try:
+        # First, try parsing as JSON (new format)
+        sources_data = json.loads(source_string)
+        return [Source.model_validate(source) for source in sources_data]
+    except (json.JSONDecodeError, ValueError):
+        logger.warning("Failed to parse source string")
+        return None

{unique_toolkit-1.19.1 → unique_toolkit-1.19.2}/unique_toolkit/agentic/tools/a2a/tool/service.py

@@ -18,7 +18,6 @@ from unique_toolkit.agentic.tools.a2a.tool._schema import (
 from unique_toolkit.agentic.tools.a2a.tool.config import (
     SubAgentToolConfig,
 )
-from unique_toolkit.agentic.tools.agent_chunks_hanlder import AgentChunksHandler
 from unique_toolkit.agentic.tools.factory import ToolFactory
 from unique_toolkit.agentic.tools.schemas import ToolCallResponse
 from unique_toolkit.agentic.tools.tool import Tool
@@ -31,7 +30,6 @@ from unique_toolkit.language_model import (
     LanguageModelFunction,
     LanguageModelToolDescription,
 )
-from unique_toolkit.language_model.schemas import LanguageModelMessage
 
 
 class SubAgentResponseSubscriber(Protocol):
@@ -199,13 +197,6 @@ class SubAgentTool(Tool[SubAgentToolConfig]):
                 content=response_text_with_references,
             )
 
-    @override
-    def get_tool_call_result_for_loop_history(
-        self,
-        tool_response: ToolCallResponse,
-        agent_chunks_handler: AgentChunksHandler,
-    ) -> LanguageModelMessage: ...  # Empty as method is deprecated
-
     async def _get_chat_id(self) -> str | None:
         if not self.config.reuse_chat:
             return None

{unique_toolkit-1.19.1 → unique_toolkit-1.19.2}/unique_toolkit/agentic/tools/mcp/tool_wrapper.py

@@ -13,7 +13,6 @@ from unique_toolkit.agentic.tools.tool_progress_reporter import (
     ToolProgressReporter,
 )
 from unique_toolkit.app.schemas import ChatEvent, McpServer, McpTool
-from unique_toolkit.language_model import LanguageModelMessage
 from unique_toolkit.language_model.schemas import (
     LanguageModelFunction,
     LanguageModelToolDescription,
@@ -99,12 +98,6 @@ class MCPToolWrapper(Tool[MCPToolConfig]):
         """Return evaluation checks based on tool response"""
         return []
 
-    def get_tool_call_result_for_loop_history(
-        self,
-        tool_response: ToolCallResponse,
-    ) -> LanguageModelMessage:
-        raise NotImplementedError("function is not supported")
-
     async def run(self, tool_call: LanguageModelFunction) -> ToolCallResponse:
         """Execute the MCP tool using SDK to call public API"""
         self.logger.info(f"Running MCP tool: {self.name}")

{unique_toolkit-1.19.1 → unique_toolkit-1.19.2}/unique_toolkit/agentic/tools/test/test_mcp_manager.py

@@ -50,11 +50,6 @@ class MockInternalSearchTool(Tool[BaseToolConfig]):
     def tool_format_information_for_system_prompt(self) -> str:
         return "Use this tool to search for content"
 
-    def get_tool_call_result_for_loop_history(self, tool_response):
-        from unique_toolkit.language_model.schemas import LanguageModelMessage
-
-        return LanguageModelMessage(role="tool", content="Mock search result")
-
     def evaluation_check_list(self):
         return []
 

{unique_toolkit-1.19.1 → unique_toolkit-1.19.2}/unique_toolkit/agentic/tools/tool.py

@@ -5,7 +5,6 @@ from typing import Any, Generic, TypeVar, cast
 from typing_extensions import deprecated
 
 from unique_toolkit.agentic.evaluation.schemas import EvaluationMetricName
-from unique_toolkit.agentic.tools.agent_chunks_hanlder import AgentChunksHandler
 from unique_toolkit.agentic.tools.config import ToolBuildConfig, ToolSelectionPolicy
 from unique_toolkit.agentic.tools.schemas import (
     BaseToolConfig,
@@ -20,7 +19,6 @@ from unique_toolkit.chat.service import (
 from unique_toolkit.language_model import LanguageModelToolDescription
 from unique_toolkit.language_model.schemas import (
     LanguageModelFunction,
-    LanguageModelMessage,
 )
 from unique_toolkit.language_model.service import LanguageModelService
 
@@ -100,15 +98,6 @@ class Tool(ABC, Generic[ConfigType]):
         """
         return ""
 
-    @deprecated("Do not use as is bound to loop agent only")
-    @abstractmethod
-    def get_tool_call_result_for_loop_history(
-        self,
-        tool_response: ToolCallResponse,
-        agent_chunks_handler: AgentChunksHandler,
-    ) -> LanguageModelMessage:
-        raise NotImplementedError
-
     @deprecated(
         "Do not use. The tool should not determine how"
         "it is checked. This should be defined by the user"

unique_toolkit-1.19.1/unique_toolkit/_common/feature_flags/schema.py

@@ -1,13 +0,0 @@
-from pydantic import BaseModel, Field
-
-from unique_toolkit.agentic.tools.config import get_configuration_dict
-
-
-class FeatureExtendedSourceSerialization(BaseModel):
-    """Mixin for experimental feature in Source serialization"""
-
-    model_config = get_configuration_dict()
-    full_sources_serialize_dump: bool = Field(
-        default=False,
-        description="Whether to include the full source object in the tool response. If True, includes the full Source object. If False, uses the old format with only source_number and content.",
-    )

