unique_toolkit 0.7.30__tar.gz → 0.7.32__tar.gz

{unique_toolkit-0.7.30 → unique_toolkit-0.7.32}/CHANGELOG.md

@@ -5,6 +5,13 @@ 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).
 
+## [0.7.32] - 2025-06-24
+- Create `classmethod` for `LanguageModelMessages` to load raw messages to root
+
+## [0.7.31] - 2025-06-19
+- Add typings to references in payload from `LanguageModelStreamResponseMessage` 
+- Add `original_index` to the base reference to reflect updated api
+
 ## [0.7.30] - 2025-06-20
 - Adding litellm models `litellm:gemini-2-5-flash`, `gemini-2-5-flash-lite-preview-06-17`, `litellm:gemini-2-5-pro`, `litellm:gemini-2-5-pro-preview-06-05`
 

{unique_toolkit-0.7.30 → unique_toolkit-0.7.32}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: unique_toolkit
-Version: 0.7.30
+Version: 0.7.32
 Summary: 
 License: Proprietary
 Author: Martin Fadler
@@ -17,7 +17,7 @@ Requires-Dist: python-dotenv (>=1.0.1,<2.0.0)
 Requires-Dist: regex (>=2024.5.15,<2025.0.0)
 Requires-Dist: tiktoken (>=0.7.0,<0.8.0)
 Requires-Dist: typing-extensions (>=4.9.0,<5.0.0)
-Requires-Dist: unique-sdk (>=0.9.33,<0.10.0)
+Requires-Dist: unique-sdk (>=0.9.36,<0.10.0)
 Description-Content-Type: text/markdown
 
 # Unique Toolkit
@@ -111,6 +111,13 @@ 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).
 
+## [0.7.32] - 2025-06-24
+- Create `classmethod` for `LanguageModelMessages` to load raw messages to root
+
+## [0.7.31] - 2025-06-19
+- Add typings to references in payload from `LanguageModelStreamResponseMessage` 
+- Add `original_index` to the base reference to reflect updated api
+
 ## [0.7.30] - 2025-06-20
 - Adding litellm models `litellm:gemini-2-5-flash`, `gemini-2-5-flash-lite-preview-06-17`, `litellm:gemini-2-5-pro`, `litellm:gemini-2-5-pro-preview-06-05`
 

{unique_toolkit-0.7.30 → unique_toolkit-0.7.32}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "unique_toolkit"
-version = "0.7.30"
+version = "0.7.32"
 description = ""
 authors = [
     "Martin Fadler <martin.fadler@unique.ch>",
@@ -20,7 +20,7 @@ numpy = "^1.26.4"
 python-dotenv = "^1.0.1"
 regex = "^2024.5.15"
 tiktoken = "^0.7.0"
-unique-sdk = "^0.9.33"
+unique-sdk = "^0.9.36"
 
 [tool.poetry.group.dev.dependencies]
 ruff = "0.11.7"

{unique_toolkit-0.7.30 → unique_toolkit-0.7.32}/unique_toolkit/chat/schemas.py

@@ -10,6 +10,8 @@ from pydantic import (
     model_validator,
 )
 
+from unique_toolkit.content.schemas import ContentReference
+
 # set config to convert camelCase to snake_case
 model_config = ConfigDict(
     alias_generator=camelize, populate_by_name=True, arbitrary_types_allowed=True
@@ -37,15 +39,6 @@ class ToolCall(BaseModel):
     function: Function
 
 
-class Reference(BaseModel):
-    model_config = model_config
-    name: str
-    url: str | None = None
-    sequence_number: int
-    source_id: str
-    source: str
-
-
 class ChatMessage(BaseModel):
     # This model should strictly meets https://github.com/Unique-AG/monorepo/blob/master/node/apps/node-chat/src/public-api/2023-12-06/dtos/message/public-message.dto.ts
     model_config = model_config
@@ -63,7 +56,7 @@ class ChatMessage(BaseModel):
     created_at: datetime | None = None
     completed_at: datetime | None = None
     updated_at: datetime | None = None
-    references: list[Reference] | None = None
+    references: list[ContentReference] | None = None
 
     # TODO make sdk return role consistently in lowercase
     # Currently needed as sdk returns role in uppercase

{unique_toolkit-0.7.30 → unique_toolkit-0.7.32}/unique_toolkit/content/schemas.py

