Update pmo_func.py

pmo_func.py

@@ -224,7 +224,7 @@ class FactChecker:
         
         print("No FactCheck result, falling back to live web analysis...")
         return self.search_and_analyze_claim(claim)
-
+        
 class img_manipulation:
     def __init__(self):
         self.GEN_AI_IMAGE = pipeline("image-classification", model="umm-maybe/AI-image-detector", device=DEVICE)
@@ -240,37 +240,40 @@ class img_manipulation:
             print(f'AI image detection error: {e}')
             return 0.0
 
-    def generated_image(self, img_pth, quality=90, scale=15):
+    def generated_image(self, img_pth, quality=90):
+        """
+        Calculates the ELA score entirely in memory without saving any files.
+        """
         try:
             with Image.open(img_pth) as orig_img:
                 orig_img = orig_img.convert('RGB')
-                temp_path = 'temp_resaved.jpg'
-                orig_img.save(temp_path, 'JPEG', quality=quality)
-                with Image.open(temp_path) as resaved_img:
+
+                # Create an in-memory buffer to hold the re-saved image
+                buffer = io.BytesIO()
+                orig_img.save(buffer, 'JPEG', quality=quality)
+                buffer.seek(0) # Rewind buffer to the beginning
+
+                with Image.open(buffer) as resaved_img:
+                    # Calculate the difference between the original and re-saved image
                     ela_image = ImageChops.difference(orig_img, resaved_img)
-            os.remove(temp_path)
+
             ela_data = np.array(ela_image)
             mean_intensity = ela_data.mean()
             scaled_score = min(100, (mean_intensity / 25.0) * 100)
+            return scaled_score
             
-            # Save the ELA image and return its path for serving
-            ela_path = "ela_result.png"
-            enhancer = ImageEnhance.Brightness(ela_image)
-            max_diff = max(1, max([ex[1] for ex in ela_image.getextrema()]))
-            ela_image_enhanced = enhancer.enhance(scale / max_diff)
-            ela_image_enhanced.save(ela_path)
-            return scaled_score, ela_path
         except Exception as e:
-            print(f'ELA generation error: {e}')
-            return 0.0, None
+            print(f'ELA calculation error: {e}')
+            return 0.0
 
     def run_image_forensics(self, image_path):
         ai_score = self.Gen_AI_IMG(image_path)
-        classic_score, ela_path = self.generated_image(image_path)
+        classic_score = self.generated_image(image_path)
+        
+        # The return dictionary no longer includes 'ela_image_path'
         return {
             "ai_generated_score_percent": ai_score,
             "classic_edit_score_percent": classic_score,
-            "ela_image_path": ela_path
         }
 
 class OCR: