address review: canonicalize through the checked io gateway, treat any occupied export name as taken

* MIDI: index controller dedup by channel and kind

Remove Duplicate Notes rescanned every controller already kept at the
same tick for each incoming one. A session whose sequence holds one tick
with many distinct channel/controller pairs made that quadratic and
blocked editing. Track where each channel+kind sits at the current tick
in a map instead; the last event per key still wins.

* MIDI: tag dedup index entries with their tick instead of clearing per tick

Clearing the map on every tick change cost O(capacity) per tick after a
dense tick. Store the tick with each entry and treat entries from an
earlier tick as absent. Add an oracle test comparing against the
previous per-tick rescan on mixed-tick input.

crates/engine/src/clip_group_file.rs

@@ -87,16 +87,23 @@ fn decode_group_audio(
     cmd_id: &str,
 ) -> Result<(Vec<(String, soundcraft_audio_io::AudioBuffer)>, BTreeMap<u64, usize>)> {
     let (mut decoded, mut index, mut file_of) = (Vec::new(), BTreeMap::new(), BTreeMap::new());
+    // The folder as the file system knows it, so a link among the group's files can't lead out.
+    let dir = if dir.as_os_str().is_empty() { Path::new(".") } else { dir };
+    let root = crate::io::canonical_path(dir)?;
     for (old, rel) in audio_entries {
         let full = dir.join(rel);
+        let escapes = |rel: &str| EngineError::BadParams(cmd_id.into(), format!("audio path escapes the group folder: {rel}"));
         if !full.starts_with(dir)
             || Path::new(rel)
                 .components()
                 .any(|c| matches!(c, std::path::Component::ParentDir | std::path::Component::RootDir | std::path::Component::Prefix(_)))
         {
-            return Err(EngineError::BadParams(cmd_id.into(), format!("audio path escapes the group folder: {rel}")));
+            return Err(escapes(rel));
+        }
+        let key = crate::io::canonical_path(&full)?;
+        if !key.starts_with(&root) {
+            return Err(escapes(rel));
         }
-        let key = crate::io::file_key(&full);
         let n = match index.get(&key) {
             Some(n) => *n,
             None => {
@@ -268,9 +275,59 @@ mod tests {
         std::fs::write(&path, file.to_string()).unwrap();
         let r = e.execute("file.import_clip_groups", &json!({"path": path, "at": 1000, "track": "Bass"}));
         assert!(matches!(r, Err(crate::EngineError::BadParams(..))), "absolute paths are refused: {r:?}");
+        // A link inside the folder that leads out of it is refused once resolved.
+        #[cfg(unix)]
+        {
+            let outside = std::env::temp_dir().join(format!("sc-grp-outside-{}.wav", std::process::id()));
+            std::fs::copy(dir.join("Wait... what_0.wav"), &outside).unwrap();
+            std::os::unix::fs::symlink(&outside, dir.join("link.wav")).unwrap();
+            for rel in file["audio"].as_object_mut().unwrap().values_mut() {
+                *rel = json!("link.wav");
+            }
+            std::fs::write(&path, file.to_string()).unwrap();
+            let r = e.execute("file.import_clip_groups", &json!({"path": path, "at": 1000, "track": "Bass"}));
+            assert!(matches!(r, Err(crate::EngineError::BadParams(..))), "links out of the folder are refused: {r:?}");
+            let _ = std::fs::remove_file(&outside);
+        }
         let _ = std::fs::remove_dir_all(&dir);
     }
 
+    /// Under a read root, a link in the group folder to a folder outside the root is refused the same way
+    /// whether or not the file asked for exists there, so the group cannot be used to probe for names.
+    #[cfg(unix)]
+    #[test]
+    fn links_out_of_the_read_root_are_refused_alike_for_existing_and_missing_files() {
+        let base = std::env::temp_dir().join(format!("sc-grp-root-{}", std::process::id()));
+        let _ = std::fs::remove_dir_all(&base);
+        let (dir, outside) = (base.join("group"), base.join("outside"));
+        std::fs::create_dir_all(&dir).unwrap();
+        std::fs::create_dir_all(&outside).unwrap();
+        let path = dir.join("drums.scgrp").to_string_lossy().to_string();
+        let mut e = crate::demo::demo_engine();
+        let kick = e.session().tracks.iter().find(|t| t.name == "Kick").map(|t| t.id).unwrap();
+        let ids: Vec<_> = e.session().track(kick).unwrap().clips().iter().take(1).map(|c| c.id.0).collect();
+        e.execute("file.export_clip_groups", &json!({"path": path, "clips": ids})).unwrap();
+        let mut file: serde_json::Value = serde_json::from_str(&std::fs::read_to_string(&path).unwrap()).unwrap();
+        let exported = file["audio"].as_object().unwrap().values().next().unwrap().as_str().unwrap().to_string();
+        std::fs::copy(dir.join(&exported), outside.join("real.wav")).unwrap();
+        std::os::unix::fs::symlink(&outside, dir.join("out")).unwrap();
+        let roots = crate::file_access::AutomationRoots::new(&[&dir], &[]).unwrap().unwrap();
+        let mut refusals = Vec::new();
+        for target in ["out/real.wav", "out/missing.wav"] {
+            for rel in file["audio"].as_object_mut().unwrap().values_mut() {
+                *rel = json!(target);
+            }
+            std::fs::write(&path, file.to_string()).unwrap();
+            let r = crate::file_access::confine(Some(&roots), || {
+                e.execute("file.import_clip_groups", &json!({"path": path, "at": 1000, "track": "Bass"}))
+            });
+            refusals.push(r.unwrap_err().to_string());
+        }
+        assert!(refusals[0].contains("outside the read root"), "{}", refusals[0]);
+        assert_eq!(refusals[0].replace("real.wav", "X"), refusals[1].replace("missing.wav", "X"), "existing and missing names are refused alike");
+        let _ = std::fs::remove_dir_all(&base);
+    }
+
     #[test]
     fn imported_clips_carry_no_auto_rename_scheme() {
         let dir = std::env::temp_dir().join(format!("sc-grp-names-{}", std::process::id()));

crates/engine/src/io.rs

@@ -85,6 +85,17 @@ pub(crate) fn make_dir(dir: impl AsRef<Path>) -> Result<()> {
     std::fs::create_dir_all(dir).map_err(|err| EngineError::Io(format!("{}: {err}", dir.display())))
 }
 
+/// `path` with its links followed, inside the automation read root when there is one. The root is
+/// checked on the path as named (links followed, a missing tail resolved by its nearest folder)
+/// before anything is resolved, so a name that leads out of the root is refused the same way
+/// whether or not it exists there.
+pub(crate) fn canonical_path(path: impl AsRef<Path>) -> Result<PathBuf> {
+    let path = path.as_ref();
+    let checked = if path.as_os_str().is_empty() { Path::new(".") } else { path };
+    crate::file_access::check_read(checked).map_err(EngineError::Io)?;
+    std::fs::canonicalize(path).map_err(|err| EngineError::Io(format!("{}: {err}", path.display())))
+}
+
 /// Write `bytes` to the file at `path`, inside the automation write root when there is one.
 pub(crate) fn write_path(path: impl AsRef<Path>, bytes: impl AsRef<[u8]>) -> Result<()> {
     let path = path.as_ref();
@@ -329,7 +340,7 @@ impl Out {
         if cfg!(target_arch = "wasm32") {
             return Ok(Out::Memory(std::io::Cursor::new(Vec::new())));
         }
-        Ok(Out::File(std::io::BufWriter::with_capacity(1 << 20, std::fs::File::create(part)?)))
+        Ok(Out::File(std::io::BufWriter::with_capacity(1 << 20, create_new(part)?)))
     }
 
     /// Puts the finished file at `path` and returns its size: the part file is synced and renamed
@@ -462,14 +473,27 @@ pub fn export_clips(e: &Engine, ids: &[soundcraft_model::ClipId], dir: &str) ->
         let stem = sanitize_name(&c.name);
         let mut path = Path::new(dir).join(format!("{stem}.wav"));
         let mut k = 2u32;
-        while path.exists() {
+        // The name is taken by creating the file, so whatever entry already has it (a link
+        // to elsewhere included) is left alone rather than written through.
+        let mut file = loop {
+            // An occupied name is taken whatever it is (a link whose target is outside the write
+            // root too): it is only looked at, never resolved, so the root check does not stop the
+            // search for a free name. The folder itself was checked above.
+            if std::fs::symlink_metadata(&path).is_err() {
+                crate::file_access::check_write(&path).map_err(EngineError::Io)?;
+                match std::fs::OpenOptions::new().write(true).create_new(true).open(&path) {
+                    Ok(f) => break f,
+                    Err(err) if err.kind() == std::io::ErrorKind::AlreadyExists => {}
+                    Err(err) => return Err(EngineError::Io(format!("{}: {err}", path.display()))),
+                }
+            }
             path = Path::new(dir).join(format!("{stem}-{k}.wav"));
             k = k.saturating_add(1);
             if k > 10_000 {
                 return Err(EngineError::Io(format!("{dir}: too many files named {stem}")));
             }
-        }
-        write_path(&path, bytes)?;
+        };
+        std::io::Write::write_all(&mut file, &bytes).map_err(|err| EngineError::Io(format!("{}: {err}", path.display())))?;
         n += 1;
     }
     Ok(n)
@@ -506,13 +530,13 @@ fn write_session(session: &Session, path: &str, overwrite: bool) -> Result<(Sess
         {
             src.path = f.join(&src.path).to_string_lossy().into_owned();
         }
-        let exists_rel = same_dir && !Path::new(&src.path).is_absolute() && dir.join(&src.path).exists();
+        let exists_rel = same_dir && !Path::new(&src.path).is_absolute() && file_found(dir.join(&src.path));
         let shared = shared_paths.get(i).copied().unwrap_or(false);
         if exists_rel && !src.unsaved && !shared {
             continue;
         }
         let in_folder = Path::new(&src.path).strip_prefix(&audio_dir).ok().and_then(Path::to_str).map(str::to_string);
-        if let Some(f) = in_folder.filter(|_| !src.unsaved && Path::new(&src.path).is_file()) {
+        if let Some(f) = in_folder.filter(|_| !src.unsaved && file_found(&src.path)) {
             src.path = format!("Audio Files/{f}");
             continue;
         }
@@ -528,7 +552,7 @@ fn write_session(session: &Session, path: &str, overwrite: bool) -> Result<(Sess
         if overwrite {
             // One file per source id so autosaves never collide or pile up.
             stem = format!("{stem}_{}", src.id.0);
-            if audio_dir.join(format!("{stem}.wav")).exists() {
+            if file_found(audio_dir.join(format!("{stem}.wav"))) {
                 src.path = format!("Audio Files/{stem}.wav");
                 src.unsaved = false;
                 files.insert(Arc::as_ptr(audio), src.path.clone());
@@ -588,11 +612,22 @@ fn sync_dir(dir: &Path) {
 
 fn write_synced(path: &Path, bytes: &[u8]) -> std::io::Result<()> {
     use std::io::Write;
-    let mut f = std::fs::File::create(path)?;
+    let mut f = create_new(path)?;
     f.write_all(bytes)?;
     f.sync_all()
 }
 
+/// Creates `path` as a new file. A leftover of that name (a part file from an interrupted write)
+/// is removed first, but an existing entry is never opened: a link planted at a predictable name
+/// in the destination folder can't redirect the write elsewhere.
+fn create_new(path: &Path) -> std::io::Result<std::fs::File> {
+    match std::fs::remove_file(path) {
+        Err(err) if err.kind() != std::io::ErrorKind::NotFound => return Err(err),
+        _ => {}
+    }
+    std::fs::OpenOptions::new().write(true).create_new(true).open(path)
+}
+
 /// Copies `from` to `to`, which must not exist yet; an existing entry, even a link, is never followed.
 fn copy_new(from: &Path, to: &Path) -> std::io::Result<()> {
     let mut src = std::fs::File::open(from)?;
@@ -687,7 +722,12 @@ fn store_wav(dir: &Path, stem: &str, bytes: &[u8]) -> Result<(String, bool)> {
                 return Ok((name, true));
             }
             Err(err) if err.kind() == std::io::ErrorKind::AlreadyExists => {
-                if std::fs::metadata(&path).is_ok_and(|m| m.len() == bytes.len() as u64) && std::fs::read(&path).is_ok_and(|b| b == bytes) {
+                // Reused only when it may be read: whether it equals the audio must not be
+                // learnt about a file the roots allow writing beside but not reading.
+                if crate::file_access::check_read(&path).is_ok()
+                    && std::fs::metadata(&path).is_ok_and(|m| m.len() == bytes.len() as u64)
+                    && std::fs::read(&path).is_ok_and(|b| b == bytes)
+                {
                     return Ok((name, false));
                 }
             }
@@ -785,6 +825,67 @@ pub fn plan_takes(e: &Engine, unsaved_dir: &Path, input_channels: usize) -> Vec<
         .collect()
 }
 