@@ -60,13 +60,23 @@ class Content(BaseModel):
 
 class ContentReference(BaseModel):
     model_config = model_config
-    id: str
-    message_id: str
+    id: str = Field(
+        default="",
+        description="The id of the content reference. Can be empty on the ChatMessage Object",
+    )
+    message_id: str = Field(
+        default="",
+        description="The id of the message that this reference belongs to. Can be empty on the ChatMessage Object",
+    )
     name: str
     sequence_number: int
     source: str
     source_id: str
     url: str
+    original_index: list[int] = Field(
+        default=[],
+        description="List of indices in the ChatMessage original_content this reference refers to. This is usually the id in the functionCallResponse. List type due to implementation in node-chat",
+    )
 
 
 class ContentSearchType(StrEnum):

{unique_toolkit-0.7.30 → unique_toolkit-0.7.32}/unique_toolkit/language_model/functions.py

@@ -7,7 +7,7 @@ import unique_sdk
 from pydantic import BaseModel
 
 from unique_toolkit.chat.schemas import ChatMessage, ChatMessageRole
-from unique_toolkit.content.schemas import ContentChunk
+from unique_toolkit.content.schemas import ContentChunk, ContentReference
 from unique_toolkit.evaluators import DOMAIN_NAME
 from unique_toolkit.language_model import (
     LanguageModelMessageRole,
@@ -20,7 +20,6 @@ from unique_toolkit.language_model import (
 )
 from unique_toolkit.language_model.infos import LanguageModelName
 from unique_toolkit.language_model.reference import (
-    PotentialReference,
     add_references_to_message,
 )
 
@@ -333,20 +332,9 @@ def _create_language_model_stream_response_with_references(
         chat_id="chat_unknown",
     )
 
-    search_context = [
-        PotentialReference(
-            id=source.id,
-            chunk_id=source.id,
-            title=source.title,
-            key=source.key or "",
-            url=source.url,
-        )
-        for source in content_chunks
-    ]
-
     message, __ = add_references_to_message(
         message=message,
-        search_context=search_context,
+        search_context=content_chunks,
     )
 
     stream_response_message = LanguageModelStreamResponseMessage(
@@ -355,7 +343,9 @@ def _create_language_model_stream_response_with_references(
         role=LanguageModelMessageRole.ASSISTANT,
         text=message.content or "",
         original_text=content,
-        references=[u.model_dump() for u in message.references or []],
+        references=[
+            ContentReference(**u.model_dump()) for u in message.references or []
+        ],
     )
 
     tool_calls = [r.function for r in response.choices[0].message.tool_calls or []]

{unique_toolkit-0.7.30 → unique_toolkit-0.7.32}/unique_toolkit/language_model/reference.py

@@ -1,27 +1,12 @@
 import re
 
-from pydantic import BaseModel
-
-from unique_toolkit.chat.schemas import ChatMessage, Reference
-
-
-class NodeReference(Reference):
-    original_index: list[int] = []
-    message_id: str | None = None
-
-
-class PotentialReference(BaseModel):
-    id: str
-    chunk_id: str | None = None
-    title: str | None = None
-    key: str
-    url: str | None = None
-    internally_stored_at: str | None = None
+from unique_toolkit.chat.schemas import ChatMessage
+from unique_toolkit.content.schemas import ContentChunk, ContentReference
 
 
 def add_references_to_message(
     message: ChatMessage,
-    search_context: list[PotentialReference],
+    search_context: list[ContentChunk],
     model: str | None = None,
 ) -> tuple[ChatMessage, bool]:
     """Add references to a message and return the updated message with change status.
@@ -41,21 +26,21 @@ def add_references_to_message(
     )
     message.content = _postprocess_message(text)
 
-    message.references = [Reference(**ref.model_dump()) for ref in ref_found]
+    message.references = ref_found
     references_changed = len(ref_found) > 0
     return message, references_changed
 
 
 def _add_references(
     text: str,
-    search_context: list[PotentialReference],
+    search_context: list[ContentChunk],
     message_id: str,
     model: str | None = None,
-) -> tuple[str, list[NodeReference]]:
+) -> tuple[str, list[ContentReference]]:
     """Add references to text and return the processed text with reference status.
 
     Returns:
-        Tuple[str, bool]: (processed_text, ref_found)
+        Tuple[str, list[Reference]]: (processed_text, ref_found)
     """
     references = _find_references(
         text=text,
@@ -169,11 +154,11 @@ def _get_max_sub_count_in_text(text: str) -> int:
 
 def _find_references(
     text: str,
-    search_context: list[PotentialReference],
+    search_context: list[ContentChunk],
     message_id: str,
-) -> list[NodeReference]:
+) -> list[ContentReference]:
     """Find references in text based on search context."""
-    references: list[NodeReference] = []
+    references: list[ContentReference] = []
     sequence_number = 1 + _get_max_sub_count_in_text(text)
 
     # Find all numbers in brackets to ensure we get references in order of occurrence
@@ -190,7 +175,7 @@ def _find_references(
             continue
 
         # Don't put the reference twice
-        reference_name = search.title or search.key
+        reference_name = search.title or search.key or f"Content {search.id}"
         found_reference = next(
             (r for r in references if r.name == reference_name), None
         )
@@ -206,7 +191,7 @@ def _find_references(
         )
 
         references.append(
-            NodeReference(
+            ContentReference(
                 name=reference_name,
                 url=url,
                 sequence_number=sequence_number,
@@ -216,6 +201,7 @@ def _find_references(
                 else search.id,
                 source="node-ingestion-chunks",
                 message_id=message_id,
+                id=search.id,
             )
         )
         sequence_number += 1
@@ -229,7 +215,7 @@ def _extract_numbers_in_brackets(text: str) -> list[int]:
     return [int(match) for match in matches]
 
 
-def _add_footnotes_to_text(text: str, references: list[NodeReference]) -> str:
+def _add_footnotes_to_text(text: str, references: list[ContentReference]) -> str:
     """Replace bracket references with superscript footnotes."""
     for reference in references:
         for original_index in reference.original_index:

{unique_toolkit-0.7.30 → unique_toolkit-0.7.32}/unique_toolkit/language_model/schemas.py

@@ -18,6 +18,7 @@ from pydantic import (
 )
 from typing_extensions import deprecated
 
+from unique_toolkit.content.schemas import ContentReference
 from unique_toolkit.language_model.utils import format_message
 
 # set config to convert camelCase to snake_case
@@ -48,7 +49,7 @@ class LanguageModelStreamResponseMessage(BaseModel):
     role: LanguageModelMessageRole
     text: str
     original_text: str | None = None
-    references: list[dict[str, list | dict | str | int | float | bool]] = []  # type: ignore
+    references: list[ContentReference] = []
 
     # TODO make sdk return role in lowercase
     # Currently needed as sdk returns role in uppercase
@@ -242,6 +243,40 @@ class LanguageModelMessages(RootModel):
         | LanguageModelUserMessage
     ]
 
+    @classmethod
+    def load_messages_to_root(cls, data: list[dict] | dict) -> Self:
+        """Convert list of dictionaries to appropriate message objects based on role."""
+        # Handle case where data is already wrapped in root
+        if isinstance(data, dict) and "root" in data:
+            messages_list = data["root"]
+        elif isinstance(data, list):
+            messages_list = data
+        else:
+            raise ValueError("Invalid data type")
+
+        # Convert the messages list
+        converted_messages = []
+        for item in messages_list:
+            if isinstance(item, dict):
+                role = item.get("role", "").lower()
+
+                # Map dictionary to appropriate message class based on role
+                if role == "system":
+                    converted_messages.append(LanguageModelSystemMessage(**item))
+                elif role == "user":
+                    converted_messages.append(LanguageModelUserMessage(**item))
+                elif role == "assistant":
+                    converted_messages.append(LanguageModelAssistantMessage(**item))
+                elif role == "tool":
+                    converted_messages.append(LanguageModelToolMessage(**item))
+                else:
+                    # Fallback to base LanguageModelMessage
+                    converted_messages.append(LanguageModelMessage(**item))
+            else:
+                # If it's already a message object, keep it as is
+                converted_messages.append(item)
+        return cls(root=converted_messages)
+
     def __str__(self):
         return "\n\n".join([str(message) for message in self.root])