API reference¶

Generated from the compiled minarrow module, which must be importable in the documentation build environment.

Arrays and tables¶

Array ¶

A minarrow array exposed to Python.

arrow_type `property` ¶

arrow_type

The mapped Arrow logical type.

bit_width `property` ¶

bit_width

The physical integer width in bits.

dtype `property` ¶

dtype

The concrete dtype.

is_view `property` ¶

is_view

Whether this array is a windowed view of a larger buffer.

name `property` ¶

name

The column name, or None for an unnamed array.

null_count `property` ¶

null_count

The number of nulls.

from_arrow `staticmethod` ¶

from_arrow(obj)

Import a PyArrow or Polars array via the Arrow C Data Interface, zero-copy for the primary buffers. An Arrow array carries no column name, so the result is unnamed.

from_cudf `staticmethod` ¶

from_cudf(obj)

Alias for from_arrow, accepting a cuDF object through the Arrow PyCapsule interface.

from_nanoarrow `staticmethod` ¶

from_nanoarrow(obj)

Alias for from_arrow, accepting any nanoarrow object through the Arrow PyCapsule interface.

from_pandas `staticmethod` ¶

from_pandas(obj)

Alias for from_arrow, accepting a pandas object through the Arrow PyCapsule interface.

from_polars `staticmethod` ¶

from_polars(obj)

Alias for from_arrow, accepting any Polars object through the Arrow PyCapsule interface.

push `method descriptor` ¶

push(value)

Appends a value to the array. When the Array is in a view state, or another object is referencing the underlying memory, it is materialised to an owned array before appending.

Suited to appending a few values. To build a large array, consider constructing it from a Python sequence in one call, Array([...]), rather than many push calls, since each call crosses the Python boundary.

push_null `method descriptor` ¶

push_null()

Appends a null to the array. When the Array is in a view state, or another object is referencing the underlying memory, it is materialised to an owned array before appending.

set `method descriptor` ¶

set(index, value)

Sets the element at index to value. Negative indices count from the end, so -1 is the last element. An out-of-range index raises IndexError. When the Array is in a view state, or another object is referencing the underlying memory, it is materialised to an owned array before the write, leaving the source unchanged.

to_arrow `method descriptor` ¶

to_arrow()

Export to a PyArrow array via the Arrow C Data Interface.

to_cudf `method descriptor` ¶

to_cudf()

Convert to a cuDF Series through the Arrow PyCapsule interface. Runs on GPU and requires the cudf package.

to_nanoarrow `method descriptor` ¶

to_nanoarrow()

Convert to a nanoarrow Array through the Arrow PyCapsule interface. Requires the nanoarrow package.

to_pandas `method descriptor` ¶

to_pandas()

Convert to a pandas Series through the Arrow PyCapsule interface. Requires pandas 3.0+ (Series.from_arrow).

to_polars `method descriptor` ¶

to_polars()

Convert to a Polars Series through the Arrow PyCapsule interface. A named array carries its name onto the Series. Requires the polars package.

Table ¶

A minarrow table exposed to Python.

columns `property` ¶

columns

The column names in order.

dtypes `property` ¶

dtypes

A name to dtype mapping over the columns, in order.

is_view `property` ¶

is_view

Whether this table is a view over another table's buffers.

n_cols `property` ¶

n_cols

The number of columns.

n_rows `property` ¶

n_rows

The number of rows.

name `property` ¶

name

The table name, or None when unnamed.

schema `property` ¶

schema

The table's schema: the column fields in order.

add_column `method descriptor` ¶

add_column(name, column)

Appends column under name. A view-backed table is materialised first. The column length must match the table's row count unless the table is empty.

from_arrow `staticmethod` ¶

from_arrow(obj)

Import a PyArrow RecordBatch via the Arrow C Data Interface.

from_cudf `staticmethod` ¶

from_cudf(obj)

Alias for from_arrow, accepting a cuDF object through the Arrow PyCapsule interface.

from_daft `staticmethod` ¶

from_daft(obj)

Alias for from_arrow, accepting any Daft object through the Arrow PyCapsule interface.

from_datafusion `staticmethod` ¶

from_datafusion(obj)

Alias for from_arrow, accepting any DataFusion object through the Arrow PyCapsule interface.

from_duckdb `staticmethod` ¶

from_duckdb(obj)

Alias for from_arrow, accepting any DuckDB object through the Arrow PyCapsule interface.

from_ibis `staticmethod` ¶

from_ibis(obj)

Alias for from_arrow, accepting an Ibis object through the Arrow PyCapsule interface.

from_nanoarrow `staticmethod` ¶

from_nanoarrow(obj)

Alias for from_arrow, accepting any nanoarrow object through the Arrow PyCapsule interface.

