Update data_preprocess.py

data_preprocess.py

@@ -532,32 +532,39 @@ from pipeline import upload_file_to_drive
 #         print(f"Text successfully saved to '{file_path}'")
 #     except Exception as e:
 #         print(f"Error saving text to docx file: {e}") 
-def save_text_to_docx(text_content: str, filename: str, drive_folder_id: str):
-    """
-    Saves a given text string into a .docx file locally, then uploads to Google Drive.
+# def save_text_to_docx(text_content: str, filename: str, drive_folder_id: str):
+#     """
+#     Saves a given text string into a .docx file locally, then uploads to Google Drive.
 
-    Args:
-        text_content (str): The text string to save.
-        filename (str): The target .docx file name, e.g. 'BRU18_merged_document.docx'.
-        drive_folder_id (str): Google Drive folder ID where to upload the file.
-    """
-    try:
-        # ✅ Save to temporary local path first
-        print("file name: ", filename)
-        print("length text content: ", len(text_content))
-        local_path = os.path.join(tempfile.gettempdir(), filename)
-        document = Document()
-        for paragraph_text in text_content.split('\n'):
-            document.add_paragraph(paragraph_text)
-        document.save(local_path)
-        print(f"✅ Text saved locally to: {local_path}")
-
-        # ✅ Upload to Drive
-        pipeline.upload_file_to_drive(local_path, filename, drive_folder_id)
-        print(f"✅ Uploaded '{filename}' to Google Drive folder ID: {drive_folder_id}")
+#     Args:
+#         text_content (str): The text string to save.
+#         filename (str): The target .docx file name, e.g. 'BRU18_merged_document.docx'.
+#         drive_folder_id (str): Google Drive folder ID where to upload the file.
+#     """
+#     try:
+#         # ✅ Save to temporary local path first
+#         print("file name: ", filename)
+#         print("length text content: ", len(text_content))
+#         local_path = os.path.join(tempfile.gettempdir(), filename)
+#         document = Document()
+#         for paragraph_text in text_content.split('\n'):
+#             document.add_paragraph(paragraph_text)
+#         document.save(local_path)
+#         print(f"✅ Text saved locally to: {local_path}")
+
+#         # ✅ Upload to Drive
+#         pipeline.upload_file_to_drive(local_path, filename, drive_folder_id)
+#         print(f"✅ Uploaded '{filename}' to Google Drive folder ID: {drive_folder_id}")
+
+#     except Exception as e:
+#         print(f"❌ Error saving or uploading DOCX: {e}")
+def save_text_to_docx(text_content: str, full_local_path: str):
+    document = Document()
+    for paragraph_text in text_content.split('\n'):
+        document.add_paragraph(paragraph_text)
+    document.save(full_local_path)
+    print(f"✅ Saved DOCX locally: {full_local_path}")
 
-    except Exception as e:
-        print(f"❌ Error saving or uploading DOCX: {e}")
 
 
 '''2 scenerios: