fix: store absolute paths in dataset.json

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

main.py

@@ -120,10 +120,10 @@ def upsert_clip_annotation(folder: str, clip_path: str, label: str) -> None:
                 entries = json.load(f)
             except (json.JSONDecodeError, ValueError):
                 entries = []
-    rel_path = os.path.relpath(clip_path, folder)
+    abs_path = os.path.abspath(clip_path)
-    entry: dict = {"path": rel_path, "label": label}
+    entry: dict = {"path": abs_path, "label": label}
     for i, e in enumerate(entries):
-        if e.get("path") == rel_path:
+        if e.get("path") == abs_path:
             entries[i] = entry
             break
     else:

tests/test_utils.py

@@ -245,7 +245,7 @@ def test_upsert_creates_file():
             entries = json.load(f)
         assert len(entries) == 1
         assert entries[0]["label"] == "dog barking"
-        assert entries[0]["path"] == "clip_001.mp4"
+        assert entries[0]["path"] == clip
 
 def test_upsert_appends_new_clips():
     with tempfile.TemporaryDirectory() as d:

tools/migrate_dataset_json.py

@@ -30,7 +30,7 @@ def load_db_records(db_path: Path) -> list[dict]:
 
 def build_entries_for_folder(folder: str, records: list[dict]) -> list[dict]:
     return [
-        {"path": os.path.relpath(rec["output_path"], folder), "label": rec["label"]}
+        {"path": os.path.abspath(rec["output_path"]), "label": rec["label"]}
         for rec in records
     ]