Redesign dataset workbench management

backend/app/modules/dataset/service.py

@@ -16,6 +16,7 @@ from ...config import settings
 DATASET_KINDS = ("images", "labels", "masks")
 IMAGE_EXTS = {".png", ".jpg", ".jpeg", ".bmp", ".tif", ".tiff"}
 LABEL_EXTS = {".txt", ".json", ".yaml", ".yml"}
+VIDEO_EXTS = {".mp4", ".avi", ".mov", ".mkv", ".wmv", ".flv", ".webm", ".m4v"}
 ARCHIVE_SUFFIXES = (".zip", ".tar", ".tar.gz", ".tgz")
 ARCHIVE_ALIASES = {
     "images": {"image", "images", "img", "imgs", "ori", "original", "originals"},
@@ -47,7 +48,7 @@ def safe_filename(value: str | None) -> str:
 
 def upload_exts_for_kind(kind: str) -> set[str]:
     if kind == "images":
-        return IMAGE_EXTS
+        return IMAGE_EXTS | VIDEO_EXTS
     if kind == "masks":
         return IMAGE_EXTS | LABEL_EXTS
     if kind == "labels":
@@ -138,6 +139,8 @@ def metadata_path(name: str) -> Path:
 def create_dataset(name: str, description: str = "") -> dict:
     safe_name = slugify(name)
     root = dataset_dir(safe_name)
+    if root.exists():
+        raise ValueError(f"dataset already exists: {safe_name}")
     for kind in DATASET_KINDS:
         (root / kind).mkdir(parents=True, exist_ok=True)
     meta = {
@@ -155,6 +158,44 @@ def create_dataset(name: str, description: str = "") -> dict:
     return describe_dataset(safe_name)
 
 
+def delete_dataset(name: str) -> dict:
+    safe_name = slugify(name)
+    root = dataset_dir(safe_name)
+    if not root.exists() or not root.is_dir():
+        raise ValueError(f"dataset not found: {safe_name}")
+    shutil.rmtree(root)
+    return {"deleted": True, "name": safe_name}
+
+
+def copy_dataset(source_name: str, target_name: str, description: str | None = None) -> dict:
+    source_safe = slugify(source_name)
+    target_safe = slugify(target_name)
+    source = dataset_dir(source_safe)
+    target = dataset_dir(target_safe)
+    if not source.exists() or not source.is_dir():
+        raise ValueError(f"dataset not found: {source_safe}")
+    if target.exists():
+        raise ValueError(f"dataset already exists: {target_safe}")
+    shutil.copytree(source, target)
+    meta = _load_meta(source_safe)
+    meta.update(
+        {
+            "name": target_safe,
+            "description": description if description is not None else f"Copy of {source_safe}",
+            "created_at": datetime.now(timezone.utc).isoformat(),
+            "copied_from": source_safe,
+            "root": str(target.relative_to(settings.project_root)),
+            "layout": {
+                "images": str((target / "images").relative_to(settings.project_root)),
+                "labels": str((target / "labels").relative_to(settings.project_root)),
+                "masks": str((target / "masks").relative_to(settings.project_root)),
+            },
+        }
+    )
+    metadata_path(target_safe).write_text(json.dumps(meta, ensure_ascii=False, indent=2), encoding="utf-8")
+    return describe_dataset(target_safe)
+
+
 def _load_meta(name: str) -> dict:
     path = metadata_path(name)
     if path.exists():
@@ -192,6 +233,22 @@ def _image_shape(path: Path) -> dict | None:
         return None
 
 
+def _file_info(path: Path, kind: str) -> dict:
+    suffix = path.suffix.lower()
+    stem = path.stem
+    return {
+        "name": path.name,
+        "stem": stem,
+        "pair_stem": normalize_mask_stem(stem) if kind == "masks" else stem,
+        "path": str(path.resolve()),
+        "relative_path": str(path.resolve().relative_to(settings.project_root)),
+        "size": path.stat().st_size,
+        "ext": suffix,
+        "previewable": suffix in IMAGE_EXTS,
+        "media_type": "image" if suffix in IMAGE_EXTS else "video" if suffix in VIDEO_EXTS else "file",
+    }
+
+
 def _validate_yolo_txt(path: Path) -> dict:
     errors = []
     classes: set[int] = set()
@@ -232,20 +289,14 @@ def describe_dataset(name: str) -> dict:
     absolute_layout = {kind: str((root / kind).resolve()) for kind in DATASET_KINDS}
     counts = {}
     samples = {}
+    files_by_kind = {}
     for kind in sorted(DATASET_KINDS):
         files = list(_iter_files(root / kind))
         counts[kind] = len(files)
-        samples[kind] = [
-            {
-                "name": path.name,
-                "path": str(path.resolve()),
-                "relative_path": str(path.resolve().relative_to(settings.project_root)),
-                "size": path.stat().st_size,
-                "previewable": path.suffix.lower() in IMAGE_EXTS,
-            }
-            for path in files[:80]
-        ]
-    return {**meta, "absolute_layout": absolute_layout, "counts": counts, "samples": samples}
+        file_items = [_file_info(path, kind) for path in files[:2000]]
+        files_by_kind[kind] = file_items
+        samples[kind] = file_items[:80]
+    return {**meta, "absolute_layout": absolute_layout, "counts": counts, "samples": samples, "files": files_by_kind}
 
 
 def validate_dataset(name: str) -> dict: