Package ai.rapids.cudf

Class ORCOptions.Builder

java.lang.Object
ai.rapids.cudf.ColumnFilterOptions.Builder<ORCOptions.Builder>
ai.rapids.cudf.ORCOptions.Builder
Enclosing class:
ORCOptions
public static class ORCOptions.Builder extends ColumnFilterOptions.Builder<ORCOptions.Builder>

  • Constructor Details

    • Builder

      public Builder()

  • Method Details

    • withNumPyTypes

      public ORCOptions.Builder withNumPyTypes(boolean useNumPyTypes)
      Specify whether the parser should implicitly promote TIMESTAMP_DAYS columns to TIMESTAMP_MILLISECONDS for compatibility with NumPy.
      Parameters:
      useNumPyTypes - true to request this conversion, false to avoid.
      Returns:
      builder for chaining

    • withTimeUnit

      public ORCOptions.Builder withTimeUnit(DType unit)
      Specify the time unit to use when returning timestamps.
      Parameters:
      unit - default unit of time specified by the user
      Returns:
      builder for chaining

    • decimal128Column

      public ORCOptions.Builder decimal128Column(String... names)
      Specify decimal columns which shall be read as DECIMAL128. Otherwise, decimal columns will be read as DECIMAL64 by default in ORC. In terms of child columns of nested types, their parents need to be prepended as prefix of the column name, in case of ambiguity. For struct columns, the names of child columns are formatted as `{struct_col_name}.{child_col_name}`. For list columns, the data(child) columns are named as `{list_col_name}.1`.
      Parameters:
      names - names of columns which read as DECIMAL128
      Returns:
      builder for chaining

    • build

      public ORCOptions build()