hashcodecs 1.2.0__tar.gz → 1.2.1__tar.gz

{hashcodecs-1.2.0 → hashcodecs-1.2.1}/CHANGELOG.md

@@ -4,6 +4,16 @@ This file records notable user-facing changes to `hashcodecs`. Version 1.0.0 sta
 
 ## [Unreleased]
 
+## [1.2.1] - 2026-08-26
+
+### What's Changed
+* fix: harden buffer and lenient decode paths by @kozistr in https://github.com/kozistr/hashcodecs-rs/pull/58
+* Refactor Murmur SIMD and centralize XXH3 long dispatch by @kozistr in https://github.com/kozistr/hashcodecs-rs/pull/59
+* Repair sdist verification and roll back unpublished 1.2.1 by @kozistr in https://github.com/kozistr/hashcodecs-rs/pull/60
+
+
+**Full Changelog**: https://github.com/kozistr/hashcodecs-rs/compare/v1.2.0...v1.2.1
+
 ## [1.2.0] - 2026-08-25
 
 ### What's Changed
@@ -117,7 +127,8 @@ This file records notable user-facing changes to `hashcodecs`. Version 1.0.0 sta
 - Initial Python and Rust APIs for Base64 and MurmurHash3.
 - Runtime SIMD dispatch and platform-specific CPython wheels.
 
-[Unreleased]: https://github.com/kozistr/hashcodecs-rs/compare/v1.2.0...HEAD
+[Unreleased]: https://github.com/kozistr/hashcodecs-rs/compare/v1.2.1...HEAD
+[1.2.1]: https://github.com/kozistr/hashcodecs-rs/compare/v1.2.0...v1.2.1
 [1.2.0]: https://github.com/kozistr/hashcodecs-rs/compare/v1.1.0...v1.2.0
 [1.1.0]: https://github.com/kozistr/hashcodecs-rs/compare/v1.0.0...v1.1.0
 [1.0.0]: https://github.com/kozistr/hashcodecs-rs/compare/v0.6.1...v1.0.0

{hashcodecs-1.2.0 → hashcodecs-1.2.1}/CITATION.cff

@@ -5,8 +5,8 @@ authors:
     given-names: Hyeongchan
     orcid: https://orcid.org/0000-0002-1729-0580
 title: "hashcodecs: SIMD-accelerated Base64, MurmurHash3, and XXH3 for Python and Rust"
-version: 1.2.0
-date-released: 2026-08-25
+version: 1.2.1
+date-released: 2026-08-26
 license: "MIT OR Apache-2.0"
 repository-code: "https://github.com/kozistr/hashcodecs-rs"
-url: "https://github.com/kozistr/hashcodecs-rs"
+url: "https://github.com/kozistr/hashcodecs-rs"

{hashcodecs-1.2.0 → hashcodecs-1.2.1}/Cargo.lock

@@ -194,7 +194,7 @@ dependencies = [
 
 [[package]]
 name = "hashcodecs"
-version = "1.2.0"
+version = "1.2.1"
 dependencies = [
  "base64",
  "base64-turbo",

{hashcodecs-1.2.0 → hashcodecs-1.2.1}/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "hashcodecs"
-version = "1.2.0"
+version = "1.2.1"
 edition = "2024"
 rust-version = "1.89"
 description = "SIMD-accelerated Base64 codecs and fast MurmurHash3 and xxHash implementations"
@@ -36,16 +36,16 @@ pyo3 = { version = "0.29.2", optional = true }
 pyo3-build-config = { version = "0.29.2", optional = true }
 
 [dev-dependencies]
-base64 = "0.23.1"
-base64-turbo = "0.3.0"
-criterion = { version = "0.8.2", default-features = false, features = ["cargo_bench_support"] }
-fastmurmur3 = "0.2.0"
-mimalloc = "0.1.52"
-mm3h = "0.1.3"
-murmur3 = "0.5"
-murmurs = "1.0.5"
-xxhash-c-sys = "0.8.7"
-xxhash-rust = { version = "0.8.18", features = ["xxh3"] }
+base64 = "=0.23.1"
+base64-turbo = "=0.3.0"
+criterion = { version = "=0.8.2", default-features = false, features = ["cargo_bench_support"] }
+fastmurmur3 = "=0.2.0"
+mimalloc = "=0.1.52"
+mm3h = "=0.1.3"
+murmur3 = "=0.5.2"
+murmurs = "=1.0.5"
+xxhash-c-sys = "=0.8.7"
+xxhash-rust = { version = "=0.8.18", features = ["xxh3"] }
 
