C API Reference#

group nanoarrow

Except where noted, objects are not thread-safe and clients should take care to serialize accesses to methods.

Because this library is intended to be vendored, it provides full type definitions and encourages clients to stack or statically allocate where convenient.

Creating schemas#

group nanoarrow-schema

These functions allocate, copy, and destroy ArrowSchema structures

Functions

void ArrowSchemaInit(struct ArrowSchema *schema)#

Initialize an ArrowSchema.

Initializes the fields and release callback of schema_out. Caller is responsible for calling the schema->release callback if NANOARROW_OK is returned.

ArrowErrorCode ArrowSchemaInitFromType(struct ArrowSchema *schema, enum ArrowType type)#

Initialize an ArrowSchema from an ArrowType.

A convenience constructor for that calls ArrowSchemaInit() and ArrowSchemaSetType() for the common case of constructing an unparameterized type. The caller is responsible for calling the schema->release callback if NANOARROW_OK is returned.

int64_t ArrowSchemaToString(const struct ArrowSchema *schema, char *out, int64_t n, char recursive)#

Get a human-readable summary of a Schema.

Writes a summary of an ArrowSchema to out (up to n - 1 characters) and returns the number of characters required for the output if n were sufficiently large. If recursive is non-zero, the result will also include children.

ArrowErrorCode ArrowSchemaSetType(struct ArrowSchema *schema, enum ArrowType type)#

Set the format field of a schema from an ArrowType.

Initializes the fields and release callback of schema_out. For NANOARROW_TYPE_LIST, NANOARROW_TYPE_LARGE_LIST, and NANOARROW_TYPE_MAP, the appropriate number of children are allocated, initialized, and named; however, the caller must ArrowSchemaSetType() on the preinitialized children. Schema must have been initialized using ArrowSchemaInit() or ArrowSchemaDeepCopy().

ArrowErrorCode ArrowSchemaSetTypeStruct(struct ArrowSchema *schema, int64_t n_children)#

Set the format field and initialize children of a struct schema.

The specified number of children are initialized; however, the caller is responsible for calling ArrowSchemaSetType() and ArrowSchemaSetName() on each child. Schema must have been initialized using ArrowSchemaInit() or ArrowSchemaDeepCopy().

ArrowErrorCode ArrowSchemaSetTypeFixedSize(struct ArrowSchema *schema, enum ArrowType type, int32_t fixed_size)#

Set the format field of a fixed-size schema.

Returns EINVAL for fixed_size <= 0 or for type that is not NANOARROW_TYPE_FIXED_SIZE_BINARY or NANOARROW_TYPE_FIXED_SIZE_LIST. For NANOARROW_TYPE_FIXED_SIZE_LIST, the appropriate number of children are allocated, initialized, and named; however, the caller must ArrowSchemaSetType() the first child. Schema must have been initialized using ArrowSchemaInit() or ArrowSchemaDeepCopy().

ArrowErrorCode ArrowSchemaSetTypeDecimal(struct ArrowSchema *schema, enum ArrowType type, int32_t decimal_precision, int32_t decimal_scale)#

Set the format field of a decimal schema.

Returns EINVAL for scale <= 0 or for type that is not NANOARROW_TYPE_DECIMAL128 or NANOARROW_TYPE_DECIMAL256. Schema must have been initialized using ArrowSchemaInit() or ArrowSchemaDeepCopy().

ArrowErrorCode ArrowSchemaSetTypeRunEndEncoded(struct ArrowSchema *schema, enum ArrowType run_end_type)#

Set the format field of a run-end encoded schema.

Returns EINVAL for run_end_type that is not NANOARROW_TYPE_INT16, NANOARROW_TYPE_INT32 or NANOARROW_TYPE_INT64. Schema must have been initialized using ArrowSchemaInit() or ArrowSchemaDeepCopy(). The caller must call ArrowSchemaSetTypeXXX(schema->children[1]) to set the value type. Note that when building arrays using the ArrowArrayAppendXXX() functions, the run-end encoded array’s logical length must be updated manually.

ArrowErrorCode ArrowSchemaSetTypeDateTime(struct ArrowSchema *schema, enum ArrowType type, enum ArrowTimeUnit time_unit, const char *timezone)#

Set the format field of a time, timestamp, or duration schema.

Returns EINVAL for type that is not NANOARROW_TYPE_TIME32, NANOARROW_TYPE_TIME64, NANOARROW_TYPE_TIMESTAMP, or NANOARROW_TYPE_DURATION. The timezone parameter must be NULL for a non-timestamp type. Schema must have been initialized using ArrowSchemaInit() or ArrowSchemaDeepCopy().

ArrowErrorCode ArrowSchemaSetTypeUnion(struct ArrowSchema *schema, enum ArrowType type, int64_t n_children)#

Set the format field of a union schema.

Returns EINVAL for a type that is not NANOARROW_TYPE_DENSE_UNION or NANOARROW_TYPE_SPARSE_UNION. The specified number of children are allocated, and initialized.

ArrowErrorCode ArrowSchemaDeepCopy(const struct ArrowSchema *schema, struct ArrowSchema *schema_out)#

Make a (recursive) copy of a schema.

Allocates and copies fields of schema into schema_out.

ArrowErrorCode ArrowSchemaSetFormat(struct ArrowSchema *schema, const char *format)#

Copy format into schema->format.

schema must have been allocated using ArrowSchemaInitFromType() or ArrowSchemaDeepCopy().

ArrowErrorCode ArrowSchemaSetName(struct ArrowSchema *schema, const char *name)#

Copy name into schema->name.

schema must have been allocated using ArrowSchemaInitFromType() or ArrowSchemaDeepCopy().

ArrowErrorCode ArrowSchemaSetMetadata(struct ArrowSchema *schema, const char *metadata)#

Copy metadata into schema->metadata.

schema must have been allocated using ArrowSchemaInitFromType() or ArrowSchemaDeepCopy.

ArrowErrorCode ArrowSchemaAllocateChildren(struct ArrowSchema *schema, int64_t n_children)#

Allocate the schema->children array.

Includes the memory for each child struct ArrowSchema. schema must have been allocated using ArrowSchemaInitFromType() or ArrowSchemaDeepCopy().

ArrowErrorCode ArrowSchemaAllocateDictionary(struct ArrowSchema *schema)#

Allocate the schema->dictionary member.

schema must have been allocated using ArrowSchemaInitFromType() or ArrowSchemaDeepCopy().

Reading schemas#

group nanoarrow-schema-view

Functions

ArrowErrorCode ArrowSchemaViewInit(struct ArrowSchemaView *schema_view, const struct ArrowSchema *schema, struct ArrowError *error)#: Initialize an ArrowSchemaView.

struct ArrowSchemaView#

#include <nanoarrow.h>

A non-owning view of a parsed ArrowSchema.

Contains more readily extractable values than a raw ArrowSchema. Clients can stack or statically allocate this structure but are encouraged to use the provided getters to ensure forward compatibility.

Public Members

const struct ArrowSchema *schema#: A pointer to the schema represented by this view.

enum ArrowType type#

The data type represented by the schema.

This value may be NANOARROW_TYPE_DICTIONARY if the schema has a non-null dictionary member; datetime types are valid values. This value will never be NANOARROW_TYPE_EXTENSION (see extension_name and/or extension_metadata to check for an extension type).

enum ArrowType storage_type#

The storage data type represented by the schema.

This value will never be NANOARROW_TYPE_DICTIONARY, NANOARROW_TYPE_EXTENSION or any datetime type. This value represents only the type required to interpret the buffers in the array.

struct ArrowLayout layout#: The storage layout represented by the schema.

struct ArrowStringView extension_name#

The extension type name if it exists.

If the ARROW:extension:name key is present in schema.metadata, extension_name.data will be non-NULL.

struct ArrowStringView extension_metadata#

The extension type metadata if it exists.

If the ARROW:extension:metadata key is present in schema.metadata, extension_metadata.data will be non-NULL.

int32_t fixed_size#

Format fixed size parameter.

This value is set when parsing a fixed-size binary or fixed-size list schema; this value is undefined for other types. For a fixed-size binary schema this value is in bytes; for a fixed-size list schema this value refers to the number of child elements for each element of the parent.

int32_t decimal_bitwidth#

Decimal bitwidth.

This value is set when parsing a decimal type schema; this value is undefined for other types.

int32_t decimal_precision#

Decimal precision.

This value is set when parsing a decimal type schema; this value is undefined for other types.

int32_t decimal_scale#

Decimal scale.

This value is set when parsing a decimal type schema; this value is undefined for other types.

enum ArrowTimeUnit time_unit#

Format time unit parameter.

This value is set when parsing a date/time type. The value is undefined for other types.

const char *timezone#

Format timezone parameter.

