unique_toolkit 0.7.10

No project description provided

Unique Toolkit

This package provides highlevel abstractions and methods on top of unique_sdk to ease application development for the Unique Platform.

The Toolkit is structured along the following domains:

• unique_toolkit.chat
• unique_toolkit.content
• unique_toolkit.embedding
• unique_toolkit.language_model
• unique_toolkit.short_term_memory

Each domain comprises a set of functions (in functions.py) and a service class (in service.py) which encapsulates the basic functionalities to interact with the domain entities, the schemas (in schemas.py) used in the service and required for interacting with the service functions, utility functions (in utils.py) which give additional functionality to interact with the domain entities (all domains except embedding) and other domain specific functionalities which are explained in the respective domain documentation.

In addition, the unique_toolkit.app module provides functions to initialize apps that interact with the Unique platform. It also includes some utility functions to run async tasks in parallel (async webserver and app implementation required).

Changelog

See the CHANGELOG.md file for details on changes and version history.

Domains

App

The unique_toolkit.app module encompasses functions for initializing and securing apps that will interact with the Unique platform.

• init_logging.py can be used to initalize the logger either with unique dictConfig or an any other dictConfig.
• init_sdk.py can be used to initialize the sdk using the correct env variables and retrieving the endpoint secret.
• schemas.py contains the Event schema which can be used to parse and validate the unique.chat.external-module.chosen event.
• verification.py can be used to verify the endpoint secret and construct the event.

Chat

The unique_toolkit.chat module encompasses all chat related functionality.

• functions.py comprises the functions to manage and load the chat history and interact with the chat ui, e.g., creating a new assistant message.
• service.py comprises the ChatService and provides an interface to manage and load the chat history and interact with the chat ui, e.g., creating a new assistant message and stream complete.
• schemas.py comprises all relevant schemas, e.g., ChatMessage, used in the ChatService.
• utils.py comprises utility functions to use and convert ChatMessage objects in assistants, e.g., convert_chat_history_to_injectable_string converts the chat history to a string that can be injected into a prompt.

Content

The unique_toolkit.content module encompasses all content related functionality. Content can be any type of textual data that is stored in the Knowledgebase on the Unique platform. During the ingestion of the content, the content is parsed, split in chunks, indexed, and stored in the database.

• functions.py comprises the functions to manage and load the chat history and interact with the chat ui, e.g., creating a new assistant message.
• service.py comprises the ContentService and provides an interface to interact with the content, e.g., search content, search content chunks, upload and download content.
• schemas.py comprises all relevant schemas, e.g., Content and ContentChunk, used in the ContentService.
• utils.py comprise utility functions to manipulate Content and ContentChunk objects, e.g., sort_content_chunks and merge_content_chunks.

Embedding

The unique_toolkit.embedding module encompasses all embedding related functionality. Embeddings are used to represent textual data in a high-dimensional space. The embeddings can be used to calculate the similarity between two texts, for instance.

• functions.py comprises the functions to embed text and calculate the similarity between two texts.
• service.py encompasses the EmbeddingService and provides an interface to interact with the embeddings, e.g., embed text and calculate the similarity between two texts.
• schemas.py comprises all relevant schemas, e.g., Embeddings, used in the EmbeddingService.

Language Model

The unique_toolkit.language_model module encompasses all language model related functionality and information on the different language models deployed through the Unique platform.

• infos.py comprises the information on all language models deployed through the Unique platform. We recommend to use the LanguageModel class, initialized with the LanguageModelName, e.g., LanguageModel(LanguageModelName.AZURE_GPT_35_TURBO_0125) to get the information on the specific language model like the name, version, token limits or retirement date.
• functions.py comprises the functions to complete and stream complete to chat.
• service.py comprises the LanguageModelService and provides an interface to interact with the language models, e.g., complete.
• schemas.py comprises all relevant schemas, e.g., LanguageModelResponse, used in the LanguageModelService.
• utils.py comprises utility functions to parse the output of the language model, e.g., convert_string_to_json finds and parses the last json object in a string.

Short Term Memory

The unique_toolkit.short_term_memory module encompasses all short term memory related functionality.

• functions.py comprises the functions to manage and load the chat history and interact with the chat ui, e.g., creating a new assistant message.
• service.py comprises the ShortTermMemoryService and provides an interface to interact with the short term memory, e.g., create memory.
• schemas.py comprises all relevant schemas, e.g., ShortTermMemory, used in the ShortTermMemoryService.

