Add case-sensitive glob support to syntax mapping

to allow us to map `BUILD` case sensitively to Python for Skylark

src/syntax_mapping.rs

@@ -17,9 +17,9 @@ use ignored_suffixes::IgnoredSuffixes;
 mod builtin;
 pub mod ignored_suffixes;
 
-fn make_glob_matcher(from: &str) -> Result<GlobMatcher> {
+fn make_glob_matcher(from: &str, case_insensitive: bool) -> Result<GlobMatcher> {
     let matcher = GlobBuilder::new(from)
-        .case_insensitive(true)
+        .case_insensitive(case_insensitive)
         .literal_separator(true)
         .build()?
         .compile_matcher();
@@ -97,7 +97,14 @@ impl<'a> SyntaxMapping<'a> {
     }
 
     pub fn insert(&mut self, from: &str, to: MappingTarget<'a>) -> Result<()> {
-        let matcher = make_glob_matcher(from)?;
+        let matcher = make_glob_matcher(from, true)?;
+        self.custom_mappings.push((matcher, to));
+        Ok(())
+    }
+
+    /// Like [`Self::insert`], but the glob pattern is matched case-sensitively.
+    pub fn insert_case_sensitive(&mut self, from: &str, to: MappingTarget<'a>) -> Result<()> {
+        let matcher = make_glob_matcher(from, false)?;
         self.custom_mappings.push((matcher, to));
         Ok(())
     }
@@ -261,4 +268,41 @@ mod tests {
             Some(MappingTarget::MapTo("alpha"))
         );
     }
+
+    #[test]
+    fn case_sensitive_custom_mappings_work() {
+        let mut map = SyntaxMapping::new();
+        map.insert_case_sensitive("MY_SPECIAL_FILE", MappingTarget::MapTo("Python"))
+            .ok();
+
+        // Exact case matches
+        assert_eq!(
+            map.get_syntax_for("/path/to/MY_SPECIAL_FILE"),
+            Some(MappingTarget::MapTo("Python"))
+        );
+        // Different case should NOT match the case-sensitive rule
+        assert_eq!(map.get_syntax_for("/path/to/my_special_file"), None);
+        assert_eq!(map.get_syntax_for("/path/to/My_Special_File"), None);
+    }
+
+    #[test]
+    fn builtin_mappings_build_is_case_sensitive() {
+        let map = SyntaxMapping::new();
+
+        // "BUILD" (uppercase) should map to Python via case-sensitive builtin
+        assert_eq!(
+            map.get_syntax_for("/path/to/BUILD"),
+            Some(MappingTarget::MapTo("Python"))
+        );
+        // "build" (lowercase) should still map to MapToUnknown
+        assert_eq!(
+            map.get_syntax_for("/path/to/build"),
+            Some(MappingTarget::MapToUnknown)
+        );
+        // Mixed case should NOT match the Python rule
+        assert_eq!(
+            map.get_syntax_for("/path/to/Build"),
+            Some(MappingTarget::MapToUnknown)
+        );
+    }
 }

src/syntax_mapping/builtin.rs

@@ -53,8 +53,9 @@ include!(concat!(
 /// A failure to compile is a fatal error.
 ///
 /// Used internally by `Lazy<Option<GlobMatcher>>`'s lazy evaluation closure.
-fn build_matcher_fixed(from: &str) -> GlobMatcher {
-    make_glob_matcher(from).expect("A builtin fixed glob matcher failed to compile")
+fn build_matcher_fixed(from: &str, case_insensitive: bool) -> GlobMatcher {
+    make_glob_matcher(from, case_insensitive)
+        .expect("A builtin fixed glob matcher failed to compile")
 }
 
 /// Join a list of matcher segments to create a glob string, replacing all
@@ -64,7 +65,7 @@ fn build_matcher_fixed(from: &str) -> GlobMatcher {
 /// to compile.
 ///
 /// Used internally by `Lazy<Option<GlobMatcher>>`'s lazy evaluation closure.
-fn build_matcher_dynamic(segs: &[MatcherSegment]) -> Option<GlobMatcher> {
+fn build_matcher_dynamic(segs: &[MatcherSegment], case_insensitive: bool) -> Option<GlobMatcher> {
     // join segments
     let mut buf = String::new();
     for seg in segs {
@@ -77,7 +78,7 @@ fn build_matcher_dynamic(segs: &[MatcherSegment]) -> Option<GlobMatcher> {
         }
     }
     // compile glob matcher
-    let matcher = make_glob_matcher(&buf).ok()?;
+    let matcher = make_glob_matcher(&buf, case_insensitive).ok()?;
     Some(matcher)
 }
 

src/syntax_mapping/builtins/README.md

@@ -20,9 +20,9 @@ syntax mappings defined by all TOML files, and embed them into the binary.
 
 ## File syntax
 
-Each TOML file should contain a single section named `mappings`, with each of
-its keys being a language identifier (first column of `bat -L`; also referred to
-as "target").
+Each TOML file should contain a single section named `mappings` and/or a single
+section named `case_sensitive_mappings`, with each of its keys being a language
+identifier (first column of `bat -L`; also referred to as "target").
 
 The value of each key should be an array of strings, with each item being a glob
 matcher. We will call each of these items a "rule".

src/syntax_mapping/builtins/common/50-bazel.toml

@@ -1,2 +1,2 @@
-[mappings]
+[case_sensitive_mappings]
 "Python" = ["BUILD"]