unique_orchestrator 0.0.1__tar.gz

unique_orchestrator-0.0.1/CHANGELOG.md

@@ -0,0 +1,10 @@
+# Changelog
+
+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).
+
+
+## [0.0.1] - 2025-08-18
+- Initial release of `orchestrator`.

unique_orchestrator-0.0.1/LICENSE

@@ -0,0 +1 @@
+`unique_toolkit` is covered by the [`Unique License v1`](https://github.com/Unique-AG/license/releases/tag/unique-license.v1), unless the/a header or a nested LICENSE specifies another license.

unique_orchestrator-0.0.1/PKG-INFO

@@ -0,0 +1,39 @@
+Metadata-Version: 2.1
+Name: unique_orchestrator
+Version: 0.0.1
+Summary: 
+License: Proprietary
+Author: Martin Fadler
+Author-email: martin.fadler@unique.ch
+Requires-Python: >=3.11,<4.0
+Classifier: License :: Other/Proprietary License
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Requires-Dist: pydantic (>=2.8.2,<3.0.0)
+Requires-Dist: pydantic-settings (>=2.10.1,<3.0.0)
+Requires-Dist: pytest (>=8.4.1,<9.0.0)
+Requires-Dist: python-dotenv (>=1.0.1,<2.0.0)
+Requires-Dist: typing-extensions (>=4.9.0,<5.0.0)
+Requires-Dist: unique-deep-research (>=0.0.10,<0.0.11)
+Requires-Dist: unique-follow-up-questions (>=0.0.4,<0.0.5)
+Requires-Dist: unique-internal-search (==0.0.5)
+Requires-Dist: unique-sdk (>=0.10.0,<0.11.0)
+Requires-Dist: unique-stock-ticker (>=0.0.5,<0.0.6)
+Requires-Dist: unique-toolkit (>=0.8.57,<0.9.0)
+Requires-Dist: unique-web-search (>=0.1.0,<0.2.0)
+Description-Content-Type: text/markdown
+
+# Internal Search Tool
+
+Internal Search Tool to find documents in the Knowledge Base
+# Changelog
+
+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).
+
+
+## [0.0.1] - 2025-08-18
+- Initial release of `orchestrator`.

unique_orchestrator-0.0.1/README.md

@@ -0,0 +1,3 @@
+# Internal Search Tool
+
+Internal Search Tool to find documents in the Knowledge Base

unique_orchestrator-0.0.1/pyproject.toml

@@ -0,0 +1,55 @@
+[tool.poetry]
+name = "unique_orchestrator"
+version = "0.0.1"
+description = ""
+authors = [
+    "Martin Fadler <martin.fadler@unique.ch>",
+    "Sadique Sheik <sadique@unique.ch>",
+    "Fabian Schläpfer <fabian@unique.ch>",
+    "Pascal Hauri <pascal@unique.ch>",
+]
+readme = ["README.md", "CHANGELOG.md"]
+license = "Proprietary"
+
+[tool.poetry.dependencies]
+python = "^3.11"
+typing-extensions = "^4.9.0"
+pydantic = "^2.8.2"
+pydantic-settings = "^2.10.1"
+python-dotenv = "^1.0.1"
+pytest = "^8.4.1"
+unique-sdk = "^0.10.0"
+unique-toolkit = "^0.8.57"
+unique-stock-ticker = "^0.0.5"
+unique-follow-up-questions = "^0.0.4"
+unique-web-search = "^0.1.0"
+unique-deep-research = "^0.0.10"
+unique-internal-search = "0.0.5"
+
+
+[tool.poetry.group.dev.dependencies]
+ruff = "^0.12.10"
+python = "^3.11"
+typing-extensions = "^4.9.0"
+pydantic = "^2.8.2"
+pydantic-settings = "^2.10.1"
+python-dotenv = "^1.0.1"
+pytest = "^8.4.1"
+unique-sdk = { path = "../unique_sdk" }
+unique-toolkit = { path = "../unique_toolkit", develop = true }
+unique-stock-ticker = { path = "../postprocessors/unique_stock_ticker", develop = true }
+unique-follow-up-questions = { path = "../postprocessors/unique_follow_up_questions", develop = true }
+unique-web-search = { path = "../tool_packages/unique_web_search", develop = true }
+unique-deep-research = { path = "../tool_packages/unique_deep_research", develop = true }
+unique-internal-search = { path = "../tool_packages/unique_internal_search", develop = true }
+
+
+[build-system]
+requires = ["poetry-core"]
+build-backend = "poetry.core.masonry.api"
+
+[tool.ruff]
+target-version = "py311"
+
+[tool.ruff.lint]
+extend-select = ["I"]

unique_orchestrator-0.0.1/unique_orchestrator/config.py

@@ -0,0 +1,505 @@
+from enum import StrEnum
+from pathlib import Path
+from typing import Annotated, Any, Generic, Literal, TypeVar
+
+from pydantic import BaseModel, Field, ValidationInfo, field_validator
+from unique_deep_research.config import DeepResearchToolConfig
+from unique_deep_research.service import DeepResearchTool
+from unique_follow_up_questions.config import FollowUpQuestionsConfig
+from unique_internal_search.config import InternalSearchConfig
+from unique_internal_search.service import InternalSearchTool
+from unique_stock_ticker.config import StockTickerConfig
+from unique_toolkit._common.default_language_model import DEFAULT_GPT_4o
+from unique_toolkit._common.validators import (
+    LMI,
+    ClipInt,
+    get_LMI_default_field,
+)
+from unique_toolkit.evals.hallucination.constants import HallucinationConfig
+from unique_toolkit.evals.schemas import EvaluationMetricName
+from unique_toolkit.history_manager.history_manager import (
+    UploadedContentConfig,
+)
+from unique_toolkit.language_model import LanguageModelName
+from unique_toolkit.language_model.infos import (
+    LanguageModelInfo,
+)
+from unique_toolkit.tools.config import get_configuration_dict
+from unique_toolkit.tools.factory import ToolFactory
+from unique_toolkit.tools.schemas import BaseToolConfig
+from unique_toolkit.tools.tool import ToolBuildConfig
+from unique_web_search.config import WebSearchConfig
+from unique_web_search.service import WebSearchTool
+
+
+class SpaceType(StrEnum):
+    UNIQUE_CUSTOM = "unique_custom"
+    UNIQUE_AI = "unique_ai"
+    UNIQUE_TRANSLATION = "unique_translation"
+    UNIQUE_MAGIC_TABLE = ""
+
+
+T = TypeVar("T", bound=SpaceType)
+
+
+class SpaceConfigBase(BaseModel, Generic[T]):
+    """Base class for space configuration."""
+
+    model_config = get_configuration_dict(frozen=True)
+    type: T = Field(description="The type of the space.")
+
+    project_name: str = Field(
+        default="Unique AI",
+        description="The project name as optained from spaces 2.0",
+    )
+
+    language_model: LMI = get_LMI_default_field(DEFAULT_GPT_4o)
+
+    custom_instructions: str = Field(
+        default="",
+        description="A custom instruction provided by the system admin.",
+    )
+
+    tools: list[ToolBuildConfig] = Field(
+        default=[
+            ToolBuildConfig(
+                name=InternalSearchTool.name,
+                configuration=InternalSearchConfig(
+                    exclude_uploaded_files=True,
+                ),
+            ),
+            ToolBuildConfig(
+                name=WebSearchTool.name,
+                configuration=WebSearchConfig(),
+            ),
+            ToolBuildConfig(
+                name=DeepResearchTool.name,
+                configuration=DeepResearchToolConfig(),
+            ),
+        ],
+    )
+
+    @field_validator("tools", mode="after")
+    @classmethod
+    def set_input_context_size(
+        cls, tools: list[ToolBuildConfig], info: ValidationInfo
+    ) -> list[ToolBuildConfig]:
+        for tool in tools:
+            if tool.name == InternalSearchTool.name:
+                tool.configuration.language_model_max_input_tokens = (  # type: ignore
+                    info.data["language_model"].token_limits.token_limit_input
+                )
+            elif tool.name == WebSearchTool.name:
+                tool.configuration.language_model_max_input_tokens = (  # type: ignore
+                    info.data["language_model"].token_limits.token_limit_input
+                )
+        return tools
+
+
+class UniqueAISpaceConfig(SpaceConfigBase):
+    """Contains configuration for the entities that a space provides."""
+
+    type: Literal[SpaceType.UNIQUE_AI] = SpaceType.UNIQUE_AI
+
+
+UniqueAISpaceConfig.model_rebuild()
+
+LIMIT_LOOP_ITERATIONS = 50
+LIMIT_MAX_TOOL_CALLS_PER_ITERATION = 50
+
+
+class LoopConfiguration(BaseModel):
+    model_config = get_configuration_dict()
+
+    max_tool_calls_per_iteration: Annotated[
+        int,
+        *ClipInt(min_value=1, max_value=LIMIT_MAX_TOOL_CALLS_PER_ITERATION),
+    ] = 10
+
+
+class EvaluationConfig(BaseModel):
+    model_config = get_configuration_dict()
+    max_review_steps: int = 3
+    hallucination_config: HallucinationConfig = HallucinationConfig()
+
+
+class LoopAgentTokenLimitsConfig(BaseModel):
+    model_config = get_configuration_dict()
+
+    language_model: LMI = LanguageModelInfo.from_name(
+        LanguageModelName.AZURE_GPT_4o_2024_1120
+    )
+
+    percent_of_max_tokens_for_history: float = Field(
+        default=0.2,
+        ge=0.0,
+        lt=1.0,
+        description="The fraction of the max input tokens that will be reserved for the history.",
+    )
+
+    @property
+    def max_history_tokens(self) -> int:
+        return int(
+            self.language_model.token_limits.token_limit_input
+            * self.percent_of_max_tokens_for_history,
+        )
+
+
+class SearchAgentConfig(BaseModel):
+    """Configure the search agent."""
+
+    model_config = get_configuration_dict(frozen=True)
+
+    language_model: LMI = LanguageModelInfo.from_name(DEFAULT_GPT_4o)
+
+    token_limits: LoopAgentTokenLimitsConfig = Field(
+        default=LoopAgentTokenLimitsConfig(percent_of_max_tokens_for_history=0.6)
+    )
+    temperature: float = 0.0
+    additional_llm_options: dict[str, Any] = Field(
+        default={},
+        description="Additional options to pass to the language model.",
+    )
+
+    # Space 2.0
+    project_name: str = Field(
+        default="Unique AI",
+        description="The project name as optained from spaces 2.0",
+    )
+
+    # Space 2.0
+    custom_instructions: str = Field(
+        default="",
+        description="A custom instruction provided by the system admin.",
+    )
+
+    thinking_steps_display: bool = False
+
+    ##############################
+    ### General Configurations
+    ##############################
+    max_loop_iterations: Annotated[
+        int, *ClipInt(min_value=1, max_value=LIMIT_LOOP_ITERATIONS)
+    ] = 8
+
+    loop_configuration: LoopConfiguration = LoopConfiguration()
+
+    tools: list[ToolBuildConfig] = Field(
+        default=[
+            ToolBuildConfig(
+                name=InternalSearchTool.name,
+                configuration=InternalSearchConfig(
+                    exclude_uploaded_files=True,
+                ),
+            ),
+            ToolBuildConfig(
+                name=WebSearchTool.name,
+                configuration=WebSearchConfig(),
+            ),
+            ToolBuildConfig(
+                name=DeepResearchTool.name,
+                configuration=DeepResearchToolConfig(),
+            ),
+        ],
+    )
+
+    system_prompt_template: str = Field(
+        default_factory=lambda: (
+            Path(__file__).parent / "prompts" / "system_prompt.jinja2"
+        ).read_text(),
+        description="The system prompt template as a Jinja2 template string.",
+    )
+
+    user_message_prompt_template: str = Field(
+        default_factory=lambda: (
+            Path(__file__).parent / "prompts" / "user_message_prompt.jinja2"
+        ).read_text(),
+        description="The user message prompt template as a Jinja2 template string.",
+    )
+
+    ##############################
+    ### Follow-up Questions
+    ##############################
+    follow_up_questions_config: FollowUpQuestionsConfig = FollowUpQuestionsConfig()
+
+    ##############################
+    ### Evaluation
+    ##############################
+    evaluation_config: EvaluationConfig = EvaluationConfig(
+        hallucination_config=HallucinationConfig(),
+        max_review_steps=0,
+    )
+
+    ##############################
+    ### Stock Ticker
+    ##############################
+    stock_ticker_config: StockTickerConfig = StockTickerConfig()
+
+    # TODO: generalize this there should only be 1 point in the code where we do the tool check.
+    def get_tool_config(self, tool: str) -> BaseToolConfig:
+        """Get the tool configuration by name."""
+        return ToolFactory.build_tool_config(tool)
+
+    # TODO: @gustavhartz, the Hallucination check should be triggered if enabled and the answer contains references.
+    force_checks_on_stream_response_references: list[EvaluationMetricName] = Field(
+        default=[EvaluationMetricName.HALLUCINATION],
+        description="A list of checks to force on references. This is used to add hallucination check to references without new tool calls.",
+    )
+
+    uploaded_content_config: UploadedContentConfig = Field(
