new flat KV structure

This commit is contained in:
Pierre
2026-02-08 21:53:42 +01:00
parent 4c3a0df4b4
commit 79d86b62d3
20 changed files with 2335 additions and 110 deletions
+49
View File
@@ -66,6 +66,9 @@ func (d *Dispatcher) dispatchToStream(event FileEvent) {
if event.OldPath != "" {
fields["oldPath"] = event.OldPath
}
if event.WatcherID != "" {
fields["watcherId"] = event.WatcherID
}
// Publish to stream
_, err := client.XAdd(EventsStream, StreamMaxLen, fields)
@@ -73,3 +76,49 @@ func (d *Dispatcher) dispatchToStream(event FileEvent) {
log.Printf("[Dispatcher] Failed to publish to stream: %v", err)
}
}
// EmitReset clears the stream and emits a reset event
// This is called when triggering a full rescan
func (d *Dispatcher) EmitReset(watcherID string) error {
d.mu.RLock()
client := d.storageClient
d.mu.RUnlock()
if client == nil || !client.IsConnected() {
return fmt.Errorf("storage client not connected")
}
// Delete the stream to clear all historical events
if err := client.ClearStream(EventsStream); err != nil {
log.Printf("[Dispatcher] Warning: failed to clear stream: %v", err)
// Continue anyway - we'll emit the reset event
}
log.Printf("[Dispatcher] Cleared stream %s for reset", EventsStream)
// Emit reset event as the first message in the fresh stream
event := FileEvent{
Type: EventTypeReset,
Path: "", // No specific path for reset
WatcherID: watcherID,
Timestamp: NowMS(),
}
// Dispatch synchronously for reset
d.dispatchToStream(event)
return nil
}
// GetStreamLength returns the current stream length
func (d *Dispatcher) GetStreamLength() (int64, error) {
d.mu.RLock()
client := d.storageClient
d.mu.RUnlock()
if client == nil || !client.IsConnected() {
return 0, fmt.Errorf("storage client not connected")
}
return client.GetStreamLength(EventsStream)
}
+146
View File
@@ -0,0 +1,146 @@
package watcher
import (
"sync"
)
// FileState tracks the known state of a file
type FileState struct {
Size int64
MtimeNano int64
MidHash string
}
// WatcherState tracks all files for a single watcher (in-memory)
type WatcherState struct {
mu sync.RWMutex
files map[string]*FileState // key: relative path
}
// NewWatcherState creates a new watcher state tracker
func NewWatcherState() *WatcherState {
return &WatcherState{
files: make(map[string]*FileState),
}
}
// Get returns the file state for a given path, or nil if not tracked
func (ws *WatcherState) Get(path string) *FileState {
ws.mu.RLock()
defer ws.mu.RUnlock()
return ws.files[path]
}
// Set updates or adds the file state for a given path
func (ws *WatcherState) Set(path string, state *FileState) {
ws.mu.Lock()
defer ws.mu.Unlock()
ws.files[path] = state
}
// Delete removes a file from the tracked state
func (ws *WatcherState) Delete(path string) {
ws.mu.Lock()
defer ws.mu.Unlock()
delete(ws.files, path)
}
// GetAll returns a snapshot of all tracked files
// The returned map is a copy - safe to modify
func (ws *WatcherState) GetAll() map[string]*FileState {
ws.mu.RLock()
defer ws.mu.RUnlock()
snapshot := make(map[string]*FileState, len(ws.files))
for path, state := range ws.files {
// Copy the state to avoid race conditions
stateCopy := *state
snapshot[path] = &stateCopy
}
return snapshot
}
// Clear removes all tracked files
func (ws *WatcherState) Clear() {
ws.mu.Lock()
defer ws.mu.Unlock()
ws.files = make(map[string]*FileState)
}
// Count returns the number of tracked files
func (ws *WatcherState) Count() int {
ws.mu.RLock()
defer ws.mu.RUnlock()
return len(ws.files)
}
// StateRegistry manages per-watcher state
type StateRegistry struct {
mu sync.RWMutex
states map[string]*WatcherState // key: watcher ID
}
// NewStateRegistry creates a new state registry
func NewStateRegistry() *StateRegistry {
return &StateRegistry{
states: make(map[string]*WatcherState),
}
}
// GetOrCreate returns the state for a watcher, creating one if needed
func (sr *StateRegistry) GetOrCreate(watcherID string) *WatcherState {
sr.mu.Lock()
defer sr.mu.Unlock()
state, exists := sr.states[watcherID]
if !exists {
state = NewWatcherState()
sr.states[watcherID] = state
}
return state
}
// Get returns the state for a watcher, or nil if not exists
func (sr *StateRegistry) Get(watcherID string) *WatcherState {
sr.mu.RLock()
defer sr.mu.RUnlock()
return sr.states[watcherID]
}
// Clear removes the state for a watcher
func (sr *StateRegistry) Clear(watcherID string) {
sr.mu.Lock()
defer sr.mu.Unlock()
if state, exists := sr.states[watcherID]; exists {
state.Clear()
}
}
// ClearAll removes all watcher states
func (sr *StateRegistry) ClearAll() {
sr.mu.Lock()
defer sr.mu.Unlock()
for _, state := range sr.states {
state.Clear()
}
}
// Delete removes a watcher's state completely
func (sr *StateRegistry) Delete(watcherID string) {
sr.mu.Lock()
defer sr.mu.Unlock()
delete(sr.states, watcherID)
}
// Count returns the number of files tracked for a watcher
func (sr *StateRegistry) Count(watcherID string) int {
sr.mu.RLock()
defer sr.mu.RUnlock()
if state, exists := sr.states[watcherID]; exists {
