Merge branch 'trishume:master' into patch-1

src/parsing/syntax_set.rs

@@ -213,6 +213,7 @@ impl SyntaxSet {
     /// This uses regexes that come with some sublime syntax grammars for matching things like
     /// shebangs and mode lines like `-*- Mode: C -*-`
     pub fn find_syntax_by_first_line<'a>(&'a self, s: &str) -> Option<&'a SyntaxReference> {
+        let s = s.strip_prefix("\u{feff}").unwrap_or(s); // Strip UTF-8 BOM
         let cache = self.first_line_cache();
         for &(ref reg, i) in cache.regexes.iter().rev() {
             if reg.search(s, 0, s.len(), None) {
@@ -1401,6 +1402,16 @@ mod tests {
         assert_prototype_only_on(&["main"], &rebuilt, &rebuilt.syntaxes()[0]);
     }
 
+    #[test]
+    fn find_syntax_set_from_line_with_bom() {
+        // Regression test for #529
+        let syntax_set = SyntaxSet::load_defaults_newlines();
+        let syntax_ref = syntax_set
+            .find_syntax_by_first_line("\u{feff}<?xml version=\"1.0\"?>")
+            .unwrap();
+        assert_eq!(syntax_ref.name, "XML");
+    }
+
     fn assert_ops_contain(ops: &[(usize, ScopeStackOp)], expected: &(usize, ScopeStackOp)) {
         assert!(
             ops.contains(expected),

tests/public_api.rs

@@ -1,11 +1,19 @@
 #[test]
 fn public_api() {
+    // NOTE: consider switching back to using `public_api::MINIMUM_NIGHTLY_RUST_VERSION` after the
+    // version is newer than the one set here
+    let nightly_version = "nightly-2023-09-18";
+    assert_eq!(
+        public_api::MINIMUM_NIGHTLY_RUST_VERSION,
+        "nightly-2023-08-25"
+    );
+
     // Install a compatible nightly toolchain if it is missing
-    rustup_toolchain::install(public_api::MINIMUM_NIGHTLY_RUST_VERSION).unwrap();
+    rustup_toolchain::install(nightly_version).unwrap();
 
     // Build rustdoc JSON
     let rustdoc_json = rustdoc_json::Builder::default()
-        .toolchain(public_api::MINIMUM_NIGHTLY_RUST_VERSION)
+        .toolchain(nightly_version)
         .build()
         .unwrap();