unique_toolkit-1.19.1/unique_toolkit/agentic/history_manager/utils.py

@@ -1,173 +0,0 @@
-import json
-import logging
-from copy import deepcopy
-
-from unique_toolkit.agentic.tools.schemas import Source
-from unique_toolkit.agentic.tools.utils.source_handling.schema import (
-    SourceFormatConfig,
-)
-from unique_toolkit.content.schemas import ContentChunk, ContentMetadata
-from unique_toolkit.language_model.schemas import (
-    LanguageModelAssistantMessage,
-    LanguageModelMessage,
-    LanguageModelToolMessage,
-)
-
-logger = logging.getLogger(__name__)
-
-
-def convert_tool_interactions_to_content_messages(
-    loop_history: list[LanguageModelMessage],
-) -> list[LanguageModelMessage]:
-    new_loop_history = []
-    copy_loop_history = deepcopy(loop_history)
-
-    for message in copy_loop_history:
-        if isinstance(message, LanguageModelAssistantMessage) and message.tool_calls:
-            new_loop_history.append(_convert_tool_call_to_content(message))
-
-        elif isinstance(message, LanguageModelToolMessage):
-            new_loop_history.append(_convert_tool_call_response_to_content(message))
-        else:
-            new_loop_history.append(message)
-
-    return new_loop_history
-
-
-def _convert_tool_call_to_content(
-    assistant_message: LanguageModelAssistantMessage,
-) -> LanguageModelAssistantMessage:
-    assert assistant_message.tool_calls is not None
-    new_content = "The assistant requested the following tool_call:"
-    for tool_call in assistant_message.tool_calls:
-        new_content += (
-            f"\n\n- {tool_call.function.name}: {tool_call.function.arguments}"
-        )
-    assistant_message.tool_calls = None
-    assistant_message.content = new_content
-
-    return assistant_message
-
-
-def _convert_tool_call_response_to_content(
-    tool_message: LanguageModelToolMessage,
-) -> LanguageModelAssistantMessage:
-    new_content = f"The assistant received the following tool_call_response: {tool_message.name}, {tool_message.content}"
-    assistant_message = LanguageModelAssistantMessage(
-        content=new_content,
-    )
-    return assistant_message
-
-
-def transform_chunks_to_string(
-    content_chunks: list[ContentChunk],
-    max_source_number: int,
-    cfg: SourceFormatConfig | None,
-    full_sources_serialize_dump: bool = False,
-) -> tuple[str, list[Source]]:
-    """Transform content chunks into a string of sources.
-
-    Args:
-        content_chunks (list[ContentChunk]): The content chunks to transform
-        max_source_number (int): The maximum source number to use
-        feature_full_sources (bool, optional): Whether to include the full source object. Defaults to False which is the old format.
-
-    Returns:
-        str: String for the tool call response
-    """
-    if not content_chunks:
-        return "No relevant sources found.", []
-    if full_sources_serialize_dump:
-        sources = [
-            Source(
-                source_number=max_source_number + i,
-                key=chunk.key,
-                id=chunk.id,
-                order=chunk.order,
-                content=chunk.text,
-                chunk_id=chunk.chunk_id,
-                metadata=(
-                    _format_metadata(chunk.metadata, cfg) or None
-                    if chunk.metadata
-                    else None
-                ),
-                url=chunk.url,
-            ).model_dump(
-                exclude_none=True,
-                exclude_defaults=True,
-                by_alias=True,
-            )
-            for i, chunk in enumerate(content_chunks)
-        ]
-    else:
-        sources = [
-            {
-                "source_number": max_source_number + i,
-                "content": chunk.text,
-                **(
-                    {"metadata": meta}
-                    if (
-                        meta := _format_metadata(chunk.metadata, cfg)
-                    )  # only add when not empty
-                    else {}
-                ),
-            }
-            for i, chunk in enumerate(content_chunks)
-        ]
-    return json.dumps(sources), sources
-
-
-def load_sources_from_string(
-    source_string: str,
-) -> list[Source] | None:
-    """Transform JSON string from language model tool message in the tool call response into Source objects"""
-
-    try:
-        # First, try parsing as JSON (new format)
-        sources_data = json.loads(source_string)
-        return [Source.model_validate(source) for source in sources_data]
-    except (json.JSONDecodeError, ValueError):
-        logger.warning("Failed to parse source string")
-        return None
-
-
-def _format_metadata(
-    metadata: ContentMetadata | None,
-    cfg: SourceFormatConfig | None,
-) -> str:
-    """
-    Build the concatenated tag string from the chunk's metadata
-    and the templates found in cfg.sections.
-    Example result:
-      "<|topic|>GenAI<|/topic|>\n<|date|>This document is from: 2025-04-29<|/date|>\n"
-    """
-    if metadata is None:
-        return ""
-
-    if cfg is None or not cfg.sections:
-        # If no configuration is provided, return empty string
-        return ""
-
-    meta_dict = metadata.model_dump(exclude_none=True, by_alias=True)
-    sections = cfg.sections
-
-    parts: list[str] = []
-    for key, template in sections.items():
-        if key in meta_dict:
-            parts.append(template.format(meta_dict[key]))
-
-    return "".join(parts)
-
-
-### In case we do not want any formatting of metadata we could use this function
-# def _filtered_metadata(
-#     meta: ContentMetadata | None,
-#     cfg: SourceFormatConfig,
-# ) -> dict[str, str] | None:
-#     if meta is None:
-#         return None
-
-#     allowed = set(cfg.sections)
-#     raw = meta.model_dump(exclude_none=True, by_alias=True)
-#     pruned = {k: v for k, v in raw.items() if k in allowed}
-#     return pruned or None