return state.Count()
}
return 0
}
+10
View File
@@ -10,6 +10,7 @@ const (
EventTypeChange FileEventType = "change"
EventTypeDelete FileEventType = "delete"
EventTypeRename FileEventType = "rename"
EventTypeReset FileEventType = "reset"
)
// FileEvent represents a file system event
@@ -20,6 +21,7 @@ type FileEvent struct {
Timestamp int64 `json:"timestamp"`
MidHash256 string `json:"midhash256,omitempty"` // midhash256 CID (SHA-256 of middle 1MB + file size)
OldPath string `json:"oldPath,omitempty"` // For rename events
WatcherID string `json:"watcherId,omitempty"` // For reset events
}
// PendingEvent tracks a file event that's being debounced
@@ -44,6 +46,14 @@ type ScanStatusResponse struct {
FileCount int `json:"fileCount,omitempty"`
}
// DiffScanResult holds the results of a differential scan
type DiffScanResult struct {
Added int // Number of new files added
Changed int // Number of files with changed size/mtime
Deleted int // Number of files that no longer exist
Total int // Total number of files currently tracked
}
// NowMS returns current time in milliseconds
func NowMS() int64 {
return time.Now().UnixMilli()
+143 -9
View File
@@ -13,10 +13,11 @@ import (
// Watcher provides file scanning and event dispatching
type Watcher struct {
config *config.Config
debouncer *Debouncer
dispatcher *Dispatcher
filesPath string
config *config.Config
debouncer *Debouncer
dispatcher *Dispatcher
filesPath string
stateRegistry *StateRegistry
mu sync.RWMutex
eventBuffer []FileEvent
@@ -27,11 +28,12 @@ func NewWatcher(cfg *config.Config, dispatcher *Dispatcher) (*Watcher, error) {
debouncer := NewDebouncer(0) // No debouncing for polling-based scanning
w := &Watcher{
config: cfg,
debouncer: debouncer,
dispatcher: dispatcher,
filesPath: cfg.FilesPath,
eventBuffer: make([]FileEvent, 0, 1000),
config: cfg,
debouncer: debouncer,
dispatcher: dispatcher,
filesPath: cfg.FilesPath,
stateRegistry: NewStateRegistry(),
eventBuffer: make([]FileEvent, 0, 1000),
}
return w, nil
@@ -110,6 +112,138 @@ func (w *Watcher) ScanDirectory(root string) int {
return count
}
// ScanDirectoryDiff scans a directory and only emits events for differences
// Returns the count of added, changed, and deleted files
func (w *Watcher) ScanDirectoryDiff(root string, watcherID string) DiffScanResult {
result := DiffScanResult{}
state := w.stateRegistry.GetOrCreate(watcherID)
// Get snapshot of known files
knownFiles := state.GetAll()
seenFiles := make(map[string]bool)
filepath.Walk(root, func(path string, info os.FileInfo, err error) error {
if err != nil {
return nil
}
// Skip directories
if info.IsDir() {
return nil
}
// Skip hidden files
if strings.HasPrefix(filepath.Base(path), ".") {
return nil
}
// Get relative path
relPath, err := filepath.Rel(w.filesPath, path)
if err != nil {
relPath = path
}
relPath = filepath.ToSlash(relPath)
seenFiles[relPath] = true
// Check current file state
currentMtime := info.ModTime().UnixNano()
currentSize := info.Size()
prevState := knownFiles[relPath]
if prevState == nil {
// New file - emit add event
event := FileEvent{
Type: EventTypeAdd,
Path: relPath,
Size: currentSize,
Timestamp: NowMS(),
}
// Compute midhash256
if hash, err := ComputeMidHash256(path); err == nil {
event.MidHash256 = hash
}
// Dispatch event
w.dispatcher.Dispatch(event)
// Update state
state.Set(relPath, &FileState{
Size: currentSize,
MtimeNano: currentMtime,
MidHash: event.MidHash256,
})
result.Added++
} else if prevState.Size != currentSize || prevState.MtimeNano != currentMtime {
// File changed - emit change event
event := FileEvent{
Type: EventTypeChange,
Path: relPath,
Size: currentSize,
Timestamp: NowMS(),
}
// Compute midhash256
if hash, err := ComputeMidHash256(path); err == nil {
event.MidHash256 = hash
}
// Dispatch event
w.dispatcher.Dispatch(event)
// Update state
state.Set(relPath, &FileState{
Size: currentSize,
MtimeNano: currentMtime,
MidHash: event.MidHash256,
})
result.Changed++
}
// If file exists and hasn't changed, do nothing
return nil
})
// Check for deleted files (in known but not seen)
for relPath := range knownFiles {
if !seenFiles[relPath] {
// File deleted - emit delete event
event := FileEvent{
Type: EventTypeDelete,
Path: relPath,
Timestamp: NowMS(),
}
w.dispatcher.Dispatch(event)
state.Delete(relPath)
result.Deleted++
}
}
result.Total = state.Count()
return result
}
// ClearState clears the tracked state for a watcher
func (w *Watcher) ClearState(watcherID string) {
w.stateRegistry.Clear(watcherID)
}
// ClearEventBuffer clears the in-memory event buffer
func (w *Watcher) ClearEventBuffer() {
w.mu.Lock()
defer w.mu.Unlock()
w.eventBuffer = make([]FileEvent, 0, 1000)
}
// GetStateRegistry returns the state registry for external access
func (w *Watcher) GetStateRegistry() *StateRegistry {
return w.stateRegistry
}
// ScanMountPath scans a specific mount path and returns file count
// This is used by the mount poller to scan individual mounts
func (w *Watcher) ScanMountPath(mountPath string) (int, error) {