zerovec/
cow.rs

1// This file is part of ICU4X. For terms of use, please see the file
2// called LICENSE at the top level of the ICU4X source tree
3// (online at: https://github.com/unicode-org/icu4x/blob/main/LICENSE ).
4
5use crate::ule::{EncodeAsVarULE, UleError, VarULE};
6#[cfg(feature = "alloc")]
7use alloc::boxed::Box;
8use core::fmt;
9use core::marker::PhantomData;
10#[cfg(feature = "alloc")]
11use core::mem::ManuallyDrop;
12use core::ops::Deref;
13use core::ptr::NonNull;
14use zerofrom::ZeroFrom;
15
16/// Copy-on-write type that efficiently represents [`VarULE`] types as their bitstream representation.
17///
18/// The primary use case for [`VarULE`] types is the ability to store complex variable-length datastructures
19/// inside variable-length collections like [`crate::VarZeroVec`].
20///
21/// Underlying this ability is the fact that [`VarULE`] types can be efficiently represented as a flat
22/// bytestream.
23///
24/// In zero-copy cases, sometimes one wishes to unconditionally use this bytestream representation, for example
25/// to save stack size. A struct with five `Cow<'a, str>`s is not as stack-efficient as a single `Cow` containing
26/// the bytestream representation of, say, `Tuple5VarULE<str, str, str, str, str>`.
27///
28/// This type helps in this case: It is logically a `Cow<'a, V>`, with some optimizations, that is guaranteed
29/// to serialize as a byte stream in machine-readable scenarios.
30///
31/// During human-readable serialization, it will fall back to the serde impls on `V`, which ought to have
32/// a human-readable variant.
33pub struct VarZeroCow<'a, V: ?Sized> {
34    /// Safety invariant: Contained slice must be a valid V
35    /// It may or may not have a lifetime valid for 'a, it must be valid for as long as this type is around.
36    raw: RawVarZeroCow,
37    marker1: PhantomData<&'a V>,
38    #[cfg(feature = "alloc")]
39    marker2: PhantomData<Box<V>>,
40}
41
42/// VarZeroCow without the `V` to simulate a dropck eyepatch
43/// (i.e., prove to rustc that the dtor is not able to observe V or 'a)
44///
45/// This is effectively `Cow<'a, [u8]>`, with the lifetime managed externally
46struct RawVarZeroCow {
47    /// Pointer to data
48    ///
49    /// # Safety Invariants
50    ///
51    /// 1. This slice must always be valid as a byte slice
52    /// 2. If `owned` is true, this slice can be freed.
53    /// 3. VarZeroCow, the only user of this type, will impose an additional invariant that the buffer is a valid V
54    buf: NonNull<[u8]>,
55    /// The buffer is `Box<[u8]>` if true
56    #[cfg(feature = "alloc")]
57    owned: bool,
58    // Safety: We do not need any PhantomDatas here, since the Drop impl does not observe borrowed data
59    // if there is any.
60}
61
62#[cfg(feature = "alloc")]
63impl Drop for RawVarZeroCow {
64    fn drop(&mut self) {
65        // Note: this drop impl NEVER observes borrowed data (which may have already been cleaned up by the time the impl is called)
66        if self.owned {
67            unsafe {
68                // Safety: (Invariant 2 on buf)
69                // since owned is true, this is a valid Box<[u8]> and can be cleaned up
70                let _ = Box::<[u8]>::from_raw(self.buf.as_ptr());
71            }
72        }
73    }
74}
75
76// This is mostly just a `Cow<[u8]>`, safe to implement Send and Sync on
77unsafe impl Send for RawVarZeroCow {}
78unsafe impl Sync for RawVarZeroCow {}
79
80impl Clone for RawVarZeroCow {
81    fn clone(&self) -> Self {
82        #[cfg(feature = "alloc")]
83        if self.is_owned() {
84            // This clones the box
85            let b: Box<[u8]> = self.as_bytes().into();
86            let b = ManuallyDrop::new(b);
87            let buf: NonNull<[u8]> = (&**b).into();
88            return Self {
89                // Invariants upheld:
90                // 1 & 3: The bytes came from `self` so they're a valid value and byte slice
91                // 2: This is owned (we cloned it), so we set owned to true.
92                buf,
93                owned: true,
94            };
95        }
96        // Unfortunately we can't just use `new_borrowed(self.deref())` since the lifetime is shorter
97        Self {
98            // Invariants upheld:
99            // 1 & 3: The bytes came from `self` so they're a valid value and byte slice
100            // 2: This is borrowed (we're sharing a borrow), so we set owned to false.
101            buf: self.buf,
102            #[cfg(feature = "alloc")]
103            owned: false,
104        }
105    }
106}
107
108impl<'a, V: ?Sized> Clone for VarZeroCow<'a, V> {
109    fn clone(&self) -> Self {
110        let raw = self.raw.clone();
111        // Invariant upheld: raw came from a valid VarZeroCow, so it
112        // is a valid V
113        unsafe { Self::from_raw(raw) }
114    }
115}
116
117impl<'a, V: VarULE + ?Sized> VarZeroCow<'a, V> {
118    /// Construct from a slice. Errors if the slice doesn't represent a valid `V`
119    pub fn parse_bytes(bytes: &'a [u8]) -> Result<Self, UleError> {
120        let val = V::parse_bytes(bytes)?;
121        Ok(Self::new_borrowed(val))
122    }
123
124    /// Construct from an owned slice. Errors if the slice doesn't represent a valid `V`
125    #[cfg(feature = "alloc")]
126    pub fn parse_owned_bytes(bytes: Box<[u8]>) -> Result<Self, UleError> {
127        V::validate_bytes(&bytes)?;
128        let bytes = ManuallyDrop::new(bytes);
129        let buf: NonNull<[u8]> = (&**bytes).into();
130        let raw = RawVarZeroCow {
131            // Invariants upheld:
132            // 1 & 3: The bytes came from `val` so they're a valid value and byte slice
133            // 2: This is owned, so we set owned to true.
134            buf,
135            owned: true,
136        };
137        Ok(Self {
138            raw,
139            marker1: PhantomData,
140            #[cfg(feature = "alloc")]
141            marker2: PhantomData,
142        })
143    }
144
145    /// Construct from a slice that is known to represent a valid `V`
146    ///
147    /// # Safety
148    ///
149    /// `bytes` must be a valid `V`, i.e. it must successfully pass through
150    /// `V::parse_bytes()` or `V::validate_bytes()`.
151    pub const unsafe fn from_bytes_unchecked(bytes: &'a [u8]) -> Self {
152        unsafe {
153            // Safety: bytes is an &T which is always non-null
154            let buf: NonNull<[u8]> = NonNull::new_unchecked(bytes as *const [u8] as *mut [u8]);
155            let raw = RawVarZeroCow {
156                // Invariants upheld:
157                // 1 & 3: Passed upstream to caller
158                // 2: This is borrowed, so we set owned to false.
159                buf,
160                #[cfg(feature = "alloc")]
161                owned: false,
162            };
163            // Invariant passed upstream to caller
164            Self::from_raw(raw)
165        }
166    }
167
168    /// Construct this from an [`EncodeAsVarULE`] version of the contained type
169    ///
170    /// Will always construct an owned version
171    #[cfg(feature = "alloc")]
172    pub fn from_encodeable<E: EncodeAsVarULE<V>>(encodeable: &E) -> Self {
173        let b = crate::ule::encode_varule_to_box(encodeable);
174        Self::new_owned(b)
175    }
176
177    /// Construct a new borrowed version of this
178    pub fn new_borrowed(val: &'a V) -> Self {
179        unsafe {
180            // Safety: val is a valid V, by type
181            Self::from_bytes_unchecked(val.as_bytes())
182        }
183    }
184
185    /// Construct a new borrowed version of this
186    #[cfg(feature = "alloc")]
187    pub fn new_owned(val: Box<V>) -> Self {
188        let val = ManuallyDrop::new(val);
189        let buf: NonNull<[u8]> = val.as_bytes().into();
190        let raw = RawVarZeroCow {
191            // Invariants upheld:
192            // 1 & 3: The bytes came from `val` so they're a valid value and byte slice
193            // 2: This is owned, so we set owned to true.
