core: replace bundle checksum with BLAKE3

BLAKE3 has a verified streaming mode (Bao), which allows variable chunk
sizes without altering the root hash. This means that a BLAKE3 hash can
serve as a long-term stable content address in location-independent
storage (as demonstrated by iroh).

Signed-off-by: Kim Altintop <kim@eagain.io>

16 files changed, 110 insertions, 76 deletions

diff --git a/Cargo.lock b/Cargo.lock
index 19c6e65..e19a8d4 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -87,6 +87,18 @@ dependencies = [
 ]
 
 [[package]]
+name = "arrayref"
+version = "0.3.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "6b4930d2cb77ce62f89ee5d5289b4ac049559b1c45539271f5ed4fdc7db34545"
+
+[[package]]
+name = "arrayvec"
+version = "0.7.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8da52d66c7071e2e3fa2a1e5c6d088fec47b593032b254f5e980de8ea54454d6"
+
+[[package]]
 name = "ascii"
 version = "1.1.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
@@ -149,6 +161,20 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "bef38d45163c2f1dde094a7dfd33ccf595c92905c8f8f4fdc18d06fb1037718a"
 
 [[package]]
+name = "blake3"
+version = "1.3.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "42ae2468a89544a466886840aa467a25b766499f4f04bf7d9fcd10ecee9fccef"
+dependencies = [
+ "arrayref",
+ "arrayvec",
+ "cc",
+ "cfg-if",
+ "constant_time_eq",
+ "digest 0.10.6",
+]
+
+[[package]]
 name = "block-buffer"
 version = "0.9.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
@@ -319,6 +345,12 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "520fbf3c07483f94e3e3ca9d0cfd913d7718ef2483d2cfd91c0d9e91474ab913"
 
 [[package]]
+name = "constant_time_eq"
+version = "0.2.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "13418e745008f7349ec7e449155f419a61b92b58a99cc3616942b926825ec76b"
+
+[[package]]
 name = "core-foundation"
 version = "0.9.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
@@ -785,10 +817,12 @@ version = "0.1.0"
 dependencies = [
  "anyhow",
  "base64",
+ "blake3",
  "clap",
  "clap_complete",
  "clap_mangen",
  "console",
+ "digest 0.10.6",
  "directories",
  "either",
  "erased-serde",
diff --git a/Cargo.toml b/Cargo.toml
index 4de6fdb..229e85b 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -16,12 +16,15 @@ sha1dc = ["sha1collisiondetection"]
 anyhow.features = ["backtrace"]
 anyhow.version = "1"
 base64.version = "0.13"
+blake3.version = "1.3.3"
+blake3.features = ["traits-preview"]
 clap.features = ["derive", "env", "string", "wrap_help"]
 clap.version = "4.0"
 clap_complete.version = "4.0"
 clap_mangen.version = "0.2"
 console.default-features = false
 console.version = "0.15"
+digest.version = "0.10"
 directories.version = "4.0"
 either.version = "1.8"
 erased-serde.version = "0.3"
diff --git a/src/bundle.rs b/src/bundle.rs
index 25eafd0..6f333ca 100644
--- a/src/bundle.rs
+++ b/src/bundle.rs
@@ -1,13 +1,16 @@
 // Copyright © 2022 Kim Altintop <kim@eagain.io>
 // SPDX-License-Identifier: GPL-2.0-only WITH openvpn-openssl-exception
 
-use std::io;
+use std::{
+    fmt::{
+        self,
+        Debug,
+        Display,
+    },
+    io,
+};
 
 use log::info;
-use sha2::{
-    Digest,
-    Sha256,
-};
 use url::Url;
 
 use crate::io::{
@@ -42,12 +45,35 @@ pub use list::{
 pub const FILE_EXTENSION: &str = "bundle";
 pub const DOT_FILE_EXTENSION: &str = ".bundle";
 
+#[derive(Clone, Copy, Eq, Hash, PartialEq, serde::Serialize, serde::Deserialize)]
+pub struct Checksum(#[serde(with = "crate::serde::display")] blake3::Hash);
+
+impl Debug for Checksum {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        let hex = self.0.to_hex();
+        let hex: &str = hex.as_str();
+
+        f.debug_tuple("Checksum").field(&hex).finish()
+    }
+}
+
+impl Display for Checksum {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        Display::fmt(&self.0, f)
+    }
+}
+
+impl From<&blake3::Hasher> for Checksum {
+    fn from(hasher: &blake3::Hasher) -> Self {
+        Self(hasher.finalize())
+    }
+}
+
 #[derive(Clone, Debug, serde::Serialize, serde::Deserialize)]
 pub struct Info {
     pub len: u64,
     pub hash: Hash,
-    #[serde(with = "hex::serde")]
-    pub checksum: [u8; 32],
+    pub checksum: Checksum,
     #[serde(default, skip_serializing_if = "Vec::is_empty")]
     pub uris: Vec<Url>,
 }
@@ -56,7 +82,7 @@ pub struct Info {
 pub struct Expect<'a> {
     pub len: u64,
     pub hash: &'a Hash,
-    pub checksum: Option<&'a [u8]>,
+    pub checksum: Option<&'a Checksum>,
 }
 
