Fix dataset sidecar write race condition, duplicate app init, and missing status/priority persistence

python/src/notes_app/api/app.py

@@ -47,6 +47,3 @@ def health() -> dict[str, str]:
 
 
 app = create_app()
-
-
-app = create_app()

python/src/notes_app/models/dataset_metadata.py

@@ -20,6 +20,8 @@ class DatasetMetadata:
     created: datetime
     modified: datetime
     tags: tuple[str, ...] = ()
+    status: str = ""
+    priority: int = 0
     format: str = ""
     path: str = ""
     size_bytes: int = 0
@@ -37,6 +39,8 @@ def from_dataset(cls, dataset: Dataset) -> "DatasetMetadata":
             created=dataset.created,
             modified=dataset.modified,
             tags=dataset.tags,
+            status=dataset.status,
+            priority=dataset.priority,
             format=dataset.format,
             path=cls._normalize_relative_path(dataset.path),
             size_bytes=dataset.size_bytes,
@@ -54,6 +58,8 @@ def to_dataset(self) -> Dataset:
             created=self.created,
             modified=self.modified,
             tags=self.tags,
+            status=self.status,
+            priority=self.priority,
             format=self.format,
             path=self._normalize_relative_path(self.path),
             size_bytes=self.size_bytes,
@@ -72,6 +78,8 @@ def to_dict(self) -> dict[str, object]:
             "created": self._to_iso(self.created),
             "modified": self._to_iso(self.modified),
             "tags": list(self.tags),
+            "status": self.status,
+            "priority": self.priority,
             "format": self.format,
             "path": self._normalize_relative_path(self.path),
             "sizeBytes": self.size_bytes,
@@ -105,6 +113,8 @@ def from_dict(cls, data: dict[str, object]) -> "DatasetMetadata":
             created=cls._from_iso(data.get("created")),
             modified=cls._from_iso(data.get("modified"), fallback=cls._from_iso(data.get("created"))),
             tags=cls._normalize_tags(data.get("tags", [])),
+            status=str(data.get("status") or ""),
+            priority=cls._to_int(data.get("priority"), default=0),
             format=str(data.get("format") or ""),
             path=cls._normalize_relative_path(str(data.get("path") or "")),
             size_bytes=cls._to_int(data.get("sizeBytes"), default=0),

python/src/notes_app/repositories/file_dataset_repository.py

@@ -4,6 +4,8 @@
 import io
 import json
 import logging
+import os
+import tempfile
 from dataclasses import replace
 from datetime import datetime
 from pathlib import Path
@@ -44,7 +46,7 @@ def save(self, item: Dataset) -> None:
         """Write (or overwrite) the sidecar for *item*."""
         normalized = replace(item, path=DatasetMetadata._normalize_relative_path(item.path))
         sidecar_path = self._sidecar_path_for(normalized)
-        sidecar_path.write_text(render_sidecar(normalized), encoding="utf-8")
+        self._write_text_atomic(sidecar_path, render_sidecar(normalized))
 
     def list_all(self) -> Iterable[Dataset]:
         results: list[Dataset] = []
@@ -130,7 +132,7 @@ def save_with_file(
 
         # Write sidecar next to the data file
         sidecar_path = self._datasets_dir / f"{relative_name}.dataset.yml"
-        sidecar_path.write_text(render_sidecar(updated), encoding="utf-8")
+        self._write_text_atomic(sidecar_path, render_sidecar(updated))
 
         return updated
 
@@ -164,7 +166,7 @@ def save_profile(self, dataset: Dataset, profile: dict[str, object]) -> None:
             raw_data = yaml.safe_load(sidecar_path.read_text(encoding="utf-8")) or {}
             data = raw_data if isinstance(raw_data, dict) else {}
             data["profile"] = profile
-            sidecar_path.write_text(yaml.safe_dump(data, sort_keys=False), encoding="utf-8")
+            self._write_text_atomic(sidecar_path, yaml.safe_dump(data, sort_keys=False))
         except (OSError, UnicodeDecodeError, yaml.YAMLError) as exc:
             _LOG.warning("Unable to save profile for sidecar %s: %s", sidecar_path.name, exc)
 
@@ -193,6 +195,27 @@ def preview(self, dataset: Dataset, limit: int) -> dict[str, object]:
     # Internal helpers
     # ------------------------------------------------------------------
 
+    @staticmethod
+    def _write_text_atomic(path: Path, text: str, encoding: str = "utf-8") -> None:
+        """Write *text* to *path* atomically using a temp file and ``os.replace``.
+
+        Ensures that readers never observe a partially-written sidecar file,
+        which would otherwise cause spurious 404 errors when a background
+        profiling thread rewrites the sidecar concurrently with a read request.
+        """
+        dir_path = path.parent
+        fd, tmp_name = tempfile.mkstemp(dir=dir_path, suffix=".tmp")
+        try:
+            with os.fdopen(fd, "w", encoding=encoding) as fh:
+                fh.write(text)
+            os.replace(tmp_name, path)
+        except Exception:
+            try:
+                os.unlink(tmp_name)
+            except OSError:
+                pass
+            raise
+
     def _sidecar_path_for(self, dataset: Dataset) -> Path:
         """Return the sidecar path that corresponds to *dataset*."""
         if dataset.path: