@tishlang/tish-format 3.15.1 → 3.15.3

package/crates/tish_core/src/json.rs

@@ -363,6 +363,18 @@ fn parse_key<'a>(input: &'a str, cache: &mut KeyCache) -> Result<(Arc<str>, &'a
     }
 }
 
+/// The four hex digits after `\\u`, consuming them (and counting their bytes); None if they
+/// aren't four hex digits.
+fn hex4(chars: &mut core::iter::Peekable<core::str::Chars<'_>>, byte_count: &mut usize) -> Option<u32> {
+    let mut hex = String::new();
+    for _ in 0..4 {
+        let c = chars.next()?;
+        hex.push(c);
+        *byte_count += c.len_utf8();
+    }
+    u32::from_str_radix(&hex, 16).ok()
+}
+
 /// Decode a JSON string that contains at least one escape (the slow path).
 fn parse_string_escaped(input: &str) -> Result<(Value, &str), String> {
     let input = &input[1..]; // skip opening quote
@@ -404,19 +416,39 @@ fn parse_string_escaped(input: &str) -> Result<(Value, &str), String> {
                         result.push('/');
                         byte_count += 1;
                     }
+                    Some('b') => {
+                        result.push('\u{8}');
+                        byte_count += 1;
+                    }
+                    Some('f') => {
+                        result.push('\u{c}');
+                        byte_count += 1;
+                    }
                     Some('u') => {
                         byte_count += 1;
-                        let mut hex = String::new();
-                        for _ in 0..4 {
-                            if let Some(c) = chars.next() {
-                                hex.push(c);
-                                byte_count += c.len_utf8();
-                            }
-                        }
-                        if let Ok(n) = u32::from_str_radix(&hex, 16) {
-                            if let Some(c) = char::from_u32(n) {
-                                result.push(c);
+                        match hex4(&mut chars, &mut byte_count) {
+                            // A UTF-16 surrogate pair (`\ud83d\ude00`) is one character. A lone
+                            // surrogate has no character of its own, so it reads as U+FFFD.
+                            Some(high @ 0xD800..=0xDBFF) => {
+                                let mut ahead = chars.clone();
+                                let mut ahead_bytes = 0;
+                                let low = if ahead.next() == Some('\\') && ahead.next() == Some('u') {
+                                    hex4(&mut ahead, &mut ahead_bytes)
+                                } else {
+                                    None
+                                };
+                                match low {
+                                    Some(low @ 0xDC00..=0xDFFF) => {
+                                        chars = ahead;
+                                        byte_count += 2 + ahead_bytes;
+                                        let n = 0x10000 + ((high - 0xD800) << 10) + (low - 0xDC00);
+                                        result.push(char::from_u32(n).unwrap_or('\u{FFFD}'));
+                                    }
+                                    _ => result.push('\u{FFFD}'),
+                                }
                             }
+                            Some(n) => result.push(char::from_u32(n).unwrap_or('\u{FFFD}')),
+                            None => {}
                         }
                     }
                     Some(c) => {
@@ -664,3 +696,31 @@ mod tests {
         }
     }
 }
+
+#[cfg(test)]
+mod escape_tests {
+    use super::json_parse;
+    use crate::Value;
+
+    fn text(json: &str) -> String {
+        match json_parse(json).unwrap() {
+            Value::String(s) => s.to_string(),
+            other => panic!("not a string: {other:?}"),
+        }
+    }
+
+    #[test]
+    fn surrogate_pairs_are_one_character() {
+        assert_eq!(text(r#""😀 ok""#), "😀 ok");
+        assert_eq!(text(r#""a😀b""#), "a😀b");
+        assert_eq!(text(r#""é""#), "é");
+    }
+
+    #[test]
+    fn lone_surrogates_and_short_escapes() {
+        assert_eq!(text(r#""\ud83d x""#), "\u{FFFD} x");
+        assert_eq!(text(r#""\ude00""#), "\u{FFFD}");
+        assert_eq!(text(r#""\ud83dA""#), "\u{FFFD}A");
+        assert_eq!(text(r#""\b\f\n""#), "\u{8}\u{c}\n");
+    }
+}

