Crate hexane 

Hexane is a columnar compression library implementing the encoding described in the Automerge Binary Format.

Data is stored in ColumnData<C> where the cursor type C selects the encoding (RLE<T>, delta, boolean, or raw). Values are batched into Slabs — Arc-wrapped byte buffers — held in a SpanTree B-tree for O(log n) positional seek, insert, and splice.

Cursor Types

Type	Item	Encoding
UIntCursor	u64	RLE + unsigned LEB128
IntCursor	i64	RLE + signed LEB128
StrCursor	str	RLE + length-prefixed UTF-8
ByteCursor	[u8]	RLE + length-prefixed bytes
BooleanCursor	bool	Boolean run-length encoding
DeltaCursor	i64	Delta-encoded integers
RawCursor	[u8]	Uncompressed raw bytes

Quick Example

use hexane::{ColumnData, UIntCursor};

let mut col: ColumnData<UIntCursor> = ColumnData::new();
col.splice(0, 0, [1u64, 2, 3, 4, 5]);
assert_eq!(col.to_vec(), vec![Some(1), Some(2), Some(3), Some(4), Some(5)]);

let bytes = col.save();
let col2: ColumnData<UIntCursor> = ColumnData::load(&bytes).unwrap();
assert_eq!(col.to_vec(), col2.to_vec());

See the README for comprehensive usage documentation.

Modules

tree

Structs

Acc

A running accumulator — the cumulative sum of per-item Agg values.

Agg

A per-item aggregate value, used for min/max slab metadata and accumulator steps.

ColAccIter

An iterator adapter over ColumnDataIter that emits the Acc value after each item.

ColGroupItem

A single item from a ColGroupIter, bundling the item with its position and the pre-item accumulator.

ColGroupIter

An iterator adapter over ColumnDataIter that emits ColGroupItem values.

ColumnData

A compressed, mutable column of optional typed values.

ColumnDataIter

An iterator over items in a ColumnData, with rich navigation capabilities.

ColumnDataIterState

Serializable snapshot of a ColumnDataIter position.

CursorIter

A low-level iterator that decodes Runs from a raw byte slice using a ColumnCursor.

Encoder

A streaming encoder that accumulates values and writes them into a compressed slab.

RawReader

A low-level reader for sequential byte access across a multi-slab tree.

RleCursor

A ColumnCursor that decodes run-length encoded data.

Run

A single RLE run: count consecutive repetitions of value.

RunIter

Slab

SlabWeight

SlabWriter

SpanTree

SpliceDel

Enums

PackError

Errors returned when decoding or validating column data.

ReadRawError

Errors returned by RawReader::read_next.

WriteOp

Traits

ColumnCursor

Core trait for cursor types that know how to decode a specific columnar encoding.

EncoderState

HasAcc

Implemented by slab weight / span-weight types that carry a cumulative Acc sum.

HasMinMax

HasPos

Implemented by slab weight / span-weight types that carry a cumulative item count.

MaybePackable

Adapter trait that lets both T and Option<T> (and &T, Option<&T>, etc.) be used as input values when writing to a ColumnData or Encoder.

Packable

Types that can be encoded into and decoded from the hexane columnar format.

SpanWeight

Functions

lebsize

The number of bytes required to encode val as a LEB128 integer

ulebsize

The number of bytes required to encode val as a uLEB128 integer

Type Aliases

BooleanCursor

A ColumnCursor that decodes boolean run-length encoded columns.

ByteCursor

RLE cursor for byte-slice columns ([u8]), with a 128-item slab size.

DeltaCursor

A ColumnCursor that decodes delta-encoded signed 64-bit integer columns.

IntCursor

RLE cursor for signed 64-bit integer columns (i64), with a 64-item slab size.

RawCursor

A ColumnCursor for uncompressed raw byte columns.

SlabTree

StrCursor

RLE cursor for UTF-8 string columns (str), with a 128-item slab size.

UIntCursor

RLE cursor for unsigned 64-bit integer columns (u64), with a 64-item slab size.