Add v3-preview module gathering a few breaking changes (#105)

This new module addresses the following breaking changes of #72: - `static` instead of `const` for pre-defined encodings - Private `Encoding` implementation with `unsafe` constructor - Use `MaybeUninit` internally and expose `_uninit` functions - Use static dispatch instead of dynamic dispatch (fixing #97)
ia0 · May 5, 2024 · 2591b46 · 2591b46
1 parent 18203b9
commit 2591b46
Show file tree

Hide file tree

Showing 11 changed files with 1,556 additions and 90 deletions.
diff --git a/lib/Cargo.toml b/lib/Cargo.toml
@@ -21,3 +21,4 @@ rustdoc-args = ["--cfg=docsrs"]
 default = ["std"]
 alloc = []
 std = ["alloc"]
+v3-preview = []
diff --git a/lib/benches/lib.rs b/lib/benches/lib.rs
@@ -1,9 +1,17 @@
 #![feature(test)]
 
-extern crate data_encoding;
 extern crate test;
 
-use data_encoding::{Specification, BASE32, BASE32_DNSCURVE, BASE64, BASE64_NOPAD, HEXLOWER};
+#[cfg(feature = "v3-preview")]
+use std::convert::TryFrom as _;
+
+#[cfg(not(feature = "v3-preview"))]
+use data_encoding as constants;
+#[cfg(feature = "v3-preview")]
+use data_encoding::v3_preview as constants;
+#[cfg(feature = "v3-preview")]
+use data_encoding::v3_preview::{Bit1, Bit2, Bit3, Bit6, Encoding, False, True};
+use data_encoding::Specification;
 use test::Bencher;
 
 #[bench]
@@ -13,6 +21,8 @@ fn base02_encode_base(b: &mut Bencher) {
     let mut spec = Specification::new();
     spec.symbols.push_str("01");
     let base = spec.encoding().unwrap();
+    #[cfg(feature = "v3-preview")]
+    let base = Encoding::<Bit1, True, False, False, False>::try_from(base).unwrap();
     b.iter(|| base.encode_mut(input, output));
 }
 
@@ -23,6 +33,8 @@ fn base02_decode_base(b: &mut Bencher) {
     let mut spec = Specification::new();
     spec.symbols.push_str("01");
     let base = spec.encoding().unwrap();
+    #[cfg(feature = "v3-preview")]
+    let base = Encoding::<Bit1, True, False, False, False>::try_from(base).unwrap();
     b.iter(|| base.decode_mut(input, output));
 }
 
@@ -33,6 +45,8 @@ fn base04_encode_base(b: &mut Bencher) {
     let mut spec = Specification::new();
     spec.symbols.push_str("0123");
     let base = spec.encoding().unwrap();
+    #[cfg(feature = "v3-preview")]
+    let base = Encoding::<Bit2, True, False, False, False>::try_from(base).unwrap();
     b.iter(|| base.encode_mut(input, output));
 }
 
@@ -43,6 +57,8 @@ fn base04_decode_base(b: &mut Bencher) {
     let mut spec = Specification::new();
     spec.symbols.push_str("0123");
     let base = spec.encoding().unwrap();
+    #[cfg(feature = "v3-preview")]
+    let base = Encoding::<Bit2, True, False, False, False>::try_from(base).unwrap();
     b.iter(|| base.decode_mut(input, output));
 }
 
@@ -53,6 +69,8 @@ fn base08_encode_base(b: &mut Bencher) {
     let mut spec = Specification::new();
     spec.symbols.push_str("01234567");
     let base = spec.encoding().unwrap();
+    #[cfg(feature = "v3-preview")]
+    let base = Encoding::<Bit3, True, False, False, False>::try_from(base).unwrap();
     b.iter(|| base.encode_mut(input, output));
 }
 
@@ -63,56 +81,58 @@ fn base08_decode_base(b: &mut Bencher) {
     let mut spec = Specification::new();
     spec.symbols.push_str("01234567");
     let base = spec.encoding().unwrap();
+    #[cfg(feature = "v3-preview")]
+    let base = Encoding::<Bit3, True, False, False, False>::try_from(base).unwrap();
     b.iter(|| base.decode_mut(input, output));
 }
 
