Functions
uint64_t	zigzag_encode (int64_t n)
	Zigzag-encode a signed 64-bit integer to an unsigned representation.

uint32_t	zigzag_encode32 (int32_t n)
	Zigzag-encode a signed 32-bit integer to an unsigned representation.

int64_t	zigzag_decode (uint64_t v)
	Zigzag-decode an unsigned 64-bit integer back to its signed representation.

int32_t	zigzag_decode32 (uint32_t v)
	Zigzag-decode an unsigned 32-bit integer back to its signed representation.

size_t	encode_uvarint (std::vector< uint8_t > &buf, uint64_t value)
	Encode an unsigned varint (LEB128) into a byte buffer.

uint64_t	decode_uvarint (const uint8_t *data, size_t &pos, size_t size)
	Decode an unsigned varint (LEB128) from a byte buffer.

int	bit_width_for (uint64_t value)
	Compute the minimum number of bits required to represent an unsigned value.

void	bit_pack_values (std::vector< uint8_t > &out, const uint64_t *values, size_t count, int bit_width)
	Bit-pack an arbitrary number of values at a given bit width.

void	bit_unpack_values (const uint8_t src, uint64_t values, size_t count, int bit_width)
	Unpack an arbitrary number of values at a given bit width from packed data.

std::vector< uint8_t >	encode_int64 (const int64_t *values, size_t count)
	Encode int64 values using the DELTA_BINARY_PACKED algorithm.

std::vector< uint8_t >	encode_int32 (const int32_t *values, size_t count)
	Encode int32 values using the DELTA_BINARY_PACKED algorithm.

std::vector< int64_t >	decode_int64 (const uint8_t *data, size_t size, size_t num_values)
	Decode DELTA_BINARY_PACKED data back to int64 values.

std::vector< int32_t >	decode_int32 (const uint8_t *data, size_t size, size_t num_values)
	Decode DELTA_BINARY_PACKED data back to int32 values.

Variables
constexpr size_t	DEFAULT_BLOCK_SIZE = 128
	Default number of delta values per block (must be a multiple of 128).

constexpr size_t	DEFAULT_MINIBLOCK_COUNT = 4
	Default number of miniblocks within each block.

constexpr size_t	VALUES_PER_MINIBLOCK = DEFAULT_BLOCK_SIZE / DEFAULT_MINIBLOCK_COUNT
	Number of delta values per miniblock (DEFAULT_BLOCK_SIZE / DEFAULT_MINIBLOCK_COUNT).

Function Documentation

◆ bit_pack_values()

void signet::forge::delta::bit_pack_values	(	std::vector< uint8_t > &	out,
		const uint64_t *	values,
		size_t	count,
		int	bit_width
	)

inline

Bit-pack an arbitrary number of values at a given bit width.

Appends ceil(count * bit_width / 8) bytes to out, packing each value LSB-first. Typically used to pack one miniblock of VALUES_PER_MINIBLOCK (32) delta-adjusted values. If bit_width is 0, no bytes are emitted.

Parameters

out	Output byte buffer to append packed data to.
values	Pointer to `count` unsigned values to pack.
count	Number of values (typically 32 for a miniblock).
bit_width	Bits per value (0–64).

See also: bit_unpack_values

Definition at line 204 of file delta.hpp.

◆ bit_unpack_values()

void signet::forge::delta::bit_unpack_values	(	const uint8_t *	src,
		uint64_t *	values,
		size_t	count,
		int	bit_width
	)

inline

Unpack an arbitrary number of values at a given bit width from packed data.

Reverses bit_pack_values(). Reads ceil(count * bit_width / 8) bytes from src and unpacks count values. If bit_width is 0, all output values are set to zero.

Parameters

src	Pointer to packed byte data (at least ceil(count*bit_width/8) bytes).
values	Output array for `count` unpacked unsigned values.
count	Number of values to unpack.
bit_width	Bits per value (0–64).

See also: bit_pack_values

Definition at line 248 of file delta.hpp.

◆ bit_width_for()

int signet::forge::delta::bit_width_for ( uint64_t value )

inline

Compute the minimum number of bits required to represent an unsigned value.

