arize 8.0.0b0__py3-none-any.whl → 8.0.0b2__py3-none-any.whl

arize/client.py

@@ -20,9 +20,6 @@ if TYPE_CHECKING:
 
 logger = logging.getLogger(__name__)
 
-# TODO(Kiko): Clean commented lines over the SDK
-# TODO(Kiko): Implement https://github.com/Arize-ai/arize/pull/59917
-
 # TODO(Kiko): Go private connect. Need a `base_domain`, such that we get:
 # - api.<base_domain>
 # - app.<base_domain>
@@ -31,29 +28,23 @@ logger = logging.getLogger(__name__)
 
 # TODO(Kiko): Enforce type checking, remove all type ignores
 
-# TODO(Kiko): Go over docstrings
-# TODO(Kiko): Missing parameter descriptions in some docstrings
-# TODO(Kiko): Missing return descriptions in some docstrings
-
 # TODO(Kiko): Go over headers on each logging call
+# TODO(Kiko): InvalidAdditionalHeadersError is unused. Have we handled extra headers?
 
 # TODO(Kiko): Need to implement 'Update existing examples in a dataset'
 
-# TODO(Kiko): why logs don't show on scripts, only on jupyter notebooks
 # TODO(Kiko): test caching in colab environment
 # TODO(Kiko): Protobuf versioning is too old
 # TODO(Kiko): Go through main APIs and add CtxAdapter where missing
 # TODO(Kiko): Search and handle other TODOs
-# TODO(Kiko): Go over **every file** and do not import anything at runtime, use `if TYPE_CHECKING`
-# with `from __future__ import annotations` (must include for Python < 3.11)
 
 
 class ArizeClient(LazySubclientsMixin):
     """Root client for the Arize SDK.
 
     The ArizeClient provides access to all Arize platform services including datasets,
-    experiments, ML models, projects, and spans. It uses SDKConfiguration internally to
-    manage configuration settings.
+    experiments, ML models, projects, and spans. It uses :class:`arize.config.SDKConfiguration`
+    internally to manage configuration settings.
 
     All parameters are optional (except api_key which must be provided via argument
     or environment variable). For each parameter, values are resolved in this order:
@@ -105,44 +96,16 @@ class ArizeClient(LazySubclientsMixin):
             "SpansClient",
         ),
     }
-    _EXTRAS: ClassVar[dict[str, tuple[str | None, tuple[str, ...]]]] = {
-        # Gate only the generated-backed ones
-        "datasets": (
-            "datasets-experiments",
-            (
-                "pydantic",
-                "openinference.semconv",
-            ),
-        ),
-        "experiments": (
-            "datasets-experiments",
-            (
-                "pydantic",
-                "wrapt",
-                # "numpy",
-                # "openinference.semconv",
-                # "opentelemetry.sdk",
-                # "opentelemetry.exporter.otlp.proto.grpc.trace_exporter",
-            ),
-        ),
-        "spans": (
-            "spans",
-            (
-                "google.protobuf",
-                "numpy",
-                "openinference.semconv",
-                "opentelemetry",
-                "pandas",
-                "pyarrow",
-                "requests",
-                "tqdm",
-            ),
-        ),
-        # Imports are gated in each method of the models client
-        # This is to allow for very lean package install if people only
-        # want to stream ML records
-        "ml": (None, ()),
-    }
+    # DISABLED: Optional dependency gating system
+    # This dict would map subclients to their optional dependencies and extra names.
+    # When enabled, it prevents loading subclients if dependencies aren't installed,
+    # showing: "Install via: pip install arize[extra-name]"
+    #
+    # To re-enable, populate with entries like:
+    # "subclient_name": ("extra-name", ("package1", "package2", "package3")),
+    # "another_subclient": (None, ()),  # No requirements
+    #
+    _EXTRAS: ClassVar[dict[str, tuple[str | None, tuple[str, ...]]]] = {}
 
     def __init__(
         self,
@@ -175,7 +138,7 @@ class ArizeClient(LazySubclientsMixin):
             api_key: Arize API key for authentication. Required - must be provided here
                 or via ARIZE_API_KEY environment variable. Raises MissingAPIKeyError if not set.
             region: Arize region (e.g., Region.US_CENTRAL, Region.EU_WEST). When specified,
-                overrides individual host/port settings. ENV: ARIZE_REGION. Default: Region.UNSPECIFIED.
+                overrides individual host/port settings. ENV: ARIZE_REGION. Default: Region.UNSET.
             api_host: Custom API endpoint host. ENV: ARIZE_API_HOST. Default: "api.arize.com".
             api_scheme: API endpoint scheme (http/https). ENV: ARIZE_API_SCHEME. Default: "https".
             otlp_host: OTLP endpoint host. ENV: ARIZE_OTLP_HOST. Default: "otlp.arize.com".
@@ -210,7 +173,8 @@ class ArizeClient(LazySubclientsMixin):
 
         Notes:
             Values provided to this class override environment variables, which in turn
-            override default values. See SDKConfiguration for detailed parameter documentation.
+            override default values. See :class:`arize.config.SDKConfiguration`
+            for detailed parameter documentation.
         """
         cfg_kwargs: dict = {}
         if api_key is not None:

arize/config.py

@@ -3,10 +3,8 @@
 import logging
 import os
 import sys
-import threading
 from dataclasses import dataclass, field, fields
 from pathlib import Path
-from typing import Any
 
 from arize.constants.config import (
     DEFAULT_API_HOST,
@@ -167,6 +165,9 @@ def _parse_bool(val: bool | str | None) -> bool:
 class SDKConfiguration:
     """Configuration for the Arize SDK with endpoint and authentication settings.
 
+    This class holds pure configuration data and does not manage client lifecycle.
+    Client creation and caching is handled by :class:`arize.ArizeClient`.
+
     This class is used internally by ArizeClient to manage SDK configuration. It is not
     recommended to use this class directly; users should interact with ArizeClient
     instead.
@@ -225,13 +226,16 @@ class SDKConfiguration:
         region: Arize region (e.g., US_CENTRAL, EU_WEST). When specified, overrides
             individual host/port settings.
             Environment variable: ARIZE_REGION.
-            Default: Region.UNSPECIFIED.
+            Default: :class:`Region.UNSET`.
         single_host: Single host to use for all endpoints. Overrides individual host settings.
             Environment variable: ARIZE_SINGLE_HOST.
             Default: "" (not set).
         single_port: Single port to use for all endpoints. Overrides individual port settings (0-65535).
             Environment variable: ARIZE_SINGLE_PORT.
             Default: 0 (not set).
+
+    Raises:
+        MissingAPIKeyError: If api_key is not provided via argument or environment variable.
     """
 
     api_key: str = field(
@@ -323,17 +327,11 @@ class SDKConfiguration:
         )
     )
 
-    # Private, excluded from comparisons & repr
-    _gen_client: Any = field(default=None, repr=False, compare=False)
-    _gen_lock: threading.Lock = field(
-        default_factory=threading.Lock, repr=False, compare=False
-    )
-
     def __post_init__(self) -> None:
         """Validate and configure SDK endpoints after initialization.
 
         Raises:
-            MissingAPIKeyError: If API key is not provided.
+            MissingAPIKeyError: If api_key is not provided via argument or environment variable.
         """
         # Validate Configuration
         if not self.api_key:
@@ -341,7 +339,7 @@ class SDKConfiguration:
 
         has_single_host = bool(self.single_host)
         has_single_port = self.single_port != 0
-        has_region = self.region is not Region.UNSPECIFIED
+        has_region = self.region is not Region.UNSET
         if (has_single_host or has_single_port) and has_region:
             logger.info(
                 "Multiple endpoint override options provided. Preference order is: "
@@ -444,28 +442,3 @@ class SDKConfiguration:
             lines.append(f"  {f.name}={val!r},")
         lines.append(")")
         return "\n".join(lines)
-
-    # TODO(Kiko): This may not be well placed in this class
-    def get_generated_client(self) -> object:
-        """Get or create the generated OpenAPI client instance."""
-        # If already cached, return immediately
-        if self._gen_client is not None:
-            return self._gen_client
-
-        # Thread-safe initialization
-        with self._gen_lock:
-            if self._gen_client is not None:
-                return self._gen_client
-
-            # Import lazily so extra dependencies can be
-            # enforced outside the configuration class
-            from arize._generated import api_client as gen
-
-            cfg = gen.Configuration(host=self.api_url)
-            if self.api_key:
-                cfg.access_token = self.api_key
-            client = gen.ApiClient(cfg)
-
-            # Bypass frozen to set the cache once
-            object.__setattr__(self, "_gen_client", client)
-            return client

arize/constants/ml.py

@@ -3,50 +3,43 @@
 import json
 from pathlib import Path
 
-# MAX_BYTES_PER_BULK_RECORD = 100000
-# MAX_DAYS_WITHIN_RANGE = 365
 MIN_PREDICTION_ID_LEN = 1
 MAX_PREDICTION_ID_LEN = 512
 MIN_DOCUMENT_ID_LEN = 1
 MAX_DOCUMENT_ID_LEN = 128
-# # The maximum number of character for tag values
+# The maximum number of character for tag values
 MAX_TAG_LENGTH = 20_000
 MAX_TAG_LENGTH_TRUNCATION = 1_000
-# # The maximum number of character for embedding raw data
+# The maximum number of character for embedding raw data
 MAX_RAW_DATA_CHARACTERS = 2_000_000
 MAX_RAW_DATA_CHARACTERS_TRUNCATION = 5_000
 # The maximum number of acceptable years in the past from current time for prediction_timestamps
 MAX_PAST_YEARS_FROM_CURRENT_TIME = 5
 # The maximum number of acceptable years in the future from current time for prediction_timestamps
 MAX_FUTURE_YEARS_FROM_CURRENT_TIME = 1
-# # The maximum number of character for llm model name
+# The maximum number of character for llm model name
 MAX_LLM_MODEL_NAME_LENGTH = 20_000
 MAX_LLM_MODEL_NAME_LENGTH_TRUNCATION = 50
-# # The maximum number of character for prompt template
+# The maximum number of character for prompt template
 MAX_PROMPT_TEMPLATE_LENGTH = 50_000
 MAX_PROMPT_TEMPLATE_LENGTH_TRUNCATION = 5_000
-# # The maximum number of character for prompt template version
+# The maximum number of character for prompt template version
 MAX_PROMPT_TEMPLATE_VERSION_LENGTH = 20_000
 MAX_PROMPT_TEMPLATE_VERSION_LENGTH_TRUNCATION = 50
-# # The maximum number of embeddings
+# The maximum number of embeddings
 MAX_NUMBER_OF_EMBEDDINGS = 30
 MAX_EMBEDDING_DIMENSIONALITY = 20_000
-# # The maximum number of classes for multi class
+# The maximum number of classes for multi class
 MAX_NUMBER_OF_MULTI_CLASS_CLASSES = 500
 MAX_MULTI_CLASS_NAME_LENGTH = 100
 # The maximum number of references in embedding similarity search params
 MAX_NUMBER_OF_SIMILARITY_REFERENCES = 10
-#
-# # Arize generated columns
-# GENERATED_PREDICTION_LABEL_COL = "arize_generated_prediction_label"
-# GENERATED_LLM_PARAMS_JSON_COL = "arize_generated_llm_params_json"
-#
-# # reserved columns for LLM run metadata
+# reserved columns for LLM run metadata
 LLM_RUN_METADATA_TOTAL_TOKEN_COUNT_TAG_NAME = "total_token_count"  # noqa: S105
 LLM_RUN_METADATA_PROMPT_TOKEN_COUNT_TAG_NAME = "prompt_token_count"  # noqa: S105
 LLM_RUN_METADATA_RESPONSE_TOKEN_COUNT_TAG_NAME = "response_token_count"  # noqa: S105
 LLM_RUN_METADATA_RESPONSE_LATENCY_MS_TAG_NAME = "response_latency_ms"
-#
+
 # all reserved tags
 RESERVED_TAG_COLS = [
     LLM_RUN_METADATA_TOTAL_TOKEN_COUNT_TAG_NAME,

arize/constants/spans.py

@@ -5,19 +5,15 @@ DEFAULT_DATETIME_FMT = "%Y-%m-%dT%H:%M:%S.%f+00:00"
 # Minumum/Maximum number of characters for span/trace/parent ids in spans
 SPAN_ID_MIN_STR_LENGTH = 12
 SPAN_ID_MAX_STR_LENGTH = 128
-# # Minumum/Maximum number of characters for span name
+# Minumum/Maximum number of characters for span name
 SPAN_NAME_MIN_STR_LENGTH = 0
 SPAN_NAME_MAX_STR_LENGTH = 50
-# # Minumum/Maximum number of characters for span status message
+# Minumum/Maximum number of characters for span status message
 SPAN_STATUS_MSG_MIN_STR_LENGTH = 0
 SPAN_STATUS_MSG_MAX_STR_LENGTH = 10_000
-# # Minumum/Maximum number of characters for span event name
+# Minumum/Maximum number of characters for span event name
 SPAN_EVENT_NAME_MAX_STR_LENGTH = 100
-# # Minumum/Maximum number of characters for span event attributes
-# SPAN_EVENT_ATTRS_MAX_STR_LENGTH = 10_000
-# # Maximum number of characters for span kind
-# SPAN_KIND_MAX_STR_LENGTH = 100
-# SPAN_EXCEPTION_TYPE_MAX_STR_LENGTH = 100
+# Minumum/Maximum number of characters for span event attributes
 SPAN_EXCEPTION_MESSAGE_MAX_STR_LENGTH = 100
 SPAN_EXCEPTION_STACK_TRACE_MAX_STR_LENGTH = 10_000
 SPAN_IO_VALUE_MAX_STR_LENGTH = 4_000_000
@@ -29,7 +25,6 @@ SPAN_LLM_MESSAGE_ROLE_MAX_STR_LENGTH = 100
 SPAN_LLM_MESSAGE_CONTENT_MAX_STR_LENGTH = 4_000_000
 SPAN_LLM_TOOL_CALL_FUNCTION_NAME_MAX_STR_LENGTH = 500
 SPAN_LLM_PROMPT_TEMPLATE_MAX_STR_LENGTH = 4_000_000
-# SPAN_LLM_PROMPT_TEMPLATE_VARIABLES_MAX_STR_LENGTH = 10_000
 SPAN_LLM_PROMPT_TEMPLATE_VERSION_MAX_STR_LENGTH = 100
 SPAN_TOOL_NAME_MAX_STR_LENGTH = 100
 SPAN_TOOL_DESCRIPTION_MAX_STR_LENGTH = 1_000
@@ -43,7 +38,7 @@ JSON_STRING_MAX_STR_LENGTH = 4_000_000
 EVAL_LABEL_MIN_STR_LENGTH = 1  # we do not accept empty strings
 EVAL_LABEL_MAX_STR_LENGTH = 100
 EVAL_EXPLANATION_MAX_STR_LENGTH = 10_000
-#
+
 # # Annotation related constants
 ANNOTATION_LABEL_MIN_STR_LENGTH = 1
 ANNOTATION_LABEL_MAX_STR_LENGTH = 100  # Max length for annotation label string

arize/datasets/client.py

@@ -24,6 +24,7 @@ from arize.utils.openinference_conversion import (
 from arize.utils.size import get_payload_size_mb
 
 if TYPE_CHECKING:
+    from arize._generated.api_client.api_client import ApiClient
     from arize.config import SDKConfiguration
 
 logger = logging.getLogger(__name__)
@@ -41,18 +42,21 @@ class DatasetsClient:
     :class:`arize.config.SDKConfiguration`.
     """
 
-    def __init__(self, *, sdk_config: SDKConfiguration) -> None:
+    def __init__(
+        self, *, sdk_config: SDKConfiguration, generated_client: ApiClient
+    ) -> None:
         """
         Args:
             sdk_config: Resolved SDK configuration.
+            generated_client: Shared generated API client instance.
         """  # noqa: D205, D212
         self._sdk_config = sdk_config
 