 #[bench]
 fn base16_encode_base(b: &mut Bencher) {
     let input = &[0u8; 4096];
     let output = &mut [0u8; 8192];
-    b.iter(|| HEXLOWER.encode_mut(input, output));
+    b.iter(|| constants::HEXLOWER.encode_mut(input, output));
 }
 
 #[bench]
 fn base16_decode_base(b: &mut Bencher) {
     let input = &[b'0'; 4096];
     let output = &mut [0u8; 2048];
-    b.iter(|| HEXLOWER.decode_mut(input, output));
+    b.iter(|| constants::HEXLOWER.decode_mut(input, output));
 }
 
 #[bench]
 fn base32_encode_base(b: &mut Bencher) {
     let input = &[0u8; 4096];
     let output = &mut [0u8; 6560];
-    b.iter(|| BASE32.encode_mut(input, output));
+    b.iter(|| constants::BASE32.encode_mut(input, output));
 }
 
 #[bench]
 fn base32_decode_base(b: &mut Bencher) {
     let input = &[b'A'; 4096];
     let output = &mut [0u8; 2560];
-    b.iter(|| BASE32.decode_mut(input, output));
+    b.iter(|| constants::BASE32.decode_mut(input, output));
 }
 
 #[bench]
 fn base64_encode_base(b: &mut Bencher) {
     let input = &[0u8; 4096];
     let output = &mut [0u8; 5462];
-    b.iter(|| BASE64_NOPAD.encode_mut(input, output));
+    b.iter(|| constants::BASE64_NOPAD.encode_mut(input, output));
 }
 
 #[bench]
 fn base64_decode_base(b: &mut Bencher) {
     let input = &[b'A'; 4096];
     let output = &mut [0u8; 3072];
-    b.iter(|| BASE64_NOPAD.decode_mut(input, output));
+    b.iter(|| constants::BASE64_NOPAD.decode_mut(input, output));
 }
 
 #[bench]
 fn base64_encode_pad(b: &mut Bencher) {
     let input = &mut [b'A'; 4096];
     let output = &mut [0u8; 5464];
-    b.iter(|| BASE64.encode_mut(input, output));
+    b.iter(|| constants::BASE64.encode_mut(input, output));
 }
 
 #[bench]
@@ -126,17 +146,19 @@ fn base64_decode_pad(b: &mut Bencher) {
         }
     }
     let output = &mut [0u8; 3072];
-    b.iter(|| BASE64.decode_mut(input, output).unwrap());
+    b.iter(|| constants::BASE64.decode_mut(input, output).unwrap());
 }
 
 #[bench]
 fn base64_encode_wrap(b: &mut Bencher) {
     let input = &[0u8; 4096];
     let output = &mut [0u8; 5608];
-    let mut spec = BASE64.specification();
+    let mut spec = data_encoding::BASE64.specification();
     spec.wrap.width = 76;
     spec.wrap.separator.push_str("\r\n");
     let base64 = spec.encoding().unwrap();
+    #[cfg(feature = "v3-preview")]
+    let base64 = Encoding::<Bit6, True, True, True, True>::try_from(base64).unwrap();
     b.iter(|| base64.encode_mut(input, output));
 }
 
@@ -148,23 +170,25 @@ fn base64_decode_wrap(b: &mut Bencher) {
         input[x + 3] = b'\n';
     }
     let output = &mut [0u8; 3072];
-    let mut spec = BASE64.specification();
+    let mut spec = data_encoding::BASE64.specification();
     spec.wrap.width = 76;
     spec.wrap.separator.push_str("\r\n");
     let base64 = spec.encoding().unwrap();
+    #[cfg(feature = "v3-preview")]
+    let base64 = Encoding::<Bit6, True, True, True, True>::try_from(base64).unwrap();
     b.iter(|| base64.decode_mut(input, output).unwrap());
 }
 
 #[bench]
 fn dnscurve_decode_base(b: &mut Bencher) {
     let input = &[b'0'; 4096];
     let output = &mut [0u8; 2560];
-    b.iter(|| BASE32_DNSCURVE.decode_mut(input, output));
+    b.iter(|| constants::BASE32_DNSCURVE.decode_mut(input, output));
 }
 