This value is set when parsing a timestamp type and represents the timezone format parameter. This value points to data within the schema and is undefined for other types.

const char *union_type_ids#

Union type ids parameter.

This value is set when parsing a union type and represents type ids parameter. This value points to data within the schema and is undefined for other types.

Creating arrays#

group nanoarrow-array

These functions allocate, copy, and destroy ArrowArray structures. Once an ArrowArray has been initialized via ArrowArrayInitFromType() or ArrowArrayInitFromSchema(), the caller is responsible for releasing it using the embedded release callback.

Functions

ArrowErrorCode ArrowArrayInitFromType(struct ArrowArray *array, enum ArrowType storage_type)#

Initialize the fields of an array.

Initializes the fields and release callback of array. Caller is responsible for calling the array->release callback if NANOARROW_OK is returned.

ArrowErrorCode ArrowArrayInitFromSchema(struct ArrowArray *array, const struct ArrowSchema *schema, struct ArrowError *error)#

Initialize the contents of an ArrowArray from an ArrowSchema.

Caller is responsible for calling the array->release callback if NANOARROW_OK is returned.

ArrowErrorCode ArrowArrayInitFromArrayView(struct ArrowArray *array, const struct ArrowArrayView *array_view, struct ArrowError *error)#

Initialize the contents of an ArrowArray from an ArrowArrayView.

Caller is responsible for calling the array->release callback if NANOARROW_OK is returned.

ArrowErrorCode ArrowArrayAllocateChildren(struct ArrowArray *array, int64_t n_children)#

Allocate the array->children array.

Includes the memory for each child struct ArrowArray, whose members are marked as released and may be subsequently initialized with ArrowArrayInitFromType() or moved from an existing ArrowArray. schema must have been allocated using ArrowArrayInitFromType().

ArrowErrorCode ArrowArrayAllocateDictionary(struct ArrowArray *array)#

Allocate the array->dictionary member.

Includes the memory for the struct ArrowArray, whose contents is marked as released and may be subsequently initialized with ArrowArrayInitFromType() or moved from an existing ArrowArray. array must have been allocated using ArrowArrayInitFromType()

void ArrowArraySetValidityBitmap(struct ArrowArray *array, struct ArrowBitmap *bitmap)#

Set the validity bitmap of an ArrowArray.

array must have been allocated using ArrowArrayInitFromType()

ArrowErrorCode ArrowArraySetBuffer(struct ArrowArray *array, int64_t i, struct ArrowBuffer *buffer)#

Set a buffer of an ArrowArray.

array must have been allocated using ArrowArrayInitFromType()

static inline struct ArrowBitmap *ArrowArrayValidityBitmap(struct ArrowArray *array)#

Get the validity bitmap of an ArrowArray.

array must have been allocated using ArrowArrayInitFromType()

static inline struct ArrowBuffer *ArrowArrayBuffer(struct ArrowArray *array, int64_t i)#

Get a buffer of an ArrowArray.

array must have been allocated using ArrowArrayInitFromType()

static inline ArrowErrorCode ArrowArrayStartAppending(struct ArrowArray *array)#

Start element-wise appending to an ArrowArray.

Initializes any values needed to use ArrowArrayAppend*() functions. All element-wise appenders append by value and return EINVAL if the exact value cannot be represented by the underlying storage type. array must have been allocated using ArrowArrayInitFromType()

ArrowErrorCode ArrowArrayReserve(struct ArrowArray *array, int64_t additional_size_elements)#

Reserve space for future appends.

For buffer sizes that can be calculated (i.e., not string data buffers or child array sizes for non-fixed-size arrays), recursively reserve space for additional elements. This is useful for reducing the number of reallocations that occur using the item-wise appenders.

static inline ArrowErrorCode ArrowArrayAppendNull(struct ArrowArray *array, int64_t n)#: Append a null value to an array.

static inline ArrowErrorCode ArrowArrayAppendEmpty(struct ArrowArray *array, int64_t n)#: Append an empty, non-null value to an array.

static inline ArrowErrorCode ArrowArrayAppendInt(struct ArrowArray *array, int64_t value)#

Append a signed integer value to an array.

Returns NANOARROW_OK if value can be exactly represented by the underlying storage type or EINVAL otherwise (e.g., value is outside the valid array range).

static inline ArrowErrorCode ArrowArrayAppendUInt(struct ArrowArray *array, uint64_t value)#

Append an unsigned integer value to an array.

Returns NANOARROW_OK if value can be exactly represented by the underlying storage type or EINVAL otherwise (e.g., value is outside the valid array range).

static inline ArrowErrorCode ArrowArrayAppendDouble(struct ArrowArray *array, double value)#

Append a double value to an array.

Returns NANOARROW_OK if value can be exactly represented by the underlying storage type or EINVAL otherwise (e.g., value is outside the valid array range or there is an attempt to append a non-integer to an array with an integer storage type).

static inline ArrowErrorCode ArrowArrayAppendBytes(struct ArrowArray *array, struct ArrowBufferView value)#

Append a string of bytes to an array.

Returns NANOARROW_OK if value can be exactly represented by the underlying storage type, EOVERFLOW if appending value would overflow the offset type (e.g., if the data buffer would be larger than 2 GB for a non-large string type), or EINVAL otherwise (e.g., the underlying array is not a binary, string, large binary, large string, or fixed-size binary array, or value is the wrong size for a fixed-size binary array).

static inline ArrowErrorCode ArrowArrayAppendString(struct ArrowArray *array, struct ArrowStringView value)#

Append a string value to an array.

static inline ArrowErrorCode ArrowArrayAppendInterval(struct ArrowArray *array, const struct ArrowInterval *value)#

Append a Interval to an array.

Returns NANOARROW_OK if value can be exactly represented by the underlying storage type or EINVAL otherwise.

static inline ArrowErrorCode ArrowArrayAppendDecimal(struct ArrowArray *array, const struct ArrowDecimal *value)#

Append a decimal value to an array.

Returns NANOARROW_OK if array is a decimal array with the appropriate bitwidth or EINVAL otherwise.

static inline ArrowErrorCode ArrowArrayFinishElement(struct ArrowArray *array)#

Finish a nested array element.

Appends a non-null element to the array based on the first child’s current length. Returns NANOARROW_OK if the item was successfully added, EOVERFLOW if the child of a list or map array would exceed INT_MAX elements, or EINVAL if the underlying storage type is not a struct, list, large list, or fixed-size list, or if there was an attempt to add a struct or fixed-size list element where the length of the child array(s) did not match the expected length.

static inline ArrowErrorCode ArrowArrayFinishUnionElement(struct ArrowArray *array, int8_t type_id)#

Finish a union array element.

Appends an element to the union type ids buffer and increments array->length. For sparse unions, up to one element is added to non type-id children. Returns EINVAL if the underlying storage type is not a union, if type_id is not valid, or if child sizes after appending are inconsistent.

static inline ArrowErrorCode ArrowArrayShrinkToFit(struct ArrowArray *array)#

Shrink buffer capacity to the size required.

Also applies shrinking to any child arrays. array must have been allocated using ArrowArrayInitFromType

ArrowErrorCode ArrowArrayFinishBuildingDefault(struct ArrowArray *array, struct ArrowError *error)#

Finish building an ArrowArray.

Flushes any pointers from internal buffers that may have been reallocated into array->buffers and checks the actual size of the buffers against the expected size based on the final length. array must have been allocated using ArrowArrayInitFromType()

ArrowErrorCode ArrowArrayFinishBuilding(struct ArrowArray *array, enum ArrowValidationLevel validation_level, struct ArrowError *error)#

Finish building an ArrowArray with explicit validation.

Finish building with an explicit validation level. This could perform less validation (i.e. NANOARROW_VALIDATION_LEVEL_NONE or NANOARROW_VALIDATION_LEVEL_MINIMAL) if CPU buffer data access is not possible or more validation (i.e., NANOARROW_VALIDATION_LEVEL_FULL) if buffer content was obtained from an untrusted or corruptible source.

Reading arrays#

group nanoarrow-array-view

These functions read and validate the contents ArrowArray structures.

Functions

void ArrowArrayViewInitFromType(struct ArrowArrayView *array_view, enum ArrowType storage_type)#: Initialize the contents of an ArrowArrayView.

static inline void ArrowArrayViewMove(struct ArrowArrayView *src, struct ArrowArrayView *dst)#

Move an ArrowArrayView.

Transfers the ArrowArrayView data and lifecycle management to another address and resets the contents of src.

ArrowErrorCode ArrowArrayViewInitFromSchema(struct ArrowArrayView *array_view, const struct ArrowSchema *schema, struct ArrowError *error)#: Initialize the contents of an ArrowArrayView from an ArrowSchema.

ArrowErrorCode ArrowArrayViewAllocateChildren(struct ArrowArrayView *array_view, int64_t n_children)#

Allocate the array_view->children array.

Includes the memory for each child struct ArrowArrayView

ArrowErrorCode ArrowArrayViewAllocateDictionary(struct ArrowArrayView *array_view)#: Allocate array_view->dictionary.

void ArrowArrayViewSetLength(struct ArrowArrayView *array_view, int64_t length)#: Set data-independent buffer sizes from length.

ArrowErrorCode ArrowArrayViewSetArray(struct ArrowArrayView *array_view, const struct ArrowArray *array, struct ArrowError *error)#: Set buffer sizes and data pointers from an ArrowArray.

ArrowErrorCode ArrowArrayViewSetArrayMinimal(struct ArrowArrayView *array_view, const struct ArrowArray *array, struct ArrowError *error)#: Set buffer sizes and data pointers from an ArrowArray except for those that require dereferencing buffer content.

ArrowErrorCode ArrowArrayViewValidate(struct ArrowArrayView *array_view, enum ArrowValidationLevel validation_level, struct ArrowError *error)#

Performs checks on the content of an ArrowArrayView.

If using ArrowArrayViewSetArray() to back array_view with an ArrowArray, the buffer sizes and some content (fist and last offset) have already been validated at the “default” level. If setting the buffer pointers and sizes otherwise, you may wish to perform checks at a different level. See documentation for ArrowValidationLevel for the details of checks performed at each level.

void ArrowArrayViewReset(struct ArrowArrayView *array_view)#: Reset the contents of an ArrowArrayView and frees resources.

static inline int8_t ArrowArrayViewIsNull(const struct ArrowArrayView *array_view, int64_t i)#: Check for a null element in an ArrowArrayView.

static inline int64_t ArrowArrayViewComputeNullCount(const struct ArrowArrayView *array_view)#: Compute null count for an ArrowArrayView.

static inline int8_t ArrowArrayViewUnionTypeId(const struct ArrowArrayView *array_view, int64_t i)#: Get the type id of a union array element.

static inline int8_t ArrowArrayViewUnionChildIndex(const struct ArrowArrayView *array_view, int64_t i)#: Get the child index of a union array element.

static inline int64_t ArrowArrayViewUnionChildOffset(const struct ArrowArrayView *array_view, int64_t i)#: Get the index to use into the relevant union child array.

static inline int64_t ArrowArrayViewGetIntUnsafe(const struct ArrowArrayView *array_view, int64_t i)#

Get an element in an ArrowArrayView as an integer.

This function does not check for null values, that values are actually integers, or that values are within a valid range for an int64.

static inline uint64_t ArrowArrayViewGetUIntUnsafe(const struct ArrowArrayView *array_view, int64_t i)#

Get an element in an ArrowArrayView as an unsigned integer.

This function does not check for null values, that values are actually integers, or that values are within a valid range for a uint64.

static inline double ArrowArrayViewGetDoubleUnsafe(const struct ArrowArrayView *array_view, int64_t i)#

Get an element in an ArrowArrayView as a double.

This function does not check for null values, or that values are within a valid range for a double.

static inline struct ArrowStringView ArrowArrayViewGetStringUnsafe(const struct ArrowArrayView *array_view, int64_t i)#

Get an element in an ArrowArrayView as an ArrowStringView.

This function does not check for null values.

static inline struct ArrowBufferView ArrowArrayViewGetBytesUnsafe(const struct ArrowArrayView *array_view, int64_t i)#

Get an element in an ArrowArrayView as an ArrowBufferView.

This function does not check for null values.

static inline void ArrowArrayViewGetDecimalUnsafe(const struct ArrowArrayView *array_view, int64_t i, struct ArrowDecimal *out)#

Get an element in an ArrowArrayView as an ArrowDecimal.

This function does not check for null values. The out parameter must be initialized with ArrowDecimalInit() with the proper parameters for this type before calling this for the first time.

Owning, growable buffers#

group nanoarrow-buffer

Functions

static inline void ArrowBufferInit(struct ArrowBuffer *buffer)#

Initialize an ArrowBuffer.

Initialize a buffer with a NULL, zero-size buffer using the default buffer allocator.

static inline ArrowErrorCode ArrowBufferSetAllocator(struct ArrowBuffer *buffer, struct ArrowBufferAllocator allocator)#

Set a newly-initialized buffer’s allocator.

Returns EINVAL if the buffer has already been allocated.

static inline void ArrowBufferReset(struct ArrowBuffer *buffer)#

Reset an ArrowBuffer.

