Package dev.hardwood.metadata

package dev.hardwood.metadata

Low-level Parquet file metadata types that mirror the Thrift definitions.

These types provide direct access to the metadata structures stored in a Parquet file footer. For a higher-level schema representation with computed definition and repetition levels, see dev.hardwood.schema.

The record types in this package (e.g. FileMetaData, RowGroup, ColumnChunk, Statistics) are read-only views of the metadata in a Parquet file: Hardwood constructs them while parsing, and callers consume them through their accessors. Their canonical constructors and component lists are not part of the supported API — do not instantiate these records directly, and do not rely on record deconstruction patterns over them. Because these types mirror the evolving Parquet Thrift definitions, future releases may add components; such additions are treated as backward-compatible under this policy even though they change the canonical constructor.

See Also:

• File Format – Metadata
• parquet.thrift

Related Packages

Package	Description
dev.hardwood	Core entry points for reading Parquet files.

Class	Description
BoundingBox
ColumnChunk	Column chunk metadata.
ColumnIndex	Column index for a column chunk, providing per-page min/max statistics for page-level filtering.
ColumnIndex.BoundaryOrder	Ordering of min/max values across pages.
ColumnMetaData	Metadata for a column chunk.
CompressionCodec	Compression codecs supported by Parquet.
ConvertedType	Legacy converted types in Parquet schema (used by PyArrow for LIST/MAP annotation).
Encoding	Encoding types for Parquet data.
FieldPath	Path from the root schema to a leaf column, represented as a list of field names.
FileMetaData	Top-level file metadata for a Parquet file.
GeospatialStatistics
LogicalType	Logical types that provide semantic meaning to physical types.
LogicalType.BsonType	BSON document stored as a byte array.
LogicalType.DateType	Calendar date (days since Unix epoch).
LogicalType.DecimalType	Decimal with fixed scale and precision.
LogicalType.EdgeInterpolationAlgorithm	Geodesic formulations to model and compute the shortest path on Earth.
LogicalType.EnumType	Enum stored as a UTF-8 string.
LogicalType.Float16Type	IEEE 754 half-precision (binary16) floating point, stored as a 2-byte FIXED_LEN_BYTE_ARRAY in little-endian byte order.
LogicalType.GeographyType	Geography type with configurable CRS
LogicalType.GeometryType	Geometry type with configurable CRS
LogicalType.IntervalType	Interval stored as a 12-byte fixed-length byte array (months, days, millis).
LogicalType.IntType	Integer type with a specific bit width and signedness.
LogicalType.JsonType	JSON document stored as a UTF-8 string.
LogicalType.ListType	List (repeated element) logical type.
LogicalType.MapType	Map (key-value pairs) logical type.
LogicalType.NullType	Column whose every value is null.
LogicalType.StringType	UTF-8 encoded string.
LogicalType.TimestampType	Timestamp with configurable precision and UTC adjustment.
LogicalType.TimeType	Time of day with configurable precision and UTC adjustment.
LogicalType.TimeUnit	Resolution of time and timestamp logical types.
LogicalType.UuidType	UUID stored as a 16-byte fixed-length byte array.
LogicalType.VariantType	Variant (self-describing, semi-structured) logical type per the Parquet Variant spec.
OffsetIndex	Offset index for a column chunk, providing page locations for direct lookup.
PageLocation	Location of a data page within a column chunk.
PhysicalType	Physical types supported by Parquet format.
RepetitionType	Field repetition types in Parquet schema.
RowGroup	Row group metadata.
SchemaElement	Schema element in Parquet file metadata.
Statistics	Column chunk statistics for min/max values, null count, and distinct count.