.github/workflows/CI.yml

@@ -25,7 +25,7 @@ jobs:
       - uses: actions/checkout@v4
         with:
           submodules: recursive
-      - uses: actions/setup-python@v4
+      - uses: actions/setup-python@v5
         with:
           python-version: ${{ matrix.python-version }}
       - name: Build wheels
@@ -35,7 +35,7 @@ jobs:
           args: --release --out dist -i ${{ matrix.python-version }}
           sccache: 'true'
       - name: Install Just
-        uses: extractions/setup-just@v1
+        uses: extractions/setup-just@v2
       - name: Run Cargo Tests
         run: |
           cargo test
@@ -89,8 +89,10 @@ jobs:
           - target: i686-unknown-linux-musl
             arch: x86
             interpreter: 3.8 3.9 3.10 3.11 3.12 pypy3.8 pypy3.9 pypy3.10
+          - target: aarch64-unknown-linux-musl
+            arch: aarch64
+            interpreter: 3.8 3.9 3.10 3.11 3.12 pypy3.8 pypy3.9 pypy3.10
         # all values: [x86_64, x86, aarch64, armhf, armv7, ppc64le, riscv64, s390x]
-        # { target: "aarch64-unknown-linux-musl", arch: "aarch64" },
         # { target: "armv7-unknown-linux-musleabihf", image_tag: "armv7" },
         # { target: "powerpc64le-unknown-linux-musl", image_tag: "ppc64le" },
     steps:

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "jellyfish"
-version = "1.0.3"
+version = "1.0.4"
 edition = "2021"
 description = "Approximate and phonetic matching of strings."
 authors = ["James Turk <dev@jamesturk.net>"]

docs/changelog.md

@@ -1,6 +1,11 @@
 Changelog
 =========
 
+1.0.4 - 28 May 2024
+-----------------------
+* `match_rating_codex` now returns consistent results for Unicode characters regardless of position (#210)
+* adds prebuilt wheels for Alpine (#209)
+
 1.0.3 - 17 November 2023
 -----------------------
 * `match_rating_codex` now raises a `ValueError` when passed non-alpha characters (#200)

src/match_rating.rs

@@ -4,6 +4,7 @@ use unicode_segmentation::UnicodeSegmentation;
 
 pub fn match_rating_codex(s: &str) -> Result<String, String> {
     // match rating only really makes sense on strings
+
     let s = &s.to_uppercase()[..];
     let v = UnicodeSegmentation::graphemes(s, true).collect::<FastVec<&str>>();
     let mut codex = String::new();
@@ -24,10 +25,12 @@ pub fn match_rating_codex(s: &str) -> Result<String, String> {
     }
 
     if codex.len() > 6 {
-        let mut newcodex = String::new();
-        newcodex.push_str(codex.get(..3).unwrap());
-        newcodex.push_str(codex.get(codex.len() - 3..).unwrap());
-        return Ok(newcodex);
+        // not safe to take a slice without conversion to chars() since there
+        // can be unicode left, this implementation matches the Python one
+        // even though MRC really shouldn't be used with unicode chars
+        let first_three: String = codex.chars().take(3).collect();
+        let last_three: String = codex.chars().rev().take(3).collect::<String>().chars().rev().collect();
+        return Ok(first_three + &last_three);
     }
 
     Ok(codex)