Merge pull request RustPython#3709 from youknowone/pystr-interned

introduce PyStrInterned

Author: youknowone

stdlib/src/array.rs

@@ -681,8 +681,8 @@ mod array {
         #[pyproperty]
         fn typecode(&self, vm: &VirtualMachine) -> PyStrRef {
             vm.ctx
-                .intern_string(self.read().typecode().to_string())
-                .into_pyref()
+                .intern_str(self.read().typecode().to_string())
+                .to_str()
         }
 
         #[pyproperty]

vm/src/builtins/bool.rs

@@ -112,10 +112,12 @@ impl PyBool {
     #[pymethod(magic)]
     fn repr(zelf: bool, vm: &VirtualMachine) -> PyStrRef {
         if zelf {
-            vm.ctx.true_str.clone()
+            vm.ctx.true_str
         } else {
-            vm.ctx.false_str.clone()
+            vm.ctx.false_str
         }
+        .to_owned()
+        .into_pyref()
     }
 
     #[pymethod(magic)]

vm/src/builtins/code.rs

@@ -82,7 +82,7 @@ impl ConstantBag for PyObjBag<'_> {
             bytecode::BorrowedConstant::Float { value } => ctx.new_float(value).into(),
             bytecode::BorrowedConstant::Complex { value } => ctx.new_complex(value).into(),
             bytecode::BorrowedConstant::Str { value } if value.len() <= 20 => {
-                ctx.intern_string(value).into_pyref().into()
+                ctx.intern_str(value).to_object()
             }
             bytecode::BorrowedConstant::Str { value } => ctx.new_str(value).into(),
             bytecode::BorrowedConstant::Bytes { value } => ctx.new_bytes(value.to_vec()).into(),
@@ -104,7 +104,7 @@ impl ConstantBag for PyObjBag<'_> {
     }
 
     fn make_name(&self, name: &str) -> PyStrRef {
-        self.0.intern_string(name).into_pyref()
+        self.0.intern_str(name).to_str()
     }
 }
 

vm/src/builtins/str.rs

@@ -227,6 +227,13 @@ impl IntoPyStrRef for &str {
     }
 }
 
+impl IntoPyStrRef for &'static crate::intern::PyStrInterned {
+    #[inline]
+    fn into_pystr_ref(self, _vm: &VirtualMachine) -> PyRef<PyStr> {
+        self.to_str()
+    }
+}
+
 #[pyclass(module = false, name = "str_iterator")]
 #[derive(Debug)]
 pub struct PyStrIterator {

vm/src/intern.rs

@@ -1,9 +1,13 @@
 use crate::{
     builtins::{PyStr, PyTypeRef},
     common::lock::PyRwLock,
-    Py, PyRef, PyRefExact,
+    convert::ToPyObject,
+    Py, PyObject, PyObjectRef, PyRef, PyRefExact,
+};
+use std::{
+    borrow::{Borrow, ToOwned},
+    ops::Deref,
 };
-use std::ops::Deref;
 
 #[derive(Debug)]
 pub struct StringPool {
@@ -28,24 +32,40 @@ impl Clone for StringPool {
 
 impl StringPool {
     #[inline]
-    pub unsafe fn intern<S: Internable>(&self, s: S, typ: PyTypeRef) -> PyRefExact<PyStr> {
-        if let Some(found) = self.inner.read().get(s.as_str()) {
-            return found.clone().inner;
+    pub unsafe fn intern<S: Internable>(&self, s: S, typ: PyTypeRef) -> &'static PyStrInterned {
+        if let Some(found) = self.interned(s.as_ref()) {
+            return found;
+        }
+
+        #[cold]
+        fn miss(zelf: &StringPool, s: PyRefExact<PyStr>) -> &'static PyStrInterned {
+            let cache = CachedPyStrRef { inner: s };
+            let inserted = zelf.inner.write().insert(cache.clone());
+            if inserted {
+                let interned = unsafe { PyStrInterned::borrow_cache(&cache) };
+                // unsafe { interned.as_object().mark_intern() };
+                interned
+            } else {
+                zelf.inner
+                    .read()
+                    .get(cache.as_str())
+                    .map(|cached| unsafe { PyStrInterned::borrow_cache(cached) })
+                    .expect("")
+            }
         }
-        let cache = CachedPyStrRef {
-            inner: s.into_pyref(typ),
-        };
-        let inserted = self.inner.write().insert(cache.clone());
-        if inserted {
-            cache.inner
-        } else {
-            self.inner
-                .read()
-                .get(cache.inner.as_str())
-                .unwrap()
-                .clone()
-                .inner
+        let str_ref = s.into_pyref_exact(typ);
+        miss(self, str_ref)
+    }
+
+    #[inline]
+    pub fn interned<S: MaybeInterned + ?Sized>(&self, s: &S) -> Option<&'static PyStrInterned> {
+        if let Some(interned) = s.as_interned() {
+            return Some(interned);
         }
+        self.inner
+            .read()
+            .get(s.as_ref())
+            .map(|cached| unsafe { PyStrInterned::borrow_cache(cached) })
     }
 }
 
@@ -70,57 +90,178 @@ impl PartialEq for CachedPyStrRef {
 impl Eq for CachedPyStrRef {}
 
 impl std::borrow::Borrow<str> for CachedPyStrRef {
+    #[inline]
     fn borrow(&self) -> &str {
         self.inner.as_str()
     }
 }
 
+impl AsRef<str> for CachedPyStrRef {
+    #[inline]
+    fn as_ref(&self) -> &str {
+        self.as_str()
+    }
+}
+
+impl CachedPyStrRef {
+    #[inline]
+    fn as_str(&self) -> &str {
+        self.inner.as_str()
+    }
+}
+
+/// The unique reference of interned PyStr
+/// Always intended to be used as a static reference
+pub struct PyStrInterned {
+    inner: Py<PyStr>,
+}
+
+impl PyStrInterned {
+    /// # Safety
+    /// the given cache must be alive while returned reference is alive
+    #[inline]
+    unsafe fn borrow_cache(cache: &CachedPyStrRef) -> &'static Self {
+        std::mem::transmute_copy(cache)
+    }
+
+    #[inline]
+    fn as_ptr(&self) -> *const Py<PyStr> {
+        self as *const _ as *const _
+    }
+
+    #[inline]
+    pub fn to_owned(&'static self) -> PyRefExact<PyStr> {
+        unsafe { (*(&self as *const _ as *const PyRefExact<PyStr>)).clone() }
+    }
+
+    #[inline]
+    pub fn to_str(&'static self) -> PyRef<PyStr> {
+        self.to_owned().into_pyref()
+    }
+
+    #[inline]
+    pub fn to_object(&'static self) -> PyObjectRef {
+        self.to_str().into()
+    }
+}
+
+impl Borrow<PyObject> for PyStrInterned {
+    #[inline(always)]
+    fn borrow(&self) -> &PyObject {
+        self.inner.borrow()
+    }
+}
+
+impl Deref for PyStrInterned {
+    type Target = Py<PyStr>;
+    #[inline(always)]
+    fn deref(&self) -> &Self::Target {
+        &self.inner
+    }
+}
+
+impl std::hash::Hash for PyStrInterned {
+    #[inline(always)]
+    fn hash<H: std::hash::Hasher>(&self, state: &mut H) {
+        std::hash::Hash::hash(&(self as *const _), state)
+    }
+}
+
+impl PartialEq for PyStrInterned {
+    #[inline(always)]
+    fn eq(&self, other: &Self) -> bool {
+        std::ptr::eq(self, other)
+    }
+}
+
+impl Eq for PyStrInterned {}
+
+impl AsRef<str> for PyStrInterned {
+    #[inline]
+    fn as_ref(&self) -> &str {
+        self.as_str()
+    }
+}
+
+impl std::fmt::Debug for PyStrInterned {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        std::fmt::Debug::fmt(self.as_str(), f)?;
+        write!(f, "@{:p}", self.as_ptr())
+    }
+}
+
+impl std::fmt::Display for PyStrInterned {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        std::fmt::Display::fmt(self.as_str(), f)
+    }
+}
+
 mod sealed {
-    use crate::{builtins::PyStr, object::PyRefExact};
+    use crate::{
+        builtins::PyStr,
+        object::{Py, PyRefExact},
+    };
 
     pub trait SealedInternable {}
 
     impl SealedInternable for String {}
-
     impl SealedInternable for &str {}
-
     impl SealedInternable for PyRefExact<PyStr> {}
+
+    pub trait SealedMaybeInterned {}
+
+    impl SealedMaybeInterned for str {}
+    impl SealedMaybeInterned for PyRefExact<PyStr> {}
+    impl SealedMaybeInterned for Py<PyStr> {}
 }
 
