Auto merge of rust-lang#16335 - lnicola:salsa-lz4-file-text, r=Veykril

internal: Compress file text using LZ4

I haven't tested properly, but this roughly looks like:

```
1246 MB
    59mb   4899 FileTextQuery

1008 MB
    20mb   4899 CompressedFileTextQuery
   555kb   1790 FileTextQuery
```

We might want to test on something more interesting, like `bevy`.

crates/base-db/Cargo.toml

@@ -12,6 +12,8 @@ rust-version.workspace = true
 doctest = false
 
 [dependencies]
+lz4_flex = { version = "0.11", default-features = false }
+
 la-arena.workspace = true
 salsa.workspace = true
 rustc-hash.workspace = true

crates/base-db/src/change.rs

@@ -7,13 +7,13 @@ use salsa::Durability;
 use triomphe::Arc;
 use vfs::FileId;
 
-use crate::{CrateGraph, SourceDatabaseExt, SourceRoot, SourceRootId};
+use crate::{CrateGraph, SourceDatabaseExt, SourceDatabaseExt2, SourceRoot, SourceRootId};
 
 /// Encapsulate a bunch of raw `.set` calls on the database.
 #[derive(Default)]
 pub struct FileChange {
     pub roots: Option<Vec<SourceRoot>>,
-    pub files_changed: Vec<(FileId, Option<Arc<str>>)>,
+    pub files_changed: Vec<(FileId, Option<String>)>,
     pub crate_graph: Option<CrateGraph>,
 }
 
@@ -42,7 +42,7 @@ impl FileChange {
         self.roots = Some(roots);
     }
 
-    pub fn change_file(&mut self, file_id: FileId, new_text: Option<Arc<str>>) {
+    pub fn change_file(&mut self, file_id: FileId, new_text: Option<String>) {
         self.files_changed.push((file_id, new_text))
     }
 
@@ -68,8 +68,8 @@ impl FileChange {
             let source_root = db.source_root(source_root_id);
             let durability = durability(&source_root);
             // XXX: can't actually remove the file, just reset the text
-            let text = text.unwrap_or_else(|| Arc::from(""));
-            db.set_file_text_with_durability(file_id, text, durability)
+            let text = text.unwrap_or_default();
+            db.set_file_text_with_durability(file_id, &text, durability)
         }
         if let Some(crate_graph) = self.crate_graph {
             db.set_crate_graph_with_durability(Arc::new(crate_graph), Durability::HIGH);

crates/base-db/src/lib.rs

@@ -7,6 +7,7 @@ mod input;
 
 use std::panic;
 
+use salsa::Durability;
 use syntax::{ast, Parse, SourceFile};
 use triomphe::Arc;
 
@@ -42,6 +43,7 @@ pub trait Upcast<T: ?Sized> {
     fn upcast(&self) -> &T;
 }
 
+pub const DEFAULT_FILE_TEXT_LRU_CAP: usize = 16;
 pub const DEFAULT_PARSE_LRU_CAP: usize = 128;
 pub const DEFAULT_BORROWCK_LRU_CAP: usize = 1024;
 
@@ -89,7 +91,10 @@ fn parse(db: &dyn SourceDatabase, file_id: FileId) -> Parse<ast::SourceFile> {
 #[salsa::query_group(SourceDatabaseExtStorage)]
 pub trait SourceDatabaseExt: SourceDatabase {
     #[salsa::input]
+    fn compressed_file_text(&self, file_id: FileId) -> Arc<[u8]>;
+
     fn file_text(&self, file_id: FileId) -> Arc<str>;
+
     /// Path to a file, relative to the root of its source root.
     /// Source root of the file.
     #[salsa::input]
@@ -101,6 +106,44 @@ pub trait SourceDatabaseExt: SourceDatabase {
     fn source_root_crates(&self, id: SourceRootId) -> Arc<[CrateId]>;
 }
 
+fn file_text(db: &dyn SourceDatabaseExt, file_id: FileId) -> Arc<str> {
+    let bytes = db.compressed_file_text(file_id);
+    let bytes =
+        lz4_flex::decompress_size_prepended(&bytes).expect("lz4 decompression should not fail");
+    let text = std::str::from_utf8(&bytes).expect("file contents should be valid UTF-8");
+    Arc::from(text)
+}
+
+pub trait SourceDatabaseExt2 {
+    fn set_file_text(&mut self, file_id: FileId, text: &str) {
+        self.set_file_text_with_durability(file_id, text, Durability::LOW);
+    }
+
+    fn set_file_text_with_durability(
+        &mut self,
+        file_id: FileId,
+        text: &str,
+        durability: Durability,
+    );
+}
+
+impl<Db: ?Sized + SourceDatabaseExt> SourceDatabaseExt2 for Db {
+    fn set_file_text_with_durability(
+        &mut self,
+        file_id: FileId,
+        text: &str,
+        durability: Durability,
+    ) {
+        let bytes = text.as_bytes();
+        let compressed = lz4_flex::compress_prepend_size(bytes);
+        self.set_compressed_file_text_with_durability(
+            file_id,
+            Arc::from(compressed.as_slice()),
+            durability,
+        )
+    }
+}
+
 fn source_root_crates(db: &dyn SourceDatabaseExt, id: SourceRootId) -> Arc<[CrateId]> {
     let graph = db.crate_graph();
     let mut crates = graph

crates/hir-def/src/nameres/tests/incremental.rs

@@ -1,6 +1,5 @@
-use base_db::{SourceDatabase, SourceDatabaseExt};
+use base_db::{SourceDatabase, SourceDatabaseExt2 as _};
 use test_fixture::WithFixture;
-use triomphe::Arc;
 
 use crate::{db::DefDatabase, nameres::tests::TestDB, AdtId, ModuleDefId};
 
