indexdatei eingebaut

editor.py

@@ -20,6 +20,7 @@ from watchdog.observers import Observer
 
 RECORDINGS_DIR = Path(os.environ.get("RECORDINGS_DIR", "/recordings"))
 FINISHED_DIR = Path(os.environ.get("FINISHED_DIR", "/finished"))
+INDEX_FILE = Path(os.environ.get("INDEX_FILE", str(FINISHED_DIR / ".event-audio-index.json")))
 WAVEFORMS_DIR = Path(os.environ.get("WAVEFORMS_DIR", "/waveforms"))
 WEB_PORT = int(os.environ.get("WEB_PORT", "8080"))
 
@@ -550,42 +551,227 @@ def sse_stream():
             SSE_CLIENTS.discard(client_queue)
 
 
-class RecordingFileEventHandler(FileSystemEventHandler):
-    """Turn recording/finished filesystem changes into browser events.
+# ---------------------------------------------------------------------------
+# Persistent file index
+# ---------------------------------------------------------------------------
 
-    Only meaningful MP3 file lifecycle events are forwarded. In particular,
-    continuous writes to an active recording are deliberately ignored so the
-    browser is not flooded while FFmpeg is recording.
+INDEX_LOCK = threading.RLock()
+FILE_INDEX = {"recordings": {}, "finished": {}}
+
+
+def _index_key(path):
+    return path.name
+
+
+def _index_section(kind):
+    return FILE_INDEX["recordings" if kind == "recording" else "finished"]
+
+
+def _entry_from_path(path, kind, analyze=False, existing=None):
+    """Create/update one index entry. FFprobe is only used when analyze=True."""
+    try:
+        stat = path.stat()
+    except OSError:
+        return None
+
+    entry = dict(existing or {})
+    entry.update({
+        "filename": path.name,
+        "size": stat.st_size,
+        "mtime": stat.st_mtime,
+        "kind": kind,
+    })
+
+    if analyze:
+        if kind == "finished":
+            tags = read_metadata(path)
+            entry.update({
+                "date": tags.get("date", ""),
+                "artist": tags.get("artist", ""),
+                "title": tags.get("title", ""),
+                "album": tags.get("album", ""),
+            })
+            entry["sort_ts"] = metadata_timestamp(path, tags)
+        else:
+            entry.update({
+                "date": "",
+                "artist": "",
+                "title": "",
+                "album": "",
+            })
+
+        entry["duration"] = get_duration(path)
+        entry["analyzed_mtime"] = stat.st_mtime
+        entry["analyzed_size"] = stat.st_size
+
+    return entry
+
+
+def _write_index_locked():
+    """Atomically persist the in-memory index."""
+    INDEX_FILE.parent.mkdir(parents=True, exist_ok=True)
+    temp = INDEX_FILE.with_name(f".{INDEX_FILE.name}.{uuid.uuid4().hex}.tmp")
+    temp.write_text(json.dumps(FILE_INDEX, ensure_ascii=False, indent=2), encoding="utf-8")
+    temp.replace(INDEX_FILE)
+
+
+def load_file_index():
+    """Load the index and reconcile it with the filesystem.
+
+    Reconciliation only stats existing files. FFprobe is run for a file when
+    there is no cached entry or its size/mtime changed since the last analysis.
     """
+    global FILE_INDEX
+    with INDEX_LOCK:
+        try:
+            data = json.loads(INDEX_FILE.read_text(encoding="utf-8"))
+            if not isinstance(data, dict):
+                raise ValueError("index root is not an object")
+            FILE_INDEX = {
+                "recordings": data.get("recordings", {}) if isinstance(data.get("recordings", {}), dict) else {},
+                "finished": data.get("finished", {}) if isinstance(data.get("finished", {}), dict) else {},
+            }
+        except (OSError, ValueError, json.JSONDecodeError):
+            FILE_INDEX = {"recordings": {}, "finished": {}}
+
+        changed = False
+        for kind, directory in (("recording", RECORDINGS_DIR), ("finished", FINISHED_DIR)):
+            section = _index_section(kind)
+            existing_names = set(section)
+            current_names = set()
+            for path in directory.glob("*.mp3"):
+                current_names.add(path.name)
+                try:
+                    stat = path.stat()
+                except OSError:
+                    continue
+                entry = section.get(path.name)
+                needs_analysis = (
+                    not isinstance(entry, dict)
+                    or entry.get("analyzed_mtime") != stat.st_mtime
+                    or entry.get("analyzed_size") != stat.st_size
+                    or "duration" not in entry
+                )
+                new_entry = _entry_from_path(path, kind, analyze=needs_analysis, existing=entry)
+                if new_entry is not None and new_entry != entry:
+                    section[path.name] = new_entry
+                    changed = True
+
+            for name in existing_names - current_names:
+                section.pop(name, None)
+                changed = True
+
+        if changed or not INDEX_FILE.exists():
+            _write_index_locked()
+
+        print(
+            f"[index] loaded {len(FILE_INDEX['recordings'])} recordings, "
+            f"{len(FILE_INDEX['finished'])} finished files",
+            flush=True,
+        )
+
+
+def index_file_created(path, kind):
+    """Add a newly created MP3 to the index. Finished files are already stable
+    when created atomically (uploads/finished outputs), while recordings may
+    still be written and are analyzed only on close_write.
+    """
+    with INDEX_LOCK:
+        entry = _entry_from_path(path, kind, analyze=(kind == "finished"))
+        if entry is None:
+            return
+        entry.setdefault("duration", None)
+        entry.setdefault("date", "")
+        entry.setdefault("artist", "")
+        entry.setdefault("title", "")
+        entry.setdefault("album", "")
+        _index_section(kind)[path.name] = entry
+        _write_index_locked()
+    broadcast_event("files_changed", {"action": "created", "filename": path.name})
+
+
+def index_file_closed(path, kind):
+    """Analyze a file once it is closed and therefore stable."""
+    with INDEX_LOCK:
+        if not path.is_file():
+            return
+        entry = _entry_from_path(path, kind, analyze=True, existing=_index_section(kind).get(path.name))
+        if entry is None:
+            return
+        _index_section(kind)[path.name] = entry
+        _write_index_locked()
+    broadcast_event("files_changed", {"action": "closed", "filename": path.name})
+
+
+def index_file_deleted(path, kind):
+    with INDEX_LOCK:
+        _index_section(kind).pop(path.name, None)
+        _write_index_locked()
+    broadcast_event("files_changed", {"action": "deleted", "filename": path.name})
+
+
+def index_file_moved(src_path, dest_path, kind):
+    with INDEX_LOCK:
+        section = _index_section(kind)
+        entry = section.pop(src_path.name, None)
+        if dest_path.is_file():
+            entry = _entry_from_path(dest_path, kind, analyze=entry is None, existing=entry)
+            if entry is not None:
+                section[dest_path.name] = entry
+        _write_index_locked()
+    broadcast_event("files_changed", {"action": "moved", "filename": dest_path.name})
+
+
+class RecordingFileEventHandler(FileSystemEventHandler):
+    """Keep the persistent file index in sync and notify connected browsers."""
+
+    def _kind_for_path(self, path):
+        try:
+            resolved = Path(path).resolve()
+            if resolved.parent == RECORDINGS_DIR.resolve():
+                return "recording"
+            if resolved.parent == FINISHED_DIR.resolve():
+                return "finished"
+        except OSError:
+            pass
+        return None
 
