add new trait to encode values to bytes

heed-traits/src/lib.rs

@@ -9,9 +9,9 @@
 
 #![warn(missing_docs)]
 
-use std::borrow::Cow;
 use std::cmp::{Ord, Ordering};
 use std::error::Error as StdError;
+use std::io;
 
 /// A boxed `Send + Sync + 'static` error.
 pub type BoxedError = Box<dyn StdError + Send + Sync + 'static>;
@@ -21,8 +21,48 @@ pub trait BytesEncode<'a> {
     /// The type to encode.
     type EItem: ?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;
+
+    /// This function can be used to hint callers of the
+    /// [`bytes_encode`][BytesEncode::bytes_encode] function to use
+    /// [`bytes_encode_into_writer`][BytesEncode::bytes_encode_into_writer] instead, if the latter
+    /// runs faster (for example if it needs less heap allocations).
+    ///
+    /// The default implementation returns `true` because the default implementation of
+    /// [`bytes_encode_into_writer`][BytesEncode::bytes_encode_into_writer] is to forward to
+    /// [`bytes_encode`][BytesEncode::bytes_encode].
+    fn zero_copy(item: &Self::EItem) -> bool {
+        // This is preferred to renaming the function parameter (to _item) because IDEs can
+        // autofill trait implementations, which will default the paramter name to _item then and
+        // this could probably also mess with clippy's renamed_function_params lint.
+        let _ = item;
+
+        true
+    }
+
     /// Encode the given item as bytes.
-    fn bytes_encode(item: &'a Self::EItem) -> Result<Cow<'a, [u8]>, BoxedError>;
+    fn bytes_encode(item: &'a Self::EItem) -> Result<Self::ReturnBytes, Self::Error>;
+
+    /// Encode the given item as bytes and write it into the writer.
+    ///
+    /// When implementing this, also take a look at [`zero_copy`][BytesEncode::zero_copy]'s
+    /// documentation.
+    ///
+    /// The default implementation forwards to [`bytes_encode`][BytesEncode::bytes_encode].
+    fn bytes_encode_into_writer<W: io::Write>(
+        item: &'a Self::EItem,
+        mut writer: W,
+    ) -> Result<(), BoxedError> {
+        let bytes = Self::bytes_encode(item)?;
+
+        writer.write_all(bytes.as_ref())?;
+
+        Ok(())
+    }
 }
 
 /// A trait that represents a decoding structure.

heed-types/src/bytes.rs

@@ -1,4 +1,4 @@
-use std::borrow::Cow;
+use std::convert::Infallible;
 
 use heed_traits::{BoxedError, BytesDecode, BytesEncode};
 
@@ -11,8 +11,12 @@ pub enum Bytes {}
 impl<'a> BytesEncode<'a> for Bytes {
     type EItem = [u8];
 
-    fn bytes_encode(item: &'a Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
-        Ok(Cow::Borrowed(item))
+    type ReturnBytes = &'a [u8];
+
+    type Error = Infallible;
+
+    fn bytes_encode(item: &'a Self::EItem) -> 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> BytesEncode<'a> for FixedSizeBytes<N> {
+    type EItem = [u8; N];
+
+    type ReturnBytes = [u8; N]; // TODO &'a [u8; N] or [u8; N]
+
+    type Error = Infallible;
+
+    fn bytes_encode(item: &'a Self::EItem) -> 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

@@ -1,4 +1,4 @@
-use std::borrow::Cow;
+use std::convert::Infallible;
 use std::marker::PhantomData;
 use std::mem::size_of;
 
@@ -11,8 +11,12 @@ pub struct U8;
 impl BytesEncode<'_> for U8 {
     type EItem = 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 bytes_encode(item: &Self::EItem) -> Result<Self::ReturnBytes, Self::Error> {
+        Ok([*item])
     }
 }
 
@@ -30,8 +34,12 @@ pub struct I8;
 impl BytesEncode<'_> for I8 {
     type EItem = i8;
 
