unique_toolkit 0.8.18__py3-none-any.whl → 0.8.19__py3-none-any.whl

unique_toolkit/_common/default_language_model.py

@@ -3,4 +3,4 @@ from unique_toolkit.language_model.infos import LanguageModelName
 DEFAULT_GPT_35_TURBO = LanguageModelName.AZURE_GPT_35_TURBO_0125
 DEFAULT_GPT_4o = LanguageModelName.AZURE_GPT_4o_2024_1120
 DEFAULT_GPT_4o_STRUCTURED_OUTPUT = LanguageModelName.AZURE_GPT_4o_2024_0806
-DEFAULT_GPT_4o_MINI = LanguageModelName.AZURE_GPT_4o_MINI_2024_0718
+DEFAULT_GPT_4o_MINI = LanguageModelName.AZURE_GPT_4o_MINI_2024_0718

unique_toolkit/_common/token/token_counting.py

@@ -5,7 +5,10 @@ import json
 from typing import Any, Callable
 
 from pydantic import BaseModel
-from unique_toolkit._common.token.image_token_counting import calculate_image_tokens_from_base64
+
+from unique_toolkit._common.token.image_token_counting import (
+    calculate_image_tokens_from_base64,
+)
 from unique_toolkit.language_model import (
     LanguageModelMessage,
     LanguageModelMessages,
@@ -13,8 +16,6 @@ from unique_toolkit.language_model import (
 )
 
 
-
-
 class SpecialToolCallingTokens(BaseModel):
     func_init: int = 0
     prop_init: int = 0
@@ -123,18 +124,14 @@ def num_tokens_for_tools(
                 )
             )
             if len(function.get("parameters", {}).get("properties", "")) > 0:
-                properties = function.get("parameters", {}).get(
-                    "properties", ""
-                )
+                properties = function.get("parameters", {}).get("properties", "")
                 func_token_count += special_token.prop_init
 
                 for key in list(properties.keys()):
                     func_token_count += special_token.prop_key
 
                     if "enum" in properties[key].keys():
-                        func_token_count += num_token_function_enum(
-                            properties, encode
-                        )
+                        func_token_count += num_token_function_enum(properties, encode)
 
                     func_token_count += len(
                         encode(
@@ -147,9 +144,7 @@ def num_tokens_for_tools(
     return func_token_count
 
 
-def handle_message_with_images(
-    message: list[dict], encode: Callable[[str], list[int]]
-):
+def handle_message_with_images(message: list[dict], encode: Callable[[str], list[int]]):
     token_count = 0
     for item in message:
         if item.get("type") == "image_url":
@@ -168,11 +163,7 @@ def messages_to_openai_messages(
         messages = LanguageModelMessages(messages)
 
     return [
-        {
-            k: v
-            for k, v in m.items()
-            if (k in ["content", "role"] and v is not None)
-        }
+        {k: v for k, v in m.items() if (k in ["content", "role"] and v is not None)}
         for m in json.loads(messages.model_dump_json())
     ]
 
@@ -190,6 +181,4 @@ def num_token_for_language_model_messages(
     messages: LanguageModelMessages | list[LanguageModelMessage],
     encode: Callable[[str], list[int]],
 ) -> int:
-    return num_tokens_from_messages(
-        messages_to_openai_messages(messages), encode
-    )
+    return num_tokens_from_messages(messages_to_openai_messages(messages), encode)

unique_toolkit/_common/validators.py

@@ -2,6 +2,7 @@ import logging
 from typing import Annotated, Any
 
 from pydantic import BeforeValidator, Field, PlainSerializer, ValidationInfo
+from pydantic.fields import FieldInfo
 
 from unique_toolkit.language_model import LanguageModelName
 from unique_toolkit.language_model.infos import (
@@ -9,9 +10,6 @@ from unique_toolkit.language_model.infos import (
     LanguageModelProvider,
 )
 
-from pydantic.fields import FieldInfo
-
-
 logger = logging.getLogger(__name__)
 
 # TODO @klcd: Inform on deprecation of str as input
@@ -28,6 +26,7 @@ LMI = Annotated[
     ),
 ]
 
+
 def get_LMI_default_field(llm_name: LanguageModelName, **kwargs) -> Any:
     return Field(
         default=LanguageModelInfo.from_name(llm_name),

unique_toolkit/debug_info_manager/debug_info_manager.py

@@ -1,4 +1,3 @@
-from unique_toolkit.content.schemas import ContentChunk, ContentReference
 from unique_toolkit.tools.schemas import ToolCallResponse
 
 

unique_toolkit/evals/config.py

@@ -6,7 +6,6 @@ from pydantic import BaseModel, ConfigDict, Field
 from unique_toolkit._common.validators import LMI
 from unique_toolkit.language_model.infos import LanguageModelInfo, LanguageModelName
 
-
 from .schemas import (
     EvaluationMetricName,
 )

unique_toolkit/evals/context_relevancy/schema.py

@@ -1,10 +1,6 @@
-from pydantic import BaseModel, Field, create_model
+from pydantic import BaseModel, ConfigDict, Field, create_model
 from pydantic.json_schema import SkipJsonSchema
 
-
-
-from pydantic import BaseModel, ConfigDict
-
 from unique_toolkit.tools.config import get_configuration_dict
 
 
@@ -12,7 +8,6 @@ class StructuredOutputModel(BaseModel):
     model_config = ConfigDict(extra="forbid")
 
 
-
 class StructuredOutputConfig(BaseModel):
     model_config = get_configuration_dict()
 

unique_toolkit/evals/context_relevancy/service.py

@@ -1,20 +1,9 @@
 import logging
 
 from pydantic import BaseModel, ValidationError
+
 from unique_toolkit.app.schemas import ChatEvent
 from unique_toolkit.chat.service import ChatService
-from unique_toolkit.language_model.infos import (
-    LanguageModelInfo,
-    LanguageModelName,
-    ModelCapabilities,
-)
-from unique_toolkit.language_model.prompt import Prompt
-from unique_toolkit.language_model.schemas import (
-    LanguageModelMessages,
-)
-from unique_toolkit.language_model.service import (
-    LanguageModelService,
-)
 from unique_toolkit.evals.config import EvaluationMetricConfig
 from unique_toolkit.evals.context_relevancy.schema import (
     EvaluationSchemaStructuredOutput,
@@ -30,7 +19,18 @@ from unique_toolkit.evals.schemas import (
     EvaluationMetricName,
     EvaluationMetricResult,
 )
-
+from unique_toolkit.language_model.infos import (
+    LanguageModelInfo,
+    LanguageModelName,
+    ModelCapabilities,
+)
+from unique_toolkit.language_model.prompt import Prompt
+from unique_toolkit.language_model.schemas import (
+    LanguageModelMessages,
+)
+from unique_toolkit.language_model.service import (
+    LanguageModelService,
+)
 
 from .prompts import (
     CONTEXT_RELEVANCY_METRIC_SYSTEM_MSG,

unique_toolkit/evals/evaluation_manager.py

@@ -1,33 +1,31 @@
-from abc import ABC
 import asyncio
-
-
-from unique_toolkit.tools.utils.execution.execution import Result, SafeTaskExecutor
+from abc import ABC
 from logging import Logger
-from unique_toolkit.evals.schemas import (
-    EvaluationAssessmentMessage,
-    EvaluationMetricName,
-    EvaluationMetricResult,
-)
+
 from unique_toolkit.chat.schemas import (
     ChatMessageAssessmentStatus,
     ChatMessageAssessmentType,
 )
 from unique_toolkit.chat.service import ChatService
+from unique_toolkit.evals.schemas import (
+    EvaluationAssessmentMessage,
+    EvaluationMetricName,
+    EvaluationMetricResult,
+)
 from unique_toolkit.language_model.schemas import (
     LanguageModelStreamResponse,
 )
+from unique_toolkit.tools.utils.execution.execution import Result, SafeTaskExecutor
 
 
 class Evaluation(ABC):
     """
     Abstract base class for evaluation metrics.
-    
+
     typical use cases include:
     - Hallucination checking
     - compliance checking
     """
-   
 
     def __init__(self, name: EvaluationMetricName):
         self.name = name
@@ -73,8 +71,8 @@ class EvaluationManager:
 
     The EvaluationManager serves as the central hub for managing and executing evaluations.
     """
-    # a hashmap to hold evaluations by their names
 
+    # a hashmap to hold evaluations by their names
 
     def __init__(
         self,

unique_toolkit/evals/hallucination/constants.py

@@ -16,7 +16,6 @@ from unique_toolkit.evals.schemas import (
 )
 from unique_toolkit.language_model.infos import LanguageModelInfo, LanguageModelName
 
-
 SYSTEM_MSG_KEY = "systemPrompt"
 USER_MSG_KEY = "userPrompt"
 SYSTEM_MSG_DEFAULT_KEY = "systemPromptDefault"

unique_toolkit/evals/hallucination/hallucination_evaluation.py

@@ -1,5 +1,3 @@
-from typing import Any
-
 from unique_toolkit.app.schemas import ChatEvent
 from unique_toolkit.chat.schemas import (
     ChatMessageAssessmentLabel,
@@ -7,6 +5,9 @@ from unique_toolkit.chat.schemas import (
     ChatMessageAssessmentType,
 )
 from unique_toolkit.evals.evaluation_manager import Evaluation
+from unique_toolkit.evals.hallucination.constants import (
+    HallucinationConfig,
+)
 from unique_toolkit.evals.hallucination.utils import check_hallucination
 from unique_toolkit.evals.schemas import (
     EvaluationAssessmentMessage,
@@ -14,17 +15,13 @@ from unique_toolkit.evals.schemas import (
     EvaluationMetricName,
     EvaluationMetricResult,
 )
-from unique_toolkit.evals.hallucination.constants import (
-    HallucinationConfig,
+from unique_toolkit.language_model.schemas import (
+    LanguageModelStreamResponse,
 )
 from unique_toolkit.reference_manager.reference_manager import (
     ReferenceManager,
 )
 
-from unique_toolkit.language_model.schemas import (
-    LanguageModelStreamResponse,
-)
-
 
 class HallucinationEvaluation(Evaluation):
     def __init__(

unique_toolkit/evals/hallucination/service.py

@@ -4,7 +4,6 @@ from unique_toolkit.app.schemas import ChatEvent
 from unique_toolkit.evals.config import EvaluationMetricConfig
 from unique_toolkit.evals.schemas import EvaluationMetricInput, EvaluationMetricResult
 
-
 from .constants import hallucination_metric_default_config
 from .utils import check_hallucination
 

unique_toolkit/evals/hallucination/utils.py

@@ -2,13 +2,6 @@ import logging
 from string import Template
 
 from unique_toolkit.content.schemas import ContentChunk
-from unique_toolkit.language_model.schemas import (
-    LanguageModelMessages,
-    LanguageModelStreamResponse,
-    LanguageModelSystemMessage,
-    LanguageModelUserMessage,
-)
-from unique_toolkit.language_model.service import LanguageModelService
 from unique_toolkit.evals.config import EvaluationMetricConfig
 from unique_toolkit.evals.exception import EvaluatorException
 from unique_toolkit.evals.output_parser import parse_eval_metric_result
@@ -17,7 +10,13 @@ from unique_toolkit.evals.schemas import (
     EvaluationMetricName,
     EvaluationMetricResult,
 )
-
+from unique_toolkit.language_model.schemas import (
+    LanguageModelMessages,
+    LanguageModelStreamResponse,
+    LanguageModelSystemMessage,
+    LanguageModelUserMessage,
+)
+from unique_toolkit.language_model.service import LanguageModelService
 
 from .constants import (
     SYSTEM_MSG_DEFAULT_KEY,

unique_toolkit/evals/output_parser.py

@@ -1,4 +1,3 @@
-from unique_toolkit.language_model.utils import convert_string_to_json
 from unique_toolkit.evals.context_relevancy.schema import (
     EvaluationSchemaStructuredOutput,
 )
@@ -7,6 +6,7 @@ from unique_toolkit.evals.schemas import (
     EvaluationMetricName,
     EvaluationMetricResult,
 )
+from unique_toolkit.language_model.utils import convert_string_to_json
 
 
 def parse_eval_metric_result(

unique_toolkit/evals/schemas.py

@@ -2,13 +2,14 @@ from enum import StrEnum
 from typing import Optional
 
 from pydantic import BaseModel, ConfigDict, Field
+
 from unique_toolkit.chat import ChatMessage
-from unique_toolkit.evals.exception import EvaluatorException
 from unique_toolkit.chat.schemas import (
     ChatMessageAssessmentLabel,
     ChatMessageAssessmentStatus,
     ChatMessageAssessmentType,
 )
+from unique_toolkit.evals.exception import EvaluatorException
 
 
 class EvaluationMetricName(StrEnum):

unique_toolkit/evals/tests/test_context_relevancy_service.py

@@ -1,17 +1,9 @@
 from unittest.mock import MagicMock, patch
 
 import pytest
+
 from unique_toolkit.app.schemas import ChatEvent
 from unique_toolkit.chat.service import LanguageModelName
-from unique_toolkit.language_model.infos import (
-    LanguageModelInfo,
-)
-from unique_toolkit.language_model.schemas import (
-    LanguageModelAssistantMessage,
-    LanguageModelCompletionChoice,
-    LanguageModelMessages,
-)
-from unique_toolkit.language_model.service import LanguageModelResponse
 from unique_toolkit.evals.config import EvaluationMetricConfig
 from unique_toolkit.evals.context_relevancy.prompts import (
     CONTEXT_RELEVANCY_METRIC_SYSTEM_MSG,
@@ -28,6 +20,15 @@ from unique_toolkit.evals.schemas import (
     EvaluationMetricName,
     EvaluationMetricResult,
 )
+from unique_toolkit.language_model.infos import (
+    LanguageModelInfo,
+)
+from unique_toolkit.language_model.schemas import (
+    LanguageModelAssistantMessage,
+    LanguageModelCompletionChoice,
+    LanguageModelMessages,
+)
+from unique_toolkit.language_model.service import LanguageModelResponse
 
 
 @pytest.fixture

unique_toolkit/evals/tests/test_output_parser.py

@@ -1,13 +1,17 @@
 import pytest
 
-from unique_toolkit.evals.context_relevancy.schema import EvaluationSchemaStructuredOutput, Fact
+from unique_toolkit.evals.context_relevancy.schema import (
+    EvaluationSchemaStructuredOutput,
+    Fact,
+)
 from unique_toolkit.evals.exception import EvaluatorException
-from unique_toolkit.evals.output_parser import parse_eval_metric_result, parse_eval_metric_result_structured_output
+from unique_toolkit.evals.output_parser import (
+    parse_eval_metric_result,
+    parse_eval_metric_result_structured_output,
+)
 from unique_toolkit.evals.schemas import EvaluationMetricName, EvaluationMetricResult
 
 
-
-
 def test_parse_eval_metric_result_success():
     # Test successful parsing with all fields
     result = '{"value": "high", "reason": "Test reason"}'

unique_toolkit/history_manager/history_construction_with_contents.py

@@ -1,26 +1,24 @@
 import base64
 import mimetypes
-
 from datetime import datetime
 from enum import StrEnum
 
 import numpy as np
 import tiktoken
-
 from pydantic import RootModel
 
-from unique_toolkit._common.token.token_counting import num_tokens_per_language_model_message
-from unique_toolkit.chat.service import ChatService
-from unique_toolkit.content.service import ContentService
-from unique_toolkit.language_model.schemas import LanguageModelMessages
+from unique_toolkit._common.token.token_counting import (
+    num_tokens_per_language_model_message,
+)
 from unique_toolkit.app import ChatEventUserMessage
 from unique_toolkit.chat.schemas import ChatMessage
 from unique_toolkit.chat.schemas import ChatMessageRole as ChatRole
+from unique_toolkit.chat.service import ChatService
 from unique_toolkit.content.schemas import Content
+from unique_toolkit.content.service import ContentService
 from unique_toolkit.language_model import LanguageModelMessageRole as LLMRole
 from unique_toolkit.language_model.infos import EncoderName
-
-
+from unique_toolkit.language_model.schemas import LanguageModelMessages
 
 # TODO: Test this once it moves into the unique toolkit
 
@@ -188,8 +186,7 @@ def file_content_serialization(
             return ""
         case FileContentSerialization.FILE_NAME:
             file_names = [
-                f"- Uploaded file: {f.key} at {f.created_at}"
-                for f in file_contents
+                f"- Uploaded file: {f.key} at {f.created_at}" for f in file_contents
             ]
             return "\n".join(
                 [
@@ -226,12 +223,8 @@ def get_full_history_with_contents(
             text = ""
 
         if len(c.contents) > 0:
-            file_contents = [
-                co for co in c.contents if is_file_content(co.key)
-            ]
-            image_contents = [
-                co for co in c.contents if is_image_content(co.key)
-            ]
+            file_contents = [co for co in c.contents if is_file_content(co.key)]
+            image_contents = [co for co in c.contents if is_image_content(co.key)]
 
             content = (
                 text
@@ -285,7 +278,6 @@ def get_full_history_as_llm_messages(
     return builder.build()
 