Releases the buffer using the allocator’s free method if the buffer’s data member is non-null, sets the data member to NULL, and sets the buffer’s size and capacity to 0.

static inline void ArrowBufferMove(struct ArrowBuffer *src, struct ArrowBuffer *dst)#

Move an ArrowBuffer.

Transfers the buffer data and lifecycle management to another address and resets buffer.

static inline ArrowErrorCode ArrowBufferResize(struct ArrowBuffer *buffer, int64_t new_size_bytes, char shrink_to_fit)#

Grow or shrink a buffer to a given size.

When shrinking the size of the buffer, the buffer is only reallocated if shrink_to_fit is non-zero.

static inline ArrowErrorCode ArrowBufferReserve(struct ArrowBuffer *buffer, int64_t additional_size_bytes)#

Ensure a buffer has at least a given additional capacity.

Ensures that the buffer has space to append at least additional_size_bytes, overallocating when required.

static inline void ArrowBufferAppendUnsafe(struct ArrowBuffer *buffer, const void *data, int64_t size_bytes)#

Write data to buffer and increment the buffer size.

This function does not check that buffer has the required capacity

static inline ArrowErrorCode ArrowBufferAppend(struct ArrowBuffer *buffer, const void *data, int64_t size_bytes)#

Write data to buffer and increment the buffer size.

This function writes and ensures that the buffer has the required capacity, possibly by reallocating the buffer. Like ArrowBufferReserve, this will overallocate when reallocation is required.

static inline ArrowErrorCode ArrowBufferAppendFill(struct ArrowBuffer *buffer, uint8_t value, int64_t size_bytes)#

Write fill to buffer and increment the buffer size.

This function writes the specified number of fill bytes and ensures that the buffer has the required capacity,

static inline ArrowErrorCode ArrowBufferAppendInt8(struct ArrowBuffer *buffer, int8_t value)#: Write an 8-bit integer to a buffer.

static inline ArrowErrorCode ArrowBufferAppendUInt8(struct ArrowBuffer *buffer, uint8_t value)#: Write an unsigned 8-bit integer to a buffer.

static inline ArrowErrorCode ArrowBufferAppendInt16(struct ArrowBuffer *buffer, int16_t value)#: Write a 16-bit integer to a buffer.

static inline ArrowErrorCode ArrowBufferAppendUInt16(struct ArrowBuffer *buffer, uint16_t value)#: Write an unsigned 16-bit integer to a buffer.

static inline ArrowErrorCode ArrowBufferAppendInt32(struct ArrowBuffer *buffer, int32_t value)#: Write a 32-bit integer to a buffer.

static inline ArrowErrorCode ArrowBufferAppendUInt32(struct ArrowBuffer *buffer, uint32_t value)#: Write an unsigned 32-bit integer to a buffer.

static inline ArrowErrorCode ArrowBufferAppendInt64(struct ArrowBuffer *buffer, int64_t value)#: Write a 64-bit integer to a buffer.

static inline ArrowErrorCode ArrowBufferAppendUInt64(struct ArrowBuffer *buffer, uint64_t value)#: Write an unsigned 64-bit integer to a buffer.

static inline ArrowErrorCode ArrowBufferAppendDouble(struct ArrowBuffer *buffer, double value)#: Write a double to a buffer.

static inline ArrowErrorCode ArrowBufferAppendFloat(struct ArrowBuffer *buffer, float value)#: Write a float to a buffer.

static inline ArrowErrorCode ArrowBufferAppendStringView(struct ArrowBuffer *buffer, struct ArrowStringView value)#: Write an ArrowStringView to a buffer.

static inline ArrowErrorCode ArrowBufferAppendBufferView(struct ArrowBuffer *buffer, struct ArrowBufferView value)#: Write an ArrowBufferView to a buffer.

