127 lines
5.9 KiB
Python
Executable File
127 lines
5.9 KiB
Python
Executable File
#! /usr/bin/env python3
|
|
|
|
import psycopg2
|
|
import requests
|
|
import io
|
|
import zipfile
|
|
import datetime
|
|
from config.db import db as config_db
|
|
|
|
class DwdScraper:
|
|
|
|
def run(self):
|
|
last_station = self.get_station_to_update()
|
|
print("checking station", last_station)
|
|
|
|
with psycopg2.connect(config_db["uri"]) as conn:
|
|
with conn.cursor() as cur:
|
|
|
|
cur.execute("SELECT id, name, lat, lon, dwd_id, dwd_last_update FROM stations WHERE dwd_id = %s;", [last_station])
|
|
current_station = cur.fetchone()
|
|
|
|
curr_station_id = current_station[0]
|
|
curr_station_dwd_id = current_station[4]
|
|
|
|
print(curr_station_dwd_id)
|
|
r = requests.get('https://opendata.dwd.de/climate_environment/CDC/observations_germany/climate/daily/kl/recent/tageswerte_KL_' + str(curr_station_dwd_id) + '_akt.zip', stream=True)
|
|
print(r.url)
|
|
if r.status_code == 200:
|
|
zip = zipfile.ZipFile(io.BytesIO(r.content))
|
|
files = zip.namelist()
|
|
print(files)
|
|
files_climate = [f for f in files if f.startswith("produkt_klima_")]
|
|
print(files_climate)
|
|
|
|
|
|
buffer = []
|
|
|
|
is_first_line = True
|
|
|
|
for line in zip.open(files_climate[0]):
|
|
l = line.decode('utf-8').strip()
|
|
if is_first_line:
|
|
is_first_line = False
|
|
continue
|
|
|
|
b = {}
|
|
|
|
b["dwd_id"], l = l.strip().split(";", 1)
|
|
b["dwd_id"] = str(b["dwd_id"])
|
|
b["date"], l = l.strip().split(";", 1)
|
|
b["date"] = str(b["date"])
|
|
b["date"] = datetime.date(int(b["date"][0:4]), int(b["date"][4:6]), int(b["date"][6:8]))
|
|
b["qn_3"], l = l.strip().split(";", 1)
|
|
b["qn_3"] = self.cleanup_value(int(b["qn_3"]))
|
|
b["fx"], l = l.strip().split(";", 1)
|
|
b["fx"] = self.cleanup_value(float(b["fx"]))
|
|
b["fm"], l = l.strip().split(";", 1)
|
|
b["fm"] = self.cleanup_value(float(b["fm"]))
|
|
b["qn_4"], l = l.strip().split(";", 1)
|
|
b["qn_4"] = self.cleanup_value(int(b["qn_4"]))
|
|
b["rsk"], l = l.strip().split(";", 1)
|
|
b["rsk"] = self.cleanup_value(float(b["rsk"]))
|
|
b["rskf"], l = l.strip().split(";", 1)
|
|
b["rskf"] = self.cleanup_value(int(b["rskf"]))
|
|
b["sdk"], l = l.strip().split(";", 1)
|
|
b["sdk"] = self.cleanup_value(float(b["sdk"]))
|
|
b["shk_tag"], l = l.strip().split(";", 1)
|
|
b["shk_tag"] = self.cleanup_value(float(b["shk_tag"]))
|
|
b["nm"], l = l.strip().split(";", 1)
|
|
b["nm"] = self.cleanup_value(float(b["nm"]))
|
|
b["vpm"], l = l.strip().split(";", 1)
|
|
b["vpm"] = self.cleanup_value(float(b["vpm"]))
|
|
b["pm"], l = l.strip().split(";", 1)
|
|
b["pm"] = self.cleanup_value(float(b["pm"]))
|
|
b["tmk"], l = l.strip().split(";", 1)
|
|
b["tmk"] = self.cleanup_value(float(b["tmk"]))
|
|
b["upm"], l = l.strip().split(";", 1)
|
|
b["upm"] = self.cleanup_value(float(b["upm"]))
|
|
b["txk"], l = l.strip().split(";", 1)
|
|
b["txk"] = self.cleanup_value(float(b["txk"]))
|
|
b["tnk"], l = l.strip().split(";", 1)
|
|
b["tnk"] = self.cleanup_value(float(b["tnk"]))
|
|
b["tgk"], l = l.strip().split(";", 1)
|
|
b["tgk"] = self.cleanup_value(float(b["tgk"]))
|
|
|
|
#print(b)
|
|
print(curr_station_id, b["date"].isoformat())
|
|
|
|
cur.execute("SELECT id FROM climate WHERE station = %s AND date = %s; ", [curr_station_id, b["date"].isoformat()])
|
|
if cur.rowcount == 0:
|
|
cur.execute("INSERT INTO climate (station, date, qn_3, fx, fm, qn_4, rsk, rskf, sdk, shk_tag, nm, vpm, pm, tmk, upm, txk, tnk, tgk) VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s);",
|
|
[curr_station_id, b["date"].isoformat(), b["qn_3"], b["fx"], b["fm"], b["qn_4"], b["rsk"], b["rskf"], b["sdk"], b["shk_tag"], b["nm"], b["vpm"], b["pm"], b["tmk"], b["upm"], b["txk"], b["tnk"], b["tgk"]])
|
|
print("imported")
|
|
else:
|
|
print("ignored")
|
|
|
|
cur.execute("UPDATE stations SET dwd_last_update = %s WHERE dwd_id = %s;", [datetime.datetime.today().isoformat(), last_station])
|
|
conn.commit()
|
|
|
|
def get_station_to_update(self):
|
|
"""
|
|
Returns DWD ID of station that should get updated soon.
|
|
"""
|
|
|
|
with psycopg2.connect(config_db["uri"]) as conn:
|
|
with conn.cursor() as cur:
|
|
|
|
cur.execute("SELECT min(dwd_last_update) FROM stations;")
|
|
last_date = cur.fetchall()[0][0]
|
|
|
|
if not last_date is None:
|
|
cur.execute("SELECT id, name, lat, lon, dwd_id, dwd_last_update FROM stations WHERE dwd_last_update = %s LIMIT 1;", [last_date])
|
|
else:
|
|
cur.execute("SELECT id, name, lat, lon, dwd_id, dwd_last_update FROM stations WHERE dwd_last_update IS NULL LIMIT 1;")
|
|
last_station = cur.fetchone()
|
|
|
|
return last_station[4]
|
|
|
|
def cleanup_value(self, v):
|
|
if int(v) == -999:
|
|
return None
|
|
|
|
return v
|
|
|
|
if __name__ == "__main__":
|
|
DwdScraper().run()
|