 /// A sealed marker trait for `DictKey` types that always become an exact instance of `str`
-pub trait Internable: sealed::SealedInternable + AsRef<Self::Key> {
-    type Key: crate::dictdatatype::DictKey + ?Sized;
-    fn as_str(&self) -> &str;
-    fn into_pyref(self, str_type: PyTypeRef) -> PyRefExact<PyStr>;
+pub trait Internable: sealed::SealedInternable + ToPyObject + AsRef<Self::Interned> {
+    type Interned: ?Sized + MaybeInterned;
+    fn into_pyref_exact(self, str_type: PyTypeRef) -> PyRefExact<PyStr>;
 }
 
 impl Internable for String {
-    type Key = str;
-    fn as_str(&self) -> &str {
-        String::as_str(self)
-    }
-    fn into_pyref(self, str_type: PyTypeRef) -> PyRefExact<PyStr> {
+    type Interned = str;
+    #[inline]
+    fn into_pyref_exact(self, str_type: PyTypeRef) -> PyRefExact<PyStr> {
         let obj = PyRef::new_ref(PyStr::from(self), str_type, None);
         unsafe { PyRefExact::new_unchecked(obj) }
     }
 }
 
 impl Internable for &str {
-    type Key = str;
-    fn as_str(&self) -> &str {
-        self
-    }
-    fn into_pyref(self, str_type: PyTypeRef) -> PyRefExact<PyStr> {
-        self.to_owned().into_pyref(str_type)
+    type Interned = str;
+    #[inline]
+    fn into_pyref_exact(self, str_type: PyTypeRef) -> PyRefExact<PyStr> {
+        self.to_owned().into_pyref_exact(str_type)
     }
 }
 
 impl Internable for PyRefExact<PyStr> {
-    type Key = Py<PyStr>;
-    fn as_str(&self) -> &str {
-        self.deref().as_str()
-    }
-    fn into_pyref(self, _str_type: PyTypeRef) -> PyRefExact<PyStr> {
+    type Interned = Py<PyStr>;
+    #[inline]
+    fn into_pyref_exact(self, _str_type: PyTypeRef) -> PyRefExact<PyStr> {
         self
     }
 }
+
+pub trait MaybeInterned:
+    AsRef<str> + crate::dictdatatype::DictKey + sealed::SealedMaybeInterned
+{
+    fn as_interned(&self) -> Option<&'static PyStrInterned>;
+}
+
+impl MaybeInterned for str {
+    #[inline(always)]
+    fn as_interned(&self) -> Option<&'static PyStrInterned> {
+        None
+    }
+}
+
+impl MaybeInterned for Py<PyStr> {
+    #[inline(always)]
+    fn as_interned(&self) -> Option<&'static PyStrInterned> {
+        None
+    }
+}

vm/src/stdlib/sys.rs

@@ -125,14 +125,14 @@ mod sys {
     fn byteorder(vm: &VirtualMachine) -> PyStrRef {
         // https://doc.rust-lang.org/reference/conditional-compilation.html#target_endian
         vm.ctx
-            .intern_string(if cfg!(target_endian = "little") {
+            .intern_str(if cfg!(target_endian = "little") {
                 "little"
             } else if cfg!(target_endian = "big") {
                 "big"
             } else {
                 "unknown"
             })
-            .into_pyref()
+            .to_str()
     }
 
     #[pyattr]
@@ -513,7 +513,7 @@ mod sys {
 
     #[pyfunction]
     fn intern(s: PyRefExact<PyStr>, vm: &VirtualMachine) -> PyRefExact<PyStr> {
-        vm.ctx.intern_string(s)
+        vm.ctx.intern_str(s).to_owned()
     }
 
     #[pyattr]