     def _is_relevant(self, path):
         path = Path(path)
-        if path.name.startswith("."):
-            return False
-        return path.suffix.lower() == ".mp3"
+        return not path.name.startswith(".") and path.suffix.lower() == ".mp3"
 
     def on_created(self, event):
         if event.is_directory or not self._is_relevant(event.src_path):
             return
-        broadcast_event("files_changed", {"action": "created", "filename": Path(event.src_path).name})
+        kind = self._kind_for_path(event.src_path)
+        if kind:
+            index_file_created(Path(event.src_path), kind)
 
     def on_deleted(self, event):
         if event.is_directory or not self._is_relevant(event.src_path):
             return
-        broadcast_event("files_changed", {"action": "deleted", "filename": Path(event.src_path).name})
+        kind = self._kind_for_path(event.src_path)
+        if kind:
+            index_file_deleted(Path(event.src_path), kind)
 
     def on_moved(self, event):
         if event.is_directory:
             return
-        if self._is_relevant(event.src_path) or self._is_relevant(event.dest_path):
-            broadcast_event("files_changed", {"action": "moved", "filename": Path(event.dest_path).name})
+        src = Path(event.src_path)
+        dest = Path(event.dest_path)
+        if not self._is_relevant(src) and not self._is_relevant(dest):
+            return
+        kind = self._kind_for_path(dest) or self._kind_for_path(src)
+        if kind:
+            index_file_moved(src, dest, kind)
 
     def on_closed(self, event):
-        # watchdog's Linux/inotify backend emits this when FFmpeg closes the
-        # recording. That is the useful "recording finished" signal.
+        # Linux/inotify emits close_write when FFmpeg has finished writing.
         if event.is_directory or not self._is_relevant(event.src_path):
             return
-        broadcast_event("files_changed", {"action": "closed", "filename": Path(event.src_path).name})
+        kind = self._kind_for_path(event.src_path)
+        if kind:
+            index_file_closed(Path(event.src_path), kind)
 
 
 def start_file_observer():
@@ -594,6 +780,8 @@ def start_file_observer():
     RECORDINGS_DIR.mkdir(parents=True, exist_ok=True)
     FINISHED_DIR.mkdir(parents=True, exist_ok=True)
 
+    load_file_index()
+
     observer = Observer()
     handler = RecordingFileEventHandler()
     observer.schedule(handler, str(RECORDINGS_DIR), recursive=False)
@@ -1442,41 +1630,46 @@ async function deleteFile(kind, filename) {
 
 @app.route("/")
 def index():
-    RECORDINGS_DIR.mkdir(parents=True, exist_ok=True)
-    FINISHED_DIR.mkdir(parents=True, exist_ok=True)
+    with INDEX_LOCK:
+        recording_entries = [dict(entry) for entry in FILE_INDEX["recordings"].values()]
+        finished_entries = [dict(entry) for entry in FILE_INDEX["finished"].values()]
 
     recordings = []
-    for path in RECORDINGS_DIR.glob("*.mp3"):
-        try:
-            stat = path.stat()
-            recordings.append({
-                "filename": path.name,
-                "datetime": format_datetime(stat.st_mtime),
-                "size": format_size(stat.st_size),
-                "duration": format_duration(get_duration(path)),
-                "mtime": stat.st_mtime,
-            })
-        except FileNotFoundError:
-            pass
+    for item in recording_entries:
+        recordings.append({
+            "filename": item["filename"],
+            "datetime": format_datetime(item.get("mtime", time.time())),
+            "size": format_size(item.get("size", 0)),
+            "duration": format_duration(item.get("duration")),
+            "mtime": item.get("mtime", 0),
+        })
 
     finished = []
-    for path in FINISHED_DIR.glob("*.mp3"):
-        try:
-            stat = path.stat()
-            tags = read_metadata(path)
-            finished.append({
-                "filename": path.name,
-                "datetime": metadata_datetime(path),
-                "artist": tags.get("artist", ""),
-                "title": tags.get("title", ""),
-                "album": tags.get("album", ""),
-                "size": format_size(stat.st_size),
-                "duration": format_duration(get_duration(path)),
-                "sort_date": tags.get("date", "") or datetime.fromtimestamp(stat.st_mtime, ZoneInfo("Europe/Berlin")).isoformat(),
-                "sort_ts": metadata_timestamp(path, tags),
-            })
-        except FileNotFoundError:
-            pass
+    for item in finished_entries:
+        date = item.get("date", "")
+        sort_ts = item.get("sort_ts")
+        if sort_ts is None:
+            sort_ts = item.get("mtime", 0)
+        dt_text = format_datetime(sort_ts)
+        if date:
+            try:
+                dt = datetime.fromisoformat(date.replace("Z", "+00:00"))
+                if dt.tzinfo is None:
+                    dt = dt.replace(tzinfo=ZoneInfo("Europe/Berlin"))
+                dt_text = dt.astimezone(ZoneInfo("Europe/Berlin")).strftime("%d.%m.%Y %H:%M:%S")
+            except ValueError:
+                pass
+        finished.append({
+            "filename": item["filename"],
+            "datetime": dt_text,
+            "artist": item.get("artist", ""),
+            "title": item.get("title", ""),
+            "album": item.get("album", ""),
+            "size": format_size(item.get("size", 0)),
+            "duration": format_duration(item.get("duration")),
+            "sort_date": date or datetime.fromtimestamp(item.get("mtime", 0), ZoneInfo("Europe/Berlin")).isoformat(),
+            "sort_ts": sort_ts,
+        })
 
     recordings.sort(key=lambda x: x["mtime"], reverse=True)
     finished.sort(key=lambda x: x["sort_date"], reverse=True)