 [[bench]]
 name = "base64"

{hashcodecs-1.2.0 → hashcodecs-1.2.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: hashcodecs
-Version: 1.2.0
+Version: 1.2.1
 Summary: SIMD-accelerated Base64, MurmurHash3, and xxHash codecs
 Project-URL: Documentation, https://hashcodecs-rs.readthedocs.io/
 Project-URL: Repository, https://github.com/kozistr/hashcodecs-rs

{hashcodecs-1.2.0 → hashcodecs-1.2.1}/docs/base64.md

@@ -30,10 +30,12 @@ The `standard_b64encode` and `urlsafe_b64encode` helpers select the respective a
 
 ## Decoding and Validation
 
-`b64decode(s, altchars=None, validate=False, *, padded=True, ignorechars=..., canonical=False)` decodes an ASCII
+`b64decode(s, altchars=None, validate=..., *, padded=True, ignorechars=..., canonical=False)` decodes an ASCII
 string or bytes-like value.
 
 - `validate=True` rejects characters outside the chosen alphabet instead of discarding them.
+- If you supply `ignorechars` without `validate`, hashcodecs uses strict mode. Pass `validate=False` to keep lenient
+  mode; hashcodecs then limits discarded bytes to `ignorechars`.
 - `padded=False` accepts an unpadded final Base64 quantum.
 - `ignorechars` specifies which non-alphabet bytes lenient decoding may ignore. Its default preserves the standard
   library's lenient behavior.

{hashcodecs-1.2.0 → hashcodecs-1.2.1}/hashcodecs/_hashcodecs.pyi

@@ -1,3 +1,5 @@
+from typing import final
+
 from _typeshed import ReadableBuffer
 
 def b64encode(
@@ -146,7 +148,7 @@ def b64decode(
     bit validation are available for protocols with tighter requirements.
 
     Args:
-        s: ASCII text or contiguous bytes-like Base64 data.
+        s: ASCII text or bytes-like Base64 data.
         altchars: Two characters replacing "+" and "/", or None for the
             standard alphabet.
         validate: Reject non-alphabet bytes when true. The default is lenient
@@ -256,7 +258,7 @@ def b64decode_into(
     destination prefix may already have been modified.
 
     Args:
-        s: ASCII text or contiguous bytes-like Base64 data.
+        s: ASCII text or bytes-like Base64 data.
         output: Destination bytearray with room for the complete result.
         altchars: Two characters replacing "+" and "/", or None for the
             standard alphabet.
@@ -348,7 +350,7 @@ def standard_b64decode(s: str | ReadableBuffer) -> bytes:
     Python's base64.standard_b64decode function.
 
     Args:
-        s: ASCII text or contiguous bytes-like Base64 data.
+        s: ASCII text or bytes-like Base64 data.
 
     Returns:
         Newly allocated decoded bytes.
@@ -368,7 +370,7 @@ def standard_b64decode_into(s: str | ReadableBuffer, output: bytearray) -> int:
     """Decode standard Base64 into a reusable bytearray.
 
     Args:
-        s: ASCII text or contiguous bytes-like Base64 data.
+        s: ASCII text or bytes-like Base64 data.
         output: Destination bytearray with room for the decoded result.
 
     Returns:
@@ -615,7 +617,7 @@ def xxh3_64_batch(items: list[ReadableBuffer], seed: int = 0) -> list[int]:
     """Compute canonical XXH3 64-bit hashes for a list of inputs.
 
     Args:
-        items: A list of contiguous bytes-like objects to hash.
+        items: A list of bytes-like objects to hash.
         seed: Initial unsigned 64-bit seed shared by every item.
 