package/crates/tish_core/src/vmref.rs

@@ -15,7 +15,7 @@
 //! | feature `send-values`     | `VmRef<T>`              | `NativeFn`                       | targets                                          |
 //! |---------------------------|-------------------------|----------------------------------|--------------------------------------------------|
 //! | **off** *(default)*       | `Rc<RefCell<T>>`        | `Rc<dyn Fn + 'static>`           | wasm32, wasi, interpreter, cranelift/llvm VMs    |
-//! | **on**                    | `Arc<Mutex<T>>`         | `Arc<dyn Fn + Send + Sync>`      | Rust native with `http` enabled (server workloads) |
+//! | **on**                    | `Arc<RwLock<T>>`        | `Arc<dyn Fn + Send + Sync>`      | Rust native with `http` enabled (server workloads) |
 //!
 //! The *API* is identical in both configurations (`borrow` / `borrow_mut`
 //! / `ptr_eq` / `Clone`), so every existing call site in the workspace
@@ -28,7 +28,7 @@
 //!   bit-identical to the pre-migration baseline.
 //! * **Rust native, non-server**: same — `send-values` only activates
 //!   when something in the dependency graph (usually `http`) needs it.
-//! * **Rust native with server**: `Arc<Mutex<T>>` pays ~3–5 ns per
+//! * **Rust native with server**: `Arc<RwLock<T>>` pays ~3–5 ns per
 //!   `borrow` in the uncontended case (single atomic CAS). On Tish's
 //!   hot paths — roughly 6–12 borrows per request — that's ~30–60 ns of
 //!   overhead. In exchange we get `N×` handler scaling across cores,
@@ -45,7 +45,7 @@
 //! ```
 //!
 //! Returned guard types (`VmReadGuard<'_, T>`, `VmWriteGuard<'_, T>`) are
-//! type aliases that pick `Ref`/`RefMut` or `MutexGuard` depending on the
+//! type aliases that pick `Ref`/`RefMut` or `RwLock` read/write guards depending on the
 //! feature. They both `Deref` (and, for write guards, `DerefMut`) to `T`
 //! just like the underlying types.
 
@@ -113,45 +113,45 @@ mod imp {
 }
 
 // --------------------------------------------------------------------------
-// Thread-safe backing store (opt-in): Arc<Mutex<T>>
+// Thread-safe backing store (opt-in): Arc<RwLock<T>>
 // --------------------------------------------------------------------------
 #[cfg(feature = "send-values")]
 mod imp {
-    use parking_lot::Mutex;
+    use parking_lot::RwLock;
     use std::sync::Arc;
 
     #[derive(Default)]
-    pub struct VmRef<T: ?Sized>(pub(super) Arc<Mutex<T>>);
+    pub struct VmRef<T: ?Sized>(pub(super) Arc<RwLock<T>>);
 
-    /// Read guard alias. On the multi-threaded path both readers and
-    /// writers share a single `MutexGuard` (exclusive access).
-    pub type ReadGuard<'a, T> = parking_lot::MutexGuard<'a, T>;
-    /// Write guard alias.
-    pub type WriteGuard<'a, T> = parking_lot::MutexGuard<'a, T>;
+    /// Read guard alias. Readers share the lock, like `RefCell::borrow`: native codegen reads the
+    /// same cell more than once within one statement (`{ a: x, b: n - x }` → two
+    /// `(*x.borrow())` temporaries alive until the statement ends), which an exclusive lock turned
+    /// into a self-deadlock.
+    pub type ReadGuard<'a, T> = parking_lot::RwLockReadGuard<'a, T>;
+    /// Write guard alias (exclusive).
+    pub type WriteGuard<'a, T> = parking_lot::RwLockWriteGuard<'a, T>;
 
     impl<T> VmRef<T> {
         #[inline]
         pub fn new(value: T) -> Self {
-            VmRef(Arc::new(Mutex::new(value)))
+            VmRef(Arc::new(RwLock::new(value)))
         }
     }
 
     impl<T: ?Sized> VmRef<T> {
-        /// Acquire the inner mutex. `parking_lot::Mutex` is used (not
-        /// `std::sync::Mutex`): its uncontended lock is a single atomic with
-        /// no pthread syscall — a profile of object/array-heavy code showed
-        /// `pthread_mutex_lock/unlock` as a top cost under send-values, since
-        /// every property/element access locks. It also has no poisoning, so
-        /// there is no `Result` to swallow (a handler panic aborts the thread
-        /// regardless).
+        /// Shared access. `read_recursive` so a thread that already holds a read guard can take
+        /// another even while a writer on another thread waits (a plain `read` would queue behind
+        /// that writer and deadlock). `parking_lot` rather than `std`: its uncontended lock is a
+        /// single atomic, and every property/element access takes one.
         #[inline]
         pub fn borrow(&self) -> ReadGuard<'_, T> {
-            self.0.lock()
+            self.0.read_recursive()
         }
 
+        /// Exclusive access.
         #[inline]
         pub fn borrow_mut(&self) -> WriteGuard<'_, T> {
-            self.0.lock()
+            self.0.write()
         }
 
         #[inline]
@@ -207,3 +207,38 @@ impl<T: fmt::Debug> fmt::Debug for VmRef<T> {
         write!(f, "RefCell {{ value: {:?} }}", &*self.borrow())
     }
 }
+
+#[cfg(all(test, feature = "send-values"))]
+mod send_values_tests {
+    use super::VmRef;
+
+    /// Native codegen keeps `(*x.borrow())` temporaries alive to the end of a statement, so one
+    /// statement can read the same cell twice: `{ a: x, b: n - x }`. That must not deadlock.
+    #[test]
+    fn two_reads_of_one_cell_at_once() {
+        let x = VmRef::new(2.0_f64);
+        let sum = *x.borrow() + *x.borrow();
+        assert_eq!(sum, 4.0);
+        *x.borrow_mut() = sum;
+        assert_eq!(*x.borrow(), 4.0);
+    }
+
+    /// Readers on other threads still see writes, and a held read guard doesn't block another
+    /// reader while a writer waits elsewhere.
+    #[test]
+    fn readers_and_writers_across_threads() {
+        let x = VmRef::new(0_u32);
+        let held = x.borrow();
+        let writer = {
+            let x = x.clone();
+            std::thread::spawn(move || {
+                *x.borrow_mut() += 1;
+            })
+        };
+        std::thread::sleep(std::time::Duration::from_millis(20));
+        assert_eq!(*x.borrow(), 0, "a second read while a writer waits");
+        drop(held);
+        writer.join().unwrap();
+        assert_eq!(*x.borrow(), 1);
+    }
+}

package/crates/tish_fmt/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "tishlang_fmt"
-version = "3.15.1"
+version = "3.15.3"
 edition = "2021"
 description = "Opinionated formatter for Tish source (parse → pretty-print)"
 license-file = { workspace = true }

package/crates/tish_native/src/build.rs

@@ -236,14 +236,7 @@ pub fn build_via_cargo_with_config(
             .as_deref()
             .map_or(true, |t| cfg!(target_os = "macos") && t.ends_with("-apple-darwin"));
     if build_config.artifact == NativeArtifact::FfiModule {
-        let guest_path = Path::new(&runtime_path)
-            .parent()
-            .map(|crates| crates.join("tish_ffi_guest"))
-            .ok_or_else(|| format!("Cannot locate tish_ffi_guest next to {}", runtime_path))?;
-        more_deps.push_str(&format!(
-            "\ntishlang_ffi_guest = {{ path = {:?} }}\n",
-            guest_path.display().to_string().replace('\\', "/")
-        ));
+        more_deps.push_str(&ffi_guest_dep_toml(&runtime_path)?);
     }
     if use_fast_alloc {
         more_deps.push_str("\nmimalloc = \"0.1\"\n");
@@ -518,6 +511,24 @@ fn runtime_dep_toml(runtime_path: &str, features_str: &str) -> String {
     }
 }
 
+/// The `tishlang_ffi_guest` dependency line for an FFI module (same source policy as the runtime).
+/// A released compiler's module must take it from the registry: as a path it would bring its own
+/// path-built `tishlang_core` next to the registry one the runtime uses, and the two `Value` types
+/// don't match.
+fn ffi_guest_dep_toml(runtime_path: &str) -> Result<String, String> {
+    if let Some(v) = released_compiler_version() {
+        return Ok(format!("\ntishlang_ffi_guest = {{ version = {:?} }}\n", v));
+    }
+    let guest_path = Path::new(runtime_path)
+        .parent()
+        .map(|crates| crates.join("tish_ffi_guest"))
+        .ok_or_else(|| format!("Cannot locate tish_ffi_guest next to {}", runtime_path))?;
+    Ok(format!(
+        "\ntishlang_ffi_guest = {{ path = {:?} }}\n",
+        guest_path.display().to_string().replace('\\', "/")
+    ))
+}
+
 /// The optional `tishlang_ui` dependency line (same source policy as the runtime).
 fn ui_dep_toml(runtime_path: &str, rust_code: &str) -> String {
     if !rust_code.contains("tishlang_ui") {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@tishlang/tish-format",
-  "version": "3.15.1",
+  "version": "3.15.3",
   "description": "Tish code formatter",
   "license": "PIF",
   "repository": {