arize 8.0.0b1__py3-none-any.whl → 8.0.0b4__py3-none-any.whl

arize/spans/validation/common/value_validation.py

@@ -11,6 +11,7 @@ from arize.constants.ml import (
     MAX_FUTURE_YEARS_FROM_CURRENT_TIME,
     MAX_PAST_YEARS_FROM_CURRENT_TIME,
 )
+from arize.exceptions.base import ValidationError
 from arize.exceptions.parameters import InvalidModelVersion, InvalidProjectName
 from arize.spans.columns import (
     SPAN_END_TIME_COL,
@@ -56,7 +57,7 @@ def check_invalid_model_version(
         model_version: The optional model version to validate.
 
     Returns:
-        List of validation errors if model version is invalid (empty if valid or None).
+        List of validation errors if model version is invalid (empty if valid or :obj:`None`).
     """
     if model_version is None:
         return []
@@ -73,7 +74,7 @@ def check_string_column_value_length(
     max_len: int,
     is_required: bool,
     must_be_json: bool = False,
-) -> list[InvalidMissingValueInColumn | InvalidStringLengthInColumn]:
+) -> list[ValidationError]:
     """Validate string column values are within length bounds and optionally valid JSON.
 
     Args:
@@ -90,7 +91,7 @@ def check_string_column_value_length(
     if col_name not in df.columns:
         return []
 
-    errors = []
+    errors: list[ValidationError] = []
     if is_required and df[col_name].isnull().any():
         errors.append(
             InvalidMissingValueInColumn(
@@ -129,7 +130,7 @@ def check_string_column_allowed_values(
     col_name: str,
     allowed_values: list[str],
     is_required: bool,
-) -> list[InvalidMissingValueInColumn | InvalidStringValueNotAllowedInColumn]:
+) -> list[ValidationError]:
     """Validate that string column values are within allowed values.
 
     Args:
@@ -144,7 +145,7 @@ def check_string_column_allowed_values(
     if col_name not in df.columns:
         return []
 
-    errors = []
+    errors: list[ValidationError] = []
     if is_required and df[col_name].isnull().any():
         errors.append(
             InvalidMissingValueInColumn(
@@ -177,7 +178,7 @@ def check_string_column_allowed_values(
 def check_float_column_valid_numbers(
     df: pd.DataFrame,
     col_name: str,
-) -> list[InvalidFloatValueInColumn]:
+) -> list[ValidationError]:
     """Check that float column contains only finite numbers, no infinity values.
 
     Args:
@@ -201,11 +202,7 @@ def check_float_column_valid_numbers(
 
 def check_value_columns_start_end_time(
     df: pd.DataFrame,
-) -> list[
-    InvalidMissingValueInColumn
-    | InvalidTimestampValueInColumn
-    | InvalidStartAndEndTimeValuesInColumn
-]:
+) -> list[ValidationError]:
     """Validate start and end time columns for timestamps and logical ordering.
 
     Args:
@@ -214,7 +211,7 @@ def check_value_columns_start_end_time(
     Returns:
         List of validation errors for missing values, invalid timestamps, or start > end.
     """
-    errors = []
+    errors: list[ValidationError] = []
     errors += check_value_timestamp(
         df=df,
         col_name=SPAN_START_TIME_COL.name,
@@ -243,7 +240,7 @@ def check_value_timestamp(
     df: pd.DataFrame,
     col_name: str,
     is_required: bool,
-) -> list[InvalidMissingValueInColumn | InvalidTimestampValueInColumn]:
+) -> list[ValidationError]:
     """Validate timestamp column values are within reasonable bounds.
 
     Args:
@@ -258,7 +255,7 @@ def check_value_timestamp(
     if col_name not in df.columns:
         return []
 
-    errors = []
+    errors: list[ValidationError] = []
     if is_required and df[col_name].isnull().any():
         errors.append(
             InvalidMissingValueInColumn(

arize/spans/validation/evals/dataframe_form_validation.py

@@ -27,10 +27,10 @@ def log_info_dataframe_extra_column_names(
     """Logs informational message about columns that don't follow evaluation naming conventions.
 
     Args:
-        df: DataFrame to check for extra column names, or None.
+        df: DataFrame to check for extra column names, or :obj:`None`.
 