 #[bench]
 fn dnscurve_encode_base(b: &mut Bencher) {
     let input = &[0u8; 4096];
     let output = &mut [0u8; 6554];
-    b.iter(|| BASE32_DNSCURVE.encode_mut(input, output));
+    b.iter(|| constants::BASE32_DNSCURVE.encode_mut(input, output));
 }
diff --git a/lib/fuzz/Cargo.toml b/lib/fuzz/Cargo.toml
@@ -29,3 +29,10 @@ name = "encode_write"
 path = "fuzz_targets/encode_write.rs"
 test = false
 doc = false
+
+[[bin]]
+name = "v3-preview"
+path = "fuzz_targets/v3-preview.rs"
+test = false
+doc = false
+required-features = ["data-encoding/v3-preview"]
diff --git a/lib/fuzz/examples/analyze.rs b/lib/fuzz/examples/analyze.rs
@@ -1,6 +1,3 @@
-extern crate data_encoding;
-extern crate data_encoding_fuzz;
-
 use std::collections::HashMap;
 use std::ops::AddAssign;
 

diff --git a/lib/fuzz/examples/debug.rs b/lib/fuzz/examples/debug.rs
@@ -1,5 +1,3 @@
-extern crate data_encoding_fuzz;
-
 use std::io::Read;
 
 use data_encoding_fuzz::generate_specification;

diff --git a/lib/fuzz/fuzz_targets/v3-preview.rs b/lib/fuzz/fuzz_targets/v3-preview.rs
@@ -0,0 +1,98 @@
+#![no_main]
+
+use data_encoding::v3_preview::{Bit1, Bit2, Bit3, Bit4, Bit5, Bit6, Encoding, False, True};
+use data_encoding_fuzz::{decode_prefix, generate_encoding};
+use libfuzzer_sys::fuzz_target;
+
+fuzz_target!(|data: &[u8]| {
+    let mut data = data;
+    let dyn_base = generate_encoding(&mut data);
+    let mut count = 0;
+    macro_rules! test {
+        ($Bit:ident, $Msb:ident, $Pad:ident, $Wrap:ident, $Ignore:ident) => {
+            if let Ok(base) = <&Encoding<$Bit, $Msb, $Pad, $Wrap, $Ignore>>::try_from(&dyn_base) {
+                count += 1;
+                let encoded = base.encode(data);
+                assert_eq!(encoded, dyn_base.encode(data));
+                assert_eq!(base.decode(encoded.as_bytes()).unwrap(), data);
+                if dyn_base.is_canonical() {
+                    let raw = decode_prefix(&dyn_base, &mut data);
+                    assert_eq!(base.encode(&raw).as_bytes(), data);
+                }
+            }
+        };
+    }
+    test!(Bit1, False, False, False, False);
+    test!(Bit1, False, False, False, True);
+    test!(Bit1, False, False, True, True);
+    test!(Bit1, False, True, False, False);
+    test!(Bit1, False, True, False, True);
+    test!(Bit1, False, True, True, True);
+    test!(Bit1, True, False, False, False);
+    test!(Bit1, True, False, False, True);
+    test!(Bit1, True, False, True, True);
+    test!(Bit1, True, True, False, False);
+    test!(Bit1, True, True, False, True);
+    test!(Bit1, True, True, True, True);
+    test!(Bit2, False, False, False, False);
+    test!(Bit2, False, False, False, True);
+    test!(Bit2, False, False, True, True);
+    test!(Bit2, False, True, False, False);
+    test!(Bit2, False, True, False, True);
+    test!(Bit2, False, True, True, True);
+    test!(Bit2, True, False, False, False);
+    test!(Bit2, True, False, False, True);
+    test!(Bit2, True, False, True, True);
+    test!(Bit2, True, True, False, False);
+    test!(Bit2, True, True, False, True);
+    test!(Bit2, True, True, True, True);
+    test!(Bit3, False, False, False, False);
+    test!(Bit3, False, False, False, True);
+    test!(Bit3, False, False, True, True);
+    test!(Bit3, False, True, False, False);
+    test!(Bit3, False, True, False, True);
+    test!(Bit3, False, True, True, True);
+    test!(Bit3, True, False, False, False);
+    test!(Bit3, True, False, False, True);
+    test!(Bit3, True, False, True, True);
+    test!(Bit3, True, True, False, False);
+    test!(Bit3, True, True, False, True);
+    test!(Bit3, True, True, True, True);
+    test!(Bit4, False, False, False, False);
+    test!(Bit4, False, False, False, True);
+    test!(Bit4, False, False, True, True);
+    test!(Bit4, False, True, False, False);
+    test!(Bit4, False, True, False, True);
+    test!(Bit4, False, True, True, True);
+    test!(Bit4, True, False, False, False);
+    test!(Bit4, True, False, False, True);
+    test!(Bit4, True, False, True, True);
+    test!(Bit4, True, True, False, False);
+    test!(Bit4, True, True, False, True);
+    test!(Bit4, True, True, True, True);
+    test!(Bit5, False, False, False, False);
+    test!(Bit5, False, False, False, True);
+    test!(Bit5, False, False, True, True);
+    test!(Bit5, False, True, False, False);
+    test!(Bit5, False, True, False, True);
+    test!(Bit5, False, True, True, True);
+    test!(Bit5, True, False, False, False);
+    test!(Bit5, True, False, False, True);
+    test!(Bit5, True, False, True, True);
+    test!(Bit5, True, True, False, False);
+    test!(Bit5, True, True, False, True);
+    test!(Bit5, True, True, True, True);
+    test!(Bit6, False, False, False, False);
+    test!(Bit6, False, False, False, True);
+    test!(Bit6, False, False, True, True);
+    test!(Bit6, False, True, False, False);
+    test!(Bit6, False, True, False, True);
+    test!(Bit6, False, True, True, True);
+    test!(Bit6, True, False, False, False);
+    test!(Bit6, True, False, False, True);
+    test!(Bit6, True, False, True, True);
+    test!(Bit6, True, True, False, False);
+    test!(Bit6, True, True, False, True);
+    test!(Bit6, True, True, True, True);
+    assert_eq!(count, 1);
+});
diff --git a/lib/fuzz/src/lib.rs b/lib/fuzz/src/lib.rs
@@ -1,5 +1,3 @@
-extern crate data_encoding;
-
 use data_encoding::{DecodePartial, Encoding, Specification};
 
