ddn.columnar.batch

The structural storage carrier beneath DataFrame.

This module owns the structural half of a table: Field and Schema (the ordered shape) and ColumnBatch (equal-height, uniquely named columns). It is deliberately free of table semantics - no join, groupBy, expression, sort or IO logic enters here - and it never copies or rechunks its columns: independently chunked columns stay independently chunked, because physical batch alignment is a codec concern, not a carrier concern.

DataFrame wraps a ColumnBatch instead of keeping a second column store; the old ddn.df.frame spellings of Field and Schema remain available as re-exports. The query engine's ColumnSchema (nullability and cardinality facts) is a different contract and stays with the expression layer.

Copyright

License

BSD-3-Clause

struct ColumnBatch Field Schema

Types 3

structField

One named, typed column of a Schema.

Fields

string nameThe column name.

DataType dtypeThe column dtype.

Methods

bool opEquals(const Field other) const @safe pure nothrowField equality: same name and dtype.

structSchema

An ordered collection of fields: the shape a DataFrame conforms to. vstack requires schema equality; readCsv takes one as input.

Fields

private Field[] _fields

Methods

Field[] fields() @safe pure nothrowReturns: the fields, in column order.

size_t length() const @safe pure nothrowReturns: the number of fields.

Nullable!Field field(string name) const @trusted pureFinds a field by name.

Nullable!size_t indexOf(string name) const @safe pureFinds the position of a field by name.

bool opEquals(const Schema other) const @safe pure nothrowSchema equality: same fields in the same order.

Constructors

this(Field[] fields)Constructs from fields (kept as a copy).

structColumnBatch

A structural batch of equal-height, uniquely named columns: the storage carrier DataFrame wraps. ColumnBatch.init is the empty batch (no columns, height 0). A typed batch of zero rows keeps its schema.

Fields

private Series[] _columns

Methods

Series[] columns() @safe pure nothrowReturns: the columns, in order (no copy; chunking is preserved).

size_t height() const @safe pure nothrowReturns: the number of rows (0 for the empty batch).

size_t width() const @safe pure nothrowReturns: the number of columns.

size_t[2] shape() const @safe pure nothrowReturns: `(height, width)`.

string[] names() @safe pure nothrowReturns: the column names, in order.

Schema schema() @safe pureReturns: the structural schema of the batch.

DataType[] dtypes() @safe pure nothrowReturns: the column dtypes, in order.

bool isEmpty() const @safe pure nothrowReturns: `true` when the batch has no rows or no columns.

Series column(string name) @safe pureParams: name = the column name

Series opIndex(string name) @safe pureReturns: the named column.

Constructors

this(Series[] columns)Constructs from columns, validating equal heights and unique names.