use std::collections::{HashMap, HashSet};
use std::path::{Path, PathBuf};
use std::sync::Arc;
use std::time::Duration;
use anyhow::{Context, Result};
use notify::{
event::{AccessKind, AccessMode},
Event, EventHandler, EventKind, RecommendedWatcher, RecursiveMode, Watcher, WatcherKind,
};
use notify_debouncer_mini::{
new_debouncer_opt, Config as DebouncerConfig, DebounceEventResult, Debouncer,
};
#[path = "watch_hash.rs"]
mod hash;
#[cfg(test)]
#[path = "watch_hash_tests.rs"]
mod hash_tests;
use hash::{hash_file, HASH_BUFFER_SIZE};
pub struct WatcherHandle {
_debouncer: Debouncer<ContentChangeWatcher>,
}
struct ContentChangeWatcher(RecommendedWatcher);
impl Watcher for ContentChangeWatcher {
fn new<F: EventHandler>(mut event_handler: F, config: notify::Config) -> notify::Result<Self> {
RecommendedWatcher::new(
move |event| {
if matches!(&event, Ok(event) if is_read_only_access_event(event)) {
return;
}
event_handler.handle_event(event);
},
config,
)
.map(Self)
}
fn watch(&mut self, path: &Path, recursive_mode: RecursiveMode) -> notify::Result<()> {
self.0.watch(path, recursive_mode)
}
fn unwatch(&mut self, path: &Path) -> notify::Result<()> {
self.0.unwatch(path)
}
fn configure(&mut self, config: notify::Config) -> notify::Result<bool> {
self.0.configure(config)
}
fn kind() -> WatcherKind {
RecommendedWatcher::kind()
}
}
fn is_read_only_access_event(event: &Event) -> bool {
matches!(
event.kind,
EventKind::Access(
AccessKind::Read
| AccessKind::Open(AccessMode::Read)
| AccessKind::Close(AccessMode::Read)
)
)
}
pub struct WatchConfig {
pub paths: Vec<PathBuf>,
pub explicit_files: Vec<PathBuf>,
pub ignore: Vec<PathBuf>,
pub debounce: Duration,
pub retry_unchanged_when: Option<Arc<dyn Fn() -> bool + Send + Sync>>,
}
pub fn spawn_watcher<F>(cfg: WatchConfig, on_event: F) -> Result<WatcherHandle>
where
F: Fn(Vec<PathBuf>) + Send + 'static,
{
let ignore: Vec<PathBuf> = cfg
.ignore
.iter()
.map(|p| std::fs::canonicalize(p).unwrap_or_else(|_| p.clone()))
.collect();
let explicit_files: Vec<PathBuf> = cfg
.explicit_files
.iter()
.filter_map(|path| normalize_explicit_file(path))
.collect();
let mut content_hashes: HashMap<PathBuf, u64> = HashMap::new();
let mut hash_buffer = [0_u8; HASH_BUFFER_SIZE];
for file in &explicit_files {
content_changed(&mut content_hashes, file, &mut hash_buffer);
}
let retry_unchanged_when = cfg.retry_unchanged_when.clone();
let explicit_filter = explicit_files.clone();
let notify_config = notify::Config::default().with_follow_symlinks(false);
let debouncer_config = DebouncerConfig::default()
.with_timeout(cfg.debounce)
.with_notify_config(notify_config);
let mut debouncer = new_debouncer_opt::<_, ContentChangeWatcher>(
debouncer_config,
move |res: DebounceEventResult| match res {
Ok(events) => {
let mut paths: Vec<PathBuf> = Vec::with_capacity(events.len());
let mut seen: HashSet<PathBuf> = HashSet::with_capacity(events.len());
for e in events {
collect_event_paths(&e.path, &ignore, &explicit_filter, &mut seen, &mut paths);
}
paths.retain(|path| !path.is_dir());
let retry_unchanged = retry_unchanged_when
.as_ref()
.is_some_and(|predicate| predicate());
paths.retain(|path| {
should_forward_path(
&mut content_hashes,
path,
retry_unchanged,
&mut hash_buffer,
)
});
if !paths.is_empty() {
on_event(paths);
}
}
Err(e) => {
eprintln!("watcher error: {e:?}");
}
},
)
.context("Cannot start file watcher")?;
let mut watched_roots = Vec::with_capacity(cfg.paths.len());
for p in &cfg.paths {
if p.exists() {
watched_roots.push(std::fs::canonicalize(p).unwrap_or_else(|_| p.clone()));
debouncer
.watcher()
.watch(p, RecursiveMode::Recursive)
.with_context(|| format!("Cannot watch {}", p.display()))?;
}
}
watch_explicit_parents(debouncer.watcher(), &explicit_files, &watched_roots)?;
Ok(WatcherHandle {
_debouncer: debouncer,
})
}
fn watch_explicit_parents(
watcher: &mut dyn Watcher,
explicit_files: &[PathBuf],
watched_roots: &[PathBuf],
) -> Result<()> {
let mut explicit_parents = HashSet::new();
for file in explicit_files {
if watched_roots.iter().any(|root| file.starts_with(root)) {
continue;
}
while let Some(parent) = file.ancestors().skip(1).find(|parent| parent.is_dir()) {
if !explicit_parents.insert(parent.to_path_buf()) {
break;
}
watcher
.watch(parent, RecursiveMode::NonRecursive)
.with_context(|| format!("Cannot watch {}", parent.display()))?;
}
}
Ok(())
}
fn normalize_explicit_file(path: &Path) -> Option<PathBuf> {
if let Ok(canonical) = std::fs::canonicalize(path) {
return Some(canonical);
}
let path = std::path::absolute(path).ok()?;
for ancestor in path.ancestors().skip(1) {
if let Ok(canonical) = std::fs::canonicalize(ancestor) {
return Some(canonical.join(path.strip_prefix(ancestor).ok()?));
}
}
None
}
fn is_explicit_file(event_path: &Path, explicit_files: &[PathBuf]) -> bool {
let normalized = std::fs::canonicalize(event_path).unwrap_or_else(|_| event_path.to_path_buf());
explicit_files.iter().any(|path| path == &normalized)
}
fn should_ignore_event(event_path: &Path, ignore: &[PathBuf], explicit_files: &[PathBuf]) -> bool {
is_ignored(event_path, ignore) && !is_explicit_file(event_path, explicit_files)
}
fn collect_event_paths(
event_path: &Path,
ignore: &[PathBuf],
explicit_files: &[PathBuf],
seen: &mut HashSet<PathBuf>,
paths: &mut Vec<PathBuf>,
) {
let is_directory = event_path.is_dir();
if is_directory || !event_path.exists() {
if let Some(directory) = normalize_explicit_file(event_path) {
for file in explicit_files {
if file != &directory && file.starts_with(&directory) && seen.insert(file.clone()) {
paths.push(file.clone());
}
}
}
}
if !is_directory
&& !should_ignore_event(event_path, ignore, explicit_files)
&& seen.insert(event_path.to_path_buf())
{
paths.push(event_path.to_path_buf());
}
}
fn is_ignored(event_path: &Path, ignore: &[PathBuf]) -> bool {
if ignore.is_empty() {
return false;
}
let canon = std::fs::canonicalize(event_path).ok();
let candidate: &Path = canon.as_deref().unwrap_or(event_path);
for root in ignore {
if root.is_absolute() {
if candidate.starts_with(root) {
return true;
}
continue;
}
let mut components = root.components();
let first = components.next();
let only_one_component = components.next().is_none();
if let (Some(first), true) = (first, only_one_component) {
let name = first.as_os_str();
if event_path.components().any(|c| c.as_os_str() == name)
|| candidate.components().any(|c| c.as_os_str() == name)
{
return true;
}
} else if candidate.starts_with(root) {
return true;
}
}
false
}
#[must_use]
pub fn default_ignore_paths() -> Vec<PathBuf> {
vec![
PathBuf::from("node_modules"),
PathBuf::from(".git"),
PathBuf::from("target"),
]
}
fn content_changed(
cache: &mut HashMap<PathBuf, u64>,
path: &Path,
buffer: &mut [u8; HASH_BUFFER_SIZE],
) -> bool {
match hash_file(path, buffer) {
Some(hash) => match cache.insert(path.to_path_buf(), hash) {
Some(previous) => previous != hash,
None => true,
},
None => {
cache.remove(path);
true
}
}
}
fn should_forward_path(
cache: &mut HashMap<PathBuf, u64>,
path: &Path,
retry_unchanged: bool,
buffer: &mut [u8; HASH_BUFFER_SIZE],
) -> bool {
content_changed(cache, path, buffer) || retry_unchanged
}
#[cfg(test)]
#[allow(clippy::disallowed_methods)]
mod tests {
use super::*;
#[derive(Default)]
struct RecordingWatcher {
paths: Vec<(PathBuf, RecursiveMode)>,
}
impl Watcher for RecordingWatcher {