     Returns:
-        None.
+        :obj:`None`.
     """
     if df is None:
         return
@@ -57,13 +57,13 @@ def log_info_dataframe_extra_column_names(
 def check_dataframe_column_content_type(
     df: pd.DataFrame,
 ) -> list[InvalidDataFrameColumnContentTypes]:
-    """Validates that evaluation DataFrame columns contain expected data types.
+    """Validates that evaluation :class:`pandas.DataFrame` columns contain expected data types.
 
     Checks that label columns contain strings, score columns contain numbers,
     and explanation columns contain strings.
 
     Args:
-        df: The DataFrame to validate.
+        df: The :class:`pandas.DataFrame` to validate.
 
     Returns:
         List of validation errors for columns with incorrect types.

arize/spans/validation/evals/evals_validation.py

@@ -55,13 +55,13 @@ def validate_argument_types(
 def validate_dataframe_form(
     evals_dataframe: pd.DataFrame,
 ) -> list[ValidationError]:
-    """Validate the structure and form of an evaluations DataFrame.
+    """Validate the structure and form of an evaluations :class:`pandas.DataFrame`.
 
     Args:
-        evals_dataframe: The DataFrame containing evaluation data to validate.
+        evals_dataframe: The :class:`pandas.DataFrame` containing evaluation data to validate.
 
     Returns:
-        List of validation errors found in the DataFrame structure.
+        List of validation errors found in the :class:`pandas.DataFrame` structure.
     """
     df_validation.log_info_dataframe_extra_column_names(evals_dataframe)
     checks = chain(
@@ -84,15 +84,15 @@ def validate_values(
     project_name: str,
     model_version: str | None = None,
 ) -> list[ValidationError]:
-    """Validate the values within an evaluations DataFrame.
+    """Validate the values within an evaluations :class:`pandas.DataFrame`.
 
     Args:
-        evals_dataframe: The DataFrame containing evaluation data to validate.
+        evals_dataframe: The :class:`pandas.DataFrame` containing evaluation data to validate.
         project_name: The project name associated with the evaluations.
         model_version: Optional model version. Defaults to None.
 
     Returns:
-        List of validation errors found in DataFrame values.
+        List of validation errors found in :class:`pandas.DataFrame` values.
     """
     checks = chain(
         # Common

arize/spans/validation/evals/value_validation.py

@@ -40,7 +40,7 @@ def check_eval_cols(
     Returns:
         List of validation errors found in evaluation columns.
     """
-    checks = []
+    checks: list[list[ValidationError]] = []
     for col in dataframe.columns:
         if col.endswith(EVAL_LABEL_SUFFIX):
             checks.append(

arize/spans/validation/metadata/argument_validation.py

@@ -39,7 +39,7 @@ def validate_argument_types(
     Returns:
         A list of validation errors, empty if none found
     """
-    errors = []
+    errors: list[ValidationError] = []
 
     # Check metadata_dataframe type
     if not isinstance(metadata_dataframe, pd.DataFrame):

arize/spans/validation/metadata/dataframe_form_validation.py

@@ -7,7 +7,7 @@ from arize.spans.columns import SPAN_SPAN_ID_COL
 
 
 class MetadataFormError(ValidationError):
-    """Raised when metadata DataFrame structure or format is invalid."""
+    """Raised when metadata :class:`pandas.DataFrame` structure or format is invalid."""
 
     def __init__(self, message: str, resolution: str) -> None:
         """Initialize the exception with metadata form error context.
@@ -41,7 +41,7 @@ def validate_dataframe_form(
     Returns:
         A list of validation errors, empty if none found
     """
-    errors = []
+    errors: list[ValidationError] = []
 
     # Check for empty dataframe
     if metadata_dataframe.empty:

arize/spans/validation/metadata/value_validation.py

@@ -34,6 +34,28 @@ class MetadataValueError(ValidationError):
         return f"{self.message} {self.resolution}"
 
 
+class InvalidPatchDocumentFormat(ValidationError):
+    """Raised when patch document format is invalid or cannot be parsed."""
+
+    def __init__(self, row_idx: int, message: str) -> None:
+        """Initialize the exception with patch document format error context.
+
+        Args:
+            row_idx: The row index where the invalid patch was found.
+            message: Detailed error message describing the format issue.
+        """
+        self.row_idx = row_idx
+        self.message = message
+
+    def __repr__(self) -> str:
+        """Return a string representation for debugging and logging."""
+        return "Invalid_Patch_Document_Format"
+
+    def error_message(self) -> str:
+        """Return the error message for this exception."""
+        return f"Row {self.row_idx}: {self.message}"
+
+
 def calculate_json_depth(obj: object, current_depth: int = 1) -> int:
     """Calculate the maximum nesting depth of a JSON object.
 
@@ -67,7 +89,7 @@ def validate_values(
     Returns:
         A list of validation errors, empty if none found
     """
-    errors = []
+    errors: list[ValidationError] = []
 
     # Skip validation if span_id column is not present
     if SPAN_SPAN_ID_COL.name not in metadata_dataframe.columns:

arize/spans/validation/spans/dataframe_form_validation.py

@@ -50,13 +50,13 @@ def log_info_dataframe_extra_column_names(
 def check_dataframe_column_content_type(
     df: pd.DataFrame,
 ) -> list[InvalidDataFrameColumnContentTypes]:
-    """Validates that span DataFrame columns contain data types matching Open Inference Specification.
+    """Validates span :class:`pandas.DataFrame` columns match OpenInference types.
 
     Checks that columns have appropriate data types: lists of dicts, dicts, numeric,
     boolean, timestamp, JSON strings, or plain strings based on column specifications.
 
     Args:
-        df: The DataFrame to validate.
+        df: The :class:`pandas.DataFrame` to validate.
 
     Returns:
         List of validation errors for columns with incorrect types.

arize/spans/validation/spans/spans_validation.py

@@ -56,13 +56,13 @@ def validate_argument_types(
 def validate_dataframe_form(
     spans_dataframe: pd.DataFrame,
 ) -> list[ValidationError]:
-    """Validate the structure and form of a spans DataFrame.
+    """Validate the structure and form of a spans :class:`pandas.DataFrame`.
 
     Args:
-        spans_dataframe: The DataFrame containing spans data to validate.
+        spans_dataframe: The :class:`pandas.DataFrame` containing spans data to validate.
 
     Returns:
-        List of validation errors found in the DataFrame structure.
+        List of validation errors found in the :class:`pandas.DataFrame` structure.
     """
     df_validation.log_info_dataframe_extra_column_names(spans_dataframe)
     checks = chain(
@@ -88,15 +88,15 @@ def validate_values(
     project_name: str,
     model_version: str | None = None,
 ) -> list[ValidationError]:
-    """Validate the values within a spans DataFrame.
+    """Validate the values within a spans :class:`pandas.DataFrame`.
 
     Args:
-        spans_dataframe: The DataFrame containing spans data to validate.
+        spans_dataframe: The :class:`pandas.DataFrame` containing spans data to validate.
         project_name: The project name associated with the spans.
         model_version: Optional model version. Defaults to None.
 
     Returns:
-        List of validation errors found in DataFrame values.
+        List of validation errors found in :class:`pandas.DataFrame` values.
     """
     checks = chain(
         # Common

arize/utils/arrow.py

@@ -1,6 +1,5 @@
 """Apache Arrow utilities for data serialization and file operations."""
 
-# type: ignore[pb2]
 from __future__ import annotations
 
 import base64
@@ -38,7 +37,7 @@ def post_arrow_table(
         pa_table: The PyArrow table containing the data.
         proto_schema: The protobuf schema for the data.
         headers: HTTP headers for the request.
-        timeout: Request timeout in seconds, or None for no timeout.
+        timeout: Request timeout in seconds, or :obj:`None` for no timeout.
         verify: Whether to verify SSL certificates.
         max_chunksize: Maximum chunk size for splitting large tables.
         tmp_dir: Temporary directory for serialization. Defaults to "".
@@ -124,6 +123,18 @@ def post_arrow_table(
 def _append_to_pyarrow_metadata(
     pa_schema: pa.Schema, new_metadata: dict[str, Any]
 ) -> object:
+    """Append metadata to a PyArrow schema without overwriting existing keys.
+
+    Args:
+        pa_schema: The PyArrow schema to add metadata to.
+        new_metadata: Dictionary of metadata key-value pairs to append.
+
+    Returns:
+        pa.Schema: A new PyArrow schema with the merged metadata.
+
+    Raises:
+        KeyError: If any keys in new_metadata conflict with existing schema metadata.
+    """
     # Ensure metadata is handled correctly, even if initially None.
     metadata = pa_schema.metadata
     if metadata is None:
@@ -145,6 +156,14 @@ def _append_to_pyarrow_metadata(
 def _write_arrow_file(
     path: str, pa_table: pa.Table, pa_schema: pa.Schema, max_chunksize: int
 ) -> None:
+    """Write a PyArrow table to an Arrow IPC file with specified schema and chunk size.
+
+    Args:
+        path: The file path where the Arrow file will be written.
+        pa_table: The PyArrow table containing the data to write.
+        pa_schema: The PyArrow schema to use for the file.
+        max_chunksize: Maximum number of rows per record batch chunk.
+    """
     with (
         pa.OSFile(path, mode="wb") as sink,
         pa.ipc.RecordBatchStreamWriter(sink, pa_schema) as writer,
@@ -153,6 +172,15 @@ def _write_arrow_file(
 
 
 def _maybe_log_project_url(response: requests.Response) -> None:
+    """Attempt to extract and log the Arize project URL from an HTTP response.
+
+    Args:
+        response: The HTTP response object from an Arize API request.
+
+    Notes:
+        Logs success message with URL if extraction succeeds, or warning if it fails.
+        This function never raises exceptions.
+    """
     try:
         url = get_arize_project_url(response)
         if url:
@@ -176,6 +204,14 @@ def _mktemp_in(directory: str) -> str:
 
 
 def _filesize(path: str) -> int:
+    """Get the size of a file in bytes.
+
+    Args:
+        path: The file path to check.
+
+    Returns:
+        int: The file size in bytes, or -1 if the file cannot be accessed.
+    """
     try:
         return os.path.getsize(path)
     except Exception:

arize/utils/cache.py

@@ -31,7 +31,7 @@ def load_cached_resource(
         format: File format for cached data. Defaults to "parquet".
 
     Returns:
-        The cached DataFrame if found and valid, None otherwise.
+        The cached :class:`pandas.DataFrame` if found and valid, :obj:`None` otherwise.
     """
     key = _get_cache_key(resource, resource_id, resource_updated_at)
     filepath = _get_abs_file_path(cache_dir, f"{key}.{format}", resource)
@@ -59,7 +59,7 @@ def cache_resource(
         resource: Resource type name (e.g., "dataset", "experiment").
         resource_id: Unique identifier for the resource.
         resource_updated_at: Optional timestamp of last resource update.
-        resource_data: DataFrame containing the resource data.
+        resource_data: :class:`pandas.DataFrame` containing the resource data.
         format: File format for cached data. Defaults to "parquet".
     """
     key = _get_cache_key(resource, resource_id, resource_updated_at)

arize/utils/dataframe.py

@@ -9,10 +9,10 @@ from arize.ml.types import BaseSchema
 
 # Resets the dataframe index if it is not a RangeIndex
 def reset_dataframe_index(dataframe: pd.DataFrame) -> None:
-    """Reset the DataFrame index in-place if it is not a RangeIndex.
+    """Reset the :class:`pandas.DataFrame` index in-place if it is not a RangeIndex.
 
     Args:
-        dataframe: The pandas DataFrame to reset.
+        dataframe: The :class:`pandas.DataFrame` to reset.
     """
     if not isinstance(dataframe.index, pd.RangeIndex):
         drop = dataframe.index.name in dataframe.columns
@@ -25,10 +25,10 @@ def remove_extraneous_columns(
     column_list: list[str] | None = None,
     regex: str | None = None,
 ) -> pd.DataFrame:
-    """Filter DataFrame to keep only relevant columns based on schema, list, or regex.
+    """Filter :class:`pandas.DataFrame` to keep only relevant columns based on schema, list, or regex.
 
     Args:
-        df: The pandas DataFrame to filter.
+        df: The :class:`pandas.DataFrame` to filter.
         schema: Optional schema defining used columns. Defaults to None.
         column_list: Optional explicit list of columns to keep. Defaults to None.
         regex: Optional regex pattern to match column names. Defaults to None.

arize/utils/online_tasks/dataframe_preprocessor.py

@@ -10,7 +10,7 @@ logger = logging.getLogger(__name__)
 
 
 class ColumnNotFoundError(Exception):
-    """Raised when a specified column is not found in the DataFrame."""
+    """Raised when a specified column is not found in the :class:`pandas.DataFrame`."""
 
     def __init__(self, attribute: str) -> None:
         """Initialize with the attribute that couldn't be mapped to a column.
@@ -27,13 +27,13 @@ class ColumnNotFoundError(Exception):
 def extract_nested_data_to_column(
     attributes: list[str], df: pd.DataFrame
 ) -> pd.DataFrame:
-    """Extract nested attributes from complex data structures into new DataFrame columns.
+    """Extract nested attributes from complex data structures into new :class:`pandas.DataFrame` columns.
 
     This function, used in Online Tasks, is typically run on data exported from Arize.
-    It prepares the DataFrame by extracting relevant attributes from complex, deeply
+    It prepares the :class:`pandas.DataFrame` by extracting relevant attributes from complex, deeply
     nested data structures, such as those found in LLM outputs or JSON-like records.
     It helps extract specific values from these nested structures by identifying the
-    longest matching column name in the DataFrame and recursively accessing the desired
+    longest matching column name in the :class:`pandas.DataFrame` and recursively accessing the desired
     attribute path within each row. This preprocessing step ensures that the extracted
     values are available as new columns, allowing evaluators to process and assess
     these values effectively.
@@ -81,9 +81,12 @@ def extract_nested_data_to_column(
         remainder = ".".join(parts[prefix_len:])
 
         # 3) Apply introspect row-by-row
+        # Type narrowing: prefix_col is guaranteed to be str after the None check above
+        prefix_col_str: str = prefix_col
+
         def apply_introspect_arize_attribute(
-            row: pd.Series,
-            prefix_col: str = prefix_col,
+            row: pd.Series,  # type: ignore[type-arg]
+            prefix_col: str = prefix_col_str,
             remainder: str = remainder,
         ) -> object:
             val = row[prefix_col]
@@ -94,8 +97,9 @@ def extract_nested_data_to_column(
             else:
                 return result if result is not None else np.nan
 
-        result_df[attribute] = result_df.apply(
-            apply_introspect_arize_attribute, axis=1
+        result_df[attribute] = result_df.apply(  # type: ignore[call-overload]
+            apply_introspect_arize_attribute,
+            axis=1,
         )
 
         new_cols.append(attribute)
@@ -127,7 +131,7 @@ def _introspect_arize_attribute(value: object, attribute: str) -> object:
         attribute: "0.message.content"
         Returns: 'The capital of China is Beijing.'
 
-      - Returns None immediately when a key or index is not found
+      - Returns :obj:`None` immediately when a key or index is not found
       - Handles integer parts for lists
       - Parses JSON strings
       - Converts NumPy arrays to lists
@@ -174,10 +178,10 @@ def _parse_value(
     2) Else if `current_value` is a dict, check if `attribute_parts_unprocessed[0]` is a key.
        If not found, try combining `attribute_parts_unprocessed[0] + '.' + attribute_parts_unprocessed[1]`...
        to handle dotted keys in the dict.
-    3) If none match, return (None, 1) to signal "not found, consume 1 part."
+    3) If none match, return (:obj:`None`, 1) to signal "not found, consume 1 part."
 
     Returns (parsed_value, num_parts_processed):
-      - parsed_value: the found value or None if not found
+      - parsed_value: the found value or :obj:`None` if not found
       - num_parts_processed: how many parts were processed (1 or more)
     """
     if not attribute_parts_unprocessed:

arize/utils/openinference_conversion.py

@@ -11,13 +11,13 @@ logger = logging.getLogger(__name__)
 
 
 def convert_datetime_columns_to_int(df: pd.DataFrame) -> pd.DataFrame:
-    """Convert datetime columns in a DataFrame to milliseconds since epoch.
+    """Convert datetime columns in a :class:`pandas.DataFrame` to milliseconds since epoch.
 