Development instructions

1. Install poetry on your system (through brew or pipx).

2. Install pyenv and install python 3.11. pyenv is recommended as otherwise poetry uses the python version used to install itself and not the user preferred python version.

3. If you then run python --version in your terminal, you should be able to see python version as specified in .python-version.

4. Then finally run poetry install to install the package and all dependencies.

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[0.7.10] - 2025-04-22

• Deprecate internal variables of services

[0.7.9] - 2025-04-17

• add AZURE_GPT_41_2025_0414 as part of the models

[0.7.8] - 2025-04-08

• add AZURE_GPT_4o_2024_1120 as part of the models

[0.7.7] - 2025-04-11

• Add tool choice parameter to chat event payload

[0.7.6] - 2025-04-08

• De provisioning o1-preview

[0.7.5] - 2025-04-07

• Skip None values when serializing to json schema for LanguageModelInfo

[0.7.4] - 2025-03-20

• add AZURE_GPT_45_PREVIEW_2025_0227 as part of the models

[0.7.3] - 2025-03-20

• Enable handling tool calls in message builder

[0.7.2] - 2025-03-17

• HotFix ContentService.search_content_chunks to use chat_id from event if provided.

[0.7.1] - 2025-03-11

• Fix Breaking change: ContentService.search_content_chunks ContentService.search_content_chunks now acceptschat_id for the specific to handle chat_only instances

[0.7.0] - 2025-03-11

• Fix the issue with ShortTermMemoryService.create_memory_async adding self.chat_id and self.message_id as part of the parameter.
• Breaking change: ContentService.search_content_on_chat now requires you pass in a chat_id for the specific chat instance

[0.6.9] - 2025-03-11

• Add o1-preview as part of the language model info, make the name consistent across board.

[0.6.8] - 2025-03-11

• Add verify_request_and_construct_event to verification.py

[0.6.7] - 2025-03-10

• Extend language model message builder

[0.6.6] - 2025-03-10

• Add o1, o1-mini and o3-mini models
• Remove deprecated gpt4 models
• Make token_limits and encoder a required attribute of LanguageModelInfo

[0.6.5] - 2025-03-04

• Add upload_content_from_bytes to ContentService
• Add download_content_to_bytes to ContentService

[0.6.3] - 2025-02-27

• Simplified imports for services. from unique_toolkit.language_model import LanguageModelService -> from unique_toolkit import LanguageModelService to reduce number of import lines.

[0.6.3] - 2025-02-26

• Add builder method to LanguageModelMessages class

[0.6.2] - 2025-02-25

• Deprecate LanguageModel in favor of LanguageModelInfo
• LanguageModelTokenLimits properties become mandatory, initialization allows
  • init with token_limit and fraction_input or input_token_limit and output_token_limit
  • only input_token_limit and output_token_limit are members of model

[0.6.1] - 2025-02-25

• [BREAKING] LanguageModelService.stream_complete and LanguageModelService.stream_complete_async are now moved to ChatService.stream_complete and ChatService.stream_complete_async. Correspondingly assistant_message_id and user_message_id are removed from LanguageModelService.
• Add create_user_message and create_user_message_async to ChatService (similar to create_assistant_message and create_assistant_message_async)

[0.6.0] - 2025-02-21

• make for each domain, its base functionality accessible from functions.py
• make it possible to instantiate the domain services directly from different event types, inhereted from common BaseEvent
• extend the functionalities in the ShortTermMemoryService by adding the find_latest_memory and create_memory functions for sync and async usage
• remove logger dependency from service classes
• marked deprecated:
  • from_chat_event in ShortTermMemoryService, use ShortTermMemoryService(event=event) instead
  • complete_async_util in LanguageModelService, use functions.complete_async instead
  • stream_complete_async in LanguageModelService, use stream_complete_to_chat_async instead
  • stream_complete in LanguageModelService, use stream_complete_to_chat instead
  • Event and nested schemas in app, use ChatEvent and ChatEventUserMessage, ChatEventAssistantMessage and ChatEventToolMessage instead

[0.5.56] - 2025-02-19

• Add MessageAssessment title field and change label values

[0.5.55] - 2025-02-18

• Log contentId for better debugging

[0.5.54] - 2025-02-10

• Add created_at, completed_at, updated_at and gpt_request to ChatMessage schema.

[0.5.53] - 2025-02-01

• Correct MessageAssessment schemas

[0.5.52] - 2025-02-01

• Add MessageAssessment schemas and functions to ChatService to handle message assessments.
• Fix LanguageModelService.complete_async_util to use the correct async method.

