| 1 | use std::num::NonZeroUsize; |
| 2 | use std::sync::{Mutex, OnceLock}; |
| 3 | |
| 4 | use lru::LruCache; |
| 5 | use regex::Regex; |
| 6 | |
| 7 | const DEFAULT_USER_REGEX_CACHE_CAPACITY: usize = 64; |
| 8 | |
| 9 | static USER_REGEX_CACHE: OnceLock<UserRegexCache> = OnceLock::new(); |
| 10 | |
| 11 | pub(crate) fn compile_user_regex(pattern: &str) -> Result<Regex, regex::Error> { |
| 12 | user_regex_cache().compile(pattern) |
| 13 | } |
| 14 | |
| 15 | fn user_regex_cache() -> &'static UserRegexCache { |
| 16 | USER_REGEX_CACHE.get_or_init(UserRegexCache::new) |
| 17 | } |
| 18 | |
| 19 | struct UserRegexCache { |
| 20 | inner: Mutex<LruCache<String, Regex>>, |
| 21 | } |
| 22 | |
| 23 | impl UserRegexCache { |
| 24 | fn new() -> Self { |
| 25 | Self::with_capacity( |
| 26 | NonZeroUsize::new(DEFAULT_USER_REGEX_CACHE_CAPACITY).expect("non-zero capacity"), |
| 27 | ) |
| 28 | } |
| 29 | |
| 30 | fn with_capacity(capacity: NonZeroUsize) -> Self { |
| 31 | Self { |
| 32 | inner: Mutex::new(LruCache::new(capacity)), |
| 33 | } |
| 34 | } |
| 35 | |
| 36 | fn compile(&self, pattern: &str) -> Result<Regex, regex::Error> { |
| 37 | let Ok(mut cache) = self.inner.lock() else { |
| 38 | return Regex::new(pattern); |
| 39 | }; |
| 40 | if let Some(regex) = cache.get(pattern) { |
| 41 | return Ok(regex.clone()); |
| 42 | } |
| 43 | |
| 44 | let regex = Regex::new(pattern)?; |
| 45 | cache.put(pattern.to_string(), regex.clone()); |
| 46 | Ok(regex) |
| 47 | } |
| 48 | |
| 49 | #[cfg(test)] |
| 50 | fn len(&self) -> usize { |
| 51 | self.inner.lock().expect("cache lock").len() |
| 52 | } |
| 53 | |
| 54 | #[cfg(test)] |
| 55 | fn contains(&self, pattern: &str) -> bool { |
| 56 | self.inner.lock().expect("cache lock").contains(pattern) |
| 57 | } |
| 58 | } |
| 59 | |
| 60 | #[cfg(test)] |
| 61 | mod tests { |
| 62 | use super::*; |
| 63 | |
| 64 | #[test] |
| 65 | fn repeated_pattern_uses_one_cache_entry() { |
| 66 | let cache = UserRegexCache::with_capacity(NonZeroUsize::new(2).unwrap()); |
| 67 | |
| 68 | let first = cache.compile("alpha|beta").expect("regex compiles"); |
| 69 | let second = cache.compile("alpha|beta").expect("regex cache hit"); |
| 70 | |
| 71 | assert!(first.is_match("alpha")); |
| 72 | assert!(second.is_match("beta")); |
| 73 | assert_eq!(cache.len(), 1); |
| 74 | } |
| 75 | |
| 76 | #[test] |
| 77 | fn capacity_evicts_least_recently_used_pattern() { |
| 78 | let cache = UserRegexCache::with_capacity(NonZeroUsize::new(2).unwrap()); |
| 79 | |
| 80 | cache.compile("one").expect("one compiles"); |
| 81 | cache.compile("two").expect("two compiles"); |
| 82 | cache.compile("one").expect("one is refreshed"); |
| 83 | cache.compile("three").expect("three compiles"); |
| 84 | |
| 85 | assert!(cache.contains("one")); |
| 86 | assert!(!cache.contains("two")); |
| 87 | assert!(cache.contains("three")); |
| 88 | } |
| 89 | |
| 90 | #[test] |
| 91 | fn invalid_pattern_is_not_cached() { |
| 92 | let cache = UserRegexCache::with_capacity(NonZeroUsize::new(2).unwrap()); |
| 93 | |
| 94 | assert!(cache.compile("[").is_err()); |
| 95 | |
| 96 | assert_eq!(cache.len(), 0); |
| 97 | } |
| 98 | } |
| 99 |