from_narwhals `staticmethod` ¶

from_narwhals(obj)

Alias for from_arrow, accepting a Narwhals object through the Arrow PyCapsule interface.

from_pandas `staticmethod` ¶

from_pandas(obj)

Alias for from_arrow, accepting a pandas object through the Arrow PyCapsule interface.

from_polars `staticmethod` ¶

from_polars(obj)

Alias for from_arrow, accepting any Polars object through the Arrow PyCapsule interface.

read_adbc `staticmethod` ¶

read_adbc(cursor)

Read an ADBC cursor's current result into a Table. Call after cursor.execute(...). cursor is an ADBC DBAPI cursor.

to_arrow `method descriptor` ¶

to_arrow()

Export to a PyArrow RecordBatch via the Arrow C Data Interface.

to_cudf `method descriptor` ¶

to_cudf()

Convert to a cuDF DataFrame through the Arrow PyCapsule interface. Runs on GPU and requires the cudf package.

to_daft `method descriptor` ¶

to_daft()

Convert to a Daft DataFrame through the Arrow PyCapsule interface. Requires the daft package.

to_datafusion `method descriptor` ¶

to_datafusion()

Convert to a DataFusion DataFrame through the Arrow PyCapsule interface. Requires the datafusion package.

to_duckdb `method descriptor` ¶

to_duckdb()

Convert to a DuckDB relation through the Arrow PyCapsule interface. Requires the duckdb package.

to_ibis `method descriptor` ¶

to_ibis()

Convert to an Ibis table expression. Ibis wraps a pyarrow Table, so this requires pyarrow and the ibis-framework package.

to_nanoarrow `method descriptor` ¶

to_nanoarrow()

Convert to a nanoarrow ArrayStream through the Arrow PyCapsule interface. Requires the nanoarrow package.

to_narwhals `method descriptor` ¶

to_narwhals()

Convert to a Narwhals DataFrame over a pyarrow Table. Narwhals wraps a native frame, so this requires pyarrow and narwhals.

to_pandas `method descriptor` ¶

to_pandas()

Convert to a pandas DataFrame through the Arrow PyCapsule interface. Requires pandas 3.0+ (DataFrame.from_arrow).

to_polars `method descriptor` ¶

to_polars()

Convert to a Polars DataFrame through the Arrow PyCapsule interface. Requires the polars package.

write_adbc `method descriptor` ¶

write_adbc(cursor, table_name, mode='create')

Ingest this table into a database through an ADBC cursor, over the Arrow PyCapsule interface. cursor is an ADBC DBAPI cursor from a driver such as adbc_driver_sqlite, adbc_driver_postgresql, adbc_driver_snowflake, adbc_driver_bigquery, or adbc_driver_flightsql. mode is one of create, append, replace, or create_append. Returns the row count.

ChunkedArray ¶

An ordered set of Array chunks that share a dtype and field. Wraps minarrow's SuperArray.

arrow_type `property` ¶

arrow_type

The mapped Arrow logical type.

chunks `property` ¶

chunks

The chunks in order.

dtype `property` ¶

dtype

The concrete dtype.

n_chunks `property` ¶

n_chunks

The number of chunks.

name `property` ¶

name

The column name, or None when the column is unnamed.

null_count `property` ¶

null_count

The total number of nulls across all chunks.

chunk `method descriptor` ¶

chunk(index)

The chunk at index, or None when out of range.

from_arrow `staticmethod` ¶

from_arrow(obj)

Import a chunked Arrow producer, such as a PyArrow ChunkedArray, through the Arrow C Data Interface.

to_arrow `method descriptor` ¶

to_arrow()

Export to a PyArrow ChunkedArray through the Arrow C Data Interface.

ChunkedTable ¶

An ordered set of Table batches that share a schema. Wraps minarrow's SuperTable.

batches `property` ¶

batches

The batches in order.

columns `property` ¶

columns

The column names in order.

n_batches `property` ¶

n_batches

The number of batches.

n_chunks `property` ¶

n_chunks

The number of batches, named to match ChunkedArray.n_chunks.

n_cols `property` ¶

n_cols

The number of columns.

n_rows `property` ¶

n_rows

The total number of rows across all batches.

name `property` ¶

name

The table name.

schema `property` ¶

schema

The schema attached at construction, or the schema derived from the batches when none was given.

batch `method descriptor` ¶

batch(index)

The batch at index, or None when out of range.

from_arrow `staticmethod` ¶

from_arrow(obj)

Import a chunked Arrow producer, such as a PyArrow Table, through the Arrow C Data Interface.

to_arrow `method descriptor` ¶

to_arrow()

Export to a PyArrow Table through the Arrow C Data Interface.