     Returns:
@@ -638,7 +640,7 @@ def xxh3_64_batch_into(items: list[ReadableBuffer], output: bytearray, seed: int
     written prefix remain unchanged.
 
     Args:
-        items: A list of contiguous bytes-like objects to hash.
+        items: A list of bytes-like objects to hash.
         output: Destination with at least 8 * len(items) bytes.
         seed: Initial unsigned 64-bit seed shared by every item.
 
@@ -663,7 +665,7 @@ def xxh3_128_batch(items: list[ReadableBuffer], seed: int = 0) -> list[int]:
     """Compute canonical XXH3 128-bit hashes for a list of inputs.
 
     Args:
-        items: A list of contiguous bytes-like objects to hash.
+        items: A list of bytes-like objects to hash.
         seed: Initial unsigned 64-bit seed shared by every item.
 
     Returns:
@@ -686,7 +688,7 @@ def xxh3_128_batch_into(items: list[ReadableBuffer], output: bytearray, seed: in
     written prefix remain unchanged.
 
     Args:
-        items: A list of contiguous bytes-like objects to hash.
+        items: A list of bytes-like objects to hash.
         output: Destination with at least 16 * len(items) bytes.
         seed: Initial unsigned 64-bit seed shared by every item.
 
@@ -707,11 +709,12 @@ def xxh3_128_batch_into(items: list[ReadableBuffer], output: bytearray, seed: in
     """
     ...
 
+@final
 class murmur3_x86_32:
     """Incremental MurmurHash3 x86 32-bit hasher.
 
     Args:
-        data: Optional initial contiguous bytes-like data.
+        data: Optional initial bytes-like data.
         seed: Initial unsigned 32-bit seed.
 
     Examples:
@@ -725,7 +728,7 @@ class murmur3_x86_32:
         """Initialize an incremental x86 32-bit hash state.
 
         Args:
-            data: Optional initial contiguous bytes-like data.
+            data: Optional initial bytes-like data.
             seed: Initial unsigned 32-bit seed.
 
         Raises:
@@ -738,7 +741,7 @@ class murmur3_x86_32:
         """Append bytes to the hash state.
 
         Args:
-            data: Contiguous bytes-like data to append.
+            data: Bytes-like data to append.
 
         Returns:
             None.
@@ -804,11 +807,12 @@ class murmur3_x86_32:
         """Return the algorithm name (murmur3_x86_32)."""
         ...
 
+@final
 class murmur3_x86_128:
     """Incremental MurmurHash3 x86 128-bit hasher.
 
     Args:
-        data: Optional initial contiguous bytes-like data.
+        data: Optional initial bytes-like data.
         seed: Initial unsigned 32-bit seed.
 
     Examples:
@@ -822,7 +826,7 @@ class murmur3_x86_128:
         """Initialize an incremental x86 128-bit hash state.
 
         Args:
-            data: Optional initial contiguous bytes-like data.
+            data: Optional initial bytes-like data.
             seed: Initial unsigned 32-bit seed.
 
         Raises:
@@ -835,7 +839,7 @@ class murmur3_x86_128:
         """Append bytes to the hash state.
 
         Args:
-            data: Contiguous bytes-like data to append.
+            data: Bytes-like data to append.
 
         Returns:
             None.
@@ -901,11 +905,12 @@ class murmur3_x86_128:
         """Return the algorithm name (murmur3_x86_128)."""
         ...
 
+@final
 class murmur3_x64_128:
     """Incremental MurmurHash3 x64 128-bit hasher.
 
     Args:
-        data: Optional initial contiguous bytes-like data.
+        data: Optional initial bytes-like data.
         seed: Initial unsigned 32-bit seed.
 
     Examples:
@@ -920,7 +925,7 @@ class murmur3_x64_128:
         """Initialize an incremental x64 128-bit hash state.
 
         Args:
-            data: Optional initial contiguous bytes-like data.
+            data: Optional initial bytes-like data.
             seed: Initial unsigned 32-bit seed.
 
         Raises:
@@ -933,7 +938,7 @@ class murmur3_x64_128:
         """Append bytes to the hash state.
 
         Args:
-            data: Contiguous bytes-like data to append.
+            data: Bytes-like data to append.
 
