ibm-watsonx-gov 1.3.3__cp313-cp313-win_amd64.whl

ibm_watsonx_gov/entities/enums.py

@@ -0,0 +1,274 @@
+# ----------------------------------------------------------------------------------------------------
+# IBM Confidential
+# Licensed Materials - Property of IBM
+# 5737-H76, 5900-A3Q
+# © Copyright IBM Corp. 2025  All Rights Reserved.
+# US Government Users Restricted Rights - Use, duplication or disclosure restricted by
+# GSA ADPSchedule Contract with IBM Corp.
+# ----------------------------------------------------------------------------------------------------
+
+from enum import Enum
+from typing import TYPE_CHECKING
+
+from pydantic_core import PydanticUndefined
+
+if TYPE_CHECKING:
+    from ibm_watsonx_gov.entities.metric import GenAIMetric
+
+
+class EvaluationProvider(Enum):
+    """Supported evaluation providers for metrics computation"""
+    UNITXT = "unitxt"
+    WATSONX_GOV = "watsonx_governance"
+    DETECTORS = "detectors"
+    IBM = "ibm"
+
+    @staticmethod
+    def values():
+        """Get all values of the enum"""
+        return [e.value for e in EvaluationProvider]
+
+
+class TaskType(Enum):
+    """Supported task types for generative AI models"""
+    QA = "question_answering"
+    CLASSIFICATION = "classification"
+    SUMMARIZATION = "summarization"
+    GENERATION = "generation"
+    EXTRACTION = "extraction"
+    RAG = "retrieval_augmented_generation"
+
+    @staticmethod
+    def values():
+        return [e.value for e in TaskType]
+
+    def get_metrics(self) -> list["GenAIMetric"]:
+        """
+        Helper method to return a list of metrics for the task type
+
+        Returns:
+            list[GenAIMetric]: list of gen AI metrics for the specified task type
+        """
+        # localized import to avoid circular import issues
+        from ibm_watsonx_gov.entities.metric import GenAIMetric
+
+        metrics_list = []
+        for metric_class in GenAIMetric.__subclasses__():
+
+            # Check if the class has required fields with no default value
+            required_fields = {}
+            metric_class_fields = metric_class.model_fields
+            for field, field_info in metric_class_fields.items():
+                if field_info.is_required() and field_info.default == PydanticUndefined:
+                    required_fields[field] = None
+
+            metric = metric_class.model_construct(**required_fields)
+            if self in metric.tasks:
+                metrics_list.append(metric)
+
+        return metrics_list
+
+
+class InputDataType(Enum):
+    """Supported input data types"""
+    STRUCTURED = "structured"
+    TEXT = "unstructured_text"
+    IMAGE = "unstructured_image"
+    MULTIMODAL = "multimodal"
+
+
+class ProblemType(Enum):
+    """Supported problem types for predictive AI models"""
+    BINARY = "binary"
+    MULTICLASS = "multiclass"
+    REGRESSION = "regression"
+
+
+class EvaluationStage(Enum):
+    """Supported evaluation stages"""
+    DEVELOPMENT = "development"
+    PRE_PRODUCTION = "pre_production"
+    PRODUCTION = "production"
+
+
+class ModelProviderType(Enum):
+    """Supported model provider types for Generative AI"""
+    IBM_WATSONX_AI = "ibm_watsonx.ai"
+    AZURE_OPENAI = "azure_openai"
+    RITS = "rits"
+    OPENAI = "openai"
+    VERTEX_AI = "vertex_ai"
+    GOOGLE_AI_STUDIO = "google_ai_studio"
+    AWS_BEDROCK = "aws_bedrock"
+    CUSTOM = "custom"
+    PORTKEY = "portkey"
+    WXO_AI_GATEWAY = "wxo_ai_gateway"
+
+class Region(Enum):
+    """Supported IBM Cloud regions"""
+    US_SOUTH = "us-south"
+    EU_DE = "eu-de"
+    AU_SYD = "au-syd"
+    CA_TOR = "ca-tor"
+    JP_TOK = "jp-tok"
+    EU_GB = "eu-gb"
+    AP_SOUTH = "ap-south"
+    US_GOV_EAST1 = "us-gov-east1"
+
+    @staticmethod
+    def values():
+        """Get all values of the enum"""
+        return [e.value for e in Region]
+
+
+class EvaluatorFields(Enum):
+    """Fields used in the evaluator"""
+    INPUT_FIELDS = "input_fields"
+    OUTPUT_FIELDS = "output_fields"
+    REFERENCE_FIELDS = "reference_fields"
+    CONTEXT_FIELDS = "context_fields"
+    RECORD_ID_FIELD = "record_id_field"
+    RECORD_TIMESTAMP_FIELD = "record_timestamp_field"
+    MESSAGE_ID_FIELD = "message_id_field"
+    CONVERSATION_ID_FIELD = "conversation_id_field"
+    TOOL_CALLS_FIELD = "tool_calls_field"
+    AVAILABLE_TOOLS_FIELD = "available_tools_field"
+    PROMPT_FIELD = "prompt_field"
+    STATUS_FIELD = "status_field"
+    START_TIME_FIELD = "start_time_field"
+    END_TIME_FIELD = "end_time_field"