     Args:
-        df: The pandas DataFrame to convert.
+        df: The :class:`pandas.DataFrame` to convert.
 
     Returns:
-        The DataFrame with datetime columns converted to integers.
+        The :class:`pandas.DataFrame` with datetime columns converted to integers.
     """
     for col in df.select_dtypes(
         include=["datetime64[ns]", "datetime64[ns, UTC]"]
@@ -27,13 +27,13 @@ def convert_datetime_columns_to_int(df: pd.DataFrame) -> pd.DataFrame:
 
 
 def convert_boolean_columns_to_str(df: pd.DataFrame) -> pd.DataFrame:
-    """Convert boolean columns in a DataFrame to string type.
+    """Convert boolean columns in a :class:`pandas.DataFrame` to string type.
 
     Args:
-        df: The pandas DataFrame to convert.
+        df: The :class:`pandas.DataFrame` to convert.
 
     Returns:
-        The DataFrame with boolean columns converted to strings.
+        The :class:`pandas.DataFrame` with boolean columns converted to strings.
     """
     for col in df.columns:
         if df[col].dtype == "bool":
@@ -45,10 +45,10 @@ def convert_default_columns_to_json_str(df: pd.DataFrame) -> pd.DataFrame:
     """Convert dictionary values in specific columns to JSON strings.
 
     Args:
-        df: The pandas DataFrame to convert.
+        df: The :class:`pandas.DataFrame` to convert.
 
     Returns:
-        The DataFrame with dictionaries in eligible columns converted to JSON strings.
+        The :class:`pandas.DataFrame` with dictionaries in eligible columns converted to JSON strings.
     """
     for col in df.columns:
         if _should_convert_json(col):
@@ -68,10 +68,10 @@ def convert_json_str_to_dict(df: pd.DataFrame) -> pd.DataFrame:
     """Convert JSON string values in specific columns to Python dictionaries.
 
     Args:
-        df: The pandas DataFrame to convert.
+        df: The :class:`pandas.DataFrame` to convert.
 
     Returns:
-        The DataFrame with JSON strings in eligible columns converted to dictionaries.
+        The :class:`pandas.DataFrame` with JSON strings in eligible columns converted to dictionaries.
     """
     for col in df.columns:
         if _should_convert_json(col):

arize/utils/proto.py

@@ -1,6 +1,5 @@
 """Protocol buffer schema utilities for tracing data."""
 
-# type: ignore[pb2]
 from arize._generated.protocol.rec import public_pb2 as pb2
 
 

arize/utils/types.py

@@ -43,7 +43,7 @@ def is_array_of(arr: Sequence[object], tp: T) -> bool:
     return isinstance(arr, np.ndarray) and all(isinstance(x, tp) for x in arr)
 
 
-def is_list_of(lst: Sequence[object], tp: T) -> bool:
+def is_list_of(lst: object, tp: T) -> bool:
     """Check if a value is a list with all elements of a specific type.
 
     Args:
@@ -70,10 +70,10 @@ def is_iterable_of(lst: Sequence[object], tp: T) -> bool:
 
 
 def is_dict_of(
-    d: dict[object, object],
-    key_allowed_types: T,
-    value_allowed_types: T = (),
-    value_list_allowed_types: T = (),
+    d: object,
+    key_allowed_types: type | tuple[type, ...],
+    value_allowed_types: type | tuple[type, ...] = (),
+    value_list_allowed_types: type | tuple[type, ...] = (),
 ) -> bool:
     """Method to check types are valid for dictionary.
 
@@ -98,7 +98,7 @@ def is_dict_of(
         and all(isinstance(k, key_allowed_types) for k in d)
         and all(
             isinstance(v, value_allowed_types)
-            or any(is_list_of(v, t) for t in value_list_allowed_types)
+            or any(is_list_of(v, t) for t in value_list_allowed_types)  # type: ignore[union-attr]
             for v in d.values()
             if value_allowed_types or value_list_allowed_types
         )

arize/version.py

@@ -1,3 +1,3 @@
 """Version information for the Arize SDK."""
 
-__version__ = "8.0.0b1"
+__version__ = "8.0.0b4"