Update backend/app/main.py

backend/app/main.py

@@ -451,9 +451,13 @@ async def upload_asset(
         file_suffix = Path(safe_filename).suffix
         unique_filename = f"{file_stem}_{datetime.utcnow().strftime('%Y%m%d_%H%M%S')}_{uuid.uuid4().hex[:8]}{file_suffix}"
         file_path = upload_dir / unique_filename
+        # Resolve to absolute path for reliable access
+        file_path = file_path.resolve()
         with open(file_path, "wb") as buffer:
             buffer.write(content)
         
+        print(f"✓ File saved to: {file_path} (absolute path)")
+        
         # Save to database (keep dummy content as requested)
         try:
             from app.models import Asset
@@ -462,7 +466,7 @@ async def upload_asset(
             
             db_asset = Asset(
                 name=file.filename,  # Keep original filename for display
-                file_path=str(file_path),  # Store unique filename on disk
+                file_path=str(file_path),  # Store absolute path for reliable access
                 file_type=file_type,
                 product_category=product_category or "ocr",
                 sub_category=sub_category if sub_category and sub_category != "none" else None,
@@ -877,9 +881,41 @@ async def get_pdf_pages(asset_id, db: Session = Depends(get_db)):
                         
                         if row:
                             print(f"✓ Found asset via direct connection (exception handler): {row[1]} (id={row[0]})")
-                            file_path = Path(row[2])
-                            if not file_path.exists():
-                                raise HTTPException(status_code=404, detail="File not found on disk")
+                            file_path_str = row[2]
+                            print(f"Debug: file_path from DB: {file_path_str}")
+                            # Resolve path - handle both relative and absolute paths
+                            file_path = Path(file_path_str)
+                            
+                            # Try multiple path resolution strategies
+                            if file_path.is_absolute():
+                                # Already absolute, use as-is
+                                if not file_path.exists():
+                                    raise HTTPException(status_code=404, detail=f"File not found on disk: {file_path}")
+                            else:
+                                # Relative path - try multiple locations
+                                possible_paths = [
+                                    Path.cwd() / file_path_str,  # From current working directory
+                                    Path("/app") / file_path_str,  # From /app (Docker)
+                                    Path("/app/uploads") / Path(file_path_str).name,  # Just filename in /app/uploads
+                                    Path(file_path_str).resolve(),  # Resolve relative to current dir
+                                ]
+                                
+                                file_path = None
+                                for possible_path in possible_paths:
+                                    if possible_path.exists():
+                                        file_path = possible_path
+                                        print(f"✓ Found file at: {file_path}")
+                                        break
+                                
+                                if not file_path:
+                                    # List what's actually in /app/uploads for debugging
+                                    uploads_dir = Path("/app/uploads")
+                                    if uploads_dir.exists():
+                                        files_in_uploads = list(uploads_dir.glob("*.pdf"))
+                                        print(f"Debug: PDF files in /app/uploads: {[str(f) for f in files_in_uploads[:5]]}")
+                                    raise HTTPException(status_code=404, detail=f"File not found. Tried: {[str(p) for p in possible_paths]}")
+                            
+                            print(f"Debug: Using file_path: {file_path} (exists: {file_path.exists()})")
                             
                             if row[3] != "document" or not str(file_path).lower().endswith('.pdf'):
                                 raise HTTPException(status_code=400, detail="File is not a PDF")
@@ -934,9 +970,41 @@ async def get_pdf_pages(asset_id, db: Session = Depends(get_db)):
         
         # If we got asset from ORM, use it
         if db_asset:
-            file_path = Path(db_asset.file_path)
-            if not file_path.exists():
-                raise HTTPException(status_code=404, detail="File not found on disk")
+            file_path_str = db_asset.file_path
+            print(f"Debug: file_path from ORM: {file_path_str}")
+            # Resolve path - handle both relative and absolute paths
+            file_path = Path(file_path_str)
+            
+            # Try multiple path resolution strategies
+            if file_path.is_absolute():
+                # Already absolute, use as-is
+                if not file_path.exists():
+                    raise HTTPException(status_code=404, detail=f"File not found on disk: {file_path}")
+            else:
+                # Relative path - try multiple locations
+                possible_paths = [
+                    Path.cwd() / file_path_str,  # From current working directory
+                    Path("/app") / file_path_str,  # From /app (Docker)
+                    Path("/app/uploads") / Path(file_path_str).name,  # Just filename in /app/uploads
+                    Path(file_path_str).resolve(),  # Resolve relative to current dir
+                ]
+                
+                file_path = None
+                for possible_path in possible_paths:
+                    if possible_path.exists():
+                        file_path = possible_path
+                        print(f"✓ Found file at: {file_path}")
+                        break
+                
+                if not file_path:
+                    # List what's actually in /app/uploads for debugging
+                    uploads_dir = Path("/app/uploads")
+                    if uploads_dir.exists():
+                        files_in_uploads = list(uploads_dir.glob("*.pdf"))
+                        print(f"Debug: PDF files in /app/uploads: {[str(f) for f in files_in_uploads[:5]]}")
+                    raise HTTPException(status_code=404, detail=f"File not found. Tried: {[str(p) for p in possible_paths]}")
+            
+            print(f"Debug: Using file_path: {file_path} (exists: {file_path.exists()})")
             
             if db_asset.file_type != "document" or not str(file_path).lower().endswith('.pdf'):
                 raise HTTPException(status_code=400, detail="File is not a PDF")
@@ -1017,9 +1085,41 @@ async def get_pdf_pages(asset_id, db: Session = Depends(get_db)):
             
             if row:
                 print(f"✓ Found asset via direct connection: {row[1]} (id={row[0]})")
-                file_path = Path(row[2])
-                if not file_path.exists():
-                    raise HTTPException(status_code=404, detail="File not found on disk")
+                file_path_str = row[2]
+                print(f"Debug: file_path from DB (fallback): {file_path_str}")
+                # Resolve path - handle both relative and absolute paths
+                file_path = Path(file_path_str)
+                
+                # Try multiple path resolution strategies
+                if file_path.is_absolute():
+                    # Already absolute, use as-is
+                    if not file_path.exists():
+                        raise HTTPException(status_code=404, detail=f"File not found on disk: {file_path}")
+                else:
+                    # Relative path - try multiple locations
+                    possible_paths = [
+                        Path.cwd() / file_path_str,  # From current working directory
+                        Path("/app") / file_path_str,  # From /app (Docker)
+                        Path("/app/uploads") / Path(file_path_str).name,  # Just filename in /app/uploads
+                        Path(file_path_str).resolve(),  # Resolve relative to current dir
+                    ]
+                    
+                    file_path = None
+                    for possible_path in possible_paths:
+                        if possible_path.exists():
+                            file_path = possible_path
+                            print(f"✓ Found file at: {file_path}")
+                            break
+                    
+                    if not file_path:
+                        # List what's actually in /app/uploads for debugging
+                        uploads_dir = Path("/app/uploads")
+                        if uploads_dir.exists():
+                            files_in_uploads = list(uploads_dir.glob("*.pdf"))
+                            print(f"Debug: PDF files in /app/uploads: {[str(f) for f in files_in_uploads[:5]]}")
+                        raise HTTPException(status_code=404, detail=f"File not found. Tried: {[str(p) for p in possible_paths]}")
+                
+                print(f"Debug: Using file_path (fallback): {file_path} (exists: {file_path.exists()})")
                 
                 if row[3] != "document" or not str(file_path).lower().endswith('.pdf'):
                     raise HTTPException(status_code=400, detail="File is not a PDF")
@@ -1307,4 +1407,4 @@ if FRONTEND_DIST.exists():
         # Serve index.html for all other routes (SPA routing)
         if INDEX_FILE.exists():
             return FileResponse(str(INDEX_FILE))
-        return {"detail": "Frontend not found"}
+        return {"detail": "Frontend not found"}