Crate kora_doc 

kora-doc

Document layer that turns Kora from a key-value cache engine into a JSON-native document database. Documents are stored in a compact binary packed format, queried via a WHERE expression parser, and indexed through four secondary index types (hash, sorted, array, unique).

Architecture

A JSON document passes through a pipeline of transformations before it reaches storage:

1. Decomposition (decompose) – a JSON object is recursively walked, each leaf field is assigned a numeric FieldId via the IdRegistry, string values are dictionary-encoded when their field cardinality is low, and the result is assembled into a PackedDoc.

2. Packed encoding (packed) – fields are stored in a flat binary buffer with an offset table sorted by field ID, enabling O(log F) single-field reads via binary search.

3. Recomposition (recompose) – the inverse of decomposition, rebuilding a serde_json::Value from packed bytes, dictionary lookups, and registry path resolution. Supports full reconstruction or field-level projection.

Key Modules

Module	Purpose
packed	Binary packed document format and builder
registry	Integer-keyed ID mapping for collections, fields, and docs
dictionary	Dictionary encoding for low-cardinality string values
decompose	JSON to packed document conversion
recompose	Packed document to JSON reconstruction
engine	Collection CRUD, index maintenance, and query execution
expr	WHERE clause recursive-descent parser
index	Hash, sorted, array, and unique secondary indexes
collection	Collection metadata and configuration
key	Binary key encoding for storage and index records

Re-exports

pub use collection::Collection;

pub use collection::CollectionConfig;

pub use collection::CollectionError;

pub use collection::CompressionProfile;

pub use decompose::DecomposeError;

pub use decompose::Decomposer;

pub use dictionary::DictionaryError;

pub use dictionary::StoredValue;

pub use dictionary::ValueDictionary;

pub use dictionary::ValueDictionaryConfig;

pub use engine::CollectionInfo;

pub use engine::DictionaryFieldInfo;

pub use engine::DictionaryInfo;

pub use engine::DocEngine;

pub use engine::DocError;

pub use engine::DocMutation;

pub use engine::InsertResult;

pub use engine::SetResult;

pub use engine::StorageInfo;

pub use expr::Expr;

pub use expr::ExprError;

pub use expr::ExprValue;

pub use index::hash32;

pub use index::CollectionIndexes;

pub use index::IndexConfig;

pub use index::IndexError;

pub use index::IndexType;

pub use key::decode_cdc_event_key;

pub use key::decode_cold_doc_key;

pub use key::decode_collection_key;

pub use key::decode_compound_index_key;

pub use key::decode_doc_key;

pub use key::decode_hashed_bucket_key;

pub use key::decode_sorted_index_key;

pub use key::encode_cdc_event_key;

pub use key::encode_cold_doc_key;

pub use key::encode_collection_key;

pub use key::encode_compound_index_key;

pub use key::encode_doc_key;

pub use key::encode_hashed_bucket_key;

pub use key::encode_sorted_index_key;

pub use key::KeyDecodeError;

pub use key::KeyTag;

pub use packed::FieldValue;

pub use packed::PackedDoc;

pub use packed::PackedDocBuilder;

pub use packed::PackedDocError;

pub use packed::PackedDocIter;

pub use recompose::RecomposeError;

pub use recompose::Recomposer;

pub use registry::CollectionId;

pub use registry::DocId;

pub use registry::FieldId;

pub use registry::IdRegistry;

pub use registry::RegistryError;

pub use registry::RegistrySegment;

pub use registry::RegistrySegmentRef;

Modules

collection

Collection metadata and configuration.

decompose

JSON-to-packed-document decomposition pipeline.

dictionary

Dictionary encoding for low-cardinality string field values.

engine

In-memory document engine providing collection CRUD, secondary indexes, and WHERE-clause query execution.

expr

Recursive-descent parser for WHERE-clause filter expressions.

index

Secondary index data structures for document collections.

key

Binary key encoding and decoding for document and index records.

packed

Compact binary packed document format.

recompose

Packed document to JSON reconstruction pipeline.

registry

Integer-keyed ID registry for collections, fields, and documents.