ob-jobs/ob_jobs/report.py

import pathlib
from datetime import timedelta, datetime

import jinja2

from ob_jobs.common import human_duration


def daily_report(templates_path, report_dir, err_df, query_df, insert_df, client_df):
    # count top 1000, avg_du top 10
    top_slow_sql_df = (
        query_df
        .sort_values(by='cnt', ascending=False)
        .head(1000)[['query', 'avg_du']]
        .sort_values('avg_du', ascending=False)
    )
    top_slow_sql_df = top_slow_sql_df[top_slow_sql_df['avg_du'] > 1000]
    top_slow_sql_df = (
        top_slow_sql_df
        .head(10)
        .reset_index()
    )
    top_slow_sql_df['avg_du'] = (top_slow_sql_df['avg_du']
                                 .map(lambda x: human_duration(timedelta(milliseconds=x)))
                                 )
    daily_report_content = render_daily_report(templates_path, top_slow_sql=list(top_slow_sql_df.iterrows()))
    today = datetime.today().strftime("%Y-%m-%d")
    daily_report_file_name = pathlib.Path(report_dir).joinpath(f'{today}.md')
    with open(daily_report_file_name, 'w') as f:
        f.write(daily_report_content)


def render_daily_report(tmp_path, top_slow_sql):
    templateLoader = jinja2.FileSystemLoader(searchpath=tmp_path)
    templateEnv = jinja2.Environment(loader=templateLoader)
    TEMPLATE_FILE = "daily_report.md"
    template = templateEnv.get_template(TEMPLATE_FILE)
    return template.render(top_slow_sql=top_slow_sql)