diff --git a/airflow/include/tasks/extract/airflow_docs.py b/airflow/include/tasks/extract/airflow_docs.py index d828d0b6..5c0fbd98 100644 --- a/airflow/include/tasks/extract/airflow_docs.py +++ b/airflow/include/tasks/extract/airflow_docs.py @@ -6,9 +6,10 @@ import pandas as pd import requests from bs4 import BeautifulSoup -from include.tasks.extract.utils.html_helpers import get_all_links from weaviate.util import generate_uuid5 +from include.tasks.extract.utils.html_helpers import get_all_links + def extract_airflow_docs(docs_base_url: str) -> list[pd.DataFrame]: """ diff --git a/airflow/include/tasks/extract/slack.py b/airflow/include/tasks/extract/slack.py index 17b36093..9bb61215 100644 --- a/airflow/include/tasks/extract/slack.py +++ b/airflow/include/tasks/extract/slack.py @@ -5,10 +5,10 @@ import numpy as np import pandas as pd import requests -from include.tasks.extract.utils.slack_helpers import get_slack_replies from weaviate.util import generate_uuid5 from airflow.providers.slack.hooks.slack import SlackHook +from include.tasks.extract.utils.slack_helpers import get_slack_replies slack_archive_host = "apache-airflow.slack-archives.org" slack_base_url = "https://{slack_archive_host}/v1/messages?size={size}&team={team}&channel={channel}" diff --git a/airflow/include/tasks/extract/stack_overflow.py b/airflow/include/tasks/extract/stack_overflow.py index 46f42d8e..d646c963 100644 --- a/airflow/include/tasks/extract/stack_overflow.py +++ b/airflow/include/tasks/extract/stack_overflow.py @@ -3,6 +3,9 @@ import datetime import pandas as pd +from stackapi import StackAPI +from weaviate.util import generate_uuid5 + from include.tasks.extract.utils.stack_overflow_helpers import ( process_stack_answers, process_stack_answers_api, @@ -12,8 +15,6 @@ process_stack_questions, process_stack_questions_api, ) -from stackapi import StackAPI -from weaviate.util import generate_uuid5 def extract_stack_overflow_archive(tag: str, stackoverflow_cutoff_date: str) -> pd.DataFrame: