cksum: Add support for blake3

RenjiSann · RenjiSann · commit 7a67af5a4c3e · 2026-03-21T13:38:52.000+01:00
diff --git a/src/uu/b2sum/src/b2sum.rs b/src/uu/b2sum/src/b2sum.rs
@@ -9,18 +9,14 @@ use clap::Command;
 
 use uu_checksum_common::{standalone_checksum_app_with_length, standalone_with_length_main};
 
-use uucore::checksum::{AlgoKind, calculate_blake2b_length_str};
+use uucore::checksum::{AlgoKind, calculate_blake_length_str};
 use uucore::error::UResult;
 use uucore::translate;
 
 #[uucore::main]
 pub fn uumain(args: impl uucore::Args) -> UResult<()> {
-    standalone_with_length_main(
-        AlgoKind::Blake2b,
-        uu_app(),
-        args,
-        calculate_blake2b_length_str,
-    )
+    let calculate_blake2b_length = |s: &str| calculate_blake_length_str(AlgoKind::Blake2b, s);
+    standalone_with_length_main(AlgoKind::Blake2b, uu_app(), args, calculate_blake2b_length)
 }
 
 #[inline]
diff --git a/src/uu/cksum/src/cksum.rs b/src/uu/cksum/src/cksum.rs
@@ -12,7 +12,7 @@ use uu_checksum_common::{ChecksumCommand, checksum_main, default_checksum_app, o
 
 use uucore::checksum::compute::OutputFormat;
 use uucore::checksum::{
-    AlgoKind, ChecksumError, calculate_blake2b_length_str, sanitize_sha2_sha3_length_str,
+    AlgoKind, ChecksumError, calculate_blake_length_str, sanitize_sha2_sha3_length_str,
 };
 use uucore::error::UResult;
 use uucore::hardware::{HasHardwareFeatures as _, SimdPolicy};
@@ -67,8 +67,10 @@ fn maybe_sanitize_length(
             Err(_) => Err(ChecksumError::InvalidLength(len.into()).into()),
         },
 
-        // For BLAKE2b, if a length is provided, validate it.
-        (Some(AlgoKind::Blake2b), Some(len)) => calculate_blake2b_length_str(len),
+        // For BLAKE, if a length is provided, validate it.
+        (Some(algo @ (AlgoKind::Blake2b | AlgoKind::Blake3)), Some(len)) => {
+            calculate_blake_length_str(algo, len)
+        }
 
         // For any other provided algorithm, check if length is 0.
         // Otherwise, this is an error.
diff --git a/src/uucore/src/lib/features/checksum/mod.rs b/src/uucore/src/lib/features/checksum/mod.rs
@@ -115,6 +115,8 @@ impl AlgoKind {
             ALGORITHM_OPTIONS_SHA384 => Sha384,
             ALGORITHM_OPTIONS_SHA512 => Sha512,
 
+            // Extensions not in GNU as of version 9.10
+            ALGORITHM_OPTIONS_BLAKE3 => Blake3,
             ALGORITHM_OPTIONS_SHAKE128 => Shake128,
             ALGORITHM_OPTIONS_SHAKE256 => Shake256,
             _ => return Err(ChecksumError::UnknownAlgorithm(algo.as_ref().to_string()).into()),
@@ -245,11 +247,11 @@ pub enum SizedAlgoKind {
     Md5,
     Sm3,
     Sha1,
-    Blake3,
     Sha2(ShaLength),
     Sha3(ShaLength),
-    // Note: we store Blake2b's length as BYTES.
+    // Note: we store Blake*'s length as BYTES.
     Blake2b(Option<usize>),
+    Blake3(Option<usize>),
     // Shake* length are stored in bits.
     Shake128(Option<usize>),
     Shake256(Option<usize>),
@@ -267,7 +269,6 @@ impl SizedAlgoKind {
                 | ak::Md5
                 | ak::Sm3
                 | ak::Sha1
-                | ak::Blake3
                 | ak::Sha224
                 | ak::Sha256
                 | ak::Sha384
@@ -282,8 +283,8 @@ impl SizedAlgoKind {
             (ak::Md5, _) => Ok(Self::Md5),
             (ak::Sm3, _) => Ok(Self::Sm3),
             (ak::Sha1, _) => Ok(Self::Sha1),
-            (ak::Blake3, _) => Ok(Self::Blake3),
 
+            (ak::Blake3, l) => Ok(Self::Blake3(l)),
             (ak::Shake128, l) => Ok(Self::Shake128(l)),
             (ak::Shake256, l) => Ok(Self::Shake256(l)),
             (ak::Sha2, Some(l)) => Ok(Self::Sha2(ShaLength::try_from(l)?)),
@@ -293,7 +294,8 @@ impl SizedAlgoKind {
             }
             // [`calculate_blake2b_length`] expects a length in bits but we
             // have a length in bytes.
-            (ak::Blake2b, Some(l)) => Ok(Self::Blake2b(calculate_blake2b_length_str(
+            (algo @ ak::Blake2b, Some(l)) => Ok(Self::Blake2b(calculate_blake_length_str(
+                algo,
                 &(8 * l).to_string(),
             )?)),
             (ak::Blake2b, None) => Ok(Self::Blake2b(None)),
@@ -310,11 +312,16 @@ impl SizedAlgoKind {
             Self::Md5 => "MD5".into(),
             Self::Sm3 => "SM3".into(),
             Self::Sha1 => "SHA1".into(),
-            Self::Blake3 => "BLAKE3".into(),
             Self::Sha2(len) => format!("SHA{}", len.as_usize()),
             Self::Sha3(len) => format!("SHA3-{}", len.as_usize()),
             Self::Blake2b(Some(byte_len)) => format!("BLAKE2b-{}", byte_len * 8),
             Self::Blake2b(None) => "BLAKE2b".into(),
+            Self::Blake3(byte_len) => {
+                format!(
+                    "BLAKE3-{}",
+                    byte_len.unwrap_or(Blake3::DEFAULT_BYTE_SIZE) * 8
+                )
+            }
             Self::Shake128(opt_bit_len) => format!(
                 "SHAKE128-{}",
                 opt_bit_len.unwrap_or(Shake128::DEFAULT_BIT_SIZE)
@@ -339,7 +346,6 @@ impl SizedAlgoKind {
             Self::Md5 => Box::new(Md5::default()),
             Self::Sm3 => Box::new(Sm3::default()),
             Self::Sha1 => Box::new(Sha1::default()),
-            Self::Blake3 => Box::new(Blake3::default()),
             Self::Sha2(Len224) => Box::new(Sha224::default()),
             Self::Sha2(Len256) => Box::new(Sha256::default()),
             Self::Sha2(Len384) => Box::new(Sha384::default()),
@@ -351,6 +357,9 @@ impl SizedAlgoKind {
             Self::Blake2b(len_opt) => {
                 Box::new(len_opt.map(Blake2b::with_output_bytes).unwrap_or_default())
             }
+            Self::Blake3(len_opt) => {
+                Box::new(len_opt.map(Blake3::with_output_bytes).unwrap_or_default())
+            }
             Self::Shake128(len_opt) => {
                 Box::new(len_opt.map(Shake128::with_output_bits).unwrap_or_default())
             }
@@ -369,7 +378,7 @@ impl SizedAlgoKind {
             Self::Md5 => 128,
             Self::Sm3 => 512,
             Self::Sha1 => 160,
-            Self::Blake3 => 256,
+            Self::Blake3(len) => len.unwrap_or(Blake3::DEFAULT_BYTE_SIZE) * 8,
             Self::Sha2(len) => len.as_usize(),
             Self::Sha3(len) => len.as_usize(),
             Self::Blake2b(len) => len.unwrap_or(Blake2b::DEFAULT_BYTE_SIZE * 8),
@@ -486,20 +495,22 @@ pub fn digest_reader<T: Read>(
     Ok((digest.result(), output_size))
 }
 
-/// Calculates the length of the digest.
-pub fn calculate_blake2b_length_str(bit_length: &str) -> UResult<Option<usize>> {
+/// Calculates the BYTE length of the digest.
+pub fn calculate_blake_length_str(algo: AlgoKind, bit_length: &str) -> UResult<Option<usize>> {
+    debug_assert!(matches!(algo, AlgoKind::Blake2b | AlgoKind::Blake3));
+
     // Blake2b's length is parsed in an u64.
     match bit_length.parse::<usize>() {
         Ok(0) => Ok(None),
 
         // Error cases
-        Ok(n) if n > 512 => {
+        Ok(n) if n > 512 && algo == AlgoKind::Blake2b => {
             show_error!("{}", ChecksumError::InvalidLength(bit_length.into()));
-            Err(ChecksumError::LengthTooBigForBlake("BLAKE2b".into()).into())
+            Err(ChecksumError::LengthTooBigForBlake(algo.to_uppercase().into()).into())
         }
         Err(e) if *e.kind() == IntErrorKind::PosOverflow => {
             show_error!("{}", ChecksumError::InvalidLength(bit_length.into()));
-            Err(ChecksumError::LengthTooBigForBlake("BLAKE2b".into()).into())
+            Err(ChecksumError::LengthTooBigForBlake(algo.to_uppercase().into()).into())
         }
         Err(_) => Err(ChecksumError::InvalidLength(bit_length.into()).into()),
 
@@ -632,10 +643,19 @@ mod tests {
 
     #[test]
     fn test_calculate_blake2b_length() {
-        assert_eq!(calculate_blake2b_length_str("0").unwrap(), None);
-        assert!(calculate_blake2b_length_str("10").is_err());
-        assert!(calculate_blake2b_length_str("520").is_err());
-        assert_eq!(calculate_blake2b_length_str("512").unwrap(), None);
-        assert_eq!(calculate_blake2b_length_str("256").unwrap(), Some(32));
+        assert_eq!(
+            calculate_blake_length_str(AlgoKind::Blake2b, "0").unwrap(),
+            None
+        );
+        assert!(calculate_blake_length_str(AlgoKind::Blake2b, "10").is_err());
+        assert!(calculate_blake_length_str(AlgoKind::Blake2b, "520").is_err());
+        assert_eq!(
+            calculate_blake_length_str(AlgoKind::Blake2b, "512").unwrap(),
+            None
+        );
+        assert_eq!(
+            calculate_blake_length_str(AlgoKind::Blake2b, "256").unwrap(),
+            Some(32)
+        );
     }
 }
diff --git a/src/uucore/src/lib/features/checksum/validate.rs b/src/uucore/src/lib/features/checksum/validate.rs
@@ -614,6 +614,7 @@ fn identify_algo_name_and_length(
     algo_name_input: Option<AlgoKind>,
     last_algo: &mut Option<String>,
 ) -> Result<(AlgoKind, Option<usize>), LineCheckError> {
+    use AlgoKind as ak;
     let algo_from_line = line_info.algo_name.clone().unwrap_or_default();
     let Ok(line_algo) = AlgoKind::from_cksum(algo_from_line.to_lowercase()) else {
         // Unknown algorithm
@@ -629,34 +630,33 @@ fn identify_algo_name_and_length(
         match (algo_name_input, line_algo) {
             (l, r) if l == r => (),
             // Edge case for SHA2, which matches SHA(224|256|384|512)
-            (
-                AlgoKind::Sha2,
-                AlgoKind::Sha224 | AlgoKind::Sha256 | AlgoKind::Sha384 | AlgoKind::Sha512,
-            ) => (),
+            (ak::Sha2, ak::Sha224 | ak::Sha256 | ak::Sha384 | ak::Sha512) => (),
             _ => return Err(LineCheckError::ImproperlyFormatted),
         }
     }
 