+        default_factory=UploadedContentConfig,
+        description="The uploaded content config.",
+    )
+
+
+# ------------------------------------------------------------
+# Space 2.0 Config
+# ------------------------------------------------------------
+
+
+class UniqueAIPromptConfig(BaseModel):
+    model_config = get_configuration_dict(frozen=True)
+
+    system_prompt_template: str = Field(
+        default_factory=lambda: (
+            Path(__file__).parent / "prompts" / "system_prompt.jinja2"
+        ).read_text(),
+        description="The system prompt template as a Jinja2 template string.",
+    )
+
+    user_message_prompt_template: str = Field(
+        default_factory=lambda: (
+            Path(__file__).parent / "prompts" / "user_message_prompt.jinja2"
+        ).read_text(),
+        description="The user message prompt template as a Jinja2 template string.",
+    )
+
+
+DeactivatedNone = Annotated[
+    None,
+    Field(title="Deactivated", description="None"),
+]
+
+
+class UniqueAIServices(BaseModel):
+    """Determine the services the agent is using
+
+    All services are optional and can be disabled by setting them to None.
+    """
+
+    model_config = get_configuration_dict(frozen=True)
+
+    follow_up_questions_config: (
+        Annotated[
+            FollowUpQuestionsConfig,
+            Field(
+                title="Active",
+            ),
+        ]
+        | DeactivatedNone
+    ) = FollowUpQuestionsConfig()
+
+    stock_ticker_config: (
+        Annotated[StockTickerConfig, Field(title="Active")] | DeactivatedNone
+    ) = StockTickerConfig()
+
+    evaluation_config: (
+        Annotated[
+            EvaluationConfig,
+            Field(title="Active"),
+        ]
+        | DeactivatedNone
+    ) = EvaluationConfig(
+        hallucination_config=HallucinationConfig(),
+        max_review_steps=0,
+    )
+
+    uploaded_content_config: UploadedContentConfig = UploadedContentConfig()
+
+
+class InputTokenDistributionConfig(BaseModel):
+    model_config = get_configuration_dict(frozen=True)
+
+    percent_for_history: float = Field(
+        default=0.2,
+        ge=0.0,
+        lt=1.0,
+        description="The fraction of the max input tokens that will be reserved for the history.",
+    )
+
+    def max_history_tokens(self, max_input_token: int) -> int:
+        return int(self.percent_for_history * max_input_token)
+
+
+class ExperimentalConfig(BaseModel):
+    """Experimental features this part of the configuration might evolve in the future continuously"""
+
+    model_config = get_configuration_dict(frozen=True)
+
+    thinking_steps_display: bool = False
+
+    # TODO: @gustavhartz, the Hallucination check should be triggered if enabled and the answer contains references.
+    force_checks_on_stream_response_references: list[EvaluationMetricName] = Field(
+        default=[EvaluationMetricName.HALLUCINATION],
+        description="A list of checks to force on references. This is used to add hallucination check to references without new tool calls.",
+    )
+
+    # TODO: The temperature should be used via the additional_llm_options
+    # then the additional_llm_options migth should eventually be closer to the LangaugeModelInfo
+    temperature: float = Field(
+        default=0.0,
+        ge=0.0,
+        le=10.0,
+        description="The temperature to use for the LLM.",
+    )
+
+    additional_llm_options: dict[str, Any] = Field(
+        default={},
+        description="Additional options to pass to the LLM.",
+    )
+
+    loop_configuration: LoopConfiguration = LoopConfiguration(
+        max_tool_calls_per_iteration=5
+    )
+
+
+class UniqueAIAgentConfig(BaseModel):
+    model_config = get_configuration_dict(frozen=True)
+
+    max_loop_iterations: int = 8
+
+    input_token_distribution: InputTokenDistributionConfig = Field(
+        default=InputTokenDistributionConfig(),
+        description="The distribution of the input tokens.",
+    )
+
+    prompt_config: UniqueAIPromptConfig = UniqueAIPromptConfig()
+
+    services: UniqueAIServices = UniqueAIServices()
+
+    experimental: ExperimentalConfig = ExperimentalConfig()
+
+
+class UniqueAIConfig(BaseModel):
+    model_config = get_configuration_dict(frozen=True)
+
+    space: UniqueAISpaceConfig = UniqueAISpaceConfig()