         Returns:
             None.

hashcodecs-1.2.1/hatch_build.py

@@ -0,0 +1,86 @@
+"""Hatchling hook that packages the in-tree PyO3 extension."""
+
+from __future__ import annotations
+
+import json
+import os
+import platform
+import subprocess
+import sys
+from pathlib import Path
+from typing import Any
+
+from hatchling.builders.hooks.plugin.interface import BuildHookInterface
+from packaging.tags import sys_tags
+
+
+class CustomBuildHook(BuildHookInterface[Any]):
+    """Build the CPython extension wheel."""
+
+    def initialize(self, version: str, build_data: dict[str, Any]) -> None:
+        if self.target_name != 'wheel':
+            return
+
+        root = Path(self.root)
+        wheel_tag = self._wheel_tag()
+        env = os.environ.copy()
+        configured_target_dir = env.get('CARGO_TARGET_DIR') or env.get('CARGO_LLVM_COV_TARGET_DIR')
+        if configured_target_dir:
+            target_dir = Path(configured_target_dir)
+            if not target_dir.is_absolute():
+                target_dir = root / target_dir
+        else:
+            target_dir = root / 'target' / 'hatch' / wheel_tag
+        env['CARGO_TARGET_DIR'] = str(target_dir)
+        env['PYO3_PYTHON'] = sys.executable
+        result = subprocess.run(
+            [
+                'cargo',
+                'build',
+                '--locked',
+                '--manifest-path',
+                str(root / 'Cargo.toml'),
+                '--release',
+                '--features',
+                'extension-module',
+                '--message-format=json-render-diagnostics',
+            ],
+            check=False,
+            cwd=root,
+            env=env,
+            stdout=subprocess.PIPE,
+            text=True,
+        )
+        result.check_returncode()
+
+        extension = self._cdylib_artifact(result.stdout)
+        build_data['force_include'] = {str(extension): f'hashcodecs/_hashcodecs{self._extension_suffix()}'}
+        build_data['pure_python'] = False
+        build_data['tag'] = wheel_tag
+
+    @staticmethod
+    def _cdylib_artifact(messages: str) -> Path:
+        suffixes = {'.dll', '.dylib', '.so'}
+        for line in messages.splitlines():
+            try:
+                message = json.loads(line)
+            except json.JSONDecodeError:
+                continue
+            if (
+                message.get('reason') == 'compiler-artifact'
+                and message.get('target', {}).get('name') == 'hashcodecs'
+                and 'cdylib' in message.get('target', {}).get('crate_types', [])
+            ):
+                for filename in message.get('filenames', []):
+                    artifact = Path(filename)
+                    if artifact.suffix in suffixes:
+                        return artifact
+        raise RuntimeError('Cargo did not report the hashcodecs cdylib artifact')
+
+    @staticmethod
+    def _extension_suffix() -> str:
+        return '.pyd' if platform.system() == 'Windows' else '.so'
+
+    @staticmethod
+    def _wheel_tag() -> str:
+        return str(next(sys_tags()))

{hashcodecs-1.2.0 → hashcodecs-1.2.1}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "hashcodecs"
-version = "1.2.0"
+version = "1.2.1"
 description = "SIMD-accelerated Base64, MurmurHash3, and xxHash codecs"
 readme = "README.md"
 requires-python = ">=3.10"
@@ -44,9 +44,9 @@ dev = [
     "ruff>=0.16,<0.17",
 ]
 benchmark = [
-    "mmh3>=5.2.1,<6",
-    "pybase64>=1.4.3,<2",
-    "xxhash>=3.6,<4",
+    "mmh3==5.2.1",
+    "pybase64==1.4.3",
+    "xxhash==3.8.1",
 ]
 
 [tool.hatch.build.targets.wheel]
@@ -71,8 +71,6 @@ include = [
     "/src",
     "/tools",
 ]
-exclude = ["/src/base64/tests.rs"]
-
 [tool.hatch.build.hooks.custom]
 path = "hatch_build.py"
 