     let bytes = if let Some(bitlen) = line_info.algo_bit_len {
         match line_algo {
-            AlgoKind::Blake2b if bitlen % 8 == 0 => Some(bitlen / 8),
-            AlgoKind::Sha2 | AlgoKind::Sha3 if [224, 256, 384, 512].contains(&bitlen) => {
-                Some(bitlen)
-            }
-            AlgoKind::Shake128 | AlgoKind::Shake256 => Some(bitlen),
+            ak::Blake2b | ak::Blake3 if bitlen % 8 == 0 => Some(bitlen / 8),
+            ak::Sha2 | ak::Sha3 if [224, 256, 384, 512].contains(&bitlen) => Some(bitlen),
+            ak::Shake128 | ak::Shake256 => Some(bitlen),
             // Either
-            //  the algo based line is provided with a bit length
-            //  with an algorithm that does not support it (only Blake2B does).
+            //  the algo based line is provided with a bit length with an
+            //  algorithm that does not support it (only Blake2b, Blake3, sha2,
+            //  and sha3 do).
             //
             //  eg: MD5-128 (foo.txt) = fffffffff
             //          ^ This is illegal
             // OR
             //  the given length is wrong because it's not a multiple of 8.
             _ => return Err(LineCheckError::ImproperlyFormatted),
         }
-    } else if line_algo == AlgoKind::Blake2b {
+    } else if line_algo == ak::Blake2b {
         // Default length with BLAKE2b,
         Some(64)
+    } else if line_algo == ak::Blake3 {
+        // Default length with BLAKE3,
+        Some(32)
     } else {
         None
     };
@@ -741,7 +741,7 @@ fn process_algo_based_line(
     // If the digest bitlen is known, we can check the format of the expected
     // checksum with it.
     let digest_char_length_hint = match (algo_kind, algo_byte_len) {
-        (AlgoKind::Blake2b, Some(byte_len)) => Some(byte_len),
+        (AlgoKind::Blake2b | AlgoKind::Blake3, Some(byte_len)) => Some(byte_len),
         (AlgoKind::Shake128 | AlgoKind::Shake256, Some(bit_len)) => Some(bit_len.div_ceil(8)),
         (AlgoKind::Shake128, None) => Some(sum::Shake128::DEFAULT_BIT_SIZE.div_ceil(8)),
         (AlgoKind::Shake256, None) => Some(sum::Shake256::DEFAULT_BIT_SIZE.div_ceil(8)),
@@ -764,6 +764,7 @@ fn process_non_algo_based_line(
     cli_algo_length: Option<usize>,
     opts: ChecksumValidateOptions,
 ) -> Result<(), LineCheckError> {
+    use AlgoKind as ak;
     let mut filename_to_check = line_info.filename.as_slice();
     if filename_to_check.starts_with(b"*")
         && line_number == 0
@@ -778,16 +779,16 @@ fn process_non_algo_based_line(
     // When a specific algorithm name is input, use it and use the provided
     // bits except when dealing with blake2b, sha2 and sha3, where we will
     // detect the length.
-    let (algo_kind, algo_byte_len) = match cli_algo_kind {
-        AlgoKind::Blake2b => (AlgoKind::Blake2b, Some(expected_checksum.len())),
-        algo @ (AlgoKind::Sha2 | AlgoKind::Sha3) => {
+    let algo_byte_len = match cli_algo_kind {
+        ak::Blake2b | ak::Blake3 => Some(expected_checksum.len()),
+        ak::Sha2 | ak::Sha3 => {
             // multiplication by 8 to get the number of bits
-            (algo, Some(expected_checksum.len() * 8))
+            Some(expected_checksum.len() * 8)
         }
-        _ => (cli_algo_kind, cli_algo_length),
+        _ => cli_algo_length,
     };
 
-    let algo = SizedAlgoKind::from_unsized(algo_kind, algo_byte_len)?;
+    let algo = SizedAlgoKind::from_unsized(cli_algo_kind, algo_byte_len)?;
 
     compute_and_check_digest_from_file(filename_to_check, &expected_checksum, algo, opts)
 }
diff --git a/src/uucore/src/lib/features/sum.rs b/src/uucore/src/lib/features/sum.rs
@@ -122,25 +122,48 @@ impl Digest for Blake2b {
     }
 }
 
-#[derive(Default)]
-pub struct Blake3(blake3::Hasher);
+pub struct Blake3 {
+    digest: blake3::Hasher,
+    byte_size: usize,
+}
+
+impl Blake3 {
+    /// Default length for the BLAKE3 digest in bytes.
+    pub const DEFAULT_BYTE_SIZE: usize = 32;
+
+    pub fn with_output_bytes(output_bytes: usize) -> Self {
+        Self {
+            digest: blake3::Hasher::new(),
+            byte_size: output_bytes,
+        }
+    }
+}
+
+impl Default for Blake3 {
+    fn default() -> Self {
+        Self {
+            digest: blake3::Hasher::default(),
+            byte_size: Self::DEFAULT_BYTE_SIZE,
+        }
+    }
+}
 
 impl Digest for Blake3 {
     fn hash_update(&mut self, input: &[u8]) {
-        self.0.update(input);
+        self.digest.update(input);
     }
 
     fn hash_finalize(&mut self, out: &mut [u8]) {
-        let hash_result = &self.0.finalize();
-        out.copy_from_slice(hash_result.as_bytes());
+        let mut hash_result = self.digest.finalize_xof();
+        hash_result.fill(out);
     }
 
     fn reset(&mut self) {
-        *self = Self::default();
+        *self = Self::with_output_bytes(self.output_bytes());
     }
 
     fn output_bits(&self) -> usize {
-        256
+        self.byte_size * 8
     }
 }
 