+#[cfg(test)]
+mod exclusive_tests {
+    use super::*;
+
+    fn scratch(name: &str) -> PathBuf {
+        let dir = std::env::temp_dir().join(format!("soundcraft-io-{name}-{}", std::process::id()));
+        let _ = std::fs::remove_dir_all(&dir);
+        std::fs::create_dir_all(&dir).unwrap();
+        dir
+    }
+
+    #[cfg(unix)]
+    #[test]
+    fn a_planted_link_at_the_part_name_is_not_written_through() {
+        let dir = scratch("part-link");
+        let (target, part) = (dir.join("elsewhere"), dir.join("session.scraft.tmp"));
+        std::os::unix::fs::symlink(&target, &part).unwrap();
+        write_synced(&part, b"new").unwrap();
+        assert!(!target.exists(), "the link's target was created");
+        assert_eq!(std::fs::read(&part).unwrap(), b"new", "the link itself was replaced by the file");
+        // A leftover regular part file is still replaced.
+        write_synced(&part, b"newer").unwrap();
+        assert_eq!(std::fs::read(&part).unwrap(), b"newer");
+        let _ = std::fs::remove_dir_all(&dir);
+    }
+
+    #[cfg(unix)]
+    #[test]
+    fn export_skips_a_name_held_by_a_dangling_link() {
+        let dir = scratch("export-link");
+        let e = crate::demo::demo_engine();
+        let (t, c) = e.session().tracks.iter().find_map(|t| t.clips().iter().find(|c| c.is_audio()).map(|c| (t.id, c.clone()))).unwrap();
+        let _ = t;
+        let target = dir.join("elsewhere.wav");
+        std::os::unix::fs::symlink(&target, dir.join(format!("{}.wav", sanitize_name(&c.name)))).unwrap();
+        assert_eq!(export_clips(&e, &[c.id], &dir.to_string_lossy()).unwrap(), 1);
+        assert!(!target.exists(), "the export followed the link");
+        assert!(dir.join(format!("{}-2.wav", sanitize_name(&c.name))).is_file(), "the next free name was used");
+        let _ = std::fs::remove_dir_all(&dir);
+    }
+
+    /// With a write root, a link out of it that holds the name is still just an occupied name.
+    #[cfg(unix)]
+    #[test]
+    fn export_skips_a_link_out_of_the_write_root_too() {
+        let dir = scratch("export-root-link");
+        let e = crate::demo::demo_engine();
+        let c = e.session().tracks.iter().find_map(|t| t.clips().iter().find(|c| c.is_audio()).cloned()).unwrap();
+        let elsewhere = scratch("export-root-elsewhere");
+        let target = elsewhere.join("target.wav");
+        std::os::unix::fs::symlink(&target, dir.join(format!("{}.wav", sanitize_name(&c.name)))).unwrap();
+        let roots = crate::file_access::AutomationRoots::new(&[], &[&dir]).unwrap().unwrap();
+        let n = crate::file_access::confine(Some(&roots), || export_clips(&e, &[c.id], &dir.to_string_lossy())).unwrap();
+        assert_eq!(n, 1);
+        assert!(!target.exists(), "the export followed the link out of the root");
+        assert!(dir.join(format!("{}-2.wav", sanitize_name(&c.name))).is_file(), "the next free name was used");
+        let _ = std::fs::remove_dir_all(&dir);
+        let _ = std::fs::remove_dir_all(&elsewhere);
+    }
+}
+
 #[cfg(test)]
 mod stem_tests {
     #[test]

crates/midi/src/sequence.rs

@@ -1,4 +1,6 @@
 use serde::{Deserialize, Serialize};
+use std::collections::HashMap;
+use std::collections::hash_map::Entry;
 
 /// A MIDI note. Times are in model ticks ([`crate::PPQ`] per quarter note).
 #[derive(Debug, Clone, Copy, PartialEq, Serialize, Deserialize)]
@@ -108,14 +110,28 @@ impl Sequence {
 
         self.ctrls.sort_by_key(|c| c.tick);
         let mut kept: Vec<CtrlEvent> = Vec::with_capacity(self.ctrls.len());
+        // The latest tick and `kept` index of each channel+kind, so a same-tick repeat is found
+        // without rescanning the tick's events (a tick can hold many thousands of distinct keys).
+        // An entry from an earlier tick counts as absent; keeping the tick instead of clearing the
+        // map per tick keeps the work per event constant after a dense tick.
+        let mut last_at: HashMap<(u8, (u8, u8)), (i64, usize)> = HashMap::new();
         for c in self.ctrls.drain(..) {
             // Within a tick, replace an earlier event of the same channel+kind.
-            // Only events at the same tick (the tail of `kept`) can match.
-            let slot =
-                kept.iter_mut().rev().take_while(|k| k.tick == c.tick).find(|k| k.channel == c.channel && kind_key(k.kind) == kind_key(c.kind));
-            match slot {
-                Some(slot) => *slot = c,
-                None => kept.push(c),
+            match last_at.entry((c.channel, kind_key(c.kind))) {
+                Entry::Occupied(mut e) => {
+                    let (tick, idx) = *e.get();
+                    match kept.get_mut(idx) {
+                        Some(slot) if tick == c.tick => *slot = c,
+                        _ => {
+                            e.insert((c.tick, kept.len()));
+                            kept.push(c);
+                        }
+                    }
+                }
+                Entry::Vacant(e) => {
+                    e.insert((c.tick, kept.len()));
+                    kept.push(c);
+                }
             }
         }
         self.ctrls = kept;

crates/midi/tests/it/ops.rs

@@ -222,3 +222,72 @@ fn sequence_serde_round_trip() {
     let json = serde_json::to_string(&seq).unwrap();
     assert_eq!(serde_json::from_str::<Sequence>(&json).unwrap(), seq);
 }
+
+#[test]
+fn remove_duplicates_handles_a_large_same_tick_controller_group() {
+    // Every channel x controller number at one tick, twice: the second copy of each wins. A
+    // rescan of the tick's kept events per incoming event would make this quadratic.
+    let keys: Vec<(u8, u8)> = (0..=255u8).flat_map(|ch| (0..=255u8).map(move |cc| (ch, cc))).collect();
+    let ctrls = [0, 1]
+        .iter()
+        .flat_map(|&pass| keys.iter().map(move |&(channel, cc)| CtrlEvent { tick: 7, channel, kind: CtrlKind::Cc(cc), value: pass }))
+        .collect();
+    let mut seq = Sequence { notes: Vec::new(), ctrls };
+    let start = std::time::Instant::now();
+    assert_eq!(seq.remove_duplicates(), keys.len());
+    assert!(start.elapsed() < std::time::Duration::from_secs(10), "{:?}", start.elapsed());
+    assert_eq!(seq.ctrls.len(), keys.len());
+    assert!(seq.ctrls.iter().all(|c| c.value == 1));
+}
+
+#[test]
+fn remove_duplicates_matches_a_per_tick_rescan_on_mixed_ticks() {
+    // The straightforward per-tick rescan the indexed version replaced, kept as an oracle.
+    fn reference(mut ctrls: Vec<CtrlEvent>) -> Vec<CtrlEvent> {
+        ctrls.sort_by_key(|c| c.tick);
+        let key = |k: CtrlKind| match k {
+            CtrlKind::Cc(n) => (0, n),
+            CtrlKind::PitchBend => (1, 0),
+            CtrlKind::ChannelPressure => (2, 0),
+            CtrlKind::PolyPressure(p) => (3, p),
+            CtrlKind::Program => (4, 0),
+        };
+        let mut kept: Vec<CtrlEvent> = Vec::new();
+        for c in ctrls {
+            match kept.iter_mut().rev().take_while(|k| k.tick == c.tick).find(|k| k.channel == c.channel && key(k.kind) == key(c.kind)) {
+                Some(slot) => *slot = c,
+                None => kept.push(c),
+            }
+        }
+        kept
+    }
+
+    // Few ticks, channels and kinds, so the same key recurs both within a tick (interleaved with
+    // other keys) and across ticks; unsorted input from a fixed xorshift stream.
+    let mut state = 0x9E37_79B9_7F4A_7C15u64;
+    let mut next = move || {
+        state ^= state << 13;
+        state ^= state >> 7;
+        state ^= state << 17;
+        state
+    };
+    let kinds = |r: u64| match r % 5 {
+        0 => CtrlKind::Cc((r >> 8) as u8 % 4),
+        1 => CtrlKind::PitchBend,
+        2 => CtrlKind::ChannelPressure,
+        3 => CtrlKind::PolyPressure((r >> 8) as u8 % 4),
+        _ => CtrlKind::Program,
+    };
+    for round in 0..200 {
+        let mut ctrls = Vec::new();
+        for i in 0..(next() % 64) {
+            let r = next();
+            ctrls.push(CtrlEvent { tick: (r % 6) as i64 - 2, channel: (r >> 16) as u8 % 3, kind: kinds(r >> 24), value: (round * 100 + i) as i32 });
+        }
+        let mut seq = Sequence { notes: Vec::new(), ctrls: ctrls.clone() };
+        let removed = seq.remove_duplicates();
+        let expected = reference(ctrls.clone());
+        assert_eq!(seq.ctrls, expected, "round {round}");
+        assert_eq!(removed, ctrls.len() - expected.len(), "round {round}");
+    }
+}