lance_encoding/constants.rs
1// SPDX-License-Identifier: Apache-2.0
2// SPDX-FileCopyrightText: Copyright The Lance Authors
3
4//! Constants for Lance encoding metadata keys
5//!
6//! These constants define the metadata keys used in Arrow field metadata
7//! to configure various encoding behaviors in Lance.
8
9// Compression-related metadata keys
10/// Metadata key for specifying compression scheme (e.g., "lz4", "zstd", "none")
11pub const COMPRESSION_META_KEY: &str = "lance-encoding:compression";
12/// Metadata key for specifying compression level (applies to schemes that support levels)
13pub const COMPRESSION_LEVEL_META_KEY: &str = "lance-encoding:compression-level";
14/// Metadata key for specifying RLE (Run-Length Encoding) threshold
15pub const RLE_THRESHOLD_META_KEY: &str = "lance-encoding:rle-threshold";
16/// Metadata key for specifying minichunk size
17pub const MINICHUNK_SIZE_META_KEY: &str = "lance-encoding:minichunk-size";
18
19// Dictionary encoding metadata keys
20/// Metadata key for specifying dictionary encoding threshold divisor
21/// Set to a large value to discourage dictionary encoding
22/// Set to a small value to encourage dictionary encoding
23pub const DICT_DIVISOR_META_KEY: &str = "lance-encoding:dict-divisor";
24/// Metadata key for dictionary encoding size ratio threshold (0.0-1.0]
25/// If estimated_dict_size/raw_size < ratio, use dictionary encoding.
26/// Example: 0.8 means use dict if encoded size < 80% of raw size
27/// Default: 0.8
28pub const DICT_SIZE_RATIO_META_KEY: &str = "lance-encoding:dict-size-ratio";
29
30// NOTE: BLOB_META_KEY is defined in lance-core to avoid circular dependency
31
32// Packed struct encoding metadata keys
33/// Legacy metadata key for packed struct encoding (deprecated)
34pub const PACKED_STRUCT_LEGACY_META_KEY: &str = "packed";
35/// Metadata key for packed struct encoding
36pub const PACKED_STRUCT_META_KEY: &str = "lance-encoding:packed";
37
38// Structural encoding metadata keys
39/// Metadata key for specifying structural encoding type
40pub const STRUCTURAL_ENCODING_META_KEY: &str = "lance-encoding:structural-encoding";
41/// Value for miniblock structural encoding
42pub const STRUCTURAL_ENCODING_MINIBLOCK: &str = "miniblock";
43/// Value for fullzip structural encoding
44pub const STRUCTURAL_ENCODING_FULLZIP: &str = "fullzip";
45
46// Byte stream split metadata keys
47/// Metadata key for byte stream split encoding configuration
48pub const BSS_META_KEY: &str = "lance-encoding:bss";
49/// Default BSS mode
50pub const DEFAULT_BSS_MODE: &str = "auto";