Replace

Transformation to replace a particular value in a column with another one

koheesio.spark.transformations.replace.Replace

Replace a particular value in a column with another one

Can handle empty strings ("") as well as NULL / None values.

Unsupported datatypes:

Following casts are not supported

will raise an error in Spark:

• binary
• boolean
• array<...>
• map<...,...>

Supported datatypes:

Following casts are supported:

• byte
• short
• integer
• long
• float
• double
• decimal
• timestamp
• date
• string
• void skipped by default

Any supported none-string datatype will be cast to string before the replacement is done.

Example

input_df:

id	string
1	hello
2	world
3

output_df = Replace(
    column="string",
    from_value="hello",
    to_value="programmer",
).transform(input_df)

output_df:

id	string
1	programmer
2	world
3

In this example, the value "hello" in the column "string" is replaced with "programmer".

from_value class-attribute instance-attribute

from_value: Optional[str] = Field(
    default=None,
    alias="from",
    description="The original value that needs to be replaced. If no value is given, all 'null' values will be replaced with the to_value",
)

to_value class-attribute instance-attribute

to_value: str = Field(
    default=...,
    alias="to",
    description="The new value to replace this with",
)

ColumnConfig

Column type configurations for the column to be replaced

limit_data_type class-attribute instance-attribute

limit_data_type = [*run_for_all_data_type, VOID]

run_for_all_data_type class-attribute instance-attribute

run_for_all_data_type = [
    BYTE,
    SHORT,
    INTEGER,
    LONG,
    FLOAT,
    DOUBLE,
    DECIMAL,
    STRING,
    TIMESTAMP,
    DATE,
]

func

func(column: Column) -> Column

Source code in src/koheesio/spark/transformations/replace.py

def func(self, column: Column) -> Column:
    return replace(column=column, from_value=self.from_value, to_value=self.to_value)

koheesio.spark.transformations.replace.replace

replace(
    column: Union[Column, str],
    to_value: str,
    from_value: Optional[str] = None,
)

Function to replace a particular value in a column with another one

Source code in src/koheesio/spark/transformations/replace.py

def replace(column: Union[Column, str], to_value: str, from_value: Optional[str] = None):
    """Function to replace a particular value in a column with another one"""
    # make sure we have a Column object
    if isinstance(column, str):
        column = col(column)

    if not from_value:
        condition = column.isNull()
    else:
        condition = column == from_value

    return when(condition, lit(to_value)).otherwise(column)