Merge pull request #3652 from cyqsimon/syntax-mapping-glob-parse-cleanup

Builtin syntax mapping: cleanup matcher glob parsing logic

CHANGELOG.md

@@ -33,6 +33,7 @@
 - Use git version of cross. See #3533 (@OctopusET)
 - Bump MSRV to 1.88, update `time` crate to 0.3.47 to fix RUSTSEC-2026-0009, see #3581 (@NORMAL-EX)
 - Allow home and end keys to be used with builtin pager, see #3651 (@keith-hall)
+- Builtin syntax mapping: cleanup matcher glob parsing logic #3652 (@cyqsimon)
 
 ## Syntaxes
 

build/syntax_mapping.rs

@@ -47,13 +47,24 @@ impl ToTokens for MappingTarget {
     }
 }
 
+/// 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,
+        case_sensitive: Option<bool>,
+    },
+}
+
 /// 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)]
+#[derive(Clone, Debug, PartialEq, Eq, Hash)]
 enum Case {
     Sensitive,
-    #[default]
     Insensitive,
 }
 impl ToTokens for Case {
@@ -66,11 +77,11 @@ impl ToTokens for Case {
     }
 }
 
-#[derive(Clone, Debug, Default, PartialEq, Eq, Hash, Deserialize)]
-#[serde(try_from = "RawMatcher")]
 /// A single matcher.
 ///
 /// Codegen converts this into a `Lazy<Option<GlobMatcher>>`.
+#[derive(Clone, Debug, PartialEq, Eq, Hash, Deserialize)]
+#[serde(try_from = "RawMatcher")]
 struct Matcher {
     segments: Vec<MatcherSegment>,
     /// Whether the glob pattern should be matched case-sensitively.
@@ -78,7 +89,7 @@ struct Matcher {
     /// Defaults to `Case::Insensitive` for backwards compatibility.
     case: Case,
 }
-/// Parse a matcher.
+/// Parse the glob pattern of a matcher.
 ///
 /// Note that this implementation is rather strict: it will greedily interpret
 /// every valid environment variable replacement as such, then immediately
@@ -92,9 +103,7 @@ struct Matcher {
 ///
 /// Revision history:
 /// - 2024-02-20: allow `{` and `}` (glob brace expansion)
-impl FromStr for Matcher {
+fn parse_glob(s: &str) -> Result<Vec<MatcherSegment>, anyhow::Error> {
-    type Err = anyhow::Error;
-    fn from_str(s: &str) -> Result<Self, Self::Err> {
     use MatcherSegment as Seg;
     static VAR_REGEX: Lazy<Regex> = Lazy::new(|| Regex::new(r"\$\{([\w\d_]+)\}").unwrap());
 
@@ -142,43 +151,31 @@ impl FromStr for Matcher {
         bail!(r#"Invalid matcher: "{s}""#);
     }
 
-        Ok(Self {
+    Ok(non_empty_segments)
-            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 {
+        const DEFAULT_CASE: Case = Case::Insensitive;
-            RawMatcher::Simple(s) => Matcher::from_str(&s),
+        match &raw {
+            RawMatcher::Simple(s) => {
+                let segments = parse_glob(s)?;
+                Ok(Self {
+                    segments,
+                    case: DEFAULT_CASE,
+                })
+            }
             RawMatcher::Full {
                 glob,
                 case_sensitive,
             } => {
-                let mut matcher = Matcher::from_str(&glob)?;
+                let segments = parse_glob(glob)?;
-                matcher.case = if case_sensitive {
+                let case = match case_sensitive {
-                    Case::Sensitive
+                    None => DEFAULT_CASE,
-                } else {
+                    Some(false) => Case::Insensitive,
-                    Case::Insensitive
+                    Some(true) => Case::Sensitive,
                 };
-                Ok(matcher)
+                Ok(Self { segments, case })
             }
         }
     }