|
| 1 | +// Copyright (c) 2024-present, fjall-rs |
| 2 | +// This source code is licensed under both the Apache 2.0 and MIT License |
| 3 | +// (found in the LICENSE-* files in the repository) |
| 4 | + |
| 5 | +use std::sync::Arc; |
| 6 | + |
| 7 | +/// Trait for extracting prefixes from keys for prefix bloom filters. |
| 8 | +/// |
| 9 | +/// A prefix extractor allows the bloom filter to index prefixes of keys |
| 10 | +/// instead of (or in addition to) the full keys. This enables efficient |
| 11 | +/// filtering for prefix-based queries. |
| 12 | +/// |
| 13 | +/// # Examples |
| 14 | +/// |
| 15 | +/// ## Simple fixed-length prefix: |
| 16 | +/// ``` |
| 17 | +/// use lsm_tree::prefix::PrefixExtractor; |
| 18 | +/// |
| 19 | +/// struct FixedPrefixExtractor(usize); |
| 20 | +/// |
| 21 | +/// impl PrefixExtractor for FixedPrefixExtractor { |
| 22 | +/// fn extract<'a>(&self, key: &'a [u8]) -> Box<dyn Iterator<Item = &'a [u8]> + 'a> { |
| 23 | +/// Box::new(std::iter::once(key.get(0..self.0).unwrap_or(key))) |
| 24 | +/// } |
| 25 | +/// |
| 26 | +/// fn name(&self) -> &str { |
| 27 | +/// "fixed_prefix" |
| 28 | +/// } |
| 29 | +/// } |
| 30 | +/// ``` |
| 31 | +/// |
| 32 | +/// ## Segmented prefixes (e.g., account_id#user_id): |
| 33 | +/// ``` |
| 34 | +/// use lsm_tree::prefix::PrefixExtractor; |
| 35 | +/// |
| 36 | +/// struct SegmentedPrefixExtractor; |
| 37 | +/// |
| 38 | +/// impl PrefixExtractor for SegmentedPrefixExtractor { |
| 39 | +/// fn extract<'a>(&self, key: &'a [u8]) -> Box<dyn Iterator<Item = &'a [u8]> + 'a> { |
| 40 | +/// let mut prefixes = vec![]; |
| 41 | +/// let mut end = 0; |
| 42 | +/// for (i, &byte) in key.iter().enumerate() { |
| 43 | +/// if byte == b'#' { |
| 44 | +/// prefixes.push(&key[0..i]); |
| 45 | +/// end = i; |
| 46 | +/// } |
| 47 | +/// } |
| 48 | +/// if end < key.len() { |
| 49 | +/// prefixes.push(key); |
| 50 | +/// } |
| 51 | +/// Box::new(prefixes.into_iter()) |
| 52 | +/// } |
| 53 | +/// |
| 54 | +/// fn name(&self) -> &str { |
| 55 | +/// "segmented_prefix" |
| 56 | +/// } |
| 57 | +/// } |
| 58 | +/// ``` |
| 59 | +pub trait PrefixExtractor: Send + Sync { |
| 60 | + /// Extracts zero or more prefixes from a key. |
| 61 | + /// |
| 62 | + /// All prefixes will be added to the bloom filter during segment construction. |
| 63 | + /// |
| 64 | + /// An empty iterator means the key is "out of domain" and won't be added to the bloom filter. |
| 65 | + fn extract<'a>(&self, key: &'a [u8]) -> Box<dyn Iterator<Item = &'a [u8]> + 'a>; |
| 66 | + |
| 67 | + /// Returns a unique name for this prefix extractor. |
| 68 | + fn name(&self) -> &str; |
| 69 | +} |
| 70 | + |
| 71 | +/// A prefix extractor that returns the full key. |
| 72 | +/// |
| 73 | +/// This is the default behavior if no prefix extractor is specified. |
| 74 | +pub struct FullKeyExtractor; |
| 75 | + |
| 76 | +impl PrefixExtractor for FullKeyExtractor { |
| 77 | + fn extract<'a>(&self, key: &'a [u8]) -> Box<dyn Iterator<Item = &'a [u8]> + 'a> { |
| 78 | + Box::new(std::iter::once(key)) |
| 79 | + } |
| 80 | + |
| 81 | + fn name(&self) -> &str { |
| 82 | + "full_key" |
| 83 | + } |
| 84 | +} |
| 85 | + |
| 86 | +/// A prefix extractor that returns a fixed-length prefix. |
| 87 | +/// |
| 88 | +/// If the key is shorter than the prefix length, returns the full key. |
| 89 | +pub struct FixedPrefixExtractor { |
| 90 | + length: usize, |
| 91 | +} |
| 92 | + |
| 93 | +impl FixedPrefixExtractor { |
| 94 | + /// Creates a new fixed-length prefix extractor. |
| 95 | + #[must_use] |
| 96 | + pub fn new(length: usize) -> Self { |
| 97 | + Self { length } |
| 98 | + } |
| 99 | +} |
| 100 | + |
| 101 | +impl PrefixExtractor for FixedPrefixExtractor { |
| 102 | + fn extract<'a>(&self, key: &'a [u8]) -> Box<dyn Iterator<Item = &'a [u8]> + 'a> { |
| 103 | + if key.len() <= self.length { |
| 104 | + Box::new(std::iter::once(key)) |
| 105 | + } else { |
| 106 | + Box::new(std::iter::once(&key[0..self.length])) |
| 107 | + } |
| 108 | + } |
| 109 | + |
| 110 | + fn name(&self) -> &str { |
| 111 | + "fixed_prefix" |
| 112 | + } |
| 113 | +} |
| 114 | + |
| 115 | +/// A prefix extractor that requires keys to be at least a certain length. |
| 116 | +/// |
| 117 | +/// Keys shorter than the required length are considered "out of domain" |
| 118 | +/// and won't be added to the bloom filter. This matches RocksDB's behavior. |
| 119 | +pub struct FixedLengthExtractor { |
| 120 | + length: usize, |
| 121 | +} |
| 122 | + |
| 123 | +impl FixedLengthExtractor { |
| 124 | + /// Creates a new fixed-length extractor. |
| 125 | + #[must_use] |
| 126 | + pub fn new(length: usize) -> Self { |
| 127 | + Self { length } |
| 128 | + } |
| 129 | +} |
| 130 | + |
| 131 | +impl PrefixExtractor for FixedLengthExtractor { |
| 132 | + fn extract<'a>(&self, key: &'a [u8]) -> Box<dyn Iterator<Item = &'a [u8]> + 'a> { |
| 133 | + if key.len() < self.length { |
| 134 | + // Key is too short - out of domain |
| 135 | + Box::new(std::iter::empty()) |
| 136 | + } else { |
| 137 | + Box::new(std::iter::once(&key[0..self.length])) |
| 138 | + } |
| 139 | + } |
| 140 | + |
| 141 | + fn name(&self) -> &str { |
| 142 | + "fixed_length" |
| 143 | + } |
| 144 | +} |
| 145 | + |
| 146 | +/// Examples of custom multi-prefix extractors. |
| 147 | +/// |
| 148 | +/// Users can implement their own prefix extractors that return multiple prefixes. |
| 149 | +/// The bloom filter will include all returned prefixes. |
| 150 | +/// |
| 151 | +/// # Example |
| 152 | +/// |
| 153 | +/// ``` |
| 154 | +/// use lsm_tree::prefix::PrefixExtractor; |
| 155 | +/// use std::sync::Arc; |
| 156 | +/// |
| 157 | +/// // Example 1: Hierarchical prefix extractor based on delimiter |
| 158 | +/// // For key "user/123/data" with delimiter '/', generates: |
| 159 | +/// // - "user" |
| 160 | +/// // - "user/123" |
| 161 | +/// // - "user/123/data" (full key) |
| 162 | +/// struct HierarchicalPrefixExtractor { |
| 163 | +/// delimiter: u8, |
| 164 | +/// } |
| 165 | +/// |
| 166 | +/// impl PrefixExtractor for HierarchicalPrefixExtractor { |
| 167 | +/// fn extract<'a>(&self, key: &'a [u8]) -> Box<dyn Iterator<Item = &'a [u8]> + 'a> { |
| 168 | +/// let delimiter = self.delimiter; |
| 169 | +/// let mut prefixes = Vec::new(); |
| 170 | +/// |
| 171 | +/// // Generate all prefixes up to each delimiter |
| 172 | +/// for (i, &byte) in key.iter().enumerate() { |
| 173 | +/// if byte == delimiter { |
| 174 | +/// prefixes.push(&key[0..i]); |
| 175 | +/// } |
| 176 | +/// } |
| 177 | +/// |
| 178 | +/// // Always include the full key |
| 179 | +/// prefixes.push(key); |
| 180 | +/// |
| 181 | +/// Box::new(prefixes.into_iter()) |
| 182 | +/// } |
| 183 | +/// |
| 184 | +/// fn name(&self) -> &str { |
| 185 | +/// "hierarchical_prefix" |
| 186 | +/// } |
| 187 | +/// } |
| 188 | +/// |
| 189 | +/// // Example 2: Domain-based extractor for email-like keys |
| 190 | +/// // For "user@example.com", generates: |
| 191 | +/// // - "example.com" (domain) |
| 192 | +/// // - "user@example.com" (full key) |
| 193 | +/// struct EmailDomainExtractor; |
| 194 | +/// |
| 195 | +/// impl PrefixExtractor for EmailDomainExtractor { |
| 196 | +/// fn extract<'a>(&self, key: &'a [u8]) -> Box<dyn Iterator<Item = &'a [u8]> + 'a> { |
| 197 | +/// if let Ok(key_str) = std::str::from_utf8(key) { |
| 198 | +/// if let Some(at_pos) = key_str.find('@') { |