fn new<F: EventHandler>(_: F, _: notify::Config) -> notify::Result<Self> {
Ok(Self::default())
}
fn watch(&mut self, path: &Path, mode: RecursiveMode) -> notify::Result<()> {
self.paths.push((path.to_path_buf(), mode));
Ok(())
}
fn unwatch(&mut self, _: &Path) -> notify::Result<()> {
Ok(())
}
fn kind() -> WatcherKind {
RecommendedWatcher::kind()
}
}
#[test]
fn missing_include_does_not_recursively_watch_ignored_ancestor() {
let root = tempfile::tempdir().unwrap();
let target = root.path().join("target");
std::fs::create_dir_all(target.join("debug/build/unrelated")).unwrap();
let file = normalize_explicit_file(&target.join("generated/new/state.json")).unwrap();
let mut watcher = RecordingWatcher::default();
watch_explicit_parents(&mut watcher, &[file], &[]).unwrap();
assert_eq!(
watcher.paths,
vec![(
std::fs::canonicalize(target).unwrap(),
RecursiveMode::NonRecursive
)]
);
}
#[test]
fn is_ignored_matches_descendant() {
let dir = tempfile::tempdir().unwrap();
let dist = dir.path().join("dist");
std::fs::create_dir(&dist).unwrap();
let nested = dist.join("a/b.html");
std::fs::create_dir_all(nested.parent().unwrap()).unwrap();
std::fs::write(&nested, "x").unwrap();
let ignore = vec![std::fs::canonicalize(&dist).unwrap()];
assert!(is_ignored(&nested, &ignore));
}
#[test]
fn explicit_manifest_bypasses_ignored_output_directory() {
let dir = tempfile::tempdir().unwrap();
let dist = dir.path().join("dist");
std::fs::create_dir(&dist).unwrap();
let manifest = dist.join("webui-projection.json");
let bundle = dist.join("index.js");
std::fs::write(&manifest, "{}").unwrap();
std::fs::write(&bundle, "export {};").unwrap();
let ignore = vec![std::fs::canonicalize(&dist).unwrap()];
let explicit = vec![std::fs::canonicalize(&manifest).unwrap()];
assert!(!should_ignore_event(&manifest, &ignore, &explicit));
assert!(should_ignore_event(&bundle, &ignore, &explicit));
}
#[test]
fn explicit_file_tracks_missing_parent_directories() -> std::io::Result<()> {
let dir = tempfile::tempdir()?;
let file = dir.path().join("target/new/state.json");
let normalized = normalize_explicit_file(&file)
.ok_or_else(|| std::io::Error::other("missing include was discarded"))?;
assert_eq!(
normalized,
std::fs::canonicalize(dir.path())?.join("target/new/state.json")
);
std::fs::create_dir_all(
file.parent()
.ok_or_else(|| std::io::Error::other("missing state parent"))?,
)?;
std::fs::write(&file, "{}")?;
assert!(!should_ignore_event(
&file,
&[PathBuf::from("target")],
&[normalized]
));
Ok(())
}
#[test]
fn directory_event_rechecks_explicit_files_created_before_recursive_watch() {
let dir = tempfile::tempdir().unwrap();
let directory = dir.path().join("target/new");
let file = directory.join("state.json");
let explicit = vec![normalize_explicit_file(&file).unwrap()];
std::fs::create_dir_all(&directory).unwrap();
std::fs::write(&file, "{}").unwrap();
let mut seen = HashSet::new();
let mut paths = Vec::new();
collect_event_paths(
&directory,
&[PathBuf::from("target")],
&explicit,
&mut seen,
&mut paths,
);
assert_eq!(paths, explicit);
}
#[test]
fn directory_rechecks_dedupe_and_skip_unchanged_explicit_files() {
let dir = tempfile::tempdir().unwrap();
let directory = dir.path().join("target/new");
std::fs::create_dir_all(&directory).unwrap();
let file = directory.join("state.json");
std::fs::write(&file, "{}").unwrap();
let explicit = vec![normalize_explicit_file(&file).unwrap()];
let mut hashes = HashMap::new();
let mut buffer = [0_u8; HASH_BUFFER_SIZE];
assert!(content_changed(&mut hashes, &explicit[0], &mut buffer));
let mut seen = HashSet::new();
let mut paths = Vec::new();
for _ in 0..2 {
collect_event_paths(
&directory,
