Convert pandas DataFrame to Prometheus (https://prometheus.io) text format

df_to_prom.py

from datetime import datetime
from itertools import product
from random import randint

import pandas as pd


def generate_df():
    # just example
    index = pd.date_range('2017-01-01', datetime.now(), freq='H')
    req_types = ('GET', 'POST')
    codes = ('404', '302', '200')
    data = [{'ts': ts, 'req_type': req_type, 'code': code, 'value': randint(0, 10)}
            for ts, req_type, code in product(index, req_types, codes)]
    return pd.DataFrame(data)


def convert(df, metric_name='arsenys_awesome_metric', labels=('code', 'req_type'), ):
    for _, row in df.iterrows():
        params = ','.join(['{}=\"{}\"'.format(label, row.get(label)) for label in labels])
        if params:
            line = '{}{{{}}}\t'.format(metric_name, params)
        else:
            line = '{}\t'

        unix_time = int((row.get('ts') - datetime(1970, 1, 1)).total_seconds()) * 1000
        line = '{}{}\t{}'.format(line, row.get('value'), unix_time)
        print(line)


if __name__ == '__main__':
    convert(generate_df())