lance_index/
lib.rs

1// SPDX-License-Identifier: Apache-2.0
2// SPDX-FileCopyrightText: Copyright The Lance Authors
3
4//! Lance secondary index library
5//!
6//! <section class="warning">
7//! This is internal crate used by <a href="https://github.com/lancedb/lance">the lance project</a>.
8//! <br/>
9//! API stability is not guaranteed.
10//! </section>
11
12use std::{any::Any, sync::Arc};
13
14use crate::frag_reuse::FRAG_REUSE_INDEX_NAME;
15use crate::mem_wal::MEM_WAL_INDEX_NAME;
16use async_trait::async_trait;
17use deepsize::DeepSizeOf;
18use lance_core::{Error, Result};
19use roaring::RoaringBitmap;
20use serde::{Deserialize, Serialize};
21use snafu::location;
22use std::convert::TryFrom;
23
24pub mod frag_reuse;
25pub mod mem_wal;
26pub mod metrics;
27pub mod optimize;
28pub mod prefilter;
29pub mod scalar;
30pub mod traits;
31pub mod vector;
32
33pub use crate::traits::*;
34
35pub const INDEX_FILE_NAME: &str = "index.idx";
36/// The name of the auxiliary index file.
37///
38/// This file is used to store additional information about the index, to improve performance.
39/// - For 'IVF_HNSW' index, it stores the partitioned PQ Storage.
40pub const INDEX_AUXILIARY_FILE_NAME: &str = "auxiliary.idx";
41pub const INDEX_METADATA_SCHEMA_KEY: &str = "lance:index";
42
43// Currently all vector indexes are version 1
44pub const VECTOR_INDEX_VERSION: u32 = 1;
45
46pub mod pb {
47    #![allow(clippy::use_self)]
48    include!(concat!(env!("OUT_DIR"), "/lance.index.pb.rs"));
49}
50
51/// Generic methods common across all types of secondary indices
52///
53#[async_trait]
54pub trait Index: Send + Sync + DeepSizeOf {
55    /// Cast to [Any].
56    fn as_any(&self) -> &dyn Any;
57
58    /// Cast to [Index]
59    fn as_index(self: Arc<Self>) -> Arc<dyn Index>;
60
61    /// Cast to [vector::VectorIndex]
62    fn as_vector_index(self: Arc<Self>) -> Result<Arc<dyn vector::VectorIndex>>;
63
64    /// Retrieve index statistics as a JSON Value
65    fn statistics(&self) -> Result<serde_json::Value>;
66
67    /// Prewarm the index.
68    ///
69    /// This will load the index into memory and cache it.
70    async fn prewarm(&self) -> Result<()>;
71
72    /// Get the type of the index
73    fn index_type(&self) -> IndexType;
74
75    /// Read through the index and determine which fragment ids are covered by the index
76    ///
77    /// This is a kind of slow operation.  It's better to use the fragment_bitmap.  This
78    /// only exists for cases where the fragment_bitmap has become corrupted or missing.
79    async fn calculate_included_frags(&self) -> Result<RoaringBitmap>;
80}
81
82/// Index Type
83#[derive(Debug, PartialEq, Eq, Copy, Hash, Clone, DeepSizeOf)]
84pub enum IndexType {
85    // Preserve 0-100 for simple indices.
86    Scalar = 0, // Legacy scalar index, alias to BTree
87
88    BTree = 1, // BTree
89
90    Bitmap = 2, // Bitmap
91
92    LabelList = 3, // LabelList
93
94    Inverted = 4, // Inverted
95
96    NGram = 5, // NGram
97
98    FragmentReuse = 6,
99
100    MemWal = 7,
101
102    ZoneMap = 8, // ZoneMap
103
104    // 100+ and up for vector index.
105    /// Flat vector index.