&[PathBuf::from("target")],
&explicit,
&mut seen,
&mut paths,
);
}
assert_eq!(paths, explicit);
assert!(!should_forward_path(
&mut hashes,
&paths[0],
false,
&mut buffer
));
std::fs::write(&file, "{\"count\":7}").unwrap();
assert!(should_forward_path(
&mut hashes,
&paths[0],
false,
&mut buffer
));
}
#[test]
fn directory_events_ignore_unrelated_files_and_recheck_removed_parents() {
let dir = tempfile::tempdir().unwrap();
let directory = dir.path().join("target/new");
let unrelated = dir.path().join("target/other");
std::fs::create_dir_all(&directory).unwrap();
std::fs::create_dir_all(&unrelated).unwrap();
let file = directory.join("state.json");
std::fs::write(&file, "{}").unwrap();
let explicit = vec![normalize_explicit_file(&file).unwrap()];
let ignore = vec![PathBuf::from("target")];
let mut seen = HashSet::new();
let mut paths = Vec::new();
collect_event_paths(&unrelated, &ignore, &explicit, &mut seen, &mut paths);
assert!(paths.is_empty());
std::fs::remove_file(&file).unwrap();
std::fs::remove_dir(&directory).unwrap();
collect_event_paths(&directory, &ignore, &explicit, &mut seen, &mut paths);
assert_eq!(paths, explicit);
}
#[test]
fn is_ignored_misses_sibling() {
let dir = tempfile::tempdir().unwrap();
let dist = dir.path().join("dist");
let other = dir.path().join("src/a.md");
std::fs::create_dir(&dist).unwrap();
std::fs::create_dir_all(other.parent().unwrap()).unwrap();
std::fs::write(&other, "x").unwrap();
let ignore = vec![std::fs::canonicalize(&dist).unwrap()];
assert!(!is_ignored(&other, &ignore));
}
#[test]
fn is_ignored_is_false_when_no_ignores() {
let p = Path::new("/anything/at/all");
assert!(!is_ignored(p, &[]));
}
#[test]
fn is_ignored_matches_relative_component_anywhere() {
let ignore = vec![PathBuf::from("target")];
assert!(is_ignored(Path::new("/repo/target/debug/build"), &ignore));
assert!(is_ignored(
Path::new("/repo/sub/crate/target/foo.rs"),
&ignore
));
assert!(!is_ignored(Path::new("/repo/src/target_name.rs"), &ignore));
}
#[test]
fn is_ignored_handles_node_modules_and_git() {
let ignore = default_ignore_paths();
assert!(is_ignored(
Path::new("/repo/node_modules/foo/index.js"),
&ignore
));
assert!(is_ignored(Path::new("/repo/.git/HEAD"), &ignore));
assert!(!is_ignored(Path::new("/repo/src/index.ts"), &ignore));
}
#[test]
fn read_only_access_events_are_not_content_changes() {
use notify::event::ModifyKind;
for kind in [
AccessKind::Read,
AccessKind::Open(AccessMode::Read),
AccessKind::Close(AccessMode::Read),
] {
assert!(is_read_only_access_event(&Event::new(EventKind::Access(
kind
))));
}
for kind in [
AccessKind::Any,
AccessKind::Open(AccessMode::Write),
AccessKind::Close(AccessMode::Write),
] {
assert!(!is_read_only_access_event(&Event::new(EventKind::Access(
kind
))));
}
assert!(!is_read_only_access_event(&Event::new(EventKind::Modify(
ModifyKind::Any
))));
}
#[cfg(unix)]
#[test]
fn is_ignored_matches_node_modules_symlink_path() {
use std::os::unix::fs::symlink;
let dir = tempfile::tempdir().unwrap();
let package = dir.path().join("packages/example");
let node_modules = dir.path().join("node_modules");
std::fs::create_dir_all(&package).unwrap();
std::fs::create_dir(&node_modules).unwrap();
std::fs::write(package.join("index.js"), "export {};").unwrap();
symlink(&package, node_modules.join("example")).unwrap();
let linked_file = node_modules.join("example/index.js");
let canonical = std::fs::canonicalize(&linked_file).unwrap();
assert!(!canonical
.components()
.any(|component| component.as_os_str() == "node_modules"));
assert!(is_ignored(&linked_file, &[PathBuf::from("node_modules")]));
}
#[test]
fn content_changed_skips_identical_saves() {
