git.delta.rocks / jrsonnet / refs/commits / f3c8ed70f69f

1#![deny(2	unsafe_op_in_unsafe_fn,3	clippy::missing_safety_doc,4	clippy::undocumented_unsafe_blocks5)]6#![warn(clippy::pedantic, clippy::nursery)]7#![allow(clippy::missing_const_for_fn)]8use std::{9	borrow::Cow,10	cell::RefCell,11	fmt::{self, Display},12	hash::{Hash, Hasher},13	ops::Deref,14	str,15};1617use hashbrown::{HashMap, hash_map::RawEntryMut};18use jrsonnet_gcmodule::{Acyclic, Trace};19use rustc_hash::FxBuildHasher;2021mod inner;22use inner::Inner;2324mod names;2526/// Interned string27///28/// Provides O(1) comparsions and hashing, cheap copy, and cheap conversion to [`IBytes`]29#[derive(Clone, PartialOrd, Ord, Eq)]30pub struct IStr(Inner);31impl Trace for IStr {32	fn is_type_tracked() -> bool {33		false34	}35}3637/// SAFETY:38///39/// `IStr` is acyclic40unsafe impl Acyclic for IStr {}4142impl IStr {43	#[must_use]44	pub fn empty() -> Self {45		"".into()46	}47	#[must_use]48	pub fn as_str(&self) -> &str {49		self as &str50	}5152	#[must_use]53	pub fn cast_bytes(self) -> IBytes {54		IBytes(self.0.clone())55	}56}5758impl Deref for IStr {59	type Target = str;6061	fn deref(&self) -> &Self::Target {62		// SAFETY: Inner::check_utf8 is called on IStr construction, data is utf-863		unsafe { self.0.as_str_unchecked() }64	}65}6667impl PartialEq for IStr {68	fn eq(&self, other: &Self) -> bool {69		// all IStr should be inlined into same pool70		Inner::ptr_eq(&self.0, &other.0)71	}72}7374impl PartialEq<str> for IStr {75	fn eq(&self, other: &str) -> bool {76		self as &str == other77	}78}7980impl Hash for IStr {81	fn hash<H: Hasher>(&self, state: &mut H) {82		// IStr is always obtained from pool, where no string have duplicate, thus every unique string has unique address83		state.write_usize(Inner::as_ptr(&self.0).cast::<()>() as usize);84	}85}8687impl Drop for IStr {88	fn drop(&mut self) {89		maybe_unpool(&self.0);90	}91}9293impl fmt::Debug for IStr {94	fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {95		fmt::Debug::fmt(self as &str, f)96	}97}9899impl Display for IStr {100	fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {101		fmt::Display::fmt(self as &str, f)102	}103}104105/// Interned byte array106#[derive(Clone, PartialOrd, Ord, Eq)]107pub struct IBytes(Inner);108impl Trace for IBytes {109	fn is_type_tracked() -> bool {110		false111	}112}113114impl IBytes {115	#[must_use]116	pub fn cast_str(self) -> Option<IStr> {117		if Inner::check_utf8(&self.0) {118			Some(IStr(self.0.clone()))119		} else {120			None121		}122	}123	/// # Safety124	/// data should be valid utf8125	unsafe fn cast_str_unchecked(self) -> IStr {126		// SAFETY: data is utf8127		unsafe { Inner::assume_utf8(&self.0) };128		IStr(self.0.clone())129	}130131	#[must_use]132	pub fn as_slice(&self) -> &[u8] {133		self.0.as_slice()134	}135}136137impl Deref for IBytes {138	type Target = [u8];139140	fn deref(&self) -> &Self::Target {141		self.0.as_slice()142	}143}144145impl PartialEq for IBytes {146	fn eq(&self, other: &Self) -> bool {147		// all IStr should be inlined into same pool148		Inner::ptr_eq(&self.0, &other.0)149	}150}151152impl Hash for IBytes {153	fn hash<H: Hasher>(&self, state: &mut H) {154		// IBytes is always obtained from pool, where no string have duplicate, thus every unique string has unique address155		state.write_usize(Inner::as_ptr(&self.0).cast::<()>() as usize);156	}157}158159impl Drop for IBytes {160	fn drop(&mut self) {161		maybe_unpool(&self.0);162	}163}164165fn maybe_unpool(inner: &Inner) {166	#[cold]167	#[inline(never)]168	fn unpool(inner: &Inner) {169		// May fail on program termination170		let _ = POOL.try_with(|pool| {171			let mut pool = pool.borrow_mut();172173			if pool.remove(inner).is_none() {174				// On some platforms (i.e i686-windows), try_with will not fail after TLS175				// destructor is called, but instead re-initialize the TLS with the empty pool.176				// Allow non-pooled Drop in this case.177				// https://github.com/CertainLach/jrsonnet/issues/98#issuecomment-1591624016178				// Another cause might be that you have improperly used jrsonnet in multi-threaded environment:179				// https://github.com/CertainLach/jrsonnet/issues/113180				debug_assert!(pool.is_empty(), "if you have landed here - you most likely did something naughty with multi-threading. jrsonnet string pooling uses thread_local pool");181			}182		});183	}184	// First reference - current object, second - POOL185	if Inner::strong_count(inner) <= 2 {186		unpool(inner);187	}188}189190impl fmt::Debug for IBytes {191	fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {192		fmt::Debug::fmt(self as &[u8], f)193	}194}195196impl<'c> From<Cow<'c, str>> for IStr {197	fn from(v: Cow<'c, str>) -> Self {198		intern_str(&v)199	}200}201impl From<&str> for IStr {202	fn from(v: &str) -> Self {203		intern_str(v)204	}205}206impl From<String> for IStr {207	fn from(s: String) -> Self {208		s.as_str().into()209	}210}211impl From<&String> for IStr {212	fn from(s: &String) -> Self {213		s.as_str().into()214	}215}216impl From<char> for IStr {217	fn from(value: char) -> Self {218		let mut buf = [0; 5];219		Self::from(&*value.encode_utf8(&mut buf))220	}221}222impl From<&[u8]> for IBytes {223	fn from(v: &[u8]) -> Self {224		intern_bytes(v)225	}226}227228type PoolMap = HashMap<Inner, (), FxBuildHasher>;229230thread_local! {231	static POOL: RefCell<PoolMap> = RefCell::new(HashMap::with_capacity_and_hasher(200, FxBuildHasher));232}233234/// Utils for embedding jrsonnet in non-rust.235///236/// Jrsonnet golang bindings require that it is possible to move jsonnet237/// VM between OS threads, and this is not possible due to usage of238/// `thread_local`. Instead, there is two methods added, one should be239/// called at the end of current thread work, and one that should be240/// used when using other thread.241pub mod interop {242	use std::mem;243244	use crate::{POOL, PoolMap};245246	/// Type-erased interned string pool247	pub enum PoolState {}248249	/// Dump current interned string pool, to be restored by250	/// `reenter_thread`251	pub fn exit_thread() -> *mut PoolState {252		Box::into_raw(Box::new(POOL.with_borrow_mut(mem::take))).cast()253	}254255	/// Reenter thread, using state dumped by `exit_thread`.256	///257	/// # Safety258	///259	/// `state` should be acquired from `exit_thread`, it is not allowed260	/// to reuse state to reenter multiple threads.261	pub unsafe fn reenter_thread(state: *mut PoolState) {262		let ptr: *mut PoolMap = state.cast();263		// SAFETY: ptr is an unique state per method safety requirements.264		let ptr: Box<PoolMap> = unsafe { Box::from_raw(ptr) };265		let ptr: PoolMap = *ptr;266		POOL.with_borrow_mut(|pool| {267			let _ = mem::replace(pool, ptr);268		});269	}270}271272#[must_use]273pub fn intern_bytes(bytes: &[u8]) -> IBytes {274	POOL.with(|pool| {275		let mut pool = pool.borrow_mut();276		let entry = pool.raw_entry_mut().from_key(bytes);277		match entry {278			RawEntryMut::Occupied(i) => IBytes(i.get_key_value().0.clone()),279			RawEntryMut::Vacant(e) => {280				let (k, ()) = e.insert(Inner::new_bytes(bytes), ());281				IBytes(k.clone())282			}283		}284	})285}286287#[must_use]288pub fn intern_str(str: &str) -> IStr {289	// SAFETY: Rust strings always utf8290	unsafe { intern_bytes(str.as_bytes()).cast_str_unchecked() }291}292293#[cfg(test)]294mod tests {295	use crate::IStr;296297	#[test]298	fn simple() {299		let a = IStr::from("a");300		let b = IStr::from("a");301302		assert_eq!(a.as_ptr(), b.as_ptr());303	}304}