Update app.py

app.py

@@ -25,8 +25,8 @@ from pathlib import Path
 
 from huggingface_hub import hf_hub_download
 
-# Add parent directory to path to allow importing from utils
-sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+# Add parent directory to path to allow importing from utils - updated for new structure
+sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
 
 # Import utilities
 from utils.image_processing import process_image, batch_process_images
@@ -34,6 +34,11 @@ from utils.file_utils import save_tags_to_file, get_default_save_locations
 from utils.ui_components import display_progress_bar, show_example_images, display_batch_results
 from utils.onnx_processing import batch_process_images_onnx
 
+# Add environment variables for HF Spaces permissions
+os.environ['MPLCONFIGDIR'] = '/tmp/matplotlib'
+os.environ['HF_HOME'] = '/tmp/huggingface'
+os.environ['TRANSFORMERS_CACHE'] = '/tmp/transformers'
+
 # Constants - matching your v1 pattern
 MODEL_REPO = "Camais03/camie-tagger-v2"
 ONNX_MODEL_FILE = "camie-tagger-v2.onnx"
@@ -42,25 +47,65 @@ METADATA_FILE = "camie-tagger-v2-metadata.json"
 VALIDATION_FILE = "full_validation_results.json"
 
 def get_model_files():
-    """Download model files from HF Hub and return paths - using /tmp for HF Spaces"""
+    """Download model files from HF Hub and return paths - optimized for HF Spaces"""
     try:
-        # Use /tmp directory which is writable in HF Spaces
-        cache_dir = "/tmp/hf_models"
+        # Use smaller /tmp directory and be more careful with large files
+        cache_dir = "/tmp/hf_cache"
         os.makedirs(cache_dir, exist_ok=True)
         
-        # Download files to /tmp directory
-        onnx_path = hf_hub_download(repo_id=MODEL_REPO, filename=ONNX_MODEL_FILE, cache_dir=cache_dir)
-        metadata_path = hf_hub_download(repo_id=MODEL_REPO, filename=METADATA_FILE, cache_dir=cache_dir)
+        # Download metadata first (small file)
+        metadata_path = hf_hub_download(
+            repo_id=MODEL_REPO, 
+            filename=METADATA_FILE, 
+            cache_dir=cache_dir,
+            resume_download=True  # Allow resuming if interrupted
+        )
+        
+        # Try streaming download for large ONNX file
+        try:
+            onnx_path = hf_hub_download(
+                repo_id=MODEL_REPO, 
+                filename=ONNX_MODEL_FILE, 
+                cache_dir=cache_dir,
+                resume_download=True,
+                force_download=False  # Use cached version if available
+            )
+        except Exception as e:
+            print(f"ONNX download failed: {e}")
+            # Fallback: try direct URL download with requests
+            import requests
+            onnx_url = f"https://huggingface.co/{MODEL_REPO}/resolve/main/{ONNX_MODEL_FILE}"
+            onnx_path = os.path.join(cache_dir, ONNX_MODEL_FILE)
+            
+            print(f"Trying direct download from: {onnx_url}")
+            response = requests.get(onnx_url, stream=True)
+            response.raise_for_status()
+            
+            with open(onnx_path, 'wb') as f:
+                for chunk in response.iter_content(chunk_size=8192):
+                    if chunk:
+                        f.write(chunk)
+            print(f"Direct download successful: {onnx_path}")
         
-        # Try to download optional files
+        # Try optional files
         try:
-            safetensors_path = hf_hub_download(repo_id=MODEL_REPO, filename=SAFETENSORS_MODEL_FILE, cache_dir=cache_dir)
+            safetensors_path = hf_hub_download(
+                repo_id=MODEL_REPO, 
+                filename=SAFETENSORS_MODEL_FILE, 
+                cache_dir=cache_dir,
+                resume_download=True
+            )
         except Exception as e:
             print(f"SafeTensors model not available: {e}")
             safetensors_path = None
             
         try:
-            validation_path = hf_hub_download(repo_id=MODEL_REPO, filename=VALIDATION_FILE, cache_dir=cache_dir)
+            validation_path = hf_hub_download(
+                repo_id=MODEL_REPO, 
+                filename=VALIDATION_FILE, 
+                cache_dir=cache_dir,
+                resume_download=True
+            )
         except Exception as e:
             print(f"Validation results not available: {e}")
             validation_path = None