Sync from GitHub

app.py

@@ -34,43 +34,55 @@ HF_RESULTS = "amaksay/inkslop-results"
 def restructure_dataset(task_dir: Path, task_id: str) -> None:
     """Restructure downloaded dataset so samples/ is at the root.
 
-    HF datasets have structure like:
-      task_dir/original/samples/... or task_dir/source_data/samples/...
+    HF datasets may have multiple sample directories:
+      task_dir/original/samples/...
+      task_dir/source_data/samples/...
 
-    We need:
+    We need to merge all into:
       task_dir/samples/...
 
     This ensures the dataset_name matches the task_id in results.
     """
-    # Find the directory containing 'samples/'
-    samples_dir = None
-    for subdir in task_dir.iterdir():
-        if subdir.is_dir() and (subdir / "samples").exists():
-            samples_dir = subdir / "samples"
-            break
-
-    if not samples_dir:
-        logger.warning(f"No samples/ found in {task_dir}, skipping restructure")
-        return
-
     # Check if already restructured
-    if (task_dir / "samples").exists():
+    target = task_dir / "samples"
+    if target.exists():
         logger.info(f"{task_id} already restructured")
         return
 
-    logger.info(f"Restructuring {task_id}: moving {samples_dir.parent.name}/samples to root")
+    # Find ALL directories containing 'samples/'
+    samples_dirs = []
+    for subdir in task_dir.iterdir():
+        if subdir.is_dir() and (subdir / "samples").exists():
+            samples_dirs.append(subdir / "samples")
 
-    # Move samples to task root
-    target = task_dir / "samples"
-    shutil.move(str(samples_dir), str(target))
+    if not samples_dirs:
+        logger.warning(f"No samples/ found in {task_dir}, skipping restructure")
+        return
 
-    # Clean up the now-empty parent directory
-    parent = samples_dir.parent
-    if parent.exists() and parent != task_dir:
-        try:
-            shutil.rmtree(parent)
-        except Exception:
-            pass  # Ignore cleanup errors
+    logger.info(f"Restructuring {task_id}: found {len(samples_dirs)} sample directories")
+
+    # Create target samples directory
+    target.mkdir(exist_ok=True)
+
+    # Move samples from each source directory
+    for samples_dir in samples_dirs:
+        logger.info(f"  Moving samples from {samples_dir.parent.name}/samples/")
+        for sample in samples_dir.iterdir():
+            if sample.is_dir():
+                dest = target / sample.name
+                if not dest.exists():
+                    shutil.move(str(sample), str(dest))
+                else:
+                    logger.warning(f"  Skipping duplicate sample: {sample.name}")
+
+    # Clean up the now-empty parent directories
+    for samples_dir in samples_dirs:
+        parent = samples_dir.parent
+        if parent.exists() and parent != task_dir:
+            try:
+                shutil.rmtree(parent)
+            except Exception:
+                pass  # Ignore cleanup errors
 
 
 def download_hf_datasets() -> Path:

requirements.txt

@@ -1,4 +1,4 @@
-gradio>=6.0.0
+gradio==6.3.0
 huggingface_hub>=0.20.0
 Pillow>=9.0.0
 numpy>=1.20.0