sibi-dst 2025.1.12__py3-none-any.whl → 2025.8.1__py3-none-any.whl

sibi_dst/utils/data_utils.py

@@ -1,5 +1,6 @@
+from __future__ import annotations
 
-from typing import Union, List
+from typing import Union, List, Dict, Any, Iterable
 
 import dask.dataframe as dd
 import pandas as pd
@@ -9,240 +10,192 @@ from .log_utils import Logger
 
 class DataUtils:
     """
-    Utility class for data transformation, manipulation, and merging.
-
-    This class provides functionalities for transforming numeric and boolean columns, merging
-    lookup data, checking DataFrame emptiness, and converting columns to datetime format in
-    Pandas or Dask DataFrames. It is designed to handle data preprocessing steps efficiently
-    for both small-scale and large-scale datasets. Logging and debug options are available
-    to trace execution and monitor operations.
-
-    :ivar logger: Logger instance for logging messages.
-    :type logger: logging.Logger
-    :ivar debug: Flag to enable or disable debug mode.
-    :type debug: bool
+    Helpers for transforming columns, safe emptiness checks, datetime coercion,
+    and joining lookup data for Pandas or Dask DataFrames.
     """
-    def __init__(self, logger=None, **kwargs):
+
+    def __init__(self, logger: Logger | None = None, **kwargs: Any) -> None:
         self.logger = logger or Logger.default_logger(logger_name=self.__class__.__name__)
-        self.debug = kwargs.get('debug', False)
+        self.debug: bool = bool(kwargs.get("debug", False))
+
+    # ---------- numeric / boolean transforms ----------
 
     @staticmethod
-    def _transform_column(series, fill_value, dtype):
-        """
-        Helper method to transform a column by converting it to numeric, filling missing values,
-        and casting to the specified dtype.
-
-        :param series: The column to transform.
-        :type series: pd.Series or dd.Series
-        :param fill_value: Value to replace missing or invalid data.
-        :type fill_value: int or float
-        :param dtype: Target data type for the column.
-        :type dtype: type
-        :return: Transformed column.
-        :rtype: pd.Series or dd.Series
-        """
-        return (
-            pd.to_numeric(series, errors="coerce")  # Convert to numeric, invalid to NaN
-            .fillna(fill_value)  # Replace NaN with fill_value
-            .astype(dtype)  # Convert to target dtype
-        )
-
-    def transform_numeric_columns(self, df: Union[pd.DataFrame, dd.DataFrame], columns: List[str], fill_value=0,
-                                  dtype=int):
+    def _transform_column_pandas(series: pd.Series, fill_value: Any, dtype: type) -> pd.Series:
+        return pd.to_numeric(series, errors="coerce").fillna(fill_value).astype(dtype)
+
+    def transform_numeric_columns(
+        self,
+        df: Union[pd.DataFrame, dd.DataFrame],
+        columns: List[str],
+        fill_value: Any = 0,
+        dtype: type = int,
+    ) -> Union[pd.DataFrame, dd.DataFrame]:
         """
-        Transform specified numeric columns in the DataFrame by converting their data types
-        to the specified dtype and replacing missing values with the given fill_value.
-
-        :param df: DataFrame to be transformed.
-        :type df: pd.DataFrame or dd.DataFrame
-        :param columns: List of column names to transform.
-        :type columns: list[str]
-        :param fill_value: Value to replace missing or invalid data. Default is 0.
-        :type fill_value: int or float
-        :param dtype: Target data type for the columns. Default is int.
-        :type dtype: type
-        :return: Transformed DataFrame.
-        :rtype: pd.DataFrame or dd.DataFrame
+        Convert selected columns to numeric → fillna → cast dtype.
+        Works for Pandas and Dask (partition-wise).
         """
         if not columns:
-            self.logger.warning("No columns specified.")
+            self.logger.warning("No columns specified for transform_numeric_columns.")
             return df
 
-        self.logger.debug(f"DataFrame type: {type(df)}")
-        columns = [col for col in columns if col in df.columns]
-
-        for col in columns:
-            df[col] = df[col].map_partitions(
-                self._transform_column, fill_value, dtype, meta=(col, dtype)
-            )
+        cols = [c for c in columns if c in df.columns]
+        if not cols:
+            self.logger.warning("None of the requested columns exist in the DataFrame.")
+            return df
 
-        return df
+        if isinstance(df, pd.DataFrame):
+            for col in cols:
+                df[col] = self._transform_column_pandas(df[col], fill_value, dtype)
+            return df
 
-    def transform_numeric_cols(self, df, columns, fill_value=0, dtype=int):
-        """
-        This function transforms the specified numeric columns in the given dataframe by converting
-        their data types to the specified dtype, with an optional parameter for replacing missing
-        values. It first checks if the provided columns exist in the dataframe, processes each column
-        to replace non-numeric values with NaN, fills NaN values with the given fill_value, and finally
-        converts the column to the specified dtype.
-
-        :param df: DataFrame to be transformed.
-        :type df: dask.dataframe.DataFrame
-        :param columns: List of column names to be transformed.
-        :type columns: list[str]
-        :param fill_value: Value used to replace missing or invalid data. Default is 0.
-        :type fill_value: int or float
-        :param dtype: Target data type for the columns after transformation. Default is int.
-        :type dtype: type
-        :return: Transformed dataframe with the specified numeric columns converted and modified.
-        :rtype: dask.dataframe.DataFrame
-        """
-        if not columns:
-            self.logger.warning('No columns specified')
-        self.logger.debug(f'Dataframe type:{type(df)}')
-        columns = [column for column in columns if column in df.columns]
-        for col in columns:
-            # Replace NaN with 0, then convert to boolean
+        # Dask path
+        for col in cols:
             df[col] = df[col].map_partitions(
-                lambda s: pd.to_numeric(s, errors='coerce')  # Convert to numeric, invalid to NaN
-                .fillna(fill_value)  # Replace NaN with 0
-                .astype(dtype),
-                meta=(col, dtype)
+                self._transform_column_pandas,
+                fill_value,
+                dtype,
+                meta=(col, dtype),
             )
-
         return df
 
-    def transform_boolean_columns(self, df: Union[pd.DataFrame, dd.DataFrame], columns: List[str], fill_value=0):
-        """
-        Convert specified columns in the DataFrame to boolean, replacing missing values with
-        the given fill_value.
-
-        :param df: DataFrame to be transformed.
-        :type df: pd.DataFrame or dd.DataFrame
-        :param columns: List of column names to transform.
-        :type columns: list[str]
-        :param fill_value: Value to replace missing or invalid data. Default is 0.
-        :type fill_value: int or float
-        :return: Transformed DataFrame.
-        :rtype: pd.DataFrame or dd.DataFrame
-        """
+    def transform_boolean_columns(
+        self,
+        df: Union[pd.DataFrame, dd.DataFrame],
+        columns: List[str],
+        fill_value: Any = 0,
+    ) -> Union[pd.DataFrame, dd.DataFrame]:
+        """Convenience wrapper: cast to boolean via numeric→fillna→astype(bool)."""
         return self.transform_numeric_columns(df, columns, fill_value=fill_value, dtype=bool)
 
-    def merge_lookup_data(self, classname, df, **kwargs):
-        """
-        Merge lookup data into the DataFrame based on specified columns.
-
-        Parameters:
-        - classname: The class instance to use for loading lookup data.
-        - df (pandas.DataFrame or dask.dataframe.DataFrame): The DataFrame.
-        - kwargs: Additional keyword arguments for configuration.
+    # ---------- lookup merge ----------
 
-        Returns:
-        - pandas.DataFrame or dask.dataframe.DataFrame: Updated DataFrame with merged lookup data.
+    def merge_lookup_data(
+        self,
+        classname,
+        df: Union[pd.DataFrame, dd.DataFrame],
+        **kwargs: Any,
+    ) -> Union[pd.DataFrame, dd.DataFrame]:
         """
-        # Return early if the DataFrame is empty
+        Merge lookup data for ids present in `source_col`.
+
+        Required kwargs:
+            - source_col
+            - lookup_col
+            - lookup_description_col
+            - source_description_alias
+
+        Optional kwargs:
+            - fillna_source_description_alias: bool = False
+            - fieldnames: tuple[str, str] = (lookup_col, lookup_description_col)
+            - column_names: list[str] = ['temp_join_col', source_description_alias]
+            - any other filters passed to `classname.load(...)`
+        """
+        # Early outs for emptiness and required args
         if self.is_dataframe_empty(df):
-            self.logger.debug("merge_lookup_data was given an empty dataFrame")
+            self.logger.debug("merge_lookup_data: input DataFrame empty — nothing to merge.")
             return df
 