Equivalent to ceil(log2(value + 1)). Returns 0 for value == 0.

Parameters

value The unsigned integer whose bit width to compute.

Returns: Minimum bits needed (0–64).

Definition at line 179 of file delta.hpp.

◆ decode_int32()

std::vector< int32_t > signet::forge::delta::decode_int32	(	const uint8_t *	data,
		size_t	size,
		size_t	num_values
	)

inline

Decode DELTA_BINARY_PACKED data back to int32 values.

Convenience overload that decodes via decode_int64() and narrows the results to int32. Values exceeding the int32 range are truncated.

Parameters

data	Pointer to the encoded DELTA_BINARY_PACKED payload.
size	Size of the encoded data in bytes.
num_values	Number of values to decode (from the Parquet page header).

Returns: Decoded int32 values.

See also: encode_int32, decode_int64

Definition at line 564 of file delta.hpp.

◆ decode_int64()

std::vector< int64_t > signet::forge::delta::decode_int64	(	const uint8_t *	data,
		size_t	size,
		size_t	num_values
	)

inline

Decode DELTA_BINARY_PACKED data back to int64 values.

Parses the block header (block_size, miniblock_count, total_count, first_value), then iterates through blocks and miniblocks, unpacking bit-packed adjusted deltas and reconstructing original values. Includes overflow protection: returns a partial result on integer overflow or corrupt bit widths.

Parameters

data	Pointer to the encoded DELTA_BINARY_PACKED payload.
size	Size of the encoded data in bytes.
num_values	Number of values to decode (from the Parquet page header).

Returns: Decoded int64 values (may be fewer than num_values on truncated or corrupt input).

Note: The total_value_count in the header is ignored; the caller-supplied num_values is authoritative.

See also: encode_int64, decode_int32

Definition at line 438 of file delta.hpp.

◆ decode_uvarint()

uint64_t signet::forge::delta::decode_uvarint	(	const uint8_t *	data,
		size_t &	pos,
		size_t	size
	)

inline

Decode an unsigned varint (LEB128) from a byte buffer.

Reads a variable-length encoded unsigned integer starting at data[pos]. Advances pos past the consumed bytes. Returns 0 if the buffer is exhausted or the shift exceeds 63 bits (overflow protection).

Parameters

data	Pointer to the encoded byte stream.
pos	Current read position (updated on return).
size	Total size of the byte stream.

Returns: The decoded unsigned integer, or 0 on failure.

See also: encode_uvarint

Definition at line 152 of file delta.hpp.

◆ encode_int32()

std::vector< uint8_t > signet::forge::delta::encode_int32	(	const int32_t *	values,
		size_t	count
	)

inline

Encode int32 values using the DELTA_BINARY_PACKED algorithm.

Convenience overload that widens int32 values to int64 and delegates to encode_int64(). The encoded wire format is identical.

Parameters

values	Pointer to the input int32 values.
count	Number of values to encode.

Returns: Encoded byte buffer containing the DELTA_BINARY_PACKED payload.

See also: decode_int32, encode_int64

Definition at line 408 of file delta.hpp.

◆ encode_int64()

std::vector< uint8_t > signet::forge::delta::encode_int64	(	const int64_t *	values,
		size_t	count
	)

inline

Encode int64 values using the DELTA_BINARY_PACKED algorithm.

Computes successive deltas, partitions them into blocks of DEFAULT_BLOCK_SIZE, each subdivided into DEFAULT_MINIBLOCK_COUNT miniblocks, and bit-packs the adjusted deltas (delta - min_delta) per miniblock. Achieves excellent compression on sorted or near-monotonic sequences (timestamps, IDs).

Parameters

values	Pointer to the input int64 values.
count	Number of values to encode.

Returns: Encoded byte buffer containing the DELTA_BINARY_PACKED payload.

Note: For count == 0, returns a valid header with total_count = 0.

See also: decode_int64, encode_int32

Definition at line 298 of file delta.hpp.