| 199 | +/// // Return both domain and full email |
| 200 | +/// let domain = &key[at_pos + 1..]; |
| 201 | +/// return Box::new(vec![domain, key].into_iter()); |
| 202 | +/// } |
| 203 | +/// } |
| 204 | +/// // If not an email format, just return the full key |
| 205 | +/// Box::new(std::iter::once(key)) |
| 206 | +/// } |
| 207 | +/// |
| 208 | +/// fn name(&self) -> &str { |
| 209 | +/// "email_domain" |
| 210 | +/// } |
| 211 | +/// } |
| 212 | +/// |
| 213 | +/// // Usage: |
| 214 | +/// # let path = tempfile::tempdir()?; |
| 215 | +/// let tree = lsm_tree::Config::new(path) |
| 216 | +/// .prefix_extractor(Arc::new(HierarchicalPrefixExtractor { delimiter: b'/' })) |
| 217 | +/// .open()?; |
| 218 | +/// # Ok::<(), Box<dyn std::error::Error>>(()) |
| 219 | +/// ``` |
| 220 | +
|
| 221 | +/// Type alias for a shared prefix extractor |
| 222 | +pub type SharedPrefixExtractor = Arc<dyn PrefixExtractor>; |
| 223 | + |
| 224 | +#[cfg(test)] |
| 225 | +mod tests { |
| 226 | + use super::*; |
| 227 | + |
| 228 | + #[test] |
| 229 | + fn test_full_key_extractor() { |
| 230 | + let extractor = FullKeyExtractor; |
| 231 | + let key = b"test_key"; |
| 232 | + let prefixes: Vec<_> = extractor.extract(key).collect(); |
| 233 | + assert_eq!(prefixes.len(), 1); |
| 234 | + assert_eq!(prefixes[0], b"test_key"); |
| 235 | + } |
| 236 | + |
| 237 | + #[test] |
| 238 | + fn test_fixed_prefix_extractor() { |
| 239 | + let extractor = FixedPrefixExtractor::new(5); |
| 240 | + |
| 241 | + // Key longer than prefix |
| 242 | + let key = b"longer_key"; |
| 243 | + let prefixes: Vec<_> = extractor.extract(key).collect(); |
| 244 | + assert_eq!(prefixes.len(), 1); |
| 245 | + assert_eq!(prefixes[0], b"longe"); |
| 246 | + |
| 247 | + // Key shorter than prefix |
| 248 | + let key = b"key"; |
| 249 | + let prefixes: Vec<_> = extractor.extract(key).collect(); |
| 250 | + assert_eq!(prefixes.len(), 1); |
| 251 | + assert_eq!(prefixes[0], b"key"); |
| 252 | + |
| 253 | + // Key exactly prefix length |
| 254 | + let key = b"exact"; |
| 255 | + let prefixes: Vec<_> = extractor.extract(key).collect(); |
| 256 | + assert_eq!(prefixes.len(), 1); |
| 257 | + assert_eq!(prefixes[0], b"exact"); |
| 258 | + } |
| 259 | + |
| 260 | + #[test] |
| 261 | + fn test_empty_key() { |
| 262 | + let full_key = FullKeyExtractor; |
| 263 | + let fixed = FixedPrefixExtractor::new(5); |
| 264 | + |
| 265 | + let key = b""; |
| 266 | + |
| 267 | + let prefixes: Vec<_> = full_key.extract(key).collect(); |
| 268 | + assert_eq!(prefixes.len(), 1); |
| 269 | + assert_eq!(prefixes[0], b""); |
| 270 | + |
| 271 | + let prefixes: Vec<_> = fixed.extract(key).collect(); |
| 272 | + assert_eq!(prefixes.len(), 1); |
| 273 | + assert_eq!(prefixes[0], b""); |
| 274 | + } |
| 275 | + |
| 276 | + #[test] |
| 277 | + fn test_fixed_length_extractor() { |
| 278 | + let extractor = FixedLengthExtractor::new(5); |
| 279 | + |
| 280 | + // Key shorter than required length - out of domain |
| 281 | + let key = b"abc"; |
| 282 | + let prefixes: Vec<_> = extractor.extract(key).collect(); |
| 283 | + assert_eq!(prefixes.len(), 0); // Empty iterator |
| 284 | + |
| 285 | + // Key exactly required length |
| 286 | + let key = b"exact"; |
| 287 | + let prefixes: Vec<_> = extractor.extract(key).collect(); |
| 288 | + assert_eq!(prefixes.len(), 1); |
| 289 | + assert_eq!(prefixes[0], b"exact"); |
| 290 | + |
| 291 | + // Key longer than required length |
| 292 | + let key = b"longer_key"; |
| 293 | + let prefixes: Vec<_> = extractor.extract(key).collect(); |
| 294 | + assert_eq!(prefixes.len(), 1); |
| 295 | + assert_eq!(prefixes[0], b"longe"); |
| 296 | + } |
| 297 | +} |
0 commit comments