+
+    agent: UniqueAIAgentConfig = UniqueAIAgentConfig()
+
+
+# ---
+# Configuration adapter SearchAgentConfig -> UniqueAISpaceConfig
+# --
+
+
+def search_agent_config_to_unique_ai_space_config(
+    search_agent_config: SearchAgentConfig,
+) -> UniqueAIConfig:
+    space = UniqueAISpaceConfig(
+        project_name=search_agent_config.project_name,
+        custom_instructions=search_agent_config.custom_instructions,
+        tools=search_agent_config.tools,
+        language_model=search_agent_config.language_model,
+        type=SpaceType.UNIQUE_AI,
+    )
+
+    prompt_config = UniqueAIPromptConfig(
+        system_prompt_template=search_agent_config.system_prompt_template,
+        user_message_prompt_template=search_agent_config.user_message_prompt_template,
+    )
+
+    services = UniqueAIServices(
+        follow_up_questions_config=search_agent_config.follow_up_questions_config,
+        evaluation_config=search_agent_config.evaluation_config,
+        stock_ticker_config=search_agent_config.stock_ticker_config,
+        uploaded_content_config=search_agent_config.uploaded_content_config,
+    )
+
+    experimental = ExperimentalConfig(
+        thinking_steps_display=search_agent_config.thinking_steps_display,
+        force_checks_on_stream_response_references=search_agent_config.force_checks_on_stream_response_references,
+        temperature=search_agent_config.temperature,
+        additional_llm_options=search_agent_config.additional_llm_options,
+        loop_configuration=search_agent_config.loop_configuration,
+    )
+
+    # Calculate remaining token percentages based on history percentage
+
+    history_percent = search_agent_config.token_limits.percent_of_max_tokens_for_history
+
+    agent = UniqueAIAgentConfig(
+        max_loop_iterations=search_agent_config.max_loop_iterations,
+        input_token_distribution=InputTokenDistributionConfig(
+            percent_for_history=history_percent,
+        ),
+        prompt_config=prompt_config,
+        services=services,
+        experimental=experimental,
+    )
+
+    return UniqueAIConfig(
+        space=space,
+        agent=agent,
+    )
+
+
+def needs_conversion_to_unique_ai_space_config(
+    configuration: dict[str, Any],
+) -> bool:
+    """Check if the configuration needs to be converted to the new UniqueAISpaceConfig."""
+    if (
+        "space_two_point_zero" in configuration
+        or "SpaceTwoPointZeroConfig" in configuration
+        or ("space" in configuration and "agent" in configuration)
+    ):
+        return False
+
+    return True
+
+
+if __name__ == "__main__":
+    import json
+
+    from unique_toolkit._common.utils.write_configuration import (
+        write_service_configuration,
+    )
+
+    write_service_configuration(
+        service_folderpath=Path(__file__).parent.parent,
+        write_folderpath=Path(__file__).parent,
+        config=UniqueAIConfig(),
+        sub_name="unique_ai_config",
+    )
+
+    # TODO: @cdkl Delete these models
+    # This model is only used to have the old and new models in the same json
+    # schema for the data migration in the node chat backend
+
+    # The types can be generated with quicktype.io with the following command:
+    # quicktype unique_ai_old_and_new_config.json \
+    #   --src-lang schema --lang typescript \
+    #   --just-types --prefer-types --explicit-unions \
+    #   -o unique_ai_old_new_configuration.ts \
+    #   --top-level UniqueAIOldAndNewConfig \
+    #   --raw-type any
+
+    # You will need to replace the `any` type with `unknown` in the generated file.
+    # On the branch `feat/unique-ai-configuration-migration-node-chat-part`.
+    # I you further update the types you will need to adapt both branches
+    # - feat/unique-ai-configuration-migration-next-admin-part
+    # - feat/unique-ai-configuration-migration-node-chat-part
+
+    class UniqueAIOldAndNewConfig(BaseModel):
+        new: UniqueAIConfig = UniqueAIConfig()
+        old: SearchAgentConfig = SearchAgentConfig()
+
+    with open(
+        Path(__file__).parent / "unique_ai_old_and_new_config.json",
+        "w",
+    ) as f:
+        json.dump(
+            UniqueAIOldAndNewConfig().model_json_schema(by_alias=True),
+            f,
+            indent=4,
+        )

