Raw C API

struct DFArrowArray

#include <datafusion.h>

Same as the ArrowArray struct in the Arrow C data interface.

See also: https://arrow.apache.org/docs/format/CDataInterface.html#the-arrowarray-structure

Public Members

int64_t length

int64_t null_count

int64_t offset

int64_t n_buffers

int64_t n_children

const void **buffers

struct DFArrowArray **children

struct DFArrowArray *dictionary

void (*release)(struct DFArrowArray *array)

void *private_data

struct DFArrowSchema

#include <datafusion.h>

Same as the ArrowSchema struct in the Arrow C data interface.

See also: https://arrow.apache.org/docs/format/CDataInterface.html#the-arrowschema-structure

Public Members

const char *format

const char *name

const char *metadata

int64_t flags

int64_t n_children

struct DFArrowSchema **children

struct DFArrowSchema *dictionary

void (*release)(struct DFArrowSchema *schema)

void *private_data

struct DFDataFrame

#include <datafusion.h>

A struct for data frame.

You get execution result as a data frame.

You need to free data frame by df_data_frame_free() when no longer needed.

struct DFError

#include <datafusion.h>

A struct that holds error information.

You can access to error information by df_error_get_code() and df_error_get_message().

You need to free error information by df_error_free() when no longer needed.

struct DFParquertWriterProperties

#include <datafusion.h>

A struct to customize how to write an Apache Parquet file.

You need to free this by df_parquet_writer_properties_free() when no longer needed.

struct DFSessionContext

#include <datafusion.h>

An entry point of DataFusion API.

You need to create DFSessionContext to use DataFusion API.

file datafusion.h

#include <stdarg.h>

#include <stdbool.h>

#include <stdint.h>

#include <stdlib.h>

Defines

DATAFUSION_MAJOR

DATAFUSION_MINOR

DATAFUSION_PATCH

Typedefs

typedef enum DFErrorCode DFErrorCode

typedef struct DFCSVReadOptions DFCSVReadOptions

typedef struct DFDataFrame DFDataFrame

typedef struct DFError DFError

typedef struct DFParquetReadOptions DFParquetReadOptions

typedef struct DFParquetWriterProperties DFParquetWriterProperties

typedef struct DFSessionContext DFSessionContext

typedef struct DFArrowSchema DFArrowSchema

typedef struct DFArrowArray DFArrowArray

Enums

enum DFErrorCode

Error category.

Values:

enumerator DF_ERROR_CODE_ARROW

enumerator DF_ERROR_CODE_PARQUET

enumerator DF_ERROR_CODE_AVRO

enumerator DF_ERROR_CODE_OBJECT_STORE

enumerator DF_ERROR_CODE_IO

enumerator DF_ERROR_CODE_SQL

enumerator DF_ERROR_CODE_NOT_IMPLEMENTED

enumerator DF_ERROR_CODE_INTERNAL

enumerator DF_ERROR_CODE_PLAN

enumerator DF_ERROR_CODE_SCHEMA

enumerator DF_ERROR_CODE_EXECUTION

enumerator DF_ERROR_CODE_RESOURCES_EXHAUSTED

enumerator DF_ERROR_CODE_EXTERNAL

enumerator DF_ERROR_CODE_JIT

enumerator DF_ERROR_CODE_CONTEXT

enumerator DF_ERROR_CODE_SUBSTRAIT

Functions

struct DFError *df_error_new(enum DFErrorCode code, const char *message)

void df_error_free(struct DFError *_error)

Free the given DFError.

Safety

This function should not be called with error that is not created by df_error_new().

This function should not be called for the same error multiple times.

Parameters:

• _error – A DFError returned by df_*() functions.

const char *df_error_get_message(struct DFError *error)

Get a message of this error.

Safety

This function should not be called with error that is not created by df_error_new().

This function should not be called with error that is freed by df_error_free().

Parameters:

• error – A DFError.

Returns:

A message of this error.

enum DFErrorCode df_error_get_code(struct DFError *error)

Get a code of this error.

Safety

This function should not be called with error that is not created by df_error_new().

This function should not be called with error that is freed by df_error_free().

Parameters:

• error – A DFError.

Returns:

A code of this error.

struct DFParquetWriterProperties *df_parquet_writer_properties_new(void)

void df_parquet_writer_properties_free(struct DFParquetWriterProperties *_properties)

void df_parquet_writer_properties_set_max_row_group_size(struct DFParquetWriterProperties *properties, uintptr_t size)

void df_data_frame_free(struct DFDataFrame *_data_frame)

Free the given DFDataFrame.

Safety

This function should not be called for the same data_frame multiple times.