 impl<'a> From<&'a Info> for Expect<'a> {
@@ -80,7 +106,7 @@ pub fn create<W>(mut out: W, repo: &git2::Repository, header: &Header) -> crate:
 where
     W: io::Write,
 {
-    let mut hasher = HashWriter::new(Sha256::new(), &mut out);
+    let mut hasher = HashWriter::new(blake3::Hasher::new(), &mut out);
     let mut writer = LenWriter::new(&mut hasher);
     let mut pack = {
         let mut pack = repo.packbuilder()?;
@@ -101,7 +127,7 @@ where
 
     let len = writer.bytes_written();
     let hash = header.hash();
-    let checksum = hasher.hash().into();
+    let checksum = Checksum::from(hasher.hasher());
 
     info!("Created patch bundle {hash}");
 
diff --git a/src/bundle/fetch.rs b/src/bundle/fetch.rs
index 4e58000..ff10e27 100644
--- a/src/bundle/fetch.rs
+++ b/src/bundle/fetch.rs
@@ -22,10 +22,6 @@ use either::Either::{
     Left,
     Right,
 };
-use sha2::{
-    Digest,
-    Sha256,
-};
 use tempfile::NamedTempFile;
 use url::Url;
 
@@ -96,13 +92,13 @@ impl Fetcher {
                 LockedFile::atomic(&path, true, LockedFile::DEFAULT_PERMISSIONS)?
             };
 
-            let mut out = HashWriter::new(Sha256::new(), &mut lck);
+            let mut out = HashWriter::new(blake3::Hasher::new(), &mut lck);
             out.write_all(&buf)?;
 
             let len = buf.len() as u64 + io::copy(&mut body.take(expect.len), &mut out)?;
-            let checksum = out.hash().into();
+            let checksum = bundle::Checksum::from(out.hasher());
             if let Some(chk) = expect.checksum {
-                ensure!(chk == checksum, "checksum mismatch");
+                ensure!(chk == &checksum, "checksum mismatch");
             }
             lck.seek(SeekFrom::Start(0))?;
             let header = Header::from_reader(&mut lck)?;
diff --git a/src/bundle/header.rs b/src/bundle/header.rs
index 6f3dfe3..d296af1 100644
--- a/src/bundle/header.rs
+++ b/src/bundle/header.rs
@@ -12,15 +12,13 @@ use std::{
     str::FromStr,
 };
 
+use digest::Digest;
 use hex::{
     FromHex,
     FromHexError,
 };
 use refs::Refname;
-use sha2::{
-    Digest,
-    Sha256,
-};
+use sha2::Sha256;
 
 use super::error;
 use crate::{
diff --git a/src/bundle/list.rs b/src/bundle/list.rs
index 21753fa..ea6296f 100644
--- a/src/bundle/list.rs
+++ b/src/bundle/list.rs
@@ -19,11 +19,9 @@ use std::{
 };
 
 use anyhow::anyhow;
+use digest::Digest;
 use once_cell::sync::Lazy;
-use sha2::{
-    Digest,
-    Sha256,
-};
+use sha2::Sha256;
 use url::Url;
 
 use crate::git::{
diff --git a/src/cmd/patch/prepare.rs b/src/cmd/patch/prepare.rs
index 06d5ec9..8f8b871 100644
--- a/src/cmd/patch/prepare.rs
+++ b/src/cmd/patch/prepare.rs
@@ -11,11 +11,9 @@ use anyhow::{
     bail,
     ensure,
 };
+use digest::Digest;
 use either::Either::Left;
-use sha2::{
-    Digest,
-    Sha256,
-};
+use sha2::Sha256;
 
 use crate::{
     bundle,
diff --git a/src/git.rs b/src/git.rs
index f837711..a627048 100644
--- a/src/git.rs
+++ b/src/git.rs
@@ -11,11 +11,9 @@ use anyhow::{
     ensure,
     Context,
 };
+use digest::Digest;
 use once_cell::sync::Lazy;
-use sha2::{
-    Digest,
-    Sha256,
-};
+use sha2::Sha256;
 
 mod commit;
 pub use commit::{
diff --git a/src/http.rs b/src/http.rs
index d52ef8f..f8cfaba 100644
--- a/src/http.rs
+++ b/src/http.rs
@@ -15,15 +15,13 @@ use std::{
     },
 };
 
+use digest::Digest;
 use log::{
     debug,
     error,
 };
 use once_cell::sync::Lazy;
-use sha2::{
-    Digest,
-    Sha256,
-};
+use sha2::Sha256;
 use threadpool::ThreadPool;
 use tiny_http::{
     Header,
diff --git a/src/io.rs b/src/io.rs
index 86f91c6..324d12c 100644
--- a/src/io.rs
+++ b/src/io.rs
@@ -1,10 +1,7 @@
 // Copyright © 2022 Kim Altintop <kim@eagain.io>
 // SPDX-License-Identifier: GPL-2.0-only WITH openvpn-openssl-exception
 