 
-
 def limit_to_token_window(
     messages: LanguageModelMessages,
     token_limit: int,
@@ -297,9 +289,7 @@ def limit_to_token_window(
         encode=encoder.encode,
     )
 
-    to_take: list[bool] = (
-        np.cumsum(token_per_message_reversed) < token_limit
-    ).tolist()
+    to_take: list[bool] = (np.cumsum(token_per_message_reversed) < token_limit).tolist()
     to_take.reverse()
 
     return LanguageModelMessages(

unique_toolkit/history_manager/history_manager.py

@@ -3,32 +3,28 @@ from typing import Annotated, Awaitable, Callable
 
 from pydantic import BaseModel, Field
 
+from unique_toolkit._common.validators import LMI
 from unique_toolkit.app.schemas import ChatEvent
-
-
-
+from unique_toolkit.history_manager.loop_token_reducer import LoopTokenReducer
+from unique_toolkit.history_manager.utils import transform_chunks_to_string
+from unique_toolkit.language_model.infos import LanguageModelInfo, LanguageModelName
 from unique_toolkit.language_model.schemas import (
     LanguageModelAssistantMessage,
     LanguageModelFunction,
-    LanguageModelMessage,   
+    LanguageModelMessage,
     LanguageModelMessages,
-    LanguageModelToolMessage
+    LanguageModelToolMessage,
 )
-
-from unique_toolkit.tools.schemas import ToolCallResponse
-from unique_toolkit.history_manager.utils import transform_chunks_to_string
-
-from unique_toolkit._common.validators import LMI
-from unique_toolkit.history_manager.loop_token_reducer import LoopTokenReducer
-from unique_toolkit.language_model.infos import LanguageModelInfo, LanguageModelName
 from unique_toolkit.reference_manager.reference_manager import ReferenceManager
 from unique_toolkit.tools.config import get_configuration_dict
+from unique_toolkit.tools.schemas import ToolCallResponse
 
 DeactivatedNone = Annotated[
     None,
     Field(title="Deactivated", description="None"),
 ]
 
+
 class UploadedContentConfig(BaseModel):
     model_config = get_configuration_dict()
 
@@ -44,8 +40,8 @@ class UploadedContentConfig(BaseModel):
         description="The fraction of the max input tokens that will be reserved for the uploaded content.",
     )
 
-class ExperimentalFeatures(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.",
@@ -53,14 +49,11 @@ class ExperimentalFeatures(BaseModel):
 
 
 class HistoryManagerConfig(BaseModel):
-
-
     experimental_features: ExperimentalFeatures = Field(
         default=ExperimentalFeatures(),
         description="Experimental features for the history manager.",
     )
 
-
     percent_of_max_tokens_for_history: float = Field(
         default=0.2,
         ge=0.0,
@@ -88,7 +81,6 @@ class HistoryManagerConfig(BaseModel):
     ) = UploadedContentConfig()
 
 
-
 class HistoryManager:
     """
     Manages the history of tool calls and conversation loops.
@@ -110,7 +102,6 @@ class HistoryManager:
     The HistoryManager serves as the backbone for managing and retrieving conversation history in a structured and efficient manner.
     """
 
-
     def __init__(
         self,
         logger: Logger,
@@ -134,7 +125,6 @@ class HistoryManager:
         self._loop_history: list[LanguageModelMessage] = []
         self._source_enumerator = 0
 
-
     def has_no_loop_messages(self) -> bool:
         return len(self._loop_history) == 0
 
@@ -201,13 +191,12 @@ class HistoryManager:
     def add_assistant_message(self, message: LanguageModelAssistantMessage) -> None:
         self._loop_history.append(message)
 
-        
     async def get_history_for_model_call(
         self,
         original_user_message: str,
         rendered_user_message_string: str,
         rendered_system_message_string: str,
-        remove_from_text: Callable[[str], Awaitable[str]]
+        remove_from_text: Callable[[str], Awaitable[str]],
     ) -> LanguageModelMessages:
         self._logger.info("Getting history for model call -> ")
 
@@ -218,4 +207,4 @@ class HistoryManager:
             loop_history=self._loop_history,
             remove_from_text=remove_from_text,
         )
-        return messages
+        return messages