         # Import at runtime so it's still lazy and extras-gated by the parent
         from arize._generated import api_client as gen
 
-        # Use the shared generated client from the config
-        self._api = gen.DatasetsApi(self._sdk_config.get_generated_client())
+        # Use the provided client directly
+        self._api = gen.DatasetsApi(generated_client)
 
     @prerelease_endpoint(key="datasets.list", stage=ReleaseStage.BETA)
     def list(
@@ -117,7 +121,7 @@ class DatasetsClient:
             space_id: Space ID to create the dataset in.
             examples: Dataset examples either as:
                 - a list of JSON-like dicts, or
-                - a pandas DataFrame (will be converted to records for REST).
+                - a :class:`pandas.DataFrame` (will be converted to records for REST).
             force_http: If True, force REST upload even if the payload exceeds the
                 configured REST payload threshold.
 
@@ -125,7 +129,7 @@ class DatasetsClient:
             The created dataset object as returned by the API.
 
         Raises:
-            TypeError: If `examples` is not a list of dicts or a pandas DataFrame.
+            TypeError: If `examples` is not a list of dicts or a :class:`pandas.DataFrame`.
             RuntimeError: If the Flight upload path is selected and the Flight request
                 fails.
             arize._generated.api_client.exceptions.ApiException: If the REST API
@@ -205,7 +209,8 @@ class DatasetsClient:
         Args:
             dataset_id: Dataset ID to delete.
 
-        Returns: This method returns None on success (common empty 204 response)
+        Returns:
+            This method returns None on success (common empty 204 response).
 
         Raises:
             arize._generated.api_client.exceptions.ApiException: If the REST API
@@ -354,14 +359,13 @@ class DatasetsClient:
                 the latest dataset version is selected.
             examples: Examples to append, provided as either:
                 - a list of JSON-like dicts, or
-                - a pandas DataFrame (converted to records before upload).
+                - a :class:`pandas.DataFrame` (converted to records before upload).
 
         Returns:
             The updated dataset object. To see the examples, use `list_examples()`.
 
         Raises:
-            AssertionError: If `examples` is not a list of dicts or a pandas
-                DataFrame.
+            AssertionError: If `examples` is not a list of dicts or a :class:`pandas.DataFrame`.
             arize._generated.api_client.exceptions.ApiException: If the REST API
                 returns an error response (e.g. 400/401/403/404/429).
         """

arize/datasets/errors.py

@@ -80,7 +80,7 @@ class RequiredColumnsError(DatasetError):
 
 
 class EmptyDatasetError(DatasetError):
-    """Raised when dataset DataFrame has no rows."""
+    """Raised when dataset :class:`pandas.DataFrame` has no rows."""
 
     def error_message(self) -> str:
         """Return the error message for this exception."""

arize/datasets/validation.py

@@ -8,12 +8,12 @@ from arize.datasets import errors as err
 def validate_dataset_df(
     df: pd.DataFrame,
 ) -> list[err.DatasetError]:
-    """Validate a dataset DataFrame for structural and content errors.
+    """Validate a dataset :class:`pandas.DataFrame` for structural and content errors.
 
     Checks for required columns, unique ID values, and non-empty data.
 
     Args:
-        df: The pandas DataFrame to validate.
+        df: The :class:`pandas.DataFrame` to validate.
 
