Add support for new index hash implementation.

Author: abrisco

Cargo.toml

@@ -45,6 +45,7 @@ http = { version = "1", optional = true }
 memchr = "2.5.0"
 rayon = { version = "1.7.0", optional = true }
 rustc-hash = "2.0.0"
+rustc-stable-hash = "0.1.1"
 semver = "1.0.17"
 serde = { version = "1.0.160", features = ["rc"] }
 serde_derive = "1.0.160"

src/dirs.rs

@@ -9,16 +9,23 @@ pub fn local_path_and_canonical_url(
     url: &str,
     cargo_home: Option<&std::path::Path>,
 ) -> Result<(std::path::PathBuf, String), Error> {
-    let (dir_name, canonical_url) = url_to_local_dir(url)?;
-
     let mut path = match cargo_home {
         Some(path) => path.to_owned(),
         None => home::cargo_home()?,
     };
 
     path.push("registry");
     path.push("index");
-    path.push(dir_name);
+
+    let (dir_name, canonical) = url_to_local_dir(url, true)?;
+    let canonical_url = if path.join(&dir_name).exists() {
+        path.push(dir_name);
+        canonical
+    } else {
+        let (dir_name, canonical) = url_to_local_dir(url, false)?;
+        path.push(dir_name);
+        canonical
+    };
 
     Ok((path, canonical_url))
 }
@@ -72,8 +79,8 @@ pub(crate) fn crate_name_to_relative_path(crate_name: &str, separator: Option<ch
 
 /// Converts a full url, eg https://github.com/rust-lang/crates.io-index, into
 /// the root directory name where cargo itself will fetch it on disk
-fn url_to_local_dir(url: &str) -> Result<(String, String), Error> {
-    fn to_hex(num: u64) -> String {
+fn url_to_local_dir(url: &str, use_stable_hash: bool) -> Result<(String, String), Error> {
+    fn legacy_to_hex(num: u64) -> String {
         const CHARS: &[u8] = b"0123456789abcdef";
 
         let bytes = &[
@@ -102,7 +109,7 @@ fn url_to_local_dir(url: &str) -> Result<(String, String), Error> {
     }
 
     #[allow(deprecated)]
-    fn hash_u64(url: &str, registry_kind: u64) -> u64 {
+    fn legacy_hash_u64(url: &str, registry_kind: u64) -> u64 {
         use std::hash::{Hash, Hasher, SipHasher};
 
         let mut hasher = SipHasher::new_with_keys(0, 0);
@@ -113,6 +120,35 @@ fn url_to_local_dir(url: &str) -> Result<(String, String), Error> {
         hasher.finish()
     }
 
+    fn stable_to_hex(num: u64) -> String {
+        hex::encode(num.to_le_bytes())
+    }
+
+    fn stable_hash_u64(url: &str, registry_kind: u64) -> u64 {
+        // Matches https://github.com/rust-lang/cargo/blob/49e5d24870a13cd3f26ad61159045788b36efcdf/src/cargo/util/hasher.rs#L6
+        use rustc_stable_hash::StableSipHasher128 as StableHasher;
+        use std::hash::{Hash, Hasher};
+
+        let mut hasher = StableHasher::new();
+        // Registry
+        registry_kind.hash(&mut hasher);
+        // Url
+        url.hash(&mut hasher);
+        Hasher::finish(&hasher)
+    }
+
+    let hash_u64 = if use_stable_hash {
+        stable_hash_u64
+    } else {
+        legacy_hash_u64
+    };
+
+    let to_hex = if use_stable_hash {
+        stable_to_hex
+    } else {
+        legacy_to_hex
+    };
+
     // SourceKind::Registry
     let mut registry_kind = 2;
 
@@ -190,35 +226,62 @@ mod test {
     fn http_index_url_matches_cargo() {
         use crate::sparse::URL;
         assert_eq!(
-            super::url_to_local_dir(URL).unwrap(),
+            super::url_to_local_dir(URL, false).unwrap(),
             ("index.crates.io-6f17d22bba15001f".to_owned(), URL.to_owned(),)
         );
+        assert_eq!(
+            super::url_to_local_dir(URL, true).unwrap(),
+            ("index.crates.io-1949cf8c6b5b557f".to_owned(), URL.to_owned(),)
+        );
 
         // I've confirmed this also works with a custom registry, unfortunately
         // that one includes a secret key as part of the url which would allow
         // anyone to publish to the registry, so uhh...here's a fake one instead
         assert_eq!(
-            super::url_to_local_dir("https://dl.cloudsmith.io/aBcW1234aBcW1234/embark/rust/cargo/index.git").unwrap(),
+            super::url_to_local_dir(
+                "https://dl.cloudsmith.io/aBcW1234aBcW1234/embark/rust/cargo/index.git",
+                false
+            )
+            .unwrap(),
             (
                 "dl.cloudsmith.io-ff79e51ddd2b38fd".to_owned(),
                 "https://dl.cloudsmith.io/aBcW1234aBcW1234/embark/rust/cargo/index.git".to_owned()
             )
         );
+        assert_eq!(
+            super::url_to_local_dir(
+                "https://dl.cloudsmith.io/aBcW1234aBcW1234/embark/rust/cargo/index.git",
+                true
+            )
+            .unwrap(),
+            (
+                "dl.cloudsmith.io-fade991491aaf150".to_owned(),
+                "https://dl.cloudsmith.io/aBcW1234aBcW1234/embark/rust/cargo/index.git".to_owned()
+            )
+        );
     }
 
     #[test]
     #[cfg(feature = "git")]
     fn git_url_matches_cargo() {
         use crate::git::URL;
         assert_eq!(
-            crate::dirs::url_to_local_dir(URL).unwrap(),
+            crate::dirs::url_to_local_dir(URL, false).unwrap(),
+            ("github.com-1ecc6299db9ec823".to_owned(), URL.to_owned())
+        );
+        assert_eq!(
+            crate::dirs::url_to_local_dir(URL, true).unwrap(),
             ("github.com-1ecc6299db9ec823".to_owned(), URL.to_owned())
         );
 
         // Ensure we actually strip off the irrelevant parts of a url, note that
         // the .git suffix is not part of the canonical url, but *is* used when hashing
         assert_eq!(
-            crate::dirs::url_to_local_dir(&format!("registry+{}.git?one=1&two=2#fragment", URL)).unwrap(),
+            crate::dirs::url_to_local_dir(&format!("registry+{}.git?one=1&two=2#fragment", URL), false).unwrap(),
+            ("github.com-c786010fb7ef2e6e".to_owned(), URL.to_owned())
+        );
+        assert_eq!(
+            crate::dirs::url_to_local_dir(&format!("registry+{}.git?one=1&two=2#fragment", URL), true).unwrap(),
             ("github.com-c786010fb7ef2e6e".to_owned(), URL.to_owned())
         );
     }