Parameters:

• _data_frame – A DFDataFrame.

void df_data_frame_show(struct DFDataFrame *data_frame, struct DFError **error)

Show the given data frame contents to the standard output.

Parameters:

• data_frame – A DFDataFrame to be shown.

• error – Return location for a DFError or NULL.

bool df_data_frame_write_parquet(struct DFDataFrame *data_frame, const char *path, const struct DFParquetWriterProperties *writer_properties, struct DFError **error)

Write the given data frame contents as Apache Parquet format.

Parameters:

• data_frame – A DFDataFrame to be written.

• path – An output path.

• writer_properties – Properties how to write Apache Parquet files.

• error – Return location for a DFError or NULL.

int64_t df_data_frame_export(struct DFDataFrame *data_frame, struct DFArrowSchema **c_abi_schema_out, struct DFArrowArray ***c_abi_record_batches_out, struct DFError **error)

struct DFSessionContext *df_session_context_new(void)

Create a new DFSessionContext.

It should be freed by df_session_context_free() when no longer needed.

Returns:

A newly created DFSessionContext.

void df_session_context_free(struct DFSessionContext *_context)

Free the given DFSessionContext.

Safety

This function should not be called with context that is not created by df_session_context_new().

This function should not be called for the same context multiple times.

Parameters:

• _context – A DFSessionContext created by df_session_context_new().

struct DFDataFrame *df_session_context_sql(struct DFSessionContext *context, const char *sql, struct DFError **error)

bool df_session_context_deregister(struct DFSessionContext *context, const char *name, struct DFError **error)

bool df_session_context_register_record_batches(struct DFSessionContext *context, const char *name, struct DFArrowSchema *c_abi_schema, struct DFArrowArray **c_abi_record_batches, size_t n_record_batches, struct DFError **error)

struct DFCSVReadOptions *df_csv_read_options_new(void)

void df_csv_read_options_free(struct DFCSVReadOptions *_options)

void df_csv_read_options_set_has_header(struct DFCSVReadOptions *options, bool has_header)

bool df_csv_read_options_get_has_header(struct DFCSVReadOptions *options)

void df_csv_read_options_set_delimiter(struct DFCSVReadOptions *options, uint8_t delimiter)

uint8_t df_csv_read_options_get_delimiter(struct DFCSVReadOptions *options)

bool df_csv_read_options_set_schema(struct DFCSVReadOptions *options, struct DFArrowSchema *schema, struct DFError **error)

struct DFArrowSchema *df_csv_read_options_get_schema(struct DFCSVReadOptions *options, struct DFError **error)

void df_csv_read_options_set_schema_infer_max_records(struct DFCSVReadOptions *options, uintptr_t n)

uintptr_t df_csv_read_options_get_schema_infer_max_records(struct DFCSVReadOptions *options)

bool df_csv_read_options_set_file_extension(struct DFCSVReadOptions *options, const char *file_extension, struct DFError **error)

char *df_csv_read_options_get_file_extension(struct DFCSVReadOptions *options)

bool df_csv_read_options_set_table_partition_columns(struct DFCSVReadOptions *options, struct DFArrowSchema *schema, struct DFError **error)

struct DFArrowSchema *df_csv_read_options_get_table_partition_columns(struct DFCSVReadOptions *options, struct DFError **error)

bool df_session_context_register_csv(struct DFSessionContext *context, const char *name, const char *url, struct DFCSVReadOptions *options, struct DFError **error)

struct DFParquetReadOptions *df_parquet_read_options_new(void)

void df_parquet_read_options_free(struct DFParquetReadOptions *_options)

bool df_parquet_read_options_set_file_extension(struct DFParquetReadOptions *options, const char *file_extension, struct DFError **error)

char *df_parquet_read_options_get_file_extension(struct DFParquetReadOptions *options)

bool df_parquet_read_options_set_table_partition_columns(struct DFParquetReadOptions *options, struct DFArrowSchema *schema, struct DFError **error)

struct DFArrowSchema *df_parquet_read_options_get_table_partition_columns(struct DFParquetReadOptions *options, struct DFError **error)

void df_parquet_read_options_set_pruning(struct DFParquetReadOptions *options, bool pruning)

void df_parquet_read_options_unset_pruning(struct DFParquetReadOptions *options)

bool df_parquet_read_options_is_set_pruning(struct DFParquetReadOptions *options)

bool df_parquet_read_options_get_pruning(struct DFParquetReadOptions *options)

bool df_session_context_register_parquet(struct DFSessionContext *context, const char *name, const char *url, struct DFParquetReadOptions *options, struct DFError **error)