Update vision_analyzer.py

vision_analyzer.py

@@ -23,28 +23,49 @@ import torch
 from transformers import AutoProcessor, AutoModelForCausalLM
 # from transformers import AutoProcessor, AutoModelForCausalLM
 
-import os
+# import os
 
-# Try to ensure unrar is available
-def setup_unrar():
-    # Try system installation (may not work on all Spaces)
-    if not os.path.exists("/usr/bin/unrar"):
-        os.system("apt-get update > /dev/null 2>&1 && apt-get install -y unrar > /dev/null 2>&1 || true")
+# # Try to ensure unrar is available
+# def setup_unrar():
+#     # Try system installation (may not work on all Spaces)
+#     if not os.path.exists("/usr/bin/unrar"):
+#         os.system("apt-get update > /dev/null 2>&1 && apt-get install -y unrar > /dev/null 2>&1 || true")
     
-    # Setup Python alternatives
-    try:
-        import rarfile
-        if os.path.exists("/usr/bin/unrar"):
-            rarfile.UNRAR_TOOL = "/usr/bin/unrar"
-    except ImportError:
-        os.system("pip install rarfile")
+#     # Setup Python alternatives
+#     try:
+#         import rarfile
+#         if os.path.exists("/usr/bin/unrar"):
+#             rarfile.UNRAR_TOOL = "/usr/bin/unrar"
+#     except ImportError:
+#         os.system("pip install rarfile")
     
-    try:
-        import patoolib
-    except ImportError:
-        os.system("pip install patool")
+#     try:
+#         import patoolib
+#     except ImportError:
+#         os.system("pip install patool")
+
+# setup_unrar()
 
-setup_unrar()
+
+
+# Initialize rarfile with correct paths
+import rarfile
+rarfile.UNRAR_TOOL = "unrar"  # Try system unrar if exists
+
+# Set alternative paths where unrar might be installed
+possible_unrar_paths = [
+    "/usr/bin/unrar",
+    "/usr/local/bin/unrar",
+    "/bin/unrar",
+    "/app/bin/unrar"
+]
+
+for path in possible_unrar_paths:
+    if os.path.exists(path):
+        rarfile.UNRAR_TOOL = path
+        break
+else:
+    log_message("⚠️ System unrar not found - using Python implementation")
 
 # Attempt to install flash-attn
 try:
@@ -151,6 +172,45 @@ def log_message(message: str):
     if len(processing_status["logs"]) > 100:
         processing_status["logs"] = processing_status["logs"][-100:]
 
+def extract_with_retry(rar_path: str, output_dir: str, max_retries: int = 2) -> bool:
+    """Extract RAR with retry and recovery, handling multi-part archives"""
+    filename = os.path.basename(rar_path)
+    
+    for attempt in range(max_retries):
+        try:
+            # Try with rarfile first
+            with rarfile.RarFile(rar_path) as rf:
+                rf.extractall(output_dir)
+            return True
+            
+        except rarfile.NeedFirstVolume:
+            # Handle multi-part RARs
+            base_name = filename.split(".part")[0] if ".part" in filename else filename.replace(".rar", "")
+            first_part = f"{base_name}.part01.rar"
+            first_part_path = os.path.join(os.path.dirname(rar_path), first_part)
+            
+            if not os.path.exists(first_part_path):
+                log_message(f"⚠️ Multi-part RAR detected but first part not found: {first_part}")
+                return False
+                
+            with rarfile.RarFile(first_part_path) as rf:
+                rf.extractall(output_dir)
+            return True
+            
+        except Exception as e:
+            if attempt == max_retries - 1:
+                # Final fallback to patool
+                try:
+                    import patoolib
+                    patoolib.extract_archive(rar_path, outdir=output_dir)
+                    return True
+                except Exception as patool_error:
+                    log_message(f"❌ All extraction methods failed: {str(patool_error)}")
+                    return False
+            time.sleep(1)
+    
+    return False
+
 def log_failed_file(filename: str, error: str):
     """Log failed files to persistent file"""
     with open(FAILED_FILES_LOG, "a") as f: