wasm: Cache objects via data hash (XXH3)

Cargo.lock

@@ -3280,7 +3280,7 @@ dependencies = [
 
 [[package]]
 name = "objdiff-cli"
-version = "3.0.0-beta.4"
+version = "3.0.0-beta.5"
 dependencies = [
  "anyhow",
  "argp",
@@ -3303,7 +3303,7 @@ dependencies = [
 
 [[package]]
 name = "objdiff-core"
-version = "3.0.0-beta.4"
+version = "3.0.0-beta.5"
 dependencies = [
  "anyhow",
  "arm-attr",
@@ -3356,7 +3356,7 @@ dependencies = [
 
 [[package]]
 name = "objdiff-gui"
-version = "3.0.0-beta.4"
+version = "3.0.0-beta.5"
 dependencies = [
  "anyhow",
  "cfg-if",
@@ -3392,7 +3392,7 @@ dependencies = [
 
 [[package]]
 name = "objdiff-wasm"
-version = "3.0.0-beta.4"
+version = "3.0.0-beta.5"
 dependencies = [
  "log",
  "objdiff-core",
@@ -3400,6 +3400,7 @@ dependencies = [
  "talc",
  "wit-bindgen",
  "wit-deps",
+ "xxhash-rust",
 ]
 
 [[package]]
@@ -6547,6 +6548,12 @@ version = "0.8.25"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "c5b940ebc25896e71dd073bad2dbaa2abfe97b0a391415e22ad1326d9c54e3c4"
 
+[[package]]
+name = "xxhash-rust"
+version = "0.8.15"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "fdd20c5420375476fbd4394763288da7eb0cc0b8c11deed431a91562af7335d3"
+
 [[package]]
 name = "yaxpeax-arch"
 version = "0.3.2"

Cargo.toml

@@ -14,7 +14,7 @@ strip = "debuginfo"
 codegen-units = 1
 
 [workspace.package]
-version = "3.0.0-beta.4"
+version = "3.0.0-beta.5"
 authors = ["Luke Street <luke@street.dev>"]
 edition = "2024"
 license = "MIT OR Apache-2.0"

objdiff-wasm/Cargo.toml

@@ -25,6 +25,7 @@ std = ["objdiff-core/std"]
 [dependencies]
 log = { version = "0.4", default-features = false }
 regex = { version = "1.11", default-features = false, features = ["unicode-case"] }
+xxhash-rust = { version = "0.8", default-features = false, features = ["xxh3"] }
 
 [dependencies.objdiff-core]
 path = "../objdiff-core"

objdiff-wasm/package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "objdiff-wasm",
-  "version": "3.0.0-beta.4",
+  "version": "3.0.0-beta.5",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
       "name": "objdiff-wasm",
-      "version": "3.0.0-beta.4",
+      "version": "3.0.0-beta.5",
       "license": "MIT OR Apache-2.0",
       "devDependencies": {
         "@biomejs/biome": "^1.9.3",

objdiff-wasm/package.json

@@ -1,6 +1,6 @@
 {
   "name": "objdiff-wasm",
-  "version": "3.0.0-beta.4",
+  "version": "3.0.0-beta.5",
   "description": "A local diffing tool for decompilation projects.",
   "author": {
     "name": "Luke Street",

objdiff-wasm/src/api.rs

@@ -1,6 +1,6 @@
 use alloc::{
     format,
-    rc::Rc,
+    rc::{Rc, Weak},
     str::FromStr,
     string::{String, ToString},
     vec::Vec,
@@ -9,6 +9,7 @@ use core::cell::RefCell;
 
 use objdiff_core::{diff, obj};
 use regex::{Regex, RegexBuilder};
+use xxhash_rust::xxh3::xxh3_64;
 
 use super::logging;
 
@@ -41,8 +42,7 @@ impl Guest for Component {
     fn version() -> String { env!("CARGO_PKG_VERSION").to_string() }
 }
 
-#[repr(transparent)]
+struct ResourceObject(Rc<obj::Object>, u64);
-struct ResourceObject(Rc<obj::Object>);
 
 struct ResourceObjectDiff(Rc<obj::Object>, diff::ObjectDiff);
 
@@ -421,13 +421,49 @@ impl GuestDiffConfig for ResourceDiffConfig {
     }
 }
 
+struct CachedObject(Weak<obj::Object>, u64);
+
+struct ObjectCache(RefCell<Vec<CachedObject>>);
+
+impl ObjectCache {
+    #[inline]
+    const fn new() -> Self { Self(RefCell::new(Vec::new())) }
+}
+
+impl core::ops::Deref for ObjectCache {
+    type Target = RefCell<Vec<CachedObject>>;
+
+    fn deref(&self) -> &Self::Target { &self.0 }
+}
+
+// Assume single-threaded environment
+unsafe impl Sync for ObjectCache {}
+
+static OBJECT_CACHE: ObjectCache = ObjectCache::new();
+
 impl GuestObject for ResourceObject {
     fn parse(data: Vec<u8>, diff_config: DiffConfigBorrow) -> Result<Object, String> {
-        let diff_config = diff_config.get::<ResourceDiffConfig>().0.borrow();
+        let hash = xxh3_64(&data);
-        obj::read::parse(&data, &diff_config)
+        let mut cached = None;
-            .map(|o| Object::new(ResourceObject(Rc::new(o))))
+        OBJECT_CACHE.borrow_mut().retain(|c| {
-            .map_err(|e| e.to_string())
+            if c.0.strong_count() == 0 {
+                return false;
             }
+            if c.1 == hash {
+                cached = c.0.upgrade();
+            }
+            true
+        });
+        if let Some(obj) = cached {
+            return Ok(Object::new(ResourceObject(obj, hash)));
+        }
+        let diff_config = diff_config.get::<ResourceDiffConfig>().0.borrow();
+        let obj = Rc::new(obj::read::parse(&data, &diff_config).map_err(|e| e.to_string())?);
+        OBJECT_CACHE.borrow_mut().push(CachedObject(Rc::downgrade(&obj), hash));
+        Ok(Object::new(ResourceObject(obj, hash)))
+    }
+
+    fn hash(&self) -> u64 { self.1 }
 }
 
 impl GuestObjectDiff for ResourceObjectDiff {

objdiff-wasm/wit/objdiff.wit

@@ -20,6 +20,8 @@ interface diff {
       data: list<u8>,
       config: borrow<diff-config>,
     ) -> result<object, string>;
+
+    hash: func() -> u64;
   }
 
   resource object-diff {