Add ability to gc node caches

Cargo.toml

@@ -9,7 +9,7 @@ edition = "2018"
 
 [dependencies]
 erasable = "1.2.1"
-rustc-hash = "1.0.1"
+hashbrown = "0.8.0"
 serde = { version = "1.0.89", optional = true, default-features = false }
 slice-dst = "1.4.1"
 smol_str = "0.1.10"

src/green/builder.rs

@@ -1,4 +1,4 @@
-use rustc_hash::FxHashSet;
+use hashbrown::HashMap;
 
 use crate::{
     green::{GreenElement, GreenNode, GreenToken, SyntaxKind},
@@ -7,8 +7,8 @@ use crate::{
 
 #[derive(Default, Debug)]
 pub struct NodeCache {
-    nodes: FxHashSet<GreenNode>,
-    tokens: FxHashSet<GreenToken>,
+    nodes: HashMap<GreenNode, ()>,
+    tokens: HashMap<GreenToken, ()>,
 }
 
 impl NodeCache {
@@ -17,7 +17,7 @@ impl NodeCache {
         I: IntoIterator<Item = GreenElement>,
         I::IntoIter: ExactSizeIterator,
     {
-        let mut node = GreenNode::new(kind, children);
+        let node = GreenNode::new(kind, children);
         // Green nodes are fully immutable, so it's ok to deduplicate them.
         // This is the same optimization that Roslyn does
         // https://github.com/KirillOsenkov/Bliki/wiki/Roslyn-Immutable-Trees
@@ -27,21 +27,33 @@ impl NodeCache {
         // 17% of the memory for green nodes!
         // Future work: make hashing faster by avoiding rehashing of subtrees.
         if node.children().len() <= 3 {
-            match self.nodes.get(&node) {
-                Some(existing) => node = existing.clone(),
-                None => assert!(self.nodes.insert(node.clone())),
-            }
+            self.nodes.raw_entry_mut().from_key(&node).or_insert(node, ()).0.clone()
+        } else {
+            node
         }
-        node
     }
 
     fn token(&mut self, kind: SyntaxKind, text: SmolStr) -> GreenToken {
-        let mut token = GreenToken::new(kind, text);
-        match self.tokens.get(&token) {
-            Some(existing) => token = existing.clone(),
-            None => assert!(self.tokens.insert(token.clone())),
+        let token = GreenToken::new(kind, text);
+        self.tokens.raw_entry_mut().from_key(&token).or_insert(token, ()).0.clone()
+    }
+
+    fn turn_node_gc(&mut self) -> bool {
+        // NB: `drain_filter` is `retain` but with an iterator of the removed elements.
+        // i.e.: elements where the predicate is FALSE are removed and iterated over.
+        self.nodes.drain_filter(|node, ()| node.strong_count() > 1).any(|_| true)
+    }
+
+    fn turn_token_gc(&mut self) -> bool {
+        self.tokens.drain_filter(|token, ()| token.strong_count() > 1).any(|_| true)
+    }
+
+    /// Garbage collect any elements in this cache that are only held by this cache.
+    pub fn gc(&mut self) {
+        while self.turn_node_gc() {
+            continue;
         }
-        token
+        self.turn_token_gc();
     }
 }
 

src/green/node.rs

@@ -68,6 +68,10 @@ impl GreenNode {
         let r: &SliceWithHeader<_, _> = &*self.data;
         r as *const _ as _
     }
+
+    pub(super) fn strong_count(&self) -> usize {
+        Thin::with(&self.data, |node| Arc::strong_count(node))
+    }
 }
 
 #[derive(Debug, Clone)]

src/green/token.rs

@@ -62,6 +62,12 @@ impl GreenToken {
     pub fn text_len(&self) -> TextSize {
         TextSize::try_from(self.text().len()).unwrap()
     }
+
+    pub(super) fn strong_count(&self) -> usize {
+        let ptr = Self::remove_tag(self.ptr);
+        let arc = unsafe { ManuallyDrop::new(Arc::from_raw(ptr.as_ptr())) };
+        Arc::strong_count(&arc)
+    }
 }
 
 impl fmt::Debug for GreenToken {

src/lib.rs

@@ -23,13 +23,13 @@ mod serde_impls;
 // these, as a custom interner might work better, but `SmolStr` is a pretty good
 // default.
 pub use smol_str::SmolStr;
-pub use text_size::{TextRange, TextSize, TextLen};
+pub use text_size::{TextLen, TextRange, TextSize};
 
 pub use crate::{
     api::{
         Language, SyntaxElement, SyntaxElementChildren, SyntaxNode, SyntaxNodeChildren, SyntaxToken,
     },
-    green::{Checkpoint, Children, GreenNode, GreenNodeBuilder, GreenToken, SyntaxKind},
+    green::{Checkpoint, Children, GreenNode, GreenNodeBuilder, GreenToken, NodeCache, SyntaxKind},
     syntax_text::SyntaxText,
     utility_types::{Direction, NodeOrToken, TokenAtOffset, WalkEvent},
 };