Merge pull request #3576 from vorburger/patch-2

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

CHANGELOG.md

@@ -13,6 +13,7 @@
 - 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 inconsistent `.deb` MUSL package names (aarch64-musl used `arm64` instead of `musl-linux-arm64`, and `musleabihf` target missed `bat-musl` prefix). Closes #3482, see #3642 (@mvanhorn)

build/syntax_mapping.rs

@@ -47,11 +47,37 @@ impl ToTokens for MappingTarget {
     }
 }
 
-#[derive(Clone, Debug, PartialEq, Eq, Hash, DeserializeFromStr)]
+/// Whether a glob pattern should be matched case-sensitively or case-insensitively.
+///
+/// Mirrors the runtime `Case` type in `src/syntax_mapping.rs`.
+#[derive(Clone, Debug, PartialEq, Eq, Hash, Default)]
+enum Case {
+    Sensitive,
+    #[default]
+    Insensitive,
+}
+impl ToTokens for Case {
+    fn to_tokens(&self, tokens: &mut TokenStream) {
+        let t = match self {
+            Self::Sensitive => quote! { Case::Sensitive },
+            Self::Insensitive => quote! { Case::Insensitive },
+        };
+        tokens.append_all(t);
+    }
+}
+
+#[derive(Clone, Debug, Default, PartialEq, Eq, Hash, Deserialize)]
+#[serde(try_from = "RawMatcher")]
 /// 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-sensitively.
+    ///
+    /// Defaults to `Case::Insensitive` for backwards compatibility.
+    case: Case,
+}
 /// Parse a matcher.
 ///
 /// Note that this implementation is rather strict: it will greedily interpret
@@ -116,18 +142,59 @@ impl FromStr for Matcher {
             bail!(r#"Invalid matcher: "{s}""#);
         }
 
-        Ok(Self(non_empty_segments))
+        Ok(Self {
+            segments: non_empty_segments,
+            case: Case::Insensitive,
+        })
+    }
+}
+
+/// Helper type for deserializing a `Matcher` from either a plain string or a
+/// `{ glob = "...", case_sensitive = true }` struct.
+#[derive(Deserialize)]
+#[serde(untagged)]
+enum RawMatcher {
+    Simple(String),
+    Full {
+        glob: String,
+        #[serde(default)]
+        case_sensitive: bool,
+    },
+}
+
+impl TryFrom<RawMatcher> for Matcher {
+    type Error = anyhow::Error;
+
+    fn try_from(raw: RawMatcher) -> Result<Self, Self::Error> {
+        match raw {
+            RawMatcher::Simple(s) => Matcher::from_str(&s),
+            RawMatcher::Full {
+                glob,
+                case_sensitive,
+            } => {
+                let mut matcher = Matcher::from_str(&glob)?;
+                matcher.case = if case_sensitive {
+                    Case::Sensitive
+                } else {
+                    Case::Insensitive
+                };
+                Ok(matcher)
+            }
+        }
     }
 }
 impl ToTokens for Matcher {
     fn to_tokens(&self, tokens: &mut TokenStream) {
-        let t = match self.0.as_slice() {
+        let case = &self.case;
+        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))) }
             }
             // 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)) }
+            }
         };
         tokens.append_all(t);
     }
@@ -175,6 +242,7 @@ 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>>,
 }
 impl MappingDefModel {

src/syntax_mapping.rs

@@ -17,9 +17,16 @@ use ignored_suffixes::IgnoredSuffixes;
 mod builtin;
 pub mod ignored_suffixes;
 
-fn make_glob_matcher(from: &str) -> Result<GlobMatcher> {
+/// Whether a glob pattern should be matched case-sensitively or case-insensitively.
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub(crate) enum Case {
+    Sensitive,
+    Insensitive,
+}
+
+fn make_glob_matcher(from: &str, case: Case) -> Result<GlobMatcher> {
     let matcher = GlobBuilder::new(from)
-        .case_insensitive(true)
+        .case_insensitive(matches!(case, Case::Insensitive))
         .literal_separator(true)
         .build()?
         .compile_matcher();
@@ -97,7 +104,14 @@ 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, Case::Insensitive)?;
+        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, Case::Sensitive)?;
         self.custom_mappings.push((matcher, to));
         Ok(())
     }
@@ -261,4 +275,41 @@ 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

@@ -3,7 +3,7 @@ use std::env;
 use globset::GlobMatcher;
 use once_cell::sync::Lazy;
 
-use crate::syntax_mapping::{make_glob_matcher, MappingTarget};
+use crate::syntax_mapping::{make_glob_matcher, Case, MappingTarget};
 
 // Static syntax mappings generated from /src/syntax_mapping/builtins/ by the
 // build script (/build/syntax_mapping.rs).
@@ -53,8 +53,8 @@ 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 {
+fn build_matcher_fixed(from: &str, case: Case) -> GlobMatcher {
-    make_glob_matcher(from).expect("A builtin fixed glob matcher failed to compile")
+    make_glob_matcher(from, case).expect("A builtin fixed glob matcher failed to compile")
 }
 
 /// Join a list of matcher segments to create a glob string, replacing all
@@ -64,7 +64,7 @@ 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: Case) -> Option<GlobMatcher> {
     // join segments
     let mut buf = String::new();
     for seg in segs {
@@ -77,7 +77,7 @@ 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).ok()?;
     Some(matcher)
 }
 

src/syntax_mapping/builtins/README.md

@@ -20,12 +20,10 @@ 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
+Each TOML file should contain a single section named `mappings`, with each of its keys being a language
-its keys being a language identifier (first column of `bat -L`; also referred to
+identifier (first column of `bat -L`; also referred to as "target").
-as "target").
 
-The value of each key should be an array of strings, with each item being a glob
+The value of each key should be an array of "rules". The rules are expected to be objects with a `glob` string and a `case_sensitive` boolean. For simplification, a rule can be just a glob string, which is shorthand for the default case insensitive mode.
-matcher. We will call each of these items a "rule".
 
 For example, if `foo-application` uses both TOML and YAML configuration files,
 we could write something like this:
@@ -98,6 +96,15 @@ like this:
 ]
 ```
 
+### Case sensitivity
+
+By default, all glob patterns are matched case-insensitively. To match a pattern case-sensitively, use the object form of the rule with the `case_sensitive` option:
+
+```toml
+[mappings]
+"Python" = [{ glob = "BUILD", case_sensitive = true }]
+```
+
 ## Ordering
 
 At compile time, all TOML files applicable to the target are processed in

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

@@ -0,0 +1,2 @@
+[mappings]
+"Python" = [{ glob = "BUILD", case_sensitive = true }]