add new trait to encode values to bytes

still lots of TODOs

Author: antonilol

heed-traits/src/lib.rs

@@ -12,19 +12,80 @@
 use std::borrow::Cow;
 use std::cmp::{Ord, Ordering};
 use std::error::Error as StdError;
+use std::fmt;
 
 /// A boxed `Send + Sync + 'static` error.
 pub type BoxedError = Box<dyn StdError + Send + Sync + 'static>;
 
 /// A trait that represents an encoding structure.
-pub trait BytesEncode<'a> {
+#[deprecated = "replaced by `ToBytes` to allow for more optimization"]
+#[allow(deprecated)] // deprecated BoxedErrorWrapper is used in a bound
+pub trait BytesEncode<'a>:
+    // TODO are these bound needed?
+    ToBytes<'a, SelfType = Self::EItem, ReturnBytes = Cow<'a, [u8]>, Error = BoxedErrorWrapper>
+{
     /// The type to encode.
     type EItem: ?Sized + 'a;
 
     /// Encode the given item as bytes.
     fn bytes_encode(item: &'a Self::EItem) -> Result<Cow<'a, [u8]>, BoxedError>;
 }
 
+/// A trait that represents an encoding structure.
+pub trait ToBytes<'a> {
+    /// The type to encode to bytes.
+    type SelfType: ?Sized + 'a;
+
+    /// The type containing the encoded bytes.
+    type ReturnBytes: Into<Vec<u8>> + AsRef<[u8]> + 'a;
+
+    /// The error type to return when decoding goes wrong.
+    type Error: StdError + Send + Sync + 'static;
+
+    /// Encode the given item as bytes.
+    fn to_bytes(item: &'a Self::SelfType) -> Result<Self::ReturnBytes, Self::Error>;
+}
+
+#[allow(deprecated)]
+impl<'a, T: BytesEncode<'a>> ToBytes<'a> for T {
+    type SelfType = <Self as BytesEncode<'a>>::EItem;
+
+    type ReturnBytes = Cow<'a, [u8]>;
+
+    type Error = BoxedErrorWrapper;
+
+    fn to_bytes(item: &'a Self::SelfType) -> Result<Self::ReturnBytes, Self::Error> {
+        Self::bytes_encode(item).map_err(BoxedErrorWrapper)
+    }
+}
+
+/// Wraps the [`BoxedError`] type alias because for complicated reasons it does not implement
+/// [`Error`][StdError]. This wrapper forwards [`Debug`][fmt::Debug], [`Display`][fmt::Display]
+/// and [`Error`][StdError] through the wrapper and the [`Box`].
+#[deprecated = "this wrapper was added for backwards compatibility of BytesEncode only"]
+pub struct BoxedErrorWrapper(BoxedError);
+
+#[allow(deprecated)]
+impl fmt::Debug for BoxedErrorWrapper {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        <BoxedError as fmt::Debug>::fmt(&self.0, f)
+    }
+}
+
+#[allow(deprecated)]
+impl fmt::Display for BoxedErrorWrapper {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        <BoxedError as fmt::Display>::fmt(&self.0, f)
+    }
+}
+
+#[allow(deprecated)]
+impl StdError for BoxedErrorWrapper {
+    fn source(&self) -> Option<&(dyn StdError + 'static)> {
+        self.0.source()
+    }
+}
+
 /// A trait that represents a decoding structure.
 pub trait BytesDecode<'a> {
     /// The type to decode.

heed-types/src/bytes.rs

@@ -1,18 +1,22 @@
-use std::borrow::Cow;
+use std::convert::Infallible;
 
-use heed_traits::{BoxedError, BytesDecode, BytesEncode};
+use heed_traits::{BoxedError, BytesDecode, ToBytes};
 
 /// Describes a byte slice `[u8]` that is totally borrowed and doesn't depend on
 /// any [memory alignment].
 ///
 /// [memory alignment]: std::mem::align_of()
 pub enum Bytes {}
 
-impl<'a> BytesEncode<'a> for Bytes {
-    type EItem = [u8];
+impl<'a> ToBytes<'a> for Bytes {
+    type SelfType = [u8];
 
-    fn bytes_encode(item: &'a Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
-        Ok(Cow::Borrowed(item))
+    type ReturnBytes = &'a [u8];
+
+    type Error = Infallible;
+
+    fn to_bytes(item: &'a Self::SelfType) -> Result<Self::ReturnBytes, Self::Error> {
+        Ok(item)
     }
 }
 
@@ -23,3 +27,26 @@ impl<'a> BytesDecode<'a> for Bytes {
         Ok(bytes)
     }
 }
+
+/// Like [`Bytes`], but always contains exactly `N` (the generic parameter) bytes.
+pub enum FixedSizeBytes<const N: usize> {}
+
+impl<'a, const N: usize> ToBytes<'a> for FixedSizeBytes<N> {
+    type SelfType = [u8; N];
+
+    type ReturnBytes = [u8; N]; // TODO &'a [u8; N] or [u8; N]
+
+    type Error = Infallible;
+
+    fn to_bytes(item: &'a Self::SelfType) -> Result<Self::ReturnBytes, Self::Error> {
+        Ok(*item)
+    }
+}
+
+impl<'a, const N: usize> BytesDecode<'a> for FixedSizeBytes<N> {
+    type DItem = [u8; N]; // TODO &'a [u8; N] or [u8; N]
+
+    fn bytes_decode(bytes: &'a [u8]) -> Result<Self::DItem, BoxedError> {
+        bytes.try_into().map_err(Into::into)
+    }
+}

heed-types/src/integer.rs

@@ -1,18 +1,22 @@
-use std::borrow::Cow;
+use std::convert::Infallible;
 use std::marker::PhantomData;
 use std::mem::size_of;
 
 use byteorder::{ByteOrder, ReadBytesExt};
-use heed_traits::{BoxedError, BytesDecode, BytesEncode};
+use heed_traits::{BoxedError, BytesDecode, ToBytes};
 
 /// Encodable version of [`u8`].
 pub struct U8;
 
-impl BytesEncode<'_> for U8 {
-    type EItem = u8;
+impl ToBytes<'_> for U8 {
+    type SelfType = u8;
 
-    fn bytes_encode(item: &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
-        Ok(Cow::from([*item].to_vec()))
+    type ReturnBytes = [u8; 1];
+
+    type Error = Infallible;
+
+    fn to_bytes(item: &Self::SelfType) -> Result<Self::ReturnBytes, Self::Error> {
+        Ok([*item])
     }
 }
 
@@ -27,11 +31,15 @@ impl BytesDecode<'_> for U8 {
 /// Encodable version of [`i8`].
 pub struct I8;
 
-impl BytesEncode<'_> for I8 {
-    type EItem = i8;
+impl ToBytes<'_> for I8 {
+    type SelfType = i8;
+
+    type ReturnBytes = [u8; 1];
 
-    fn bytes_encode(item: &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
-        Ok(Cow::from([*item as u8].to_vec()))
+    type Error = Infallible;
+
+    fn to_bytes(item: &Self::SelfType) -> Result<Self::ReturnBytes, Self::Error> {
+        Ok([*item as u8])
     }
 }
 
@@ -51,13 +59,17 @@ macro_rules! define_type {
 
         pub struct $name<O>(PhantomData<O>);
 
-        impl<O: ByteOrder> BytesEncode<'_> for $name<O> {
-            type EItem = $native;
+        impl<O: ByteOrder> ToBytes<'_> for $name<O> {
+            type SelfType = $native;
+
+            type ReturnBytes = [u8; size_of::<$native>()];
+
+            type Error = Infallible;
 
-            fn bytes_encode(item: &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
-                let mut buf = vec![0; size_of::<Self::EItem>()];
+            fn to_bytes(item: &Self::SelfType) -> Result<Self::ReturnBytes, Self::Error> {
+                let mut buf = [0; size_of::<$native>()];
                 O::$write_method(&mut buf, *item);
-                Ok(Cow::from(buf))
+                Ok(buf)
             }
         }
 

heed-types/src/serde_bincode.rs

@@ -1,21 +1,23 @@
-use std::borrow::Cow;
-
-use heed_traits::{BoxedError, BytesDecode, BytesEncode};
+use heed_traits::{BoxedError, BytesDecode, ToBytes};
 use serde::{Deserialize, Serialize};
 
 /// Describes a type that is [`Serialize`]/[`Deserialize`] and uses `bincode` to do so.
 ///
 /// It can borrow bytes from the original slice.
 pub struct SerdeBincode<T>(std::marker::PhantomData<T>);
 
-impl<'a, T: 'a> BytesEncode<'a> for SerdeBincode<T>
+impl<'a, T: 'a> ToBytes<'a> for SerdeBincode<T>
 where
     T: Serialize,
 {
-    type EItem = T;
+    type SelfType = T;
+
+    type ReturnBytes = Vec<u8>;
+
+    type Error = bincode::Error;
 
-    fn bytes_encode(item: &'a Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
-        bincode::serialize(item).map(Cow::Owned).map_err(Into::into)
+    fn to_bytes(item: &'a Self::SelfType) -> Result<Self::ReturnBytes, Self::Error> {
+        bincode::serialize(item)
     }
 }
 

heed-types/src/serde_json.rs

@@ -1,21 +1,23 @@
-use std::borrow::Cow;
-
-use heed_traits::{BoxedError, BytesDecode, BytesEncode};
+use heed_traits::{BoxedError, BytesDecode, ToBytes};
 use serde::{Deserialize, Serialize};
 
 /// Describes a type that is [`Serialize`]/[`Deserialize`] and uses `serde_json` to do so.
 ///
 /// It can borrow bytes from the original slice.
 pub struct SerdeJson<T>(std::marker::PhantomData<T>);
 
-impl<'a, T: 'a> BytesEncode<'a> for SerdeJson<T>
+impl<'a, T: 'a> ToBytes<'a> for SerdeJson<T>
 where
     T: Serialize,
 {
-    type EItem = T;
+    type SelfType = T;
+
+    type ReturnBytes = Vec<u8>;
+
+    type Error = serde_json::Error;
 
-    fn bytes_encode(item: &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
-        serde_json::to_vec(item).map(Cow::Owned).map_err(Into::into)
+    fn to_bytes(item: &'a Self::SelfType) -> Result<Self::ReturnBytes, Self::Error> {
+        serde_json::to_vec(item)
     }
 }
 

heed-types/src/serde_rmp.rs

@@ -1,21 +1,23 @@
-use std::borrow::Cow;
-
-use heed_traits::{BoxedError, BytesDecode, BytesEncode};
+use heed_traits::{BoxedError, BytesDecode, ToBytes};
 use serde::{Deserialize, Serialize};
 
 /// Describes a type that is [`Serialize`]/[`Deserialize`] and uses `rmp_serde` to do so.
 ///
 /// It can borrow bytes from the original slice.
 pub struct SerdeRmp<T>(std::marker::PhantomData<T>);
 
-impl<'a, T: 'a> BytesEncode<'a> for SerdeRmp<T>
+impl<'a, T: 'a> ToBytes<'a> for SerdeRmp<T>
 where
     T: Serialize,
 {
-    type EItem = T;
+    type SelfType = T;
+
+    type ReturnBytes = Vec<u8>;
+
+    type Error = rmp_serde::encode::Error;
 
-    fn bytes_encode(item: &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
-        rmp_serde::to_vec(item).map(Cow::Owned).map_err(Into::into)
+    fn to_bytes(item: &'a Self::SelfType) -> Result<Self::ReturnBytes, Self::Error> {
+        rmp_serde::to_vec(item)
     }
 }
 

heed-types/src/str.rs

@@ -1,23 +1,26 @@
-use std::borrow::Cow;
-use std::str;
+use std::convert::Infallible;
 
-use heed_traits::{BoxedError, BytesDecode, BytesEncode};
+use heed_traits::{BoxedError, BytesDecode, ToBytes};
 
-/// Describes a [`prim@str`].
+/// Describes a [`str`].
 pub enum Str {}
 
-impl BytesEncode<'_> for Str {
-    type EItem = str;
+impl<'a> ToBytes<'a> for Str {
+    type SelfType = str;
 
-    fn bytes_encode(item: &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
-        Ok(Cow::Borrowed(item.as_bytes()))
+    type ReturnBytes = &'a [u8];
+
+    type Error = Infallible;
+
+    fn to_bytes(item: &'a Self::SelfType) -> Result<Self::ReturnBytes, Self::Error> {
+        Ok(item.as_bytes())
     }
 }
 
 impl<'a> BytesDecode<'a> for Str {
     type DItem = &'a str;
 
     fn bytes_decode(bytes: &'a [u8]) -> Result<Self::DItem, BoxedError> {
-        str::from_utf8(bytes).map_err(Into::into)
+        std::str::from_utf8(bytes).map_err(Into::into)
     }
 }

heed-types/src/unit.rs

@@ -1,16 +1,20 @@
-use std::borrow::Cow;
+use std::convert::Infallible;
 use std::{error, fmt};
 
-use heed_traits::{BoxedError, BytesDecode, BytesEncode};
+use heed_traits::{BoxedError, BytesDecode, ToBytes};
 
 /// Describes the unit `()` type.
 pub enum Unit {}
 
-impl BytesEncode<'_> for Unit {
-    type EItem = ();
+impl ToBytes<'_> for Unit {
+    type SelfType = ();
 
-    fn bytes_encode(_item: &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
-        Ok(Cow::Borrowed(&[]))
+    type ReturnBytes = [u8; 0];
+
+    type Error = Infallible;
+
+    fn to_bytes(_item: &'_ Self::SelfType) -> Result<Self::ReturnBytes, Self::Error> {
+        Ok([])
     }
 }