app.py

import requests
import pandas as pd
import datetime
import time
import gradio as gr
import os

###########
# other API's of interest: https://medium.com/@imdipto/best-free-alternatives-to-the-wunderground-weather-api-21acb22450e