+    MODEL_USAGE_DETAIL_FIELDS = "model_usage_detail_fields"
+    INPUT_TOKEN_COUNT_FIELDS = "input_token_count_fields"
+    OUTPUT_TOKEN_COUNT_FIELDS = "output_token_count_fields"
+    USER_ID_FIELD = "user_id_field"
+
+    @staticmethod
+    def get_default_fields_mapping() -> dict["EvaluatorFields", str | list[str]]:
+        """Get the default fields mapping for the evaluator"""
+        return {
+            EvaluatorFields.INPUT_FIELDS: ["input_text"],
+            EvaluatorFields.OUTPUT_FIELDS: ["generated_text"],
+            EvaluatorFields.REFERENCE_FIELDS: ["ground_truth"],
+            EvaluatorFields.CONTEXT_FIELDS: ["context"],
+            EvaluatorFields.RECORD_ID_FIELD: "record_id",
+            EvaluatorFields.RECORD_TIMESTAMP_FIELD: "record_timestamp",
+            EvaluatorFields.TOOL_CALLS_FIELD: "tool_calls",
+            EvaluatorFields.MESSAGE_ID_FIELD: "message_id",
+            EvaluatorFields.CONVERSATION_ID_FIELD: "conversation_id",
+            EvaluatorFields.PROMPT_FIELD: "model_prompt",
+            EvaluatorFields.STATUS_FIELD: "status"
+        }
+
+
+class ContainerType(Enum):
+    PROJECT = "project"
+    SPACE = "space"
+
+
+class MistakeType(Enum):
+    """Enum for TCH Mistake Type"""
+    WRONG_NUMBER_OF_APIS = "Wrong number of APIs"
+    NEED_MORE_APIS = "Need more APIs"
+    WRONG_API_SELECTION = "Wrong API selection"
+    NEED_MORE_INFORMATION = "Need more information"
+    WRONG_PARAMETER_VALUE = "Wrong parameter value"
+    WRONG_UNITS_TRANSFORMATION = "Wrong units transformation"
+    HALLUCINATED_CALL = "Hallucinated API call"
+    HALLUCINATED_PARAM = "Hallucinated parameter"
+    PARAM_TYPE = "Incorrect parameter value type"
+    ALLOWED_VALUES = "Allowed values"
+    REQ_PARAM_MISSING = "Missing required parameter"
+    NONE_FUNC = "Not able to identify the user's intent."
+
+
+class MetricGroup(Enum):
+    RETRIEVAL_QUALITY = "retrieval_quality"
+    ANSWER_QUALITY = "answer_quality"
+    CONTENT_SAFETY = "content_safety"
+    PERFORMANCE = "performance"
+    USAGE = "usage"
+    MESSAGE_COMPLETION = "message_completion"
+    TOOL_CALL_QUALITY = "tool_call_quality"
+    READABILITY = "readability"
+    CUSTOM = "custom"
+
+    def get_metrics(self) -> list["GenAIMetric"]:
+        """
+        Helper method to return a list of metrics under each metric group
+
+        Returns:
+            list[GenAIMetric]: list of gen AI metrics for the specified metric group
+        """
+        # localized import to avoid circular import issues
+        from ibm_watsonx_gov.entities.metric import GenAIMetric
+        metrics_list = []
+        for metric_class in GenAIMetric.__subclasses__():
+
+            # Check if the class has required fields with no default value
+            required_fields = {}
+            metric_class_fields = metric_class.model_fields
+            for field, field_info in metric_class_fields.items():
+                if field_info.is_required() and field_info.default == PydanticUndefined:
+                    required_fields[field] = None
+
+            metric = metric_class.model_construct(**required_fields)
+            if metric.group == self:
+                metrics_list.append(metric)
+
+        return metrics_list
+
+
+class MessageStatus(Enum):
+    """Enum for message status."""
+    UNKNOWN = "unknown"
+    SUCCESSFUL = "successful"
+    FAILURE = "failure"
+
+    @staticmethod
+    def values():
+        return [e.value for e in MessageStatus]
+
+
+class MetricType(Enum):
+    OOTB = "ootb"
+    CUSTOM = "custom"
+
+    @staticmethod
+    def values():
+        return [e.value for e in MetricType]
+
+
+class CategoryClassificationType(Enum):
+    FAVOURABLE = "favourable"
+    UNFAVOURABLE = "unfavourable"
+    NEUTRAL = "neutral"
+
+    @staticmethod
+    def values():
+        return [e.value for e in CategoryClassificationType]
+
+
+class MetricValueType(Enum):
+    NUMERIC = "numeric"
+    CATEGORICAL = "categorical"
+
+    @staticmethod
+    def values():
+        return [e.value for e in MetricValueType]
+
+
+class GraniteGuardianRisks(Enum):
+    HARM = "harm"
+    SOCIAL_BIAS = "social_bias"
+    PROFANITY = "profanity"
+    SEXUAL_CONTENT = "sexual_content"
+    UNETHICAL_BEHAVIOR = "unethical_behavior"
+    VIOLENCE = "violence"
+    HARM_ENGAGEMENT = "harm_engagement"
+    EVASIVENESS = "evasiveness"
+    JAILBREAK = "jailbreak"
+    TOOL_CALL_ACCURACY = "tool_call_accuracy"
+
+    @staticmethod
+    def values():
+        return [e.value for e in GraniteGuardianRisks]