Corona/scrape_corona.py

"""
Quellen:
Johns Hopkins University:
https://gisanddata.maps.arcgis.com/apps/opsdashboard/index.html#/bda7594740fd40299423467b48e9ecf6
https://services1.arcgis.com/0MSEUqKaxRlEPj5g/arcgis/rest/services/ncov_cases/FeatureServer/1/query?f=json&where=Confirmed%20%3E%200&returnGeometry=false&spatialRel=esriSpatialRelIntersects&outFields=*&orderByFields=Confirmed%20desc%2CCountry_Region%20asc%2CProvince_State%20asc&resultOffset=0&resultRecordCount=250&cacheHint=true

WHO:
https://experience.arcgis.com/experience/685d0ace521648f8a5beeeee1b9125cd
https://services.arcgis.com/5T5nSi527N4F7luB/arcgis/rest/services/Cases_by_country_Plg_V3/FeatureServer/0/query?f=json&where=1%3D1&returnGeometry=false&spatialRel=esriSpatialRelIntersects&outFields=*&orderByFields=cum_conf%20desc&resultOffset=0&resultRecordCount=125&cacheHint=true

Deutschland RKI:
https://experience.arcgis.com/experience/478220a4c454480e823b17327b2bf1d4
https://services7.arcgis.com/mOBPykOjAyBO2ZKk/arcgis/rest/services/Coronaf%C3%A4lle_in_den_Bundesl%C3%A4ndern/FeatureServer/0/query?f=json&where=1%3D1&returnGeometry=false&spatialRel=esriSpatialRelIntersects&outFields=*&orderByFields=faelle_100000_EW%20desc&resultOffset=0&resultRecordCount=50&cacheHint=true
"""


import json
import requests
import db_model_postgrest as db
import toml
import os
import datetime

QUELLEN = {
    "jhu": "https://services1.arcgis.com/0MSEUqKaxRlEPj5g/arcgis/rest/services/ncov_cases/FeatureServer/1/query?f"
           "=json&where=Confirmed%20%3E%200&returnGeometry=false&spatialRel=esriSpatialRelIntersects&outFields"
           "=*&orderByFields=Confirmed%20desc%2CCountry_Region%20asc%2CProvince_State%20asc&resultOffset=0"
           "&resultRecordCount=250&cacheHint=true",
    "who": "https://services.arcgis.com/5T5nSi527N4F7luB/arcgis/rest/services/Cases_by_country_Plg_V3/FeatureServer/0"
           "/query?f=json&where=1%3D1&returnGeometry=false&spatialRel=esriSpatialRelIntersects&outFields"
           "=*&orderByFields=cum_conf%20desc&resultOffset=0&resultRecordCount=125&cacheHint=true",
    "rki": "https://services7.arcgis.com/mOBPykOjAyBO2ZKk/arcgis/rest/services/Coronaf%C3%A4lle_in_den_Bundesl%C3"
           "%A4ndern/FeatureServer/0/query?f=json&where=1%3D1&returnGeometry=false&spatialRel"
           "=esriSpatialRelIntersects&outFields=*&orderByFields=faelle_100000_EW%20desc&resultOffset=0"
           "&resultRecordCount=50&cacheHint=true "
}


def config_laden():
    configfile = os.path.join(SKRIPTPFAD, "config.toml")
    with open(configfile) as file:
        return toml.loads(file.read())


SKRIPTPFAD = os.path.abspath(os.path.dirname(__file__))
CONFIG = config_laden()


def daten_abrufen(url):
    response = requests.get(url)
    json_text = json.loads(response.text)
    return json_text["features"]


def jhu_daten_erstellen(quelle, daten):
    daten_dict = {}
    for data in daten:
        datum = data["attributes"]
        country = datum["Country_Region"].replace(" ", "")
        if country in daten_dict:
            datensatz = daten_dict[country]
            datensatz.ts = max(datensatz.ts, datetime.datetime.fromtimestamp(datum["Last_Update"] / 1000))
            datensatz.confirmed = datensatz.confirmed + datum["Confirmed"]
            datensatz.deaths = datensatz.deaths + datum["Deaths"]
            datensatz.recoverd = datensatz.recoverd + datum["Recovered"]
            datensatz.active = datensatz.confirmed - datensatz.deaths - datensatz.recoverd
            daten_dict[country] = datensatz
        else:
            daten_dict[country] = (db.CoronaDaten(
                datetime.datetime.fromtimestamp(datum["Last_Update"] / 1000),
                country.lower().capitalize(),
                quelle,
                datum["Confirmed"],
                datum["Deaths"],
                datum["Recovered"],
                datum["Confirmed"] - datum["Deaths"] - datum["Recovered"]
            ))
    return daten_dict


def who_daten_erstellen(quelle, daten):
    daten_dict = {}
    for data in daten:
        datum = data["attributes"]
        daten_dict[datum["ADM0_NAME"]] = (db.CoronaDaten(
            datetime.datetime.fromtimestamp(datum["DateOfReport"] / 1000),
            datum["ADM0_NAME"].lower().capitalize(),
            quelle,
            datum["cum_conf"],
            datum["cum_death"],
            -1,
            -1
        ))
    return daten_dict


def rki_daten_erstellen(quelle, daten):
    daten_dict = {}
    for data in daten:
        datum = data["attributes"]
        daten_dict[datum["LAN_ew_GEN"]] = (db.CoronaDaten(
            datetime.datetime.fromtimestamp(datum["Aktualisierung"] / 1000),
            datum["LAN_ew_GEN"].lower().capitalize(),
            quelle,
            datum["Fallzahl"],
            datum["Death"],
            -1,
            -1
        ))
    return daten_dict


def daten_erstellen(quelle, daten):
    if quelle == "jhu":
        daten_db = jhu_daten_erstellen(quelle, daten)
    elif quelle == "who":
        daten_db = who_daten_erstellen(quelle, daten)
    elif quelle == "rki":
        daten_db = rki_daten_erstellen(quelle, daten)
    else:
        raise ValueError("Quelle nicht bekannt")
    return daten_db


def daten_trennen(daten):
    daten_liste = []
    for _, value in daten.items():
        value.ts = value.ts.strftime("%Y-%m-%d %H:%M:%S")
        daten_liste.append(value)
    return daten_liste


def headers_erstellen():
    headers = {"Authorization": "Bearer {token}".format(token=CONFIG["token"]),
               "Prefer": "resolution=ignore-duplicates"}
    return headers


def main():
    headers = headers_erstellen()
    for quelle, url in QUELLEN.items():
        json_daten = daten_abrufen(url)
        daten = daten_erstellen(quelle, json_daten)
        daten_server = daten_trennen(daten)
        db.sende_daten(CONFIG["url"], CONFIG["tablename_daten"], headers, daten_server)


if __name__ == "__main__":
    main()