194            buf,
195            #[cfg(feature = "alloc")]
196            owned: true,
197        };
198        // The bytes came from `val`, so it's a valid value
199        unsafe { Self::from_raw(raw) }
200    }
201}
202
203impl<'a, V: ?Sized> VarZeroCow<'a, V> {
204    /// Whether or not this is owned
205    pub fn is_owned(&self) -> bool {
206        self.raw.is_owned()
207    }
208
209    /// Get the byte representation of this type
210    ///
211    /// Is also always a valid `V` and can be passed to
212    /// `V::from_bytes_unchecked()`
213    pub fn as_bytes(&self) -> &[u8] {
214        // The valid V invariant comes from Invariant 2
215        self.raw.as_bytes()
216    }
217
218    /// Invariant: `raw` must wrap a valid V, either owned or borrowed for 'a
219    const unsafe fn from_raw(raw: RawVarZeroCow) -> Self {
220        Self {
221            // Invariant passed up to caller
222            raw,
223            marker1: PhantomData,
224            #[cfg(feature = "alloc")]
225            marker2: PhantomData,
226        }
227    }
228}
229
230impl RawVarZeroCow {
231    /// Whether or not this is owned
232    #[inline]
233    pub fn is_owned(&self) -> bool {
234        #[cfg(feature = "alloc")]
235        return self.owned;
236        #[cfg(not(feature = "alloc"))]
237        return false;
238    }
239
240    /// Get the byte representation of this type
241    #[inline]
242    pub fn as_bytes(&self) -> &[u8] {
243        // Safety: Invariant 1 on self.buf
244        unsafe { self.buf.as_ref() }
245    }
246}
247
248impl<'a, V: VarULE + ?Sized> Deref for VarZeroCow<'a, V> {
249    type Target = V;
250    fn deref(&self) -> &V {
251        // Safety: From invariant 2 on self.buf
252        unsafe { V::from_bytes_unchecked(self.as_bytes()) }
253    }
254}
255
256impl<'a, V: VarULE + ?Sized> From<&'a V> for VarZeroCow<'a, V> {
257    fn from(other: &'a V) -> Self {
258        Self::new_borrowed(other)
259    }
260}
261
262#[cfg(feature = "alloc")]
263impl<'a, V: VarULE + ?Sized> From<Box<V>> for VarZeroCow<'a, V> {
264    fn from(other: Box<V>) -> Self {
265        Self::new_owned(other)
266    }
267}
268
269impl<'a, V: VarULE + ?Sized + fmt::Debug> fmt::Debug for VarZeroCow<'a, V> {
270    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> Result<(), fmt::Error> {
271        self.deref().fmt(f)
272    }
273}
274
275// We need manual impls since `#[derive()]` is disallowed on packed types
276impl<'a, V: VarULE + ?Sized + PartialEq> PartialEq for VarZeroCow<'a, V> {
277    fn eq(&self, other: &Self) -> bool {
278        self.deref().eq(other.deref())
279    }
280}
281
282impl<'a, V: VarULE + ?Sized + Eq> Eq for VarZeroCow<'a, V> {}
283
284impl<'a, V: VarULE + ?Sized + PartialOrd> PartialOrd for VarZeroCow<'a, V> {
285    fn partial_cmp(&self, other: &Self) -> Option<core::cmp::Ordering> {
286        self.deref().partial_cmp(other.deref())
287    }
288}
289
290impl<'a, V: VarULE + ?Sized + Ord> Ord for VarZeroCow<'a, V> {
291    fn cmp(&self, other: &Self) -> core::cmp::Ordering {
292        self.deref().cmp(other.deref())
293    }
294}
295
296// # Safety
297//
298// encode_var_ule_len: Produces the length of the contained bytes, which are known to be a valid V by invariant
299//
300// encode_var_ule_write: Writes the contained bytes, which are known to be a valid V by invariant
301unsafe impl<'a, V: VarULE + ?Sized> EncodeAsVarULE<V> for VarZeroCow<'a, V> {
302    fn encode_var_ule_as_slices<R>(&self, _: impl FnOnce(&[&[u8]]) -> R) -> R {
303        // unnecessary if the other two are implemented
304        unreachable!()
305    }
306
307    #[inline]
308    fn encode_var_ule_len(&self) -> usize {
309        self.as_bytes().len()
310    }
311
312    #[inline]
313    fn encode_var_ule_write(&self, dst: &mut [u8]) {
314        dst.copy_from_slice(self.as_bytes())
315    }
316}
317
318#[cfg(feature = "serde")]
319impl<'a, V: VarULE + ?Sized + serde::Serialize> serde::Serialize for VarZeroCow<'a, V> {
320    fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
321    where
322        S: serde::Serializer,