unique_orchestrator-0.0.1/unique_orchestrator/prompts/generic_reference_prompt.jinja2

@@ -0,0 +1,46 @@
+# Generic Reference Guidelines - ONLY FOLLOW IF NOTHING ELSE APPLICABLE FOR THE TOOL
+Whenever you use information retrieved with a tool, you must adhere to strict reference guidelines. You must strictly reference each fact used with the `source_number` of the corresponding passage, in the following format: '[source<source_number>]'.
+
+Example:
+- The stock price of Apple Inc. is $150 [source0] and the company's revenue increased by 10% [source1].
+- Moreover, the company's market capitalization is $2 trillion [source2][source3].
+- Our internal documents tell us to invest[source4] (Internal)
+
+A fact is preferably referenced by ONLY ONE source, e.g [sourceX], which should be the most relevant source for the fact.
+Follow these guidelines closely and be sure to use the proper `source_number` when referencing facts.
+Make sure that your reference follow the format [sourceX] and that the source number is correct.
+Source is written in singular form and the number is written in digits.
+
+IT IS VERY IMPORTANT TO FOLLOW THESE GUIDELINES!!
+NEVER CITE A source_number THAT YOU DON'T SEE IN THE TOOL CALL RESPONSE!!!
+The source_number in old assistant messages are no longer valid.
+EXAMPLE: If you see [source34] and [source35] in the assistant message, you can't use [source34] again in the next assistant message, this has to be the number you find in the message with role 'tool'.
+BE AWARE:All tool calls have been filtered to remove uncited sources. Tool calls return much more data than you see
+
+### Internal Document Answering Protocol for Employee Questions
+When assisting employees using internal documents, follow
+this structured approach to ensure precise, well-grounded,
+and context-aware responses:
+
+#### 1. Locate and Prioritize Relevant Internal Sources
+Give strong preference to:
+- **Most relevant documents**, such as:
+- **Documents authored by or involving** the employee or team in question
+- **Cross-validated sources**, especially when multiple documents agree
+  - Project trackers, design docs, decision logs, and OKRs
+  - Recently updated or active files
+
+#### 2. Source Reliability Guidelines
+- Prioritize information that is:
+  - **Directly written by domain experts or stakeholders**
+  - **Part of approved or finalized documentation**
+  - **Recently modified or reviewed**, if recency matters
+- Be cautious with:
+  - Outdated drafts
+  - Undocumented opinions or partial records
+
+#### 3. Acknowledge Limitations
+- If no relevant information is found, or documents conflict, clearly state this
+- Indicate where further clarification or investigation may be required
+
+ALWAYS CITE WHEN YOU REFERENCE INFORMATION FROM THE TOOL CALL RESPONSE!!!