feat: Map BUILD to Python (Starlark) for Bazel (fixes #3575)

CHANGELOG.md

@@ -12,6 +12,7 @@
 - Add `--quiet-empty` (`-E`) flag to suppress output when input is empty. Closes #1936, see #3563 (@NORMAL-EX)
 - Improve native man pages and command help syntax highlighting by stripping overstriking, see #3517 (@akirk)
 - Add `--fallback-syntax`/`--fallback-language` to apply syntax highlighting only when auto-detection fails, see #1341 (@Xavrir)
+- Map `BUILD` case sensitively to Python (Starlark) for Bazel, see #3576 (@vorburger)
 
 ## Bugfixes
 - Fix `BAT_CONFIG_DIR` pointing at system config directory causing duplicate flag errors. Closes #3589, see #3620 (@Xavrir)

build/syntax_mapping.rs

@@ -51,7 +51,13 @@ impl ToTokens for MappingTarget {
 /// A single matcher.
 ///
 /// Codegen converts this into a `Lazy<Option<GlobMatcher>>`.
-struct Matcher(Vec<MatcherSegment>);
+struct Matcher {
+    segments: Vec<MatcherSegment>,
+    /// Whether the glob pattern should be matched case-insensitively.
+    ///
+    /// Defaults to `true` (case-insensitive) for backwards compatibility.
+    case_insensitive: bool,
+}
 /// Parse a matcher.
 ///
 /// Note that this implementation is rather strict: it will greedily interpret
@@ -116,18 +122,24 @@ impl FromStr for Matcher {
             bail!(r#"Invalid matcher: "{s}""#);
         }
 
-        Ok(Self(non_empty_segments))
+        Ok(Self {
+            segments: non_empty_segments,
+            case_insensitive: true,
+        })
     }
 }
 impl ToTokens for Matcher {
     fn to_tokens(&self, tokens: &mut TokenStream) {
-        let t = match self.0.as_slice() {
+        let case_insensitive = self.case_insensitive;
+        let t = match self.segments.as_slice() {
             [] => unreachable!("0-length matcher should never be created"),
             [MatcherSegment::Text(text)] => {
-                quote! { Lazy::new(|| Some(build_matcher_fixed(#text))) }
+                quote! { Lazy::new(|| Some(build_matcher_fixed(#text, #case_insensitive))) }
             }
             // parser logic ensures that this case can only happen when there are dynamic segments
-            segs @ [_, ..] => quote! { Lazy::new(|| build_matcher_dynamic(&[ #(#segs),* ])) },
+            segs @ [_, ..] => {
+                quote! { Lazy::new(|| build_matcher_dynamic(&[ #(#segs),* ], #case_insensitive)) }
+            }
         };
         tokens.append_all(t);
     }
@@ -175,7 +187,12 @@ impl MatcherSegment {
 /// A struct that models a single .toml file in /src/syntax_mapping/builtins/.
 #[derive(Clone, Debug, Deserialize)]
 struct MappingDefModel {
+    #[serde(default)]
     mappings: IndexMap<MappingTarget, Vec<Matcher>>,
+    /// Case-sensitive mappings. Unlike `mappings`, these glob patterns are
+    /// matched case-sensitively.
+    #[serde(default)]
+    case_sensitive_mappings: IndexMap<MappingTarget, Vec<Matcher>>,
 }
 impl MappingDefModel {
     fn into_mapping_list(self) -> MappingList {
@@ -188,6 +205,19 @@ impl MappingDefModel {
                     .map(|matcher| (matcher, target.clone()))
                     .collect::<Vec<_>>()
             })
+            .chain(
+                self.case_sensitive_mappings
+                    .into_iter()
+                    .flat_map(|(target, matchers)| {
+                        matchers
+                            .into_iter()
+                            .map(|mut matcher| {
+                                matcher.case_insensitive = false;
+                                (matcher, target.clone())
+                            })
+                            .collect::<Vec<_>>()
+                    }),
+            )
             .collect();
         MappingList(list)
     }

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

@@ -0,0 +1,2 @@
+[case_sensitive_mappings]
+"Python" = ["BUILD"]