Bitmap utilities#

group nanoarrow-bitmap

Functions

static inline int8_t ArrowBitGet(const uint8_t *bits, int64_t i)#: Extract a boolean value from a bitmap.

static inline void ArrowBitSet(uint8_t *bits, int64_t i)#: Set a boolean value to a bitmap to true.

static inline void ArrowBitClear(uint8_t *bits, int64_t i)#: Set a boolean value to a bitmap to false.

static inline void ArrowBitSetTo(uint8_t *bits, int64_t i, uint8_t value)#: Set a boolean value to a bitmap.

static inline void ArrowBitsSetTo(uint8_t *bits, int64_t start_offset, int64_t length, uint8_t bits_are_set)#: Set a boolean value to a range in a bitmap.

static inline int64_t ArrowBitCountSet(const uint8_t *bits, int64_t i_from, int64_t i_to)#: Count true values in a bitmap.

static inline void ArrowBitsUnpackInt8(const uint8_t *bits, int64_t start_offset, int64_t length, int8_t *out)#: Extract int8 boolean values from a range in a bitmap.

static inline void ArrowBitsUnpackInt32(const uint8_t *bits, int64_t start_offset, int64_t length, int32_t *out)#: Extract int32 boolean values from a range in a bitmap.

static inline void ArrowBitmapInit(struct ArrowBitmap *bitmap)#

Initialize an ArrowBitmap.

Initialize the builder’s buffer, empty its cache, and reset the size to zero

static inline void ArrowBitmapMove(struct ArrowBitmap *src, struct ArrowBitmap *dst)#

Move an ArrowBitmap.

Transfers the underlying buffer data and lifecycle management to another address and resets the bitmap.

static inline ArrowErrorCode ArrowBitmapReserve(struct ArrowBitmap *bitmap, int64_t additional_size_bits)#

Ensure a bitmap builder has at least a given additional capacity.

Ensures that the buffer has space to append at least additional_size_bits, overallocating when required.

static inline ArrowErrorCode ArrowBitmapResize(struct ArrowBitmap *bitmap, int64_t new_size_bits, char shrink_to_fit)#

Grow or shrink a bitmap to a given size.

When shrinking the size of the bitmap, the bitmap is only reallocated if shrink_to_fit is non-zero.

static inline ArrowErrorCode ArrowBitmapAppend(struct ArrowBitmap *bitmap, uint8_t bits_are_set, int64_t length)#: Reserve space for and append zero or more of the same boolean value to a bitmap.

static inline void ArrowBitmapAppendUnsafe(struct ArrowBitmap *bitmap, uint8_t bits_are_set, int64_t length)#: Append zero or more of the same boolean value to a bitmap.

static inline void ArrowBitmapAppendInt8Unsafe(struct ArrowBitmap *bitmap, const int8_t *values, int64_t n_values)#

Append boolean values encoded as int8_t to a bitmap.

The values must all be 0 or 1.

static inline void ArrowBitmapAppendInt32Unsafe(struct ArrowBitmap *bitmap, const int32_t *values, int64_t n_values)#

Append boolean values encoded as int32_t to a bitmap.

The values must all be 0 or 1.

static inline void ArrowBitmapReset(struct ArrowBitmap *bitmap)#

Reset a bitmap builder.

Releases any memory held by buffer, empties the cache, and resets the size to zero

Create, read, and modify schema metadata#

group nanoarrow-metadata

Functions

ArrowErrorCode ArrowMetadataReaderInit(struct ArrowMetadataReader *reader, const char *metadata)#: Initialize an ArrowMetadataReader.

ArrowErrorCode ArrowMetadataReaderRead(struct ArrowMetadataReader *reader, struct ArrowStringView *key_out, struct ArrowStringView *value_out)#: Read the next key/value pair from an ArrowMetadataReader.

int64_t ArrowMetadataSizeOf(const char *metadata)#: The number of bytes in in a key/value metadata string.

char ArrowMetadataHasKey(const char *metadata, struct ArrowStringView key)#: Check for a key in schema metadata.

ArrowErrorCode ArrowMetadataGetValue(const char *metadata, struct ArrowStringView key, struct ArrowStringView *value_out)#

Extract a value from schema metadata.

If key does not exist in metadata, value_out is unmodified

ArrowErrorCode ArrowMetadataBuilderInit(struct ArrowBuffer *buffer, const char *metadata)#

Initialize a builder for schema metadata from key/value pairs.

metadata can be an existing metadata string or NULL to initialize an empty metadata string.

ArrowErrorCode ArrowMetadataBuilderAppend(struct ArrowBuffer *buffer, struct ArrowStringView key, struct ArrowStringView value)#: Append a key/value pair to a buffer containing serialized metadata.

ArrowErrorCode ArrowMetadataBuilderSet(struct ArrowBuffer *buffer, struct ArrowStringView key, struct ArrowStringView value)#

Set a key/value pair to a buffer containing serialized metadata.

Ensures that the only entry for key in the metadata is set to value. This function maintains the existing position of (the first instance of) key if present in the data.

ArrowErrorCode ArrowMetadataBuilderRemove(struct ArrowBuffer *buffer, struct ArrowStringView key)#: Remove a key from a buffer containing serialized metadata.

struct ArrowMetadataReader#

#include <nanoarrow.h>

Reader for key/value pairs in schema metadata.

The ArrowMetadataReader does not own any data and is only valid for the lifetime of the underlying metadata pointer.

Public Members

const char *metadata#: A metadata string from a schema->metadata field.

int64_t offset#: The current offset into the metadata string.

int32_t remaining_keys#: The number of remaining keys.

Memory management#

group nanoarrow-malloc

Non-buffer members of a struct ArrowSchema and struct ArrowArray must be allocated using ArrowMalloc() or ArrowRealloc() and freed using ArrowFree() for schemas and arrays allocated here. Buffer members are allocated using an ArrowBufferAllocator.

Functions

void *ArrowMalloc(int64_t size)#: Allocate like malloc()

void *ArrowRealloc(void *ptr, int64_t size)#: Reallocate like realloc()

void ArrowFree(void *ptr)#: Free a pointer allocated using ArrowMalloc() or ArrowRealloc().

struct ArrowBufferAllocator ArrowBufferAllocatorDefault(void)#

Return the default allocator.

The default allocator uses ArrowMalloc(), ArrowRealloc(), and ArrowFree().

struct ArrowBufferAllocator ArrowBufferDeallocator(ArrowBufferDeallocatorCallback, void *private_data)#

Create a custom deallocator.

Creates a buffer allocator with only a free method that can be used to attach a custom deallocator to an ArrowBuffer. This may be used to avoid copying an existing buffer that was not allocated using the infrastructure provided here (e.g., by an R or Python object).

Error handling#

group nanoarrow-errors

Functions generally return an errno-compatible error code; functions that need to communicate more verbose error information accept a pointer to an ArrowError. This can be stack or statically allocated. The content of the message is undefined unless an error code has been returned. If a nanoarrow function is passed a non-null ArrowError pointer, the ArrowError pointed to by the argument will be propagated with a null-terminated error message. It is safe to pass a NULL ArrowError anywhere in the nanoarrow API.

Except where documented, it is generally not safe to continue after a function has returned a non-zero ArrowErrorCode. The NANOARROW_RETURN_NOT_OK and NANOARROW_ASSERT_OK macros are provided to help propagate errors. C++ clients can use the helpers provided in the nanoarrow.hpp header to facilitate using C++ idioms for memory management and error propgagtion.

Functions

NANOARROW_CHECK_PRINTF_ATTRIBUTE int ArrowErrorSet(struct ArrowError *error, const char *fmt, ...)#

Set the contents of an error using printf syntax.

If error is NULL, this function does nothing and returns NANOARROW_OK.

Utility data structures#

group nanoarrow-utils

Functions

const char *ArrowNanoarrowVersion(void)#: Return a version string in the form “major.minor.patch”.

int ArrowNanoarrowVersionInt(void)#: Return an integer that can be used to compare versions sequentially.

void ArrowLayoutInit(struct ArrowLayout *layout, enum ArrowType storage_type)#: Initialize a description of buffer arrangements from a storage type.

static inline struct ArrowStringView ArrowCharView(const char *value)#: Create a string view from a null-terminated string.

ArrowErrorCode ArrowDecimalSetDigits(struct ArrowDecimal *decimal, struct ArrowStringView value)#: Sets the integer value of an ArrowDecimal from a string.

ArrowErrorCode ArrowDecimalAppendDigitsToBuffer(const struct ArrowDecimal *decimal, struct ArrowBuffer *buffer)#: Get the integer value of an ArrowDecimal as string.

static inline uint16_t ArrowFloatToHalfFloat(float value)#: Get the half float value of a float.

static inline float ArrowHalfFloatToFloat(uint16_t value)#: Get the float value of a half float.

static inline int64_t ArrowResolveChunk64(int64_t index, const int64_t *offsets, int64_t lo, int64_t hi)#

Resolve a chunk index from increasing int64_t offsets.

Given a buffer of increasing int64_t offsets that begin with 0 (e.g., offset buffer of a large type, run ends of a chunked array implementation), resolve a value v where lo <= v < hi such that offsets[v] <= index < offsets[v + 1].

Arrow C Data Interface#

Warning

doxygengroup: Cannot find group “nanoarrow-arrow-cdata” in doxygen xml output for project “nanoarrow_c” from directory: ../../src/apidoc/xml