     Returns:
         A list of DatasetError objects found during validation. Empty list if valid.

arize/embeddings/auto_generator.py

@@ -62,7 +62,7 @@ class EmbeddingGenerator:
 
     @classmethod
     def list_default_models(cls) -> pd.DataFrame:
-        """Return a DataFrame of default models for each use case."""
+        """Return a :class:`pandas.DataFrame` of default models for each use case."""
         df = pd.DataFrame(
             {
                 "Area": ["NLP", "NLP", "CV", "CV", "STRUCTURED"],
@@ -87,7 +87,7 @@ class EmbeddingGenerator:
 
     @classmethod
     def list_pretrained_models(cls) -> pd.DataFrame:
-        """Return a DataFrame of all available pretrained models."""
+        """Return a :class:`pandas.DataFrame` of all available pretrained models."""
         data = {
             "Task": ["NLP" for _ in NLP_PRETRAINED_MODELS]
             + ["CV" for _ in CV_PRETRAINED_MODELS],

arize/embeddings/errors.py

@@ -2,7 +2,7 @@
 
 
 class InvalidIndexError(Exception):
-    """Raised when DataFrame or Series has an invalid index."""
+    """Raised when :class:`pandas.DataFrame` or Series has an invalid index."""
 
     def __repr__(self) -> str:
         """Return a string representation for debugging and logging."""
@@ -16,7 +16,7 @@ class InvalidIndexError(Exception):
         """Initialize the exception with field name context.
 
         Args:
-            field_name: Name of the DataFrame or Series field with invalid index.
+            field_name: Name of the :class:`pandas.DataFrame` or Series field with invalid index.
         """
         self.field_name = field_name
 

arize/embeddings/tabular_generators.py

@@ -172,5 +172,5 @@ class EmbeddingGeneratorForTabularFeatures(NLPEmbeddingGenerator):
 
     @staticmethod
     def list_pretrained_models() -> pd.DataFrame:
-        """Return a DataFrame of available pretrained tabular models."""
+        """Return a :class:`pandas.DataFrame` of available pretrained tabular models."""
         return pd.DataFrame({"Model Name": sorted(TABULAR_PRETRAINED_MODELS)})

arize/exceptions/base.py

@@ -39,21 +39,6 @@ class ValidationFailure(Exception):
         self.errors = errors
 
 
-# ----------------------
-# Minimum required checks
-# ----------------------
-# class InvalidColumnNameEmptyString(ValidationError):
-#     def __repr__(self) -> str:
-#         return "Invalid_Column_Name_Empty_String"
-#
-#     def error_message(self) -> str:
-#         return (
-#             "Empty column name found: ''. The schema cannot point to columns in the "
-#             "dataframe denoted by an empty string. You can see the columns used in the "
-#             "schema by running schema.get_used_columns()"
-#         )
-
-
 class InvalidFieldTypeConversion(ValidationError):
     """Raised when fields cannot be converted to required type."""
 
@@ -79,31 +64,6 @@ class InvalidFieldTypeConversion(ValidationError):
         )
 
 
-# class InvalidFieldTypeEmbeddingFeatures(ValidationError):
-#     def __repr__(self) -> str:
-#         return "Invalid_Input_Type_Embedding_Features"
-#
-#     def __init__(self) -> None:
-#         pass
-#
-#     def error_message(self) -> str:
-#         return (
-#             "schema.embedding_feature_column_names should be a dictionary mapping strings "
-#             "to EmbeddingColumnNames objects"
-#         )
-
-
-# class InvalidFieldTypePromptResponse(ValidationError):
-#     def __repr__(self) -> str:
-#         return "Invalid_Input_Type_Prompt_Response"
-#
-#     def __init__(self, name: str) -> None:
-#         self.name = name
-#
-#     def error_message(self) -> str:
-#         return f"'{self.name}' must be of type str or EmbeddingColumnNames"
-
-
 class InvalidDataFrameIndex(ValidationError):
     """Raised when DataFrame has an invalid index that needs to be reset."""
 
@@ -117,15 +77,3 @@ class InvalidDataFrameIndex(ValidationError):
             "The index of the dataframe is invalid; "
             "reset the index by using df.reset_index(drop=True, inplace=True)"
         )
-
-
-# class InvalidSchemaType(ValidationError):
-#     def __repr__(self) -> str:
-#         return "Invalid_Schema_Type"
-#
-#     def __init__(self, schema_type: str, environment: Environments) -> None:
-#         self.schema_type = schema_type
-#         self.environment = environment
-#
-#     def error_message(self) -> str:
-#         return f"Cannot use a {self.schema_type} for a model with environment: {self.environment}"