[0.5.51] - 2025-01-30

• Add missing structured output arguments in complete_async

[0.5.50] - 2025-01-30

• Add the possibility to define completion output structure through a pydantic class

[0.5.49] - 2025-01-24

• Add parsed and refusal to LanguageModelAssistantMessage to support structured output

[0.5.48] - 2025-01-19

• Added the possibility define tool parameters with a json schema (Useful when generating tool parameters from a pydantic object)

[0.5.47] - 2025-01-07

• Added a message builder to build language model messages conveniently without importing all different messages.
• Move tool_calls to assistant message as not needed anywhere else.

[0.5.46] - 2025-01-07

• Added AZURE_GPT_35_TURBO_0125 as new model into toolkit.

[0.5.45] - 2025-01-03

• Added ShortTermMemoryService class to handle short term memory.

[0.5.44] - 2024-12-18

• Add event_constructor to verify_signature_and_construct_event to allow for custom event construction.

[0.5.43] - 2024-12-13

• Add Prompt class to handle templated prompts that can be formatted into LanguageModelSystemMessage and LanguageModelUserMessage.

[0.5.42] - 2024-12-11

• Update LanguageModelTokenLimits with fix avoiding floats for token

[0.5.41] - 2024-12-11

• Update LanguageModelTokenLimits includes a fraction_input now to always have input/output token limits available.

[0.5.40] - 2024-12-11

• Add other_options to LanguageModelService.complete, LanguageModelService.complete_async, LanguageModelService.stream_complete and LanguageModelService.stream_complete_async

[0.5.39] - 2024-12-09

• Add contentIds to Search.create and Search.create_async
• Use metadata_filter from event in ContentService.search_content_chunks and ContentService.async_search_content_chunks as default.

[0.5.38] - 2024-11-26

• Added string representation to LanguageModelMessage and LanguageModelMessages class

[0.5.37] - 2024-11-26

• content parameter in ChatService.modify_assistant_message and ChatService.modify_assistant_message_async is now optional
• Added optional parameter original_content to ChatService.modify_assistant_message and ChatService.modify_assistant_message_async
• Added optional parameter original_content to ChatService.create_assistant_message and ChatService.create_assistant_message_async

[0.5.36] - 2024-11-19

• Add possibility to return the response from the download file from chat request
• Add possibility to not specify a filename and use filename from response headers

[0.5.35] - 2024-11-18

• Add the possibilty to upload files without triggering ingestion by setting skip_ingestion to True in ContentService.upload_content

[0.5.34] - 2024-11-15

• Add content_id_to_translate to EventAdditionalParameters

[0.5.33] - 2024-10-30

• Force randomizing tool_call_id. This is helpful to better identify the tool_calls.

[0.5.32] - 2024-10-30

• Extending LanguageModelName with GPT-4o-2024-0806. This model is invoked using AZURE_GPT_4o_2024_0806.

[0.5.31] - 2024-10-29

• Adding support for function calling. Assistant message for tool calls can be directly created with LanguageModelFunctionCall.create_assistant_message_from_tool_calls. Better separation of schemas for different types of LanguageModelMessages.

[0.5.30] - 2024-10-28

• Correctly use temperature parameter in LanguageModelService.complete and LanguageModelService.complete_async methods

[0.5.29] - 2024-10-28

• Allow numbers in LanguageModelTool name

[0.5.28] - 2024-10-23

• Correctly use temperature parameter in LanguageModelService.stream_complete and LanguageModelService.stream_complete_async methods

[0.5.27] - 2024-10-22

• Add encoder_name to to language model info
• Verify tool name for LanguageModelTool to conform with frontent requirements.
• Add search_on_chat to ContentService

[0.5.26] - 2024-10-16

• Bump unique_sdk version

[0.5.25] - 2024-09-26

• Add evaluators for hallucination and context relevancy evaluation

[0.5.24] - 2024-09-26

• Add originalText to _construct_message_modify_params and _construct_message_create_params. This addition makes sure that the originalText on the database is populated with the text

[0.5.23] - 2024-09-23

• Add set_completed_at as a boolen parameter to the following functions: modify_user_message, modify_user_message_async, modify_assistant_message, modify_assistant_message_async, create_assistant_message and create_assistant_message. This parameter allows the completedAt timestamp on the database to be updated when set to True.

