Radiosondy.info recovery to SondeHub format

main.py

import pandas as pd
from datetime import datetime,timezone
from dateutil import parser
import requests

now = datetime.now(timezone.utc)
url = r'https://radiosondy.info/dyn/get_status.php'
data = pd.read_html(url)[0].values.tolist()
for row in data:
    if (row[6] == "FOUND" or row[6] == "NEED ATTENTION"): # Only process manual updates
        serial = row[0]
        recovered = False
        recovered_by = ""
        description = ""
        recovered_utc = parser.parse(row[8].split(" ", 1)[1]) # Get report time
        difference = now - recovered_utc
        if (difference.total_seconds() > 7200): # Only check last 2 hours
            break
        if (row[7] == row[7]): # NaN check
            recovered_by = row[7]
        else:
            recovered_by = row[8].split()[0] # use callsign from update time is reported not found 
        if (row[6] == "FOUND"):
            recovered = True
        if (row[9] == row[9]): # NaN check
            description = row[9]
        description += " (Radiosondy.info)"
        detailed_url = "https://radiosondy.info/sonde_archive.php?sondenumber=" + serial # need to check each sonde to find last reported location
        detailed_data = pd.read_html(detailed_url)[0].values.tolist()[0]
        lat = float(detailed_data[6].split(", ")[0])
        lon = float(detailed_data[6].split(", ")[1])
        alt = 0
        data = {"datetime": recovered_utc.isoformat(), "serial": serial, "lat": lat, "lon": lon, "alt": alt, "recovered": recovered, "recovered_by": recovered_by, "description": description}
        if (lat != 0 or lon != 0):
            check_url = "https://api.v2.sondehub.org/recovered?serial=" + serial # Check if sonde has already been uploaded
            r = requests.get(check_url)
            if (len(r.json()) == 0):
                r = requests.put('https://api.v2.sondehub.org/recovered', json = data)
                print("Uploaded: " + serial)