 pub fn generate_encoding(data: &mut &[u8]) -> Encoding {

diff --git a/lib/src/lib.rs b/lib/src/lib.rs
@@ -184,6 +184,9 @@ macro_rules! check {
     };
 }
 
+#[cfg(feature = "v3-preview")]
+pub mod v3_preview;
+
 trait Static<T: Copy>: Copy {
     fn val(self) -> T;
 }
@@ -255,12 +258,12 @@ macro_rules! dispatch {
     ($body: expr) => { $body };
 }
 
-unsafe fn chunk_unchecked(x: &[u8], n: usize, i: usize) -> &[u8] {
+unsafe fn chunk_unchecked<T>(x: &[T], n: usize, i: usize) -> &[T] {
     debug_assert!((i + 1) * n <= x.len());
     unsafe { core::slice::from_raw_parts(x.as_ptr().add(n * i), n) }
 }
 
-unsafe fn chunk_mut_unchecked(x: &mut [u8], n: usize, i: usize) -> &mut [u8] {
+unsafe fn chunk_mut_unchecked<T>(x: &mut [T], n: usize, i: usize) -> &mut [T] {
     debug_assert!((i + 1) * n <= x.len());
     unsafe { core::slice::from_raw_parts_mut(x.as_mut_ptr().add(n * i), n) }
 }
@@ -273,6 +276,7 @@ fn floor(x: usize, m: usize) -> usize {
     x / m * m
 }
 
+#[inline]
 fn vectorize<F: FnMut(usize)>(n: usize, bs: usize, mut f: F) {
     for k in 0 .. n / bs {
         for i in k * bs .. (k + 1) * bs {
@@ -362,6 +366,7 @@ fn order(msb: bool, n: usize, i: usize) -> usize {
     }
 }
 
+#[inline]
 fn enc(bit: usize) -> usize {
     match bit {
         1 | 2 | 4 => 1,
@@ -371,6 +376,7 @@ fn enc(bit: usize) -> usize {
     }
 }
 
+#[inline]
 fn dec(bit: usize) -> usize {
     enc(bit) * 8 / bit
 }
@@ -880,6 +886,7 @@ pub type InternalEncoding = &'static [u8];
 // - width % dec(bit) == 0
 // - for all x in separator values[x] is IGNORE
 #[derive(Debug, Clone, PartialEq, Eq)]
+#[repr(transparent)]
 pub struct Encoding(#[doc(hidden)] pub InternalEncoding);
 
 /// How to translate characters when decoding