@@ -17,7 +16,7 @@ fn check_def_map_is_not_recomputed(ra_fixture_initial: &str, ra_fixture_change:
         });
         assert!(format!("{events:?}").contains("crate_def_map"), "{events:#?}")
     }
-    db.set_file_text(pos.file_id, Arc::from(ra_fixture_change));
+    db.set_file_text(pos.file_id, ra_fixture_change);
 
     {
         let events = db.log_executed(|| {
@@ -267,7 +266,7 @@ fn quux() { 92 }
 m!(Y);
 m!(Z);
 "#;
-    db.set_file_text(pos.file_id, Arc::from(new_text));
+    db.set_file_text(pos.file_id, new_text);
 
     {
         let events = db.log_executed(|| {

crates/hir-expand/src/change.rs

@@ -48,7 +48,7 @@ impl ChangeWithProcMacros {
         }
     }
 
-    pub fn change_file(&mut self, file_id: FileId, new_text: Option<Arc<str>>) {
+    pub fn change_file(&mut self, file_id: FileId, new_text: Option<String>) {
         self.source_change.change_file(file_id, new_text)
     }
 

crates/hir-ty/src/tests.rs

@@ -12,7 +12,7 @@ mod traits;
 
 use std::env;
 
-use base_db::{FileRange, SourceDatabaseExt};
+use base_db::{FileRange, SourceDatabaseExt2 as _};
 use expect_test::Expect;
 use hir_def::{
     body::{Body, BodySourceMap, SyntheticSyntax},
@@ -584,7 +584,7 @@ fn salsa_bug() {
         }
     ";
 
-    db.set_file_text(pos.file_id, Arc::from(new_text));
+    db.set_file_text(pos.file_id, new_text);
 
     let module = db.module_for_file(pos.file_id);
     let crate_def_map = module.def_map(&db);

crates/hir-ty/src/tests/incremental.rs

@@ -1,6 +1,5 @@
-use base_db::SourceDatabaseExt;
+use base_db::SourceDatabaseExt2 as _;
 use test_fixture::WithFixture;
-use triomphe::Arc;
 
 use crate::{db::HirDatabase, test_db::TestDB};
 
@@ -33,7 +32,7 @@ fn foo() -> i32 {
     1
 }";
 
-    db.set_file_text(pos.file_id, Arc::from(new_text));
+    db.set_file_text(pos.file_id, new_text);
 
     {
         let events = db.log_executed(|| {
@@ -85,7 +84,7 @@ fn baz() -> i32 {
 }
 ";
 
-    db.set_file_text(pos.file_id, Arc::from(new_text));
+    db.set_file_text(pos.file_id, new_text);
 
     {
         let events = db.log_executed(|| {

crates/ide-db/src/apply_change.rs

@@ -205,6 +205,7 @@ impl RootDatabase {
 
             // SourceDatabaseExt
             base_db::FileTextQuery
+            base_db::CompressedFileTextQuery
             base_db::FileSourceRootQuery
             base_db::SourceRootQuery
             base_db::SourceRootCratesQuery

crates/ide-db/src/lib.rs

@@ -51,6 +51,7 @@ use std::{fmt, mem::ManuallyDrop};
 use base_db::{
     salsa::{self, Durability},
     AnchoredPath, CrateId, FileId, FileLoader, FileLoaderDelegate, SourceDatabase, Upcast,
+    DEFAULT_FILE_TEXT_LRU_CAP,
 };
 use hir::db::{DefDatabase, ExpandDatabase, HirDatabase};
 use triomphe::Arc;
@@ -157,6 +158,7 @@ impl RootDatabase {
 
     pub fn update_base_query_lru_capacities(&mut self, lru_capacity: Option<usize>) {
         let lru_capacity = lru_capacity.unwrap_or(base_db::DEFAULT_PARSE_LRU_CAP);
+        base_db::FileTextQuery.in_db_mut(self).set_lru_capacity(DEFAULT_FILE_TEXT_LRU_CAP);
         base_db::ParseQuery.in_db_mut(self).set_lru_capacity(lru_capacity);
         // macro expansions are usually rather small, so we can afford to keep more of them alive
         hir::db::ParseMacroExpansionQuery.in_db_mut(self).set_lru_capacity(4 * lru_capacity);
@@ -166,6 +168,7 @@ impl RootDatabase {
     pub fn update_lru_capacities(&mut self, lru_capacities: &FxHashMap<Box<str>, usize>) {
         use hir::db as hir_db;
 