{hashcodecs-1.2.0 → hashcodecs-1.2.1}/src/backend.rs

@@ -101,6 +101,7 @@ fn detect() -> Capabilities {
     {
         x86_capabilities(
             std::is_x86_feature_detected!("avx512vbmi"),
+            std::is_x86_feature_detected!("avx512bw"),
             std::is_x86_feature_detected!("avx512f"),
             std::is_x86_feature_detected!("avx2"),
             std::is_x86_feature_detected!("sse4.1"),
@@ -121,6 +122,7 @@ fn detect() -> Capabilities {
 #[cfg(any(test, target_arch = "x86", target_arch = "x86_64"))]
 fn x86_capabilities(
     avx512_vbmi: bool,
+    avx512_bw: bool,
     avx512: bool,
     avx2: bool,
     sse41: bool,
@@ -143,7 +145,7 @@ fn x86_capabilities(
     if avx512 && avx2 && ssse3 {
         simd |= SimdBackend::Avx512.bit();
     }
-    if avx512 && avx512_vbmi && avx2 && ssse3 {
+    if avx512 && avx512_bw && avx512_vbmi && avx2 && ssse3 {
         simd |= SimdBackend::Avx512Vbmi.bit();
     }
     Capabilities { simd, bmi2 }
@@ -163,40 +165,43 @@ mod tests {
 
     #[test]
     fn x86_detection_tracks_independent_feature_sets() {
-        let scalar = x86_capabilities(false, false, false, false, false, false);
+        let scalar = x86_capabilities(false, false, false, false, false, false, false);
         assert_eq!(scalar.best(&[SimdBackend::Avx2]), SimdBackend::Scalar);
 
-        let ssse3 = x86_capabilities(false, false, false, false, true, false);
+        let ssse3 = x86_capabilities(false, false, false, false, false, true, false);
         assert!(ssse3.supports(SimdBackend::Ssse3));
         assert!(!ssse3.supports(SimdBackend::Sse41));
 
-        let incomplete_sse41 = x86_capabilities(false, false, false, true, false, false);
+        let incomplete_sse41 = x86_capabilities(false, false, false, false, true, false, false);
         assert!(!incomplete_sse41.supports(SimdBackend::Sse41));
 
-        let sse41 = x86_capabilities(false, false, false, true, true, false);
+        let sse41 = x86_capabilities(false, false, false, false, true, true, false);
         assert!(sse41.supports(SimdBackend::Sse41));
 
-        let incomplete_avx2 = x86_capabilities(false, false, true, false, false, true);
+        let incomplete_avx2 = x86_capabilities(false, false, false, true, false, false, true);
         assert!(!incomplete_avx2.supports(SimdBackend::Avx2));
-        let avx2 = x86_capabilities(false, false, true, false, true, true);
+        let avx2 = x86_capabilities(false, false, false, true, false, true, true);
         assert!(avx2.supports(SimdBackend::Avx2));
         assert!(!avx2.supports(SimdBackend::Sse41));
         assert!(avx2.has_bmi2());
 
-        let avx512 = x86_capabilities(false, true, false, false, false, false);
+        let avx512 = x86_capabilities(false, false, true, false, false, false, false);
         assert!(!avx512.supports(SimdBackend::Avx512));
         assert!(!avx512.supports(SimdBackend::Avx2));
         assert!(!avx512.supports(SimdBackend::Avx512Vbmi));
 
-        let avx512 = x86_capabilities(false, true, true, false, true, false);
+        let avx512 = x86_capabilities(false, false, true, true, false, true, false);
         assert!(avx512.supports(SimdBackend::Avx512));
         assert!(avx512.supports(SimdBackend::Avx2));
         assert!(!avx512.supports(SimdBackend::Avx512Vbmi));
 
-        let incomplete_vbmi = x86_capabilities(true, false, false, false, false, false);
+        let incomplete_bw = x86_capabilities(true, false, true, true, true, true, true);
+        assert!(!incomplete_bw.supports(SimdBackend::Avx512Vbmi));
+
+        let incomplete_vbmi = x86_capabilities(false, true, true, true, true, true, true);
         assert!(!incomplete_vbmi.supports(SimdBackend::Avx512Vbmi));
 
-        let vbmi = x86_capabilities(true, true, true, true, true, true);
+        let vbmi = x86_capabilities(true, true, true, true, true, true, true);
         assert_eq!(
             vbmi.best(&[
                 SimdBackend::Avx512Vbmi,

{hashcodecs-1.2.0 → hashcodecs-1.2.1}/src/base64/decode/avx512.rs

@@ -31,7 +31,7 @@ const fn decode_shuffle() -> [u8; 64] {
     shuffle
 }
 
-#[target_feature(enable = "avx512vbmi")]
+#[target_feature(enable = "avx512vbmi,avx512bw")]
 pub(in crate::base64) unsafe fn decode<A: Decoder, S: Store>(
     input: &[u8],
     output: *mut u8,
@@ -100,7 +100,7 @@ pub(in crate::base64) unsafe fn decode<A: Decoder, S: Store>(
     }
 }
 
-#[target_feature(enable = "avx512vbmi")]
+#[target_feature(enable = "avx512vbmi,avx512bw")]
 #[inline]
 unsafe fn decode_64(
     input: *const u8,

{hashcodecs-1.2.0 → hashcodecs-1.2.1}/src/base64/decode.rs

@@ -24,7 +24,8 @@ use super::{
 ///
 /// The input must be quartet-aligned, use only the standard alphabet, and have
 /// valid trailing padding. Unlike Python's lenient decoder, this Rust API does
-/// not ignore whitespace or other non-alphabet bytes.
+/// not ignore whitespace or other non-alphabet bytes. Non-zero unused bits in
+/// the final quantum are accepted rather than treated as a canonicality error.
 ///
 /// # Arguments
 ///
@@ -54,7 +55,8 @@ pub fn b64decode(input: &[u8]) -> Result<Vec<u8>, Base64Error> {
 /// Decodes padded RFC 4648 Base64 with the URL-safe alphabet.
 ///
 /// The input must be quartet-aligned, use only the URL-safe alphabet, and have
-/// valid trailing padding.
+/// valid trailing padding. Non-zero unused bits in the final quantum are
+/// accepted rather than treated as a canonicality error.
 ///
 /// # Arguments
 ///
@@ -259,6 +261,7 @@ pub(crate) fn decode_unpadded_layout(input: &[u8]) -> Result<DecodeLayout, Base6
         .and_then(|length| length.checked_add(tail_len))
         .ok_or(Base64Error::InvalidInput)?;
     Ok(DecodeLayout {
+        input_len: input.len(),
         padding: 0,
         output_len,
     })
@@ -286,7 +289,8 @@ pub(crate) fn decode_to_slice_with_layout_and_alphabet(
     layout: DecodeLayout,
     alphabet: DecodeAlphabet,
 ) -> Result<(), Base64Error> {
-    debug_assert_eq!(output.len(), layout.output_len);
+    assert_eq!(input.len(), layout.input_len);
+    assert_eq!(output.len(), layout.output_len);
     // The slice has exactly the required initialized storage, so only bounded
     // stores are permitted.
     unsafe { decode_to_ptr_with_layout(input, output.as_mut_ptr(), layout, alphabet, false) }
@@ -300,7 +304,8 @@ pub(crate) fn decode_to_slice_with_unpadded_layout_and_alphabet(
     layout: DecodeLayout,
     alphabet: DecodeAlphabet,
 ) -> Result<(), Base64Error> {
-    debug_assert_eq!(output.len(), layout.output_len);
+    assert_eq!(input.len(), layout.input_len);
+    assert_eq!(output.len(), layout.output_len);
     unsafe {
         decode_to_ptr_with_unpadded_layout_mode(input, output.as_mut_ptr(), layout, alphabet, false)
     }
@@ -314,7 +319,8 @@ pub(crate) fn decode_to_slice_with_unpadded_layout_and_alphabet_transactional(
     layout: DecodeLayout,
     alphabet: DecodeAlphabet,
 ) -> Result<(), Base64Error> {
-    debug_assert_eq!(output.len(), layout.output_len);
+    assert_eq!(input.len(), layout.input_len);
+    assert_eq!(output.len(), layout.output_len);
     unsafe {
         decode_to_ptr_with_unpadded_layout_mode(input, output.as_mut_ptr(), layout, alphabet, true)
     }
@@ -350,7 +356,8 @@ pub(crate) fn decode_to_slice_with_layout_and_alphabet_transactional(
     layout: DecodeLayout,
     alphabet: DecodeAlphabet,
 ) -> Result<(), Base64Error> {
-    debug_assert_eq!(output.len(), layout.output_len);
+    assert_eq!(input.len(), layout.input_len);
+    assert_eq!(output.len(), layout.output_len);
     // Transactional SIMD error handling writes only complete, validated blocks,
     // so a lenient caller can safely fall back without modifying the suffix.
     unsafe {
@@ -434,6 +441,7 @@ unsafe fn decode_to_ptr_with_unpadded_layout_mode(
 ) -> Result<(), Base64Error> {
     let prefix_len = input.len() / 4 * 4;
     let prefix_layout = DecodeLayout {
+        input_len: prefix_len,
         padding: 0,
         output_len: prefix_len / 4 * 3,
     };
@@ -468,6 +476,7 @@ unsafe fn decode_to_ptr_with_unpadded_layout_mode(
 pub(crate) fn decode_layout(input: &[u8]) -> Result<DecodeLayout, Base64Error> {
     if input.is_empty() {
         return Ok(DecodeLayout {
+            input_len: 0,
             padding: 0,
             output_len: 0,
         });
@@ -482,6 +491,7 @@ pub(crate) fn decode_layout(input: &[u8]) -> Result<DecodeLayout, Base64Error> {
         _ => 0,
     };
     Ok(DecodeLayout {
+        input_len: input.len(),
         padding,
         output_len: input.len() / 4 * 3 - padding,
     })
@@ -489,8 +499,17 @@ pub(crate) fn decode_layout(input: &[u8]) -> Result<DecodeLayout, Base64Error> {
 