-        # Extract and validate required parameters
-        required_params = ['source_col', 'lookup_col', 'lookup_description_col', 'source_description_alias']
-        missing_params = [param for param in required_params if param not in kwargs]
-        if missing_params:
-            raise ValueError(f"Missing required parameters: {', '.join(missing_params)}")
+        required = ["source_col", "lookup_col", "lookup_description_col", "source_description_alias"]
+        missing = [k for k in required if k not in kwargs]
+        if missing:
+            raise ValueError(f"Missing required parameters: {', '.join(missing)}")
 
-        source_col = kwargs.pop('source_col')
-        lookup_col = kwargs.pop('lookup_col')
-        lookup_description_col = kwargs.pop('lookup_description_col')
-        source_description_alias = kwargs.pop('source_description_alias')
+        source_col = kwargs.pop("source_col")
+        lookup_col = kwargs.pop("lookup_col")
+        lookup_description_col = kwargs.pop("lookup_description_col")
+        source_description_alias = kwargs.pop("source_description_alias")
 
-        # Optional parameters with default values
-        fillna_source_description_alias = kwargs.pop('fillna_source_description_alias', False)
-        fieldnames = kwargs.pop('fieldnames', (lookup_col, lookup_description_col))
-        column_names = kwargs.pop('column_names', ['temp_join_col', source_description_alias])
+        fillna_alias = bool(kwargs.pop("fillna_source_description_alias", False))
+        fieldnames = kwargs.pop("fieldnames", (lookup_col, lookup_description_col))
+        column_names = kwargs.pop("column_names", ["temp_join_col", source_description_alias])
 
         if source_col not in df.columns:
-            self.logger.debug(f"{source_col} not in DataFrame columns")
+            self.logger.debug(f"merge_lookup_data: '{source_col}' not found in frame — skipping merge.")
             return df
 
-        # Get unique IDs from source column
-        ids = df[source_col].dropna().unique()
-        # Compute if it's a Dask Series
-        if isinstance(ids, dd.Series):
-            ids = ids.compute()
+        # Collect ids safely
+        try:
+            ids_series = df[source_col].dropna()
+            if isinstance(df, dd.DataFrame):
+                # Dask: unique() is lazy → compute smallish result
+                ids = ids_series.unique().compute()
+            else:
+                ids = ids_series.unique()
+            ids = sorted(ids.tolist() if not isinstance(ids, list) else ids)
+        except Exception as e:
+            self.logger.error(f"merge_lookup_data: failed extracting ids from '{source_col}': {e}")
+            return df
 
-        # Check if any IDs are found
-        if not len(ids):
-            self.logger.debug(f"No IDs found in the source column: {source_col}")
+        if not ids:
+            self.logger.debug(f"merge_lookup_data: no ids found in '{source_col}'.")
             return df
 
-        # Convert to a list only if necessary and sort
-        if not isinstance(ids, list):
-            ids = ids.tolist()
-        ids = sorted(ids)
-        # Prepare kwargs for loading lookup data
-        load_kwargs = kwargs.copy()
-        load_kwargs.update({
-            'fieldnames': fieldnames,
-            'column_names': column_names,
-            f'{lookup_col}__in': ids
-        })
-        # Load lookup data
+        # Load lookup data (expected to be small after filtering)
+        load_kwargs = {
+            **kwargs,
+            "fieldnames": fieldnames,
+            "column_names": column_names,
+            f"{lookup_col}__in": ids,
+        }
+
         lookup_instance = classname(debug=self.debug, logger=self.logger)
         result = lookup_instance.load(**load_kwargs)
-        if len(result.index) == 0:
-            self.logger.debug(f"No IDs found in the source column: {source_col}")
-            return df
-        # Determine the join column on the result DataFrame
-        temp_join_col = 'temp_join_col' if 'temp_join_col' in column_names else lookup_col
 
-        # Merge DataFrames
-        df = df.merge(result, how='left', left_on=source_col, right_on=temp_join_col)
+        # If lookup returns Dask, compute to pandas (broadcastable) or keep small Dask?
+        if isinstance(result, dd.DataFrame):
+            # we expect this to be small after filtering by ids; materialize
+            result = result.compute()
 
-        if fillna_source_description_alias and source_description_alias in df.columns:
-            df[source_description_alias] = df[source_description_alias].fillna('')
+        if not isinstance(result, pd.DataFrame):
+            raise TypeError("merge_lookup_data: lookup 'load' must return a pandas or dask DataFrame.")
 
-        # Drop temp_join_col if present
-        df = df.drop(columns='temp_join_col', errors='ignore')
+        if result.empty:
+            self.logger.debug("merge_lookup_data: lookup returned 0 rows — nothing to merge.")
+            return df
 
-        return df
+        # Determine join key in the lookup result
+        temp_join_col = "temp_join_col" if "temp_join_col" in column_names else lookup_col
 
-    def is_dataframe_empty(self, df):
-        """
-        Check if a DataFrame (Pandas or Dask) is empty.
+        # Perform merge (Dask can merge with a small pandas right side)
+        merged = df.merge(result, how="left", left_on=source_col, right_on=temp_join_col)
+
+        if fillna_alias and source_description_alias in merged.columns:
+            if isinstance(merged, dd.DataFrame):
+                merged[source_description_alias] = merged[source_description_alias].fillna("")
+            else:
+                merged[source_description_alias] = merged[source_description_alias].fillna("")
 
-        Parameters:
-        - df (pandas.DataFrame or dask.dataframe.DataFrame): The DataFrame.
+        # Drop helper join column if present
+        merged = merged.drop(columns="temp_join_col", errors="ignore")
+        return merged
 
-        Returns:
-        - bool: True if the DataFrame is empty, False otherwise.
+    # ---------- emptiness & datetime ----------
+
+    def is_dataframe_empty(self, df: Union[pd.DataFrame, dd.DataFrame]) -> bool:
+        """
+        Safe emptiness check. For Dask, uses head(1) to avoid full compute.
         """
         if isinstance(df, dd.DataFrame):
             try:
-                return len(df.index) == 0
+                head = df.head(1, npartitions=-1, compute=True)
+                return head.empty
             except Exception as e:
-                self.logger.error(f"Error while processing Dask DataFrame: {e}")
+                self.logger.error(f"is_dataframe_empty: Dask head() failed: {e}")
                 return False
-        elif isinstance(df, pd.DataFrame):
+        if isinstance(df, pd.DataFrame):
             return df.empty
-        else:
-            self.logger.error("Input must be a pandas or dask DataFrame.")
-            return False
+        self.logger.error("is_dataframe_empty: input must be a pandas or dask DataFrame.")
+        return False
 
     @staticmethod
-    def convert_to_datetime_dask(df, date_fields):
+    def convert_to_datetime_dask(df: dd.DataFrame, date_fields: Iterable[str]) -> dd.DataFrame:
         """
-        Convert specified columns in a Dask DataFrame to datetime, handling errors gracefully.
-
-        Parameters:
-        - df (dask.dataframe.DataFrame): The Dask DataFrame containing the columns.
-        - date_fields (list of str): List of column names to convert to datetime.
-
-        Returns:
-        - dask.dataframe.DataFrame: Updated DataFrame with specified columns converted to datetime.
+        Partition-wise datetime coercion with errors='coerce'.
         """
         for col in date_fields:
             if col in df.columns:
-                df[col] = df[col].map_partitions(pd.to_datetime, errors="coerce", meta=(col, "datetime64[ns]"))
-        return df
-
+                df[col] = df[col].map_partitions(
+                    pd.to_datetime, errors="coerce", meta=(col, "datetime64[ns]")
+                )
+        return df