-    fn bytes_encode(item: &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
-        Ok(Cow::from([*item as u8].to_vec()))
+    type ReturnBytes = [u8; 1];
+
+    type Error = Infallible;
+
+    fn bytes_encode(item: &Self::EItem) -> Result<Self::ReturnBytes, Self::Error> {
+        Ok([*item as u8])
     }
 }
 
@@ -54,10 +62,14 @@ macro_rules! define_type {
         impl<O: ByteOrder> BytesEncode<'_> for $name<O> {
             type EItem = $native;
 
-            fn bytes_encode(item: &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
-                let mut buf = vec![0; size_of::<Self::EItem>()];
+            type ReturnBytes = [u8; size_of::<$native>()];
+
+            type Error = Infallible;
+
+            fn bytes_encode(item: &Self::EItem) -> Result<Self::ReturnBytes, Self::Error> {
+                let mut buf = [0; size_of::<$native>()];
                 O::$write_method(&mut buf, *item);
-                Ok(Cow::from(buf))
+                Ok(buf)
             }
         }
 

heed-types/src/serde_bincode.rs

@@ -1,5 +1,3 @@
-use std::borrow::Cow;
-
 use heed_traits::{BoxedError, BytesDecode, BytesEncode};
 use serde::{Deserialize, Serialize};
 
@@ -14,8 +12,24 @@ where
 {
     type EItem = T;
 
-    fn bytes_encode(item: &'a Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
-        bincode::serialize(item).map(Cow::Owned).map_err(Into::into)
+    type ReturnBytes = Vec<u8>;
+
+    type Error = bincode::Error;
+
+    fn zero_copy(_item: &Self::EItem) -> bool {
+        false
+    }
+
+    fn bytes_encode(item: &Self::EItem) -> Result<Self::ReturnBytes, Self::Error> {
+        bincode::serialize(item)
+    }
+
+    fn bytes_encode_into_writer<W: std::io::Write>(
+        item: &'a Self::EItem,
+        writer: W,
+    ) -> Result<(), BoxedError> {
+        bincode::serialize_into(writer, item)?;
+        Ok(())
     }
 }
 

heed-types/src/serde_json.rs

@@ -1,5 +1,3 @@
-use std::borrow::Cow;
-
 use heed_traits::{BoxedError, BytesDecode, BytesEncode};
 use serde::{Deserialize, Serialize};
 
@@ -14,8 +12,24 @@ where
 {
     type EItem = T;
 
-    fn bytes_encode(item: &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
-        serde_json::to_vec(item).map(Cow::Owned).map_err(Into::into)
+    type ReturnBytes = Vec<u8>;
+
+    type Error = serde_json::Error;
+
+    fn zero_copy(_item: &Self::EItem) -> bool {
+        false
+    }
+
+    fn bytes_encode(item: &Self::EItem) -> Result<Self::ReturnBytes, Self::Error> {
+        serde_json::to_vec(item)
+    }
+
+    fn bytes_encode_into_writer<W: std::io::Write>(
+        item: &'a Self::EItem,
+        writer: W,
+    ) -> Result<(), BoxedError> {
+        serde_json::to_writer(writer, item)?;
+        Ok(())
     }
 }
 

heed-types/src/serde_rmp.rs

@@ -1,5 +1,3 @@
-use std::borrow::Cow;
-
 use heed_traits::{BoxedError, BytesDecode, BytesEncode};
 use serde::{Deserialize, Serialize};
 
@@ -14,8 +12,24 @@ where
 {
     type EItem = T;
 
-    fn bytes_encode(item: &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
-        rmp_serde::to_vec(item).map(Cow::Owned).map_err(Into::into)
+    type ReturnBytes = Vec<u8>;
+
+    type Error = rmp_serde::encode::Error;
+
+    fn zero_copy(_item: &Self::EItem) -> bool {
+        false
+    }
+
+    fn bytes_encode(item: &Self::EItem) -> Result<Self::ReturnBytes, Self::Error> {
+        rmp_serde::to_vec(item)
+    }
+
+    fn bytes_encode_into_writer<W: std::io::Write>(
+        item: &'a Self::EItem,
+        mut writer: W,
+    ) -> Result<(), BoxedError> {
+        rmp_serde::encode::write(&mut writer, item)?;
+        Ok(())
     }
 }
 