 #[derive(Clone, Copy)]
 pub(crate) struct DecodeLayout {
-    pub(crate) padding: usize,
-    pub(crate) output_len: usize,
+    input_len: usize,
+    padding: usize,
+    output_len: usize,
+}
+
+impl DecodeLayout {
+    #[inline(always)]
+    #[cfg(any(feature = "python", test))]
+    pub(crate) fn output_len(self) -> usize {
+        self.output_len
+    }
 }
 
 pub(crate) unsafe fn decode_quad_ptr(

{hashcodecs-1.2.0 → hashcodecs-1.2.1}/src/base64/encode/avx512.rs

@@ -28,7 +28,7 @@ const fn encode_shuffle() -> [u8; 64] {
     shuffle
 }
 
-#[target_feature(enable = "avx512vbmi")]
+#[target_feature(enable = "avx512vbmi,avx512bw")]
 pub(in crate::base64) unsafe fn encode<const URLSAFE: bool>(
     input: &[u8],
     output: *mut u8,
@@ -116,7 +116,7 @@ pub(in crate::base64) unsafe fn encode<const URLSAFE: bool>(
     }
 }
 
-#[target_feature(enable = "avx512vbmi")]
+#[target_feature(enable = "avx512vbmi,avx512bw")]
 #[inline]
 unsafe fn encode_48(
     input: *const u8,

{hashcodecs-1.2.0 → hashcodecs-1.2.1}/src/base64/tests/aarch64.rs

@@ -489,10 +489,10 @@ fn invalid_scalar_checkpoint_tail<const URLSAFE: bool, const MIXED: bool>() {
 
             let output_offset = (input_offset * 7 + invalid_index) & 15;
             let output_start = GUARD + output_offset;
-            let mut guarded_output = vec![CANARY; output_start + layout.output_len + GUARD];
+            let mut guarded_output = vec![CANARY; output_start + layout.output_len() + GUARD];
             let result = super::super::decode_to_slice_with_layout_and_alphabet(
                 input,
-                &mut guarded_output[output_start..output_start + layout.output_len],
+                &mut guarded_output[output_start..output_start + layout.output_len()],
                 layout,
                 decode_alphabet::<URLSAFE, MIXED>(),
             );