let dir = tempfile::tempdir().unwrap();
let file = dir.path().join("a.css");
std::fs::write(&file, "a { color: red; }").unwrap();
let mut cache = HashMap::new();
let mut buffer = [0_u8; HASH_BUFFER_SIZE];
assert!(content_changed(&mut cache, &file, &mut buffer));
assert!(!content_changed(&mut cache, &file, &mut buffer));
assert!(!content_changed(&mut cache, &file, &mut buffer));
std::fs::write(&file, "a { color: blue; }").unwrap();
assert!(content_changed(&mut cache, &file, &mut buffer));
assert!(!content_changed(&mut cache, &file, &mut buffer));
std::fs::remove_file(&file).unwrap();
assert!(content_changed(&mut cache, &file, &mut buffer));
std::fs::write(&file, "a { color: blue; }").unwrap();
assert!(content_changed(&mut cache, &file, &mut buffer));
}
#[test]
fn unchanged_save_can_retry_when_error_active() {
let dir = tempfile::tempdir().unwrap();
let file = dir.path().join("a.css");
std::fs::write(&file, "a { color: red; }").unwrap();
let mut cache = HashMap::new();
let mut buffer = [0_u8; HASH_BUFFER_SIZE];
assert!(should_forward_path(&mut cache, &file, false, &mut buffer));
assert!(!should_forward_path(&mut cache, &file, false, &mut buffer));
assert!(should_forward_path(&mut cache, &file, true, &mut buffer));
assert!(!should_forward_path(&mut cache, &file, false, &mut buffer));
std::fs::write(&file, "a { color: blue; }").unwrap();
assert!(should_forward_path(&mut cache, &file, true, &mut buffer));
assert!(!should_forward_path(&mut cache, &file, false, &mut buffer));
}
#[test]
fn unhashable_paths_forget_cached_content() -> std::io::Result<()> {
let dir = tempfile::tempdir()?;
let file = dir.path().join("a.css");
let mut cache = HashMap::new();
let mut buffer = [0_u8; HASH_BUFFER_SIZE];
for replacement in ["deleted", "directory", "oversized"] {
std::fs::write(&file, "original")?;
assert!(content_changed(&mut cache, &file, &mut buffer));
assert!(!content_changed(&mut cache, &file, &mut buffer));
std::fs::remove_file(&file)?;
match replacement {
"directory" => std::fs::create_dir(&file)?,
"oversized" => std::fs::File::create(&file)?.set_len(8 * 1024 * 1024 + 1)?,
_ => {}
}
for _ in 0..2 {
assert!(content_changed(&mut cache, &file, &mut buffer));
assert!(!cache.contains_key(&file));
}
match replacement {
"directory" => std::fs::remove_dir(&file)?,
"oversized" => std::fs::remove_file(&file)?,
_ => {}
}
}
std::fs::write(&file, "original")?;
assert!(content_changed(&mut cache, &file, &mut buffer));
Ok(())
}
#[test]
fn hash_buffer_is_reused_across_files_and_event_batches() -> std::io::Result<()> {
let dir = tempfile::tempdir()?;
let paths = [
dir.path().join("large.css"),
dir.path().join("empty.css"),
dir.path().join("small.css"),
];
for (path, size) in paths.iter().zip([3 * HASH_BUFFER_SIZE + 1, 0, 7]) {
std::fs::write(path, vec![b'x'; size])?;
}
let mut cache = HashMap::new();
let mut buffer = [0_u8; HASH_BUFFER_SIZE];
for changed in [true, false, false] {
for path in &paths {
assert_eq!(content_changed(&mut cache, path, &mut buffer), changed);
}
}
Ok(())
}
#[cfg(unix)]
#[test]
fn unreadable_symlink_loop_forgets_cached_content() -> std::io::Result<()> {
let dir = tempfile::tempdir()?;
let file = dir.path().join("unreadable.css");
std::fs::write(&file, "original")?;
let mut cache = HashMap::new();
let mut buffer = [0_u8; HASH_BUFFER_SIZE];
assert!(content_changed(&mut cache, &file, &mut buffer));
std::fs::remove_file(&file)?;
std::os::unix::fs::symlink("unreadable.css", &file)?;
assert!(content_changed(&mut cache, &file, &mut buffer));
assert!(!cache.contains_key(&file));
std::fs::remove_file(&file)?;
std::fs::write(&file, "original")?;
assert!(content_changed(&mut cache, &file, &mut buffer));
Ok(())
}
}