Spritpreis_Crawler/main.py
2021-11-20 12:20:30 +01:00

100 lines
2.2 KiB
Python

import logging
import os
import pymysql
import requests
from dotenv import load_dotenv
import sql_connction_handler as sql_handler
API_URL = 'https://creativecommons.tankerkoenig.de/json/detail.php'
def get_prices_for_station(station_id, api_id: str, api_key) -> []:
"""
Crawls the prices for the given station and returns them in a tuple: (E5, E10, Diesel)
:param api_id: The api id of the station
:return: The tuple with the prices
"""
url = f'{API_URL}?id={api_id}&apikey={api_key}'
station_details = requests.get(url).json()
if station_details['ok'] is True:
details = station_details['station']
return [
(station_id, details['e5'], 'E5'),
(station_id, details['e10'], 'E10'),
(station_id, details['diesel'], 'Diesel'),
]
return []
def save_prices_to_sql(prices: [()], conn: pymysql.Connection):
"""
Saves the list of price tuples to sql
:param prices: The list of prices to save to sql
:param conn: The pymysql connection
"""
stmt = 'INSERT INTO prices (station, price, fuel_type) VALUES (%s, %s, %s)'
cur = conn.cursor()
success = cur.executemany(stmt, prices)
if success:
conn.commit()
cur.close()
return
def get_stations_from_sql(conn: pymysql.Connection) -> []:
"""
Gets all stations that should be crawled from sql
:param conn: The pymysql connection
:return: The list of stations
"""
stmt = 'SELECT station_id, tankerkoenig_id FROM stations WHERE active_crawling = 1'
cur = conn.cursor()
cur.execute(stmt)
res = cur.fetchall()
cur.close()
stations = []
for station in res:
stations.append({
'id': station[0],
'api_id': station[1]
})
return stations
def crawl_and_save():
"""
Main function, crawls data from api and saves it to sql
"""
conn = sql_handler.get_connection()
api_key = os.getenv('TANKERKOENIG_API_KEY')
if not conn or not api_key:
logging.warning('Could not fetch prices as either SQL connection or API key are missing!')
return
stations = get_stations_from_sql(conn)
station_prices = []
for station in stations:
prices = get_prices_for_station(station['id'], station['api_id'], api_key)
station_prices.extend(prices)
save_prices_to_sql(station_prices, conn)
if __name__ == '__main__':
load_dotenv()
crawl_and_save()