From fc4013a43fdbb5b8dc9de8de6b9f801ed37f4122 Mon Sep 17 00:00:00 2001
From: Kerollmops <clement@meilisearch.com>
Date: Wed, 1 Jul 2020 10:35:07 +0200
Subject: [PATCH] Fix the ARC cache

---
 src/bin/indexer.rs |  7 ++++---
 src/cache.rs       | 42 ++++++------------------------------------
 2 files changed, 10 insertions(+), 39 deletions(-)

diff --git a/src/bin/indexer.rs b/src/bin/indexer.rs
index d3bf30793..cc0604ea4 100644
--- a/src/bin/indexer.rs
+++ b/src/bin/indexer.rs
@@ -2,6 +2,7 @@ use std::collections::hash_map::Entry;
 use std::collections::{HashMap, BTreeSet};
 use std::convert::{TryFrom, TryInto};
 use std::io;
+use std::ops::BitOr;
 use std::path::PathBuf;
 use std::sync::atomic::{AtomicUsize, Ordering};
 
@@ -74,7 +75,7 @@ fn index_csv<R: io::Read>(wtxn: &mut heed::RwTxn, mut rdr: csv::Reader<R>, index
                         None => {
                             let mut ids = index.word_positions.get(wtxn, &word)?.unwrap_or_default();
                             ids.insert(position);
-                            for (word, ids) in word_positions.insert(word.clone(), ids) {
+                            for (word, ids) in word_positions.insert(word.clone(), ids, RoaringBitmap::bitor) {
                                 index.word_positions.put(wtxn, &word, &ids)?;
                             }
                         }
@@ -90,7 +91,7 @@ fn index_csv<R: io::Read>(wtxn: &mut heed::RwTxn, mut rdr: csv::Reader<R>, index
                         None => {
                             let mut ids = index.word_position_docids.get(wtxn, &key)?.unwrap_or_default();
                             ids.insert(position);
-                            for ((word, position), ids) in word_position_docids.insert((word.clone(), position), ids) {
+                            for ((word, position), ids) in word_position_docids.insert((word.clone(), position), ids, RoaringBitmap::bitor) {
                                 let mut key = word.as_bytes().to_vec();
                                 key.extend_from_slice(&position.to_be_bytes());
                                 index.word_position_docids.put(wtxn, &key, &ids)?;
@@ -123,7 +124,7 @@ fn index_csv<R: io::Read>(wtxn: &mut heed::RwTxn, mut rdr: csv::Reader<R>, index
     let iter = index.word_positions.as_polymorph().iter::<_, Str, DecodeIgnore>(wtxn)?;
     for result in iter {
         let (word, ()) = result?;
-        new_words.insert(word.clone());
+        new_words.insert(word);
     }
 
     let new_words_fst = fst::Set::from_iter(new_words)?;
diff --git a/src/cache.rs b/src/cache.rs
index d90aa4294..ddff719fa 100644
--- a/src/cache.rs
+++ b/src/cache.rs
@@ -34,25 +34,6 @@ impl<K: PartialEq> PartialEq for KeyRef<K> {
 
 impl<K: Eq> Eq for KeyRef<K> {}
 
-#[cfg(feature = "nightly")]
-#[doc(hidden)]
-pub auto trait NotKeyRef {}
-
-#[cfg(feature = "nightly")]
-impl<K> !NotKeyRef for KeyRef<K> {}
-
-#[cfg(feature = "nightly")]
-impl<K, D> Borrow<D> for KeyRef<K>
-where
-    K: Borrow<D>,
-    D: NotKeyRef + ?Sized,
-{
-    fn borrow(&self) -> &D {
-        unsafe { &*self.k }.borrow()
-    }
-}
-
-#[cfg(not(feature = "nightly"))]
 impl<K> Borrow<K> for KeyRef<K> {
     fn borrow(&self) -> &K {
         unsafe { &*self.k }
@@ -88,7 +69,7 @@ impl<K, V> LruEntry<K, V> {
     }
 }
 
-/// An LRU Cache
+/// An LRU Cache.
 pub struct LruCache<K, V> {
     map: FastMap8<KeyRef<K>, Box<LruEntry<K, V>>>,
     cap: usize,
@@ -100,13 +81,6 @@ pub struct LruCache<K, V> {
 
 impl<K: Hash + Eq, V> LruCache<K, V> {
     /// Creates a new LRU Cache that holds at most `cap` items.
-    ///
-    /// # Example
-    ///
-    /// ```
-    /// use lru::LruCache;
-    /// let mut cache: LruCache<isize, &str> = LruCache::new(10);
-    /// ```
     pub fn new(cap: usize) -> LruCache<K, V> {
         let mut map = FastMap8::default();
         map.reserve(cap);
@@ -114,13 +88,6 @@ impl<K: Hash + Eq, V> LruCache<K, V> {
     }
 
     /// Creates a new LRU Cache that never automatically evicts items.
-    ///
-    /// # Example
-    ///
-    /// ```
-    /// use lru::LruCache;
-    /// let mut cache: LruCache<isize, &str> = LruCache::unbounded();
-    /// ```
     pub fn unbounded() -> LruCache<K, V> {
         LruCache::construct(usize::MAX, HashMap::default())
     }
@@ -443,13 +410,16 @@ where
         }
     }
 
-    pub fn insert(&mut self, key: K, value: V) -> Vec<(K, V)> {
+    pub fn insert<F>(&mut self, key: K, value: V, mut merge: F) -> Vec<(K, V)>
+    where F: FnMut(V, V) -> V
+    {
         let mut evicted = Vec::new();
         if self.frequent_set.contains_key(&key) {
             evicted.extend(self.frequent_set.insert(key, value));
             return evicted;
         }
-        if self.recent_set.remove(&key).is_some() {
+        if let Some(prev_value) = self.recent_set.remove(&key) {
+            let value = (merge)(prev_value, value);
             evicted.extend(self.frequent_set.insert(key, value));
             return evicted;
         }