Update app.py

app.py

@@ -35,6 +35,33 @@ def decode_image(image_base64: str):
     img = cv2.imdecode(np_arr, cv2.IMREAD_COLOR)
     return img
 
+def clean_ppstructure_result(result):
+    """
+    Recursively converts numpy arrays in the PPStructure result to standard lists.
+    Also optionally removes the 'img' field (raw numpy image crop) to keep standard JSON payloads light.
+    """
+    if isinstance(result, list):
+        return [clean_ppstructure_result(item) for item in result]
+    elif isinstance(result, dict):
+        cleaned_dict = {}
+        for key, value in result.items():
+            # PPStructure often includes the raw image crop as a numpy array in the 'img' key.
+            # Returning this in standard JSON is usually a bad idea (too heavy).
+            # Comment out the next two lines if you actually NEED raw pixel data in your JSON.
+            if key == 'img':
+                 continue
+            
+            cleaned_dict[key] = clean_ppstructure_result(value)
+        return cleaned_dict
+    elif isinstance(result, np.ndarray):
+        return result.tolist() # crucial: converts numpy array to standard list
+    elif isinstance(result, (np.float32, np.float64)):
+        return float(result) # crucial: converts numpy floats to standard python floats
+    elif isinstance(result, (np.int32, np.int64)):
+        return int(result)
+    else:
+        return result
+
 # --- API Endpoints ---
 
 @app.get("/")
@@ -74,12 +101,17 @@ def run_structure_analysis(ocr_input: OcrInput):
         
         print ("** Request for structure analysis received. Running structure_engine.**")
         # Use the structure_engine here
-        result = structure_engine.predict(img)
+         # Use the structure_engine here
+        raw_result = structure_engine.predict(img)
         
+        # --- CRITICAL STEP: Clean the result before returning ---
+        json_safe_result = clean_ppstructure_result(raw_result)
+        # --------------------------------------------------------
+
         print ("** Structure Analysis Complete. Returning the result.**")
-        print (result)
-        # Note: The result format is a list of dicts describing regions (Text, Table, Image, etc.)
-        return {"result": result}
+        # print(json_safe_result) # Helpful for debugging, but can be verbose in production logs
+
+        return {"result": json_safe_result}
         
     except Exception as e:
         return {"error": f"An error occurred in /structure: {str(e)}"}