+        base_db::FileTextQuery.in_db_mut(self).set_lru_capacity(DEFAULT_FILE_TEXT_LRU_CAP);
         base_db::ParseQuery.in_db_mut(self).set_lru_capacity(
             lru_capacities
                 .get(stringify!(ParseQuery))
@@ -199,7 +202,7 @@ impl RootDatabase {
             // base_db::ProcMacrosQuery
 
             // SourceDatabaseExt
-            // base_db::FileTextQuery
+            base_db::FileTextQuery
             // base_db::FileSourceRootQuery
             // base_db::SourceRootQuery
             base_db::SourceRootCratesQuery

crates/ide/src/lib.rs

@@ -259,7 +259,7 @@ impl Analysis {
             false,
             CrateOrigin::Local { repo: None, name: None },
         );
-        change.change_file(file_id, Some(Arc::from(text)));
+        change.change_file(file_id, Some(text));
         change.set_crate_graph(crate_graph);
         change.set_target_data_layouts(vec![Err("fixture has no layout".into())]);
         change.set_toolchains(vec![None]);

crates/load-cargo/src/lib.rs

@@ -361,8 +361,8 @@ fn load_crate_graph(
     let changes = vfs.take_changes();
     for file in changes {
         if let vfs::Change::Create(v) | vfs::Change::Modify(v) = file.change {
-            if let Ok(text) = std::str::from_utf8(&v) {
-                analysis_change.change_file(file.file_id, Some(text.into()))
+            if let Ok(text) = String::from_utf8(v) {
+                analysis_change.change_file(file.file_id, Some(text))
             }
         }
     }

crates/rust-analyzer/src/cli/rustc_tests.rs

@@ -134,7 +134,7 @@ impl Tester {
         let should_have_no_error = text.contains("// check-pass")
             || text.contains("// build-pass")
             || text.contains("// run-pass");
-        change.change_file(self.root_file, Some(Arc::from(text)));
+        change.change_file(self.root_file, Some(text));
         self.host.apply_change(change);
         let diagnostic_config = DiagnosticsConfig::test_sample();
 

crates/rust-analyzer/src/global_state.rs

@@ -330,7 +330,7 @@ impl GlobalState {
                         // FIXME: Consider doing normalization in the `vfs` instead? That allows
                         // getting rid of some locking
                         let (text, line_endings) = LineEndings::normalize(text);
-                        (Arc::from(text), line_endings)
+                        (text, line_endings)
                     })
                 } else {
                     None

crates/rust-analyzer/src/integrated_benchmarks.rs

@@ -20,7 +20,6 @@ use ide_db::{
 };
 use project_model::CargoConfig;
 use test_utils::project_root;
-use triomphe::Arc;
 use vfs::{AbsPathBuf, VfsPath};
 
 use load_cargo::{load_workspace_at, LoadCargoConfig, ProcMacroServerChoice};
@@ -70,7 +69,7 @@ fn integrated_highlighting_benchmark() {
         let mut text = host.analysis().file_text(file_id).unwrap().to_string();
         text.push_str("\npub fn _dummy() {}\n");
         let mut change = ChangeWithProcMacros::new();
-        change.change_file(file_id, Some(Arc::from(text)));
+        change.change_file(file_id, Some(text));
         host.apply_change(change);
     }
 
@@ -125,7 +124,7 @@ fn integrated_completion_benchmark() {
             patch(&mut text, "db.struct_data(self.id)", "sel;\ndb.struct_data(self.id)")
                 + "sel".len();
         let mut change = ChangeWithProcMacros::new();
-        change.change_file(file_id, Some(Arc::from(text)));
+        change.change_file(file_id, Some(text));
         host.apply_change(change);
         completion_offset
     };
@@ -168,7 +167,7 @@ fn integrated_completion_benchmark() {
             patch(&mut text, "sel;\ndb.struct_data(self.id)", ";sel;\ndb.struct_data(self.id)")
                 + ";sel".len();
         let mut change = ChangeWithProcMacros::new();
-        change.change_file(file_id, Some(Arc::from(text)));
+        change.change_file(file_id, Some(text));
         host.apply_change(change);
         completion_offset
     };
@@ -210,7 +209,7 @@ fn integrated_completion_benchmark() {
             patch(&mut text, "sel;\ndb.struct_data(self.id)", "self.;\ndb.struct_data(self.id)")
                 + "self.".len();
         let mut change = ChangeWithProcMacros::new();
-        change.change_file(file_id, Some(Arc::from(text)));
+        change.change_file(file_id, Some(text));
         host.apply_change(change);
         completion_offset
     };
@@ -307,7 +306,7 @@ fn integrated_diagnostics_benchmark() {
         let mut text = host.analysis().file_text(file_id).unwrap().to_string();
         patch(&mut text, "db.struct_data(self.id)", "();\ndb.struct_data(self.id)");
         let mut change = ChangeWithProcMacros::new();
-        change.change_file(file_id, Some(Arc::from(text)));
+        change.change_file(file_id, Some(text));
         host.apply_change(change);
     };