Add regex extractor

crates/pile-config/src/objectpath/tokenizer.rs

@@ -21,7 +21,52 @@ impl Tokenizer {
 		let mut tokens = Vec::new();
 
 		let mut window_start = None;
+		// Paren depth: while > 0, `.` / `[` / `]` / `$` are part of the ident.
+		let mut paren_depth: usize = 0;
+		// When true, the current char is escaped by a preceding `\` and is
+		// treated as a plain ident character with no special meaning.
+		let mut skip_next = false;
+
 		for (i, c) in source.char_indices() {
+			if skip_next {
+				skip_next = false;
+				// Escaped char: just extend the ident window (already opened by `\`).
+				continue;
+			}
+
+			if c == '\\' {
+				if window_start.is_none() {
+					window_start = Some(i);
+				}
+				skip_next = true;
+				continue;
+			}
+
+			if paren_depth > 0 {
+				// Inside parens: only track depth changes, everything else is ident.
+				match c {
+					'(' => {
+						if window_start.is_none() {
+							window_start = Some(i);
+						}
+						paren_depth += 1;
+					}
+					')' => {
+						if window_start.is_none() {
+							window_start = Some(i);
+						}
+						paren_depth -= 1;
+					}
+					x if x.is_ascii() => {
+						if window_start.is_none() {
+							window_start = Some(i);
+						}
+					}
+					char => return Err(PathParseError::NonAsciiChar { position: i, char }),
+				}
+				continue;
+			}
+
 			match c {
 				'$' => {
 					if let Some(s) = window_start.take() {
@@ -51,10 +96,26 @@ impl Tokenizer {
 					tokens.push((i, Token::SqbClose));
 				}
 
-				x if x.is_ascii() => match window_start {
-					None => window_start = Some(i),
-					Some(_) => continue,
-				},
+				'(' => {
+					if window_start.is_none() {
+						window_start = Some(i);
+					}
+					paren_depth += 1;
+				}
+
+				')' => {
+					if window_start.is_none() {
+						window_start = Some(i);
+					}
+					// paren_depth is 0 here — stray `)` is an ident char and
+					// parse_field will surface the error later.
+				}
+
+				x if x.is_ascii() => {
+					if window_start.is_none() {
+						window_start = Some(i);
+					}
+				}
 
 				char => return Err(PathParseError::NonAsciiChar { position: i, char }),
 			}

crates/pile-config/src/pattern/parser.rs

@@ -122,6 +122,14 @@ mod tests {
 		GroupSegment::Literal(s.into())
 	}
 
+	#[test]
+	fn regex() {
+		assert_eq!(
+			parse("{$.split(/)[-1].regex((.*).pub \\((.*)\\).pdf)[0]}").unwrap(),
+			vec![(0, path("$.split(/)[-1].regex((.*).pub \\((.*)\\).pdf)[0]"))]
+		);
+	}
+
 	#[test]
 	fn single_path() {
 		assert_eq!(parse("{$.foo}").unwrap(), vec![(0, path("$.foo"))]);