-use sha2::{
-    digest::generic_array::GenericArray,
-    Digest,
-};
+use digest::Digest;
 
 /// Created by [`Lines::until_blank`], stops iteration at the first blank line.
 pub struct UntilBlank<B> {
@@ -94,8 +91,8 @@ impl<D, W> HashWriter<D, W>
 where
     D: Digest,
 {
-    pub fn hash(self) -> GenericArray<u8, D::OutputSize> {
-        self.hasher.finalize()
+    pub fn hasher(&self) -> &D {
+        &self.hasher
     }
 }
 
diff --git a/src/metadata.rs b/src/metadata.rs
index 2da268f..378f462 100644
--- a/src/metadata.rs
+++ b/src/metadata.rs
@@ -15,11 +15,9 @@ use std::{
     ops::DerefMut,
 };
 
+use digest::Digest;
 use serde::ser::SerializeSeq;
-use sha2::{
-    Digest,
-    Sha512,
-};
+use sha2::Sha512;
 use time::{
     Duration,
     OffsetDateTime,
diff --git a/src/metadata/drop.rs b/src/metadata/drop.rs
index ae9aefd..601a157 100644
--- a/src/metadata/drop.rs
+++ b/src/metadata/drop.rs
@@ -13,11 +13,9 @@ use std::{
     ops::Deref,
 };
 
+use digest::Digest;
 use log::warn;
-use sha2::{
-    Digest,
-    Sha512,
-};
+use sha2::Sha512;
 use signature::Verifier;
 
 use super::{
diff --git a/src/metadata/identity.rs b/src/metadata/identity.rs
index 8ded144..0b45731 100644
--- a/src/metadata/identity.rs
+++ b/src/metadata/identity.rs
@@ -20,10 +20,10 @@ use anyhow::{
     anyhow,
     ensure,
 };
+use digest::Digest;
 use hex::FromHex;
 use log::warn;
 use sha2::{
-    Digest,
     Sha256,
     Sha512,
 };
diff --git a/src/patches.rs b/src/patches.rs
index 8623e4e..5efcf90 100644
--- a/src/patches.rs
+++ b/src/patches.rs
@@ -15,16 +15,14 @@ use anyhow::{
     bail,
 };
 
-use hex::FromHex;
-use once_cell::sync::Lazy;
-use sha2::{
-    digest::{
-        generic_array::GenericArray,
-        typenum::U32,
-    },
+use digest::{
+    generic_array::GenericArray,
+    typenum::U32,
     Digest,
-    Sha256,
 };
+use hex::FromHex;
+use once_cell::sync::Lazy;
+use sha2::Sha256;
 
 use crate::{
     git::Refname,
diff --git a/src/patches/bundle.rs b/src/patches/bundle.rs
index 296b24a..52397a1 100644
--- a/src/patches/bundle.rs
+++ b/src/patches/bundle.rs
@@ -21,11 +21,9 @@ use anyhow::{
     ensure,
     Context,
 };
+use digest::Digest;
 use multipart::client::lazy::Multipart;
-use sha2::{
-    Digest,
-    Sha256,
-};
+use sha2::Sha256;
 use tempfile::NamedTempFile;
 use url::Url;
 
@@ -105,14 +103,14 @@ impl Bundle {
         let encryption = pack.encryption()?;
         drop(pack);
         let mut file = File::open(&path)?;
-        let mut sha2 = Sha256::new();
+        let mut hasher = blake3::Hasher::new();
 
-        let len = io::copy(&mut file, &mut sha2)?;
+        let len = io::copy(&mut file, &mut hasher)?;
         let hash = header.hash();
         ensure!(expect.hash == &hash, "header hash mismatch");
-        let checksum = sha2.finalize().into();
+        let checksum = bundle::Checksum::from(&hasher);
         if let Some(expect) = expect.checksum {
-            ensure!(expect == checksum, "claimed and actual hash differ");
+            ensure!(expect == &checksum, "claimed and actual hash differ");
         }
 
         let info = bundle::Info {
@@ -138,10 +136,10 @@ impl Bundle {
     {
         std::fs::create_dir_all(&to)?;
         let mut tmp = NamedTempFile::new_in(&to)?;
-        let mut out = HashWriter::new(Sha256::new(), &mut tmp);
+        let mut out = HashWriter::new(blake3::Hasher::new(), &mut tmp);
 
         let len = io::copy(&mut from, &mut out)?;
-        let checksum = out.hash().into();
+        let checksum = bundle::Checksum::from(out.hasher());
 
         let (header, mut pack) = split(tmp.path())?;
         let hash = header.hash();
diff --git a/src/patches/record.rs b/src/patches/record.rs
index 6a95973..9bf7856 100644
--- a/src/patches/record.rs
+++ b/src/patches/record.rs
@@ -25,16 +25,12 @@ use anyhow::{
     ensure,
     Context,
 };
-
+use digest::Digest;
 use hex::{
     FromHex,
     ToHex,
 };
-
-use sha2::{
-    Digest,
-    Sha256,
-};
+use sha2::Sha256;
 use signature::{
     Signature as _,
     Verifier,