ansi: strip DCS sequences

Treat DCS as non-printing, including the DCS wrapper used by tmux passthrough.

Adds a regression test for tmux-wrapped OSC 8 hyperlinks.

Author: khoek

src/ansi.rs

@@ -31,6 +31,41 @@ fn consume_osc_end_exclusive(it: &mut Peekable<CharIndices<'_>>, start: usize) -
     end
 }
 
+fn consume_dcs_end_exclusive(it: &mut Peekable<CharIndices<'_>>, start: usize) -> usize {
+    let mut end = start + 1;
+    let Some((idx, 'P')) = it.next() else {
+        return end;
+    };
+    end = idx + 1;
+
+    while let Some((idx, c)) = it.next() {
+        end = idx + c.len_utf8();
+        match c {
+            '\u{9c}' => return end,
+            '\u{1b}' => {
+                let Some((next_idx, next)) = it.peek() else {
+                    return end;
+                };
+                match next {
+                    '\u{1b}' => {
+                        end = *next_idx + 1;
+                        it.next();
+                    }
+                    '\\' => {
+                        end = *next_idx + 1;
+                        it.next();
+                        return end;
+                    }
+                    _ => {}
+                }
+            }
+            _ => {}
+        }
+    }
+
+    end
+}
+
 fn find_dfa_end_exclusive_after_entry(it: &mut Peekable<CharIndices<'_>>) -> Option<usize> {
     let mut state = State::S1;
     let mut maybe_end = None;
@@ -197,6 +232,7 @@ fn find_ansi_code_exclusive(it: &mut Peekable<CharIndices>) -> Option<(usize, us
                 it.next();
                 match it.peek() {
                     Some((_, ']')) => return Some((start, consume_osc_end_exclusive(it, start))),
+                    Some((_, 'P')) => return Some((start, consume_dcs_end_exclusive(it, start))),
                     _ => {
                         if let Some(end) = find_dfa_end_exclusive_after_entry(it) {
                             return Some((start, end));
@@ -338,6 +374,7 @@ mod tests {
         Regex::new(concat!(
             r"(?s)(?:",
             r"\x1b\].*?(?:\x07|\x9c|\x1b\\|\z)|",
+            r"\x1bP(?:[^\x1b\x9c]|\x1b\x1b|\x1b[^\x1b\\])*?(?:\x9c|\x1b\\|\x1b\z|\z)|",
             r"[\x1b\x9b]([()][012AB]|[\[()#;?]*(?:[0-9]{1,4}(?:;[0-9]{0,4})*)?",
             r"[0-9A-PRZcf-nqry=><])",
             r")",
@@ -467,6 +504,13 @@ mod tests {
         assert_eq!(strip_ansi_codes(s).as_ref(), "hello");
     }
 
+    #[test]
+    fn strip_tmux_passthrough_dcs() {
+        let open = "\x1bPtmux;\x1b\x1b\x1b]8;;file:///tmp/test\x1b\x1b\\\x1b\\";
+        let close = "\x1bPtmux;\x1b\x1b\x1b]8;;\x1b\x1b\\\x1b\\";
+        assert_eq!(strip_ansi_codes(&format!("{open}hello{close}")).as_ref(), "hello");
+    }
+
     #[test]
     fn test_ansi_iter_re_vt100() {
         let s = "\x1b(0lpq\x1b)Benglish";