From 704d34d1d902e70b07108df06783b9bc552678e2 Mon Sep 17 00:00:00 2001 From: Jun-te Kim Date: Mon, 10 Mar 2025 16:28:19 +0000 Subject: [PATCH] save current work --- etl/main.py | 13 ------------- etl/scraper/scraper.py | 1 + etl/utils/sharepoint/sharepoint.py | 15 ++++++++++++++- 3 files changed, 15 insertions(+), 14 deletions(-) diff --git a/etl/main.py b/etl/main.py index 1083233..08c1284 100644 --- a/etl/main.py +++ b/etl/main.py @@ -3,7 +3,6 @@ from pdfReader.pdfReaderToText import pdfReaderToText from etl.scraper.scraper import SharePointScraper, SharePointInstaller from pprint import pprint, pformat import logging -import tempfile from etl.utils.logger import Logger from etl.validator.validator import DomnaSharePointValidator @@ -37,18 +36,6 @@ def main(): # logger.info(south_coast_scraper.surveyor_to_housing_assosications) - - - -def create_temp_file(dict_content): - with tempfile.NamedTemporaryFile(mode='w+', delete=False, suffix='.txt') as temp_file: - formatted_content = pformat(dict_content) - temp_file.write(formatted_content + "\n") - temp_file_path = temp_file.name - - logger.info(f"Temporary file created at: {temp_file_path}") - - if __name__ == "__main__": main() diff --git a/etl/scraper/scraper.py b/etl/scraper/scraper.py index 824d3ea..50ac9c6 100644 --- a/etl/scraper/scraper.py +++ b/etl/scraper/scraper.py @@ -212,5 +212,6 @@ class SharePointScraper(): filtered_files = [f for f in file_names_to_download if not f.endswith(tuple(avoid))] self.logger.warning(filtered_files) + diff --git a/etl/utils/sharepoint/sharepoint.py b/etl/utils/sharepoint/sharepoint.py index a2bb7d0..5b5a921 100644 --- a/etl/utils/sharepoint/sharepoint.py +++ b/etl/utils/sharepoint/sharepoint.py @@ -12,6 +12,8 @@ import time import logging from io import BytesIO from etl.utils.logger import Logger +import tempfile +import os logger = Logger(name="SharePoint.py", level=logging.DEBUG).get_logger() @@ -267,4 +269,15 @@ class SharePointClient: file_content.seek(0) # Reset the file pointer to the beginning - return file_content \ No newline at end of file + return file_content + + def create_temp_file(self, content, path): + # Ensure the directory exists + os.makedirs(os.path.dirname(path), exist_ok=True) + + # Write content to the specified temporary file + with open(path, 'w+') as temp_file: + temp_file.write(content + "\n") + + logger.info(f"Temporary file created at: {path}") + return path \ No newline at end of file