◆ encode_uvarint()

size_t signet::forge::delta::encode_uvarint	(	std::vector< uint8_t > &	buf,
		uint64_t	value
	)

inline

Encode an unsigned varint (LEB128) into a byte buffer.

Appends the variable-length encoding of value to buf, using the same unsigned LEB128 format as the Thrift compact protocol. Each byte uses 7 data bits and 1 continuation bit (MSB).

Parameters

buf	Output byte buffer to append to.
value	The unsigned integer to encode.

Returns: Number of bytes written (1–10).

See also: decode_uvarint

Definition at line 131 of file delta.hpp.

◆ zigzag_decode()

int64_t signet::forge::delta::zigzag_decode ( uint64_t v )

inline

Zigzag-decode an unsigned 64-bit integer back to its signed representation.

Reverses zigzag_encode(): (v >> 1) ^ (~(v & 1) + 1).

Parameters

v	The zigzag-encoded unsigned 64-bit value.

Returns: The original signed 64-bit integer.

See also: zigzag_encode

Definition at line 102 of file delta.hpp.

◆ zigzag_decode32()

int32_t signet::forge::delta::zigzag_decode32 ( uint32_t v )

inline

Zigzag-decode an unsigned 32-bit integer back to its signed representation.

32-bit variant of zigzag_decode(). Uses (v >> 1) ^ (~(v & 1) + 1).

Parameters

v	The zigzag-encoded unsigned 32-bit value.

Returns: The original signed 32-bit integer.

See also: zigzag_encode32

Definition at line 113 of file delta.hpp.

◆ zigzag_encode()

uint64_t signet::forge::delta::zigzag_encode ( int64_t n )

inline

Zigzag-encode a signed 64-bit integer to an unsigned representation.

Maps signed integers to unsigned using the formula (n << 1) ^ (n >> 63), so that small-magnitude values (positive or negative) map to small unsigned values. This is critical for varint efficiency.

Parameters

n	The signed 64-bit integer to encode.

Returns: The zigzag-encoded unsigned 64-bit value.

See also: zigzag_decode

Definition at line 79 of file delta.hpp.

◆ zigzag_encode32()

uint32_t signet::forge::delta::zigzag_encode32 ( int32_t n )

inline

Zigzag-encode a signed 32-bit integer to an unsigned representation.

32-bit variant of zigzag_encode(). Uses the formula (n << 1) ^ (n >> 31).

Parameters

n	The signed 32-bit integer to encode.

Returns: The zigzag-encoded unsigned 32-bit value.

See also: zigzag_decode32

Definition at line 91 of file delta.hpp.

Variable Documentation

◆ DEFAULT_BLOCK_SIZE

constexpr size_t signet::forge::delta::DEFAULT_BLOCK_SIZE = 128

inlineconstexpr

Default number of delta values per block (must be a multiple of 128).

Definition at line 58 of file delta.hpp.

◆ DEFAULT_MINIBLOCK_COUNT

constexpr size_t signet::forge::delta::DEFAULT_MINIBLOCK_COUNT = 4

inlineconstexpr

Default number of miniblocks within each block.

Definition at line 61 of file delta.hpp.

◆ VALUES_PER_MINIBLOCK

constexpr size_t signet::forge::delta::VALUES_PER_MINIBLOCK = DEFAULT_BLOCK_SIZE / DEFAULT_MINIBLOCK_COUNT

inlineconstexpr

Number of delta values per miniblock (DEFAULT_BLOCK_SIZE / DEFAULT_MINIBLOCK_COUNT).

Definition at line 64 of file delta.hpp.

Functions

Variables

Function Documentation

◆ bit_pack_values()

◆ bit_unpack_values()

◆ bit_width_for()

◆ decode_int32()

◆ decode_int64()

◆ decode_uvarint()

◆ encode_int32()

◆ encode_int64()

◆ encode_uvarint()

◆ zigzag_decode()

◆ zigzag_decode32()

◆ zigzag_encode()

◆ zigzag_encode32()

Variable Documentation

◆ DEFAULT_BLOCK_SIZE

◆ DEFAULT_MINIBLOCK_COUNT

◆ VALUES_PER_MINIBLOCK