Package org.apache.juneau.marshall.parquet

Class ParquetSchemaBuilder

java.lang.Object
org.apache.juneau.marshall.parquet.ParquetSchemaBuilder
public final class ParquetSchemaBuilder extends Object
Builds Parquet schema (list of SchemaElement) from Juneau bean ClassMeta.

Maps Java types to Parquet physical and logical types per the Parquet specification.

  • Constructor Details

    • ParquetSchemaBuilder

      public ParquetSchemaBuilder(MarshallingContext marshallingContext, boolean writeDatesAsTimestamp, ParquetCycleHandling cycleHandling)
      Constructor.
      Parameters:
      marshallingContext - The bean context for type resolution.
      writeDatesAsTimestamp - If true, Date/Calendar/Temporal map to INT64+TIMESTAMP_MILLIS; else to BYTE_ARRAY+STRING.
      cycleHandling - How to handle structural cycles when building schema.

    • ParquetSchemaBuilder

      public ParquetSchemaBuilder(MarshallingContext marshallingContext, boolean writeDatesAsTimestamp, ParquetCycleHandling cycleHandling, int maxRecursionDepth)
      Constructor.
      Parameters:
      marshallingContext - The bean context for type resolution.
      writeDatesAsTimestamp - If true, Date/Calendar/Temporal map to INT64+TIMESTAMP_MILLIS; else to BYTE_ARRAY+STRING.
      cycleHandling - How to handle structural cycles when building schema.
      maxRecursionDepth - Maximum expansion depth for self-referential bean types (GAP-13).

    • ParquetSchemaBuilder

      public ParquetSchemaBuilder(MarshallingContext marshallingContext, boolean writeDatesAsTimestamp, ParquetCycleHandling cycleHandling, int maxRecursionDepth, boolean nativeLogicalTypes)
      Constructor.
      Parameters:
      marshallingContext - The bean context for type resolution.
      writeDatesAsTimestamp - If true, Date/Calendar/Temporal map to INT64+TIMESTAMP_MILLIS; else to BYTE_ARRAY+STRING.
      cycleHandling - How to handle structural cycles when building schema.
      maxRecursionDepth - Maximum expansion depth for self-referential bean types (GAP-13).
      nativeLogicalTypes - If true, emit binary-native DECIMAL/DATE/TIME/TIMESTAMP-micros logical types instead of the default string / TIMESTAMP-millis normalization (GAP-9/10).

  • Method Details

    • buildSchema

      public List<ParquetSchemaElement> buildSchema(ClassMeta<?> cm)
      Builds the Parquet schema for the given bean class.
      Parameters:
      cm - The bean class metadata.
      Returns:
      The depth-first list of schema elements (root first).

    • buildSchema

      public List<ParquetSchemaElement> buildSchema(ClassMeta<?> cm, Object sampleBean)
      Builds the Parquet schema, using a sample bean to resolve collection element types when generics are erased.
      Parameters:
      cm - The bean class metadata.
      sampleBean - Optional sample instance; when provided, collection element types are resolved from actual values.
      Returns:
      The depth-first list of schema elements (root first).

    • buildSchemaFromMap

      public List<ParquetSchemaElement> buildSchemaFromMap(Map<?,?> map)
      Builds the Parquet schema for a Map root (single row, columns from map keys).
      Parameters:
      map - The map instance (keys become column names).
      Returns:
      The depth-first list of schema elements (root first).

    • buildSchemaForKeyValuePairs

      public List<ParquetSchemaElement> buildSchemaForKeyValuePairs(Object keySample, Object valueSample)
      Builds the Parquet schema for a Map root with non-String keys (flat list-of-pairs: root.key, root.value).

      Each map entry becomes one Parquet row. Keys and values use native Parquet types.

      Parameters:
      keySample - Sample key for type inference.
      valueSample - Sample value for type inference.
      Returns:
      The depth-first list of schema elements (root first).

    • getLeafColumns

      public static List<ParquetSchemaElement> getLeafColumns(List<ParquetSchemaElement> schema)
      Returns leaf (primitive) schema elements suitable for column chunk writing.
      Parameters:
      schema - The full schema from buildSchema.
      Returns:
      List of leaf elements with non-null paths.