sync from monorepo @ 2452e92e

crates/dirigent_fermata/src/core/extract.rs

@@ -0,0 +1,50 @@
+use regex::Regex;
+use std::sync::OnceLock;
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub enum Confidence {
+    /// Absolute path or path with explicit separator.
+    High,
+    /// Bare filename with extension; could be a word.
+    Low,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct PathCandidate {
+    pub text: String,
+    pub confidence: Confidence,
+}
+
+/// Heuristically extract path-like substrings from arbitrary text.
+/// Confident matches (absolute paths, paths containing separators) → `High`.
+/// Bare filenames with an extension → `Low` (advisory only).
+pub fn extract_path_candidates(text: &str) -> Vec<PathCandidate> {
+    static UNIX_ABS: OnceLock<Regex> = OnceLock::new();
+    static WIN_ABS: OnceLock<Regex> = OnceLock::new();
+    static REL_WITH_SEP: OnceLock<Regex> = OnceLock::new();
+    static BARE_NAME: OnceLock<Regex> = OnceLock::new();
+
+    let unix_abs = UNIX_ABS.get_or_init(|| Regex::new(r"(?m)(?:^|\s)(/[\w./~\-_]+)").unwrap());
+    let win_abs = WIN_ABS.get_or_init(|| Regex::new(r#"(?m)(?:^|\s)([A-Za-z]:\\[\w.\\\-_]+)"#).unwrap());
+    let rel = REL_WITH_SEP.get_or_init(|| Regex::new(r"(?m)(?:^|\s)((?:\./|\.\./|[\w\-_]+/)[\w./\-_]+)").unwrap());
+    let bare = BARE_NAME.get_or_init(|| Regex::new(r"(?m)(?:^|\s)([\w\-_]+\.[A-Za-z]{1,8})(?:\s|[.,;:!?]|$)").unwrap());
+
+    let mut out = Vec::new();
+    let mut seen = std::collections::HashSet::new();
+
+    for re in [unix_abs, win_abs, rel] {
+        for cap in re.captures_iter(text) {
+            let m = cap.get(1).unwrap().as_str().trim_end_matches(['.', ',', ';', ':', '!', '?']);
+            if seen.insert(m.to_string()) {
+                out.push(PathCandidate { text: m.to_string(), confidence: Confidence::High });
+            }
+        }
+    }
+    for cap in bare.captures_iter(text) {
+        let m = cap.get(1).unwrap().as_str();
+        if seen.insert(m.to_string()) {
+            out.push(PathCandidate { text: m.to_string(), confidence: Confidence::Low });
+        }
+    }
+    out
+}