[0.5.22] - 2024-09-17

• Add LanguageModelToolMessage as additional LanguageModelMessage

[0.5.21] - 2024-09-16

• Add tool as new role to ChatMessage, as well as tool_calls and tool_call_id as additional parameters

[0.5.20] - 2024-09-16

• LanguageModelService now supports complete_util_async that can be called without instantiating the class, currently being used in the Hallucination service and evaluation API

[0.5.19] - 2024-09-11

• LanguageModelMessage now supports content as a list of dictionary. Useful when adding image_url content along user message.

[0.5.18] - 2024-09-03

• Adds option to use metadata_filter with search.
• Adds user_metadata, tool_parameters and metadata_filter to EventPayload.
• Adds update_debug_info and modify_user_message (and the corresponding async variants) to ChatService.

[0.5.17] - 2024-08-30

• Add option to initiate LanguageModel with a string.
• Add option to call LanguageModelService functions also with a string instead of LanguageModelName enum for parameter model_name.

[0.5.16] - 2024-08-29

• Fix ContentService.upload_content function.

[0.5.15] - 2024-08-27

• Possibility to specify root directory in ContentService.download_content

[0.5.14] - 2024-08-26

• Add AZURE_GPT_4o_MINI_2024_0718 to language model infos

[0.5.13] - 2024-08-19

• Added items to LanguageModelToolParameterProperty schema to add support for parameters with list types.
• Added returns to LanguageModelTool schema to describe the return types of tool calls.

[0.5.12] - 2024-08-7

• added completedAt datetime to unique_sdk.Message.modify and unique_sdk.Message.modify_async
• added original_text and language to EventUserMessage

[0.5.11] - 2024-08-6

• made all domain specific functions and classes directly importable from unique_toolkit.[DOMAIN_NAME]
• renamed RerankerConfig to ContentRerankerConfig
• renamed get_cosine_similarity to calculate_cosine_similarity and moved it to unique_toolkit.embedding.utils
• moved calculate_tokens from unique_toolkit.content.utils to unique_toolkit.embedding.utils
• disabled deprecation warning in LanguageModel
• added additional_parameters to event
• removed ChatState and use Event instead

[0.5.10] - 2024-08-6

• fix content schema

[0.5.9] - 2024-08-6

• added created_at and updated_at to content schema

[0.5.8] - 2024-08-1

• RerankerConfig serialization alias added

[0.5.7] - 2024-07-31

• Replace mocked async service calls with async calls in unique_sdk
• Change async methods name from async_* to *_async
• Remove chat_only and scope_ids attributes from ChatState class
• Replace AsyncExecutor by simpler utility function run_async_tasks_parallel

[0.5.6] - 2024-07-30

• Bug fix: ContentService.search_content_chunks and it's async equivalent now accept None as a valid parameter value for scope_ids.

[0.5.5] - 2024-07-30

• Added parameters to ContentService.search_content_chunks and ContentService.async_search_content_chunks
  • reranker_config to optinally rerank the search results
  • search_language to specify a language for full-text-search

[0.5.4] - 2024-07-26

• correct ChatMessage schema

[0.5.3] - 2024-07-25

• downgrade numpy version to ^1.26.4 to be compatible with langchain libraries (require numpy<2.0)

[0.5.2] - 2024-07-25

• correct event schema

[0.5.1] - 2024-07-23

• correct documentation

[0.5.0] - 2024-07-23

Added

• Added unique_toolkit.app module with the following components:

  • init_logging.py for initializing the logger.
  • init_sdk.py for initializing the SDK with environment variables.
  • schemas.py containing the Event schema.
  • verification.py for verifying the endpoint secret and constructing the event.

• Added unique_toolkit.chat module with the following components:

  • state.py containing the ChatState class.
  • service.py containing the ChatService class for managing chat interactions.
  • schemas.py containing relevant schemas such as ChatMessage.
  • utils.py with utility functions for chat interactions.

• Added unique_toolkit.content module with the following components:

  • service.py containing the ContentService class for interacting with content.
  • schemas.py containing relevant schemas such as Content and ContentChunk.
  • utils.py with utility functions for manipulating content objects.

• Added unique_toolkit.embedding module with the following components:

  • service.py containing the EmbeddingService class for working with embeddings.
  • schemas.py containing relevant schemas such as Embeddings.

• Added unique_toolkit.language_model module with the following components:

  • infos.py containing information on language models deployed on the Unique platform.
  • service.py containing the LanguageModelService class for interacting with language models.
  • schemas.py containing relevant schemas such as LanguageModelResponse.
  • utils.py with utility functions for parsing language model output.

[0.0.2] - 2024-07-10

• Initial release of unique_toolkit.