106    Vector = 100, // Legacy vector index, alias to IvfPq
107    IvfFlat = 101,
108    IvfSq = 102,
109    IvfPq = 103,
110    IvfHnswSq = 104,
111    IvfHnswPq = 105,
112    IvfHnswFlat = 106,
113}
114
115impl std::fmt::Display for IndexType {
116    fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
117        match self {
118            Self::Scalar | Self::BTree => write!(f, "BTree"),
119            Self::Bitmap => write!(f, "Bitmap"),
120            Self::LabelList => write!(f, "LabelList"),
121            Self::Inverted => write!(f, "Inverted"),
122            Self::NGram => write!(f, "NGram"),
123            Self::FragmentReuse => write!(f, "FragmentReuse"),
124            Self::MemWal => write!(f, "MemWal"),
125            Self::ZoneMap => write!(f, "ZoneMap"),
126            Self::Vector | Self::IvfPq => write!(f, "IVF_PQ"),
127            Self::IvfFlat => write!(f, "IVF_FLAT"),
128            Self::IvfSq => write!(f, "IVF_SQ"),
129            Self::IvfHnswSq => write!(f, "IVF_HNSW_SQ"),
130            Self::IvfHnswPq => write!(f, "IVF_HNSW_PQ"),
131            Self::IvfHnswFlat => write!(f, "IVF_HNSW_FLAT"),
132        }
133    }
134}
135
136impl TryFrom<i32> for IndexType {
137    type Error = Error;
138
139    fn try_from(value: i32) -> Result<Self> {
140        match value {
141            v if v == Self::Scalar as i32 => Ok(Self::Scalar),
142            v if v == Self::BTree as i32 => Ok(Self::BTree),
143            v if v == Self::Bitmap as i32 => Ok(Self::Bitmap),
144            v if v == Self::LabelList as i32 => Ok(Self::LabelList),
145            v if v == Self::NGram as i32 => Ok(Self::NGram),
146            v if v == Self::Inverted as i32 => Ok(Self::Inverted),
147            v if v == Self::FragmentReuse as i32 => Ok(Self::FragmentReuse),
148            v if v == Self::MemWal as i32 => Ok(Self::MemWal),
149            v if v == Self::ZoneMap as i32 => Ok(Self::ZoneMap),
150            v if v == Self::Vector as i32 => Ok(Self::Vector),
151            v if v == Self::IvfFlat as i32 => Ok(Self::IvfFlat),
152            v if v == Self::IvfSq as i32 => Ok(Self::IvfSq),
153            v if v == Self::IvfPq as i32 => Ok(Self::IvfPq),
154            v if v == Self::IvfHnswSq as i32 => Ok(Self::IvfHnswSq),
155            v if v == Self::IvfHnswPq as i32 => Ok(Self::IvfHnswPq),
156            v if v == Self::IvfHnswFlat as i32 => Ok(Self::IvfHnswFlat),
157            _ => Err(Error::InvalidInput {
158                source: format!("the input value {} is not a valid IndexType", value).into(),
159                location: location!(),
160            }),
161        }
162    }
163}
164
165impl IndexType {
166    pub fn is_scalar(&self) -> bool {
167        matches!(
168            self,
169            Self::Scalar
170                | Self::BTree
171                | Self::Bitmap
172                | Self::LabelList
173                | Self::Inverted
174                | Self::NGram
175                | Self::ZoneMap
176        )
177    }
178
179    pub fn is_vector(&self) -> bool {
180        matches!(
181            self,
182            Self::Vector
183                | Self::IvfPq
184                | Self::IvfHnswSq
185                | Self::IvfHnswPq
186                | Self::IvfHnswFlat
187                | Self::IvfFlat
188                | Self::IvfSq
189        )
190    }
191
192    pub fn is_system(&self) -> bool {
193        matches!(self, Self::FragmentReuse | Self::MemWal)
194    }
195}
196
197pub trait IndexParams: Send + Sync {
198    fn as_any(&self) -> &dyn Any;
199
200    fn index_name(&self) -> &str;
201}
202
203#[derive(Serialize, Deserialize, Debug)]
204pub struct IndexMetadata {
205    #[serde(rename = "type")]
206    pub index_type: String,
207    pub distance_type: String,
208}
209
210pub fn is_system_index(index_meta: &lance_table::format::Index) -> bool {
211    index_meta.name == FRAG_REUSE_INDEX_NAME || index_meta.name == MEM_WAL_INDEX_NAME
212}
213
214pub fn infer_system_index_type(index_meta: &lance_table::format::Index) -> Option<IndexType> {
215    if index_meta.name == FRAG_REUSE_INDEX_NAME {
216        Some(IndexType::FragmentReuse)
217    } else if index_meta.name == MEM_WAL_INDEX_NAME {
218        Some(IndexType::MemWal)
219    } else {
220        None
221    }
222}
lance_index/lib.rs

lance_index/
lib.rs