323    {
324        if serializer.is_human_readable() {
325            <V as serde::Serialize>::serialize(self.deref(), serializer)
326        } else {
327            serializer.serialize_bytes(self.as_bytes())
328        }
329    }
330}
331
332#[cfg(feature = "serde")]
333impl<'a, 'de: 'a, V: VarULE + ?Sized> serde::Deserialize<'de> for VarZeroCow<'a, V>
334where
335    Box<V>: serde::Deserialize<'de>,
336{
337    fn deserialize<Des>(deserializer: Des) -> Result<Self, Des::Error>
338    where
339        Des: serde::Deserializer<'de>,
340    {
341        if deserializer.is_human_readable() {
342            let b = Box::<V>::deserialize(deserializer)?;
343            Ok(Self::new_owned(b))
344        } else {
345            let bytes = <&[u8]>::deserialize(deserializer)?;
346            Self::parse_bytes(bytes).map_err(serde::de::Error::custom)
347        }
348    }
349}
350
351#[cfg(feature = "databake")]
352impl<'a, V: VarULE + ?Sized> databake::Bake for VarZeroCow<'a, V> {
353    fn bake(&self, env: &databake::CrateEnv) -> databake::TokenStream {
354        env.insert("zerovec");
355        let bytes = self.as_bytes().bake(env);
356        databake::quote! {
357            // Safety: Known to come from a valid V since self.as_bytes() is always a valid V
358            unsafe {
359                zerovec::VarZeroCow::from_bytes_unchecked(#bytes)
360            }
361        }
362    }
363}
364
365#[cfg(feature = "databake")]
366impl<'a, V: VarULE + ?Sized> databake::BakeSize for VarZeroCow<'a, V> {
367    fn borrows_size(&self) -> usize {
368        self.as_bytes().len()
369    }
370}
371
372impl<'a, V: VarULE + ?Sized> ZeroFrom<'a, V> for VarZeroCow<'a, V> {
373    #[inline]
374    fn zero_from(other: &'a V) -> Self {
375        Self::new_borrowed(other)
376    }
377}
378
379impl<'a, 'b, V: VarULE + ?Sized> ZeroFrom<'a, VarZeroCow<'b, V>> for VarZeroCow<'a, V> {
380    #[inline]
381    fn zero_from(other: &'a VarZeroCow<'b, V>) -> Self {
382        Self::new_borrowed(other)
383    }
384}
385
386#[cfg(test)]
387mod tests {
388    use super::VarZeroCow;
389    use crate::ule::tuplevar::Tuple3VarULE;
390    use crate::vecs::VarZeroSlice;
391    #[test]
392    fn test_cow_roundtrip() {
393        type Messy = Tuple3VarULE<str, [u8], VarZeroSlice<str>>;
394        let vec = vec!["one", "two", "three"];
395        let messy: VarZeroCow<Messy> =
396            VarZeroCow::from_encodeable(&("hello", &b"g\xFF\xFFdbye"[..], vec));
397
398        assert_eq!(messy.a(), "hello");
399        assert_eq!(messy.b(), b"g\xFF\xFFdbye");
400        assert_eq!(&messy.c()[1], "two");
401
402        #[cfg(feature = "serde")]
403        {
404            let bincode = bincode::serialize(&messy).unwrap();
405            let deserialized: VarZeroCow<Messy> = bincode::deserialize(&bincode).unwrap();
406            assert_eq!(
407                messy, deserialized,
408                "Single element roundtrips with bincode"
409            );
410            assert!(!deserialized.is_owned());
411
412            let json = serde_json::to_string(&messy).unwrap();
413            let deserialized: VarZeroCow<Messy> = serde_json::from_str(&json).unwrap();
414            assert_eq!(messy, deserialized, "Single element roundtrips with serde");
415        }
416    }
417
418    struct TwoCows<'a> {
419        cow1: VarZeroCow<'a, str>,
420        cow2: VarZeroCow<'a, str>,
421    }
422
423    #[test]
424    fn test_eyepatch_works() {
425        // This code should compile
426        let mut two = TwoCows {
427            cow1: VarZeroCow::new_borrowed("hello"),
428            cow2: VarZeroCow::new_owned("world".into()),
429        };
430        let three = VarZeroCow::new_borrowed(&*two.cow2);
431        two.cow1 = three;
432
433        // Without the eyepatch, dropck will be worried that the dtor of two.cow1 can observe the
434        // data it borrowed from two.cow2, which may have already been deleted
435
436        // This test will fail if you add an empty `impl<'a, V: ?Sized> Drop for VarZeroCow<'a, V>`
437    }
438}
zerovec/cow.rs

zerovec/
cow.rs