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};
 
-/// Describes a [`prim@str`].
+/// Describes a [`str`].
 pub enum Str {}
 
-impl BytesEncode<'_> for Str {
+impl<'a> BytesEncode<'a> for Str {
     type EItem = 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 bytes_encode(item: &'a Self::EItem) -> 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

@@ -1,4 +1,4 @@
-use std::borrow::Cow;
+use std::convert::Infallible;
 use std::{error, fmt};
 
 use heed_traits::{BoxedError, BytesDecode, BytesEncode};
@@ -9,8 +9,12 @@ pub enum Unit {}
 impl BytesEncode<'_> for Unit {
     type EItem = ();
 
-    fn bytes_encode(_item: &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
-        Ok(Cow::Borrowed(&[]))
+    type ReturnBytes = [u8; 0];
+
+    type Error = Infallible;
+
+    fn bytes_encode(&(): &Self::EItem) -> Result<Self::ReturnBytes, Self::Error> {
+        Ok([])
     }
 }
 

heed/src/cookbook.rs

@@ -146,15 +146,15 @@
 //! to create codecs to encode prefixes when possible instead of using a slice of bytes.
 //!
 //! ```
-//! use std::borrow::Cow;
+//! use std::convert::Infallible;
 //! use std::error::Error;
 //! use std::fs;
 //! use std::path::Path;
 //!
 //! use heed::types::*;
 //! use heed::{BoxedError, BytesDecode, BytesEncode, Database, EnvOpenOptions};
 //!
-//! #[derive(Debug, PartialEq, Eq)]
+//! #[derive(Debug, Clone, Copy, PartialEq, Eq)]
 //! pub enum Level {
 //!     Debug,
 //!     Warn,
@@ -172,18 +172,20 @@
 //! impl<'a> BytesEncode<'a> for LogKeyCodec {
 //!     type EItem = LogKey;
 //!
+//!     type ReturnBytes = [u8; 5];
+//!
+//!     type Error = Infallible;
+//!
 //!     /// Encodes the u32 timestamp in big endian followed by the log level with a single byte.
-//!     fn bytes_encode(log: &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
-//!         let (timestamp_bytes, level_byte) = match log {
-//!             LogKey { timestamp, level: Level::Debug } => (timestamp.to_be_bytes(), 0),
-//!             LogKey { timestamp, level: Level::Warn } => (timestamp.to_be_bytes(), 1),
-//!             LogKey { timestamp, level: Level::Error } => (timestamp.to_be_bytes(), 2),
-//!         };
+//!     fn bytes_encode(log: &Self::EItem) -> Result<Self::ReturnBytes, Self::Error> {
+//!         let mut output = [0; 5];
 //!
-//!         let mut output = Vec::new();
-//!         output.extend_from_slice(&timestamp_bytes);
-//!         output.push(level_byte);
-//!         Ok(Cow::Owned(output))
+//!         let [timestamp @ .., level] = &mut output;
+//!
+//!         *timestamp = log.timestamp.to_be_bytes();
+//!         *level = log.level as u8;
+//!
+//!         Ok(output)
 //!     }
 //! }
 //!
@@ -218,9 +220,14 @@
 //! impl<'a> BytesEncode<'a> for LogAtHalfTimestampCodec {
 //!     type EItem = u32;
 //!
+//!     type ReturnBytes = [u8; 2];
+//!
+//!     type Error = Infallible;
+//!
 //!     /// This method encodes only the prefix of the keys in this particular case, the timestamp.
-//!     fn bytes_encode(half_timestamp: &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
-//!         Ok(Cow::Owned(half_timestamp.to_be_bytes()[..2].to_vec()))
+//!     fn bytes_encode(half_timestamp: &Self::EItem) -> Result<Self::ReturnBytes, Self::Error> {
+//!         let [bytes @ .., _, _] = half_timestamp.to_be_bytes();
+//!         Ok(bytes)
 //!     }
 //! }
 //!