Package org.apache.lucene.document.column

package org.apache.lucene.document.column

Column-oriented batch indexing API.

IndexWriter.addBatch(org.apache.lucene.document.column.ColumnBatch) accepts a ColumnBatch: a fixed number of documents presented field-by-field rather than document-by-document. Each field is a Column that exposes its values via cursor iterators rather than concrete IndexableField instances per document.

Column subtypes

• LongColumn — single- or multi-valued long values for NUMERIC / SORTED_NUMERIC doc values, 1‑D numeric points (int / long / float / double), and stored numeric fields.
• BinaryColumn — variable-length binary values for BINARY, SORTED, and SORTED_SET doc values, term inversion, multi-dimensional or arbitrary-width points, and stored binary or string fields.
• DictionaryColumn — pre-defined term dictionary plus per-doc ordinals for SORTED and SORTED_SET doc values, term inversion, and stored binary or string fields.
• VectorColumn — KNN vectors (FLOAT32 or BYTE encoding); vector-only field type.
• TokenStreamColumn — caller-supplied TokenStreams for term inversion (the columnar analogue of a custom token stream on a Field); inverted-index-only field type.

Cursors

A Column declares its Column.Density (DENSE or SPARSE) and exposes its values via cursors:

• A tuple cursor (e.g. LongTupleCursor, ObjectTupleCursor) yields (batchDocID, value) pairs in non-decreasing doc-id order. Always available.
• A bulk values cursor (e.g. LongValuesCursor) feeds dense data directly into the underlying writer. Required when Column.density() is DENSE and consulted only in that case.

Each call that requests a cursor returns a fresh cursor positioned at the first value, so columns can be consumed multiple times — once in the row-oriented pass for stored fields and term inversion, and again in the column-oriented pass for doc values, points, and vectors.

WARNING: This API is experimental and might change in incompatible ways in the next release.

Related Packages

Package	Description
org.apache.lucene.document	The logical representation of a Document for indexing and searching.

Class	Description
BinaryColumn	A Column that provides variable-size binary values via a tuple cursor.
BytesRefValuesCursor	A values cursor over a dense BinaryColumn.
Column	A single field's values across multiple documents in a ColumnBatch.
Column.Density	Whether a column has a value for every document in the batch.
ColumnBatch	A column-oriented batch of documents for indexing.
ColumnFieldAdapter	Lightweight adapter that presents a Column's current cursor value as an IndexableField so it can be fed through the row-oriented indexing pass (stored fields and term inversion).
ColumnValidation	Static validation and bounds-checking helpers for the columnar indexing path.
DictionaryColumn	A Column that provides string or binary values via a pre-defined term dictionary plus per-doc ordinals into that dictionary.
LongColumn	A Column that provides long values.
LongColumn.NumericKind	The numeric interpretation of the column's long values.
LongTupleCursor	A tuple cursor over a LongColumn.
LongValuesCursor	A values cursor over a dense LongColumn.
ObjectTupleCursor<T>	A tuple cursor over a Column whose values are objects.
OrdinalsCursor	A dense values cursor over a DictionaryColumn.
OrdinalsTupleCursor	A tuple cursor over a DictionaryColumn.
TokenStreamColumn	A Column that provides caller-supplied TokenStreams for term inversion.
VectorColumn<T>	A Column that provides KNN vector values via a tuple cursor.