Upload app.py

app.py

@@ -88,43 +88,45 @@ def read_document(file):
     else:
         return process_txt(file)
 
-def save_redacted_file(original_file, redacted_text):
-    """Saves redacted text back in the original format and returns a valid file path."""
-    
+def save_redacted_file(original_file, redacted_text, selected_entities, redaction_method):
     ext = original_file.name.split(".")[-1].lower()
-    
-    # Create a temporary file with a proper path
-    temp_dir = tempfile.gettempdir()  # OS-specific temp folder
+    temp_dir = tempfile.gettempdir()
     safe_filename = f"redacted_{os.path.basename(original_file.name)}"
     redacted_file_path = os.path.join(temp_dir, safe_filename)
-
-    if ext == "pdf":
-        # Assuming a simple text-based redacted PDF (can be further improved for real PDF processing)
-        with open(redacted_file_path, "w", encoding="utf-8") as f:
-            f.write(redacted_text)  # Saving as text-based PDF
-    elif ext == "docx":
-        doc = Document()
-        for line in redacted_text.split("\n"):
-            doc.add_paragraph(line)
-        doc.save(redacted_file_path)  # Save DOCX in temp folder
+    
+    if ext == "docx":
+        doc = Document(original_file.name)
+        for para in doc.paragraphs:
+            para.text = redact_text(para.text, selected_entities, redaction_method)  # Use redaction_method passed from UI
+        doc.save(redacted_file_path)
     elif ext == "pptx":
-        ppt = Presentation()
-        slide = ppt.slides.add_slide(ppt.slide_layouts[5])  # Blank slide
-        textbox = slide.shapes.add_textbox(0, 0, ppt.slide_width, ppt.slide_height)
-        textbox.text = redacted_text
+        ppt = Presentation(original_file.name)
+        
+        # Loop through each slide in the original PPTX and add redacted text
+        for slide_num, slide in enumerate(ppt.slides):
+            # Loop through all shapes on the slide
+            for shape in slide.shapes:
+                if hasattr(shape, "text"):
+                    # Redact the text inside the shape
+                    redacted_text_in_shape = redact_text(shape.text, selected_entities, redaction_method)
+                    shape.text = redacted_text_in_shape  # Apply the redacted text back to the shape
+                    
         ppt.save(redacted_file_path)
-    else:  # TXT
+    else:
         with open(redacted_file_path, "w", encoding="utf-8") as f:
             f.write(redacted_text)
-
-    return redacted_file_path  # Return the safe file path
+    
+    return redacted_file_path
 
 
 def process_file(file, selected_entities, redaction_method):
+    """Handles file upload, redacts selected PII, and returns redacted file."""
     text = read_document(file)
     redacted_text = redact_text(text, selected_entities, redaction_method)
-    redacted_file_path = save_redacted_file(file, redacted_text)
-    return redacted_text, redacted_file_path
+    redacted_file_path = save_redacted_file(file, redacted_text, selected_entities, redaction_method)
+
+    return redacted_text, redacted_file_path  # Returning only a valid file path
+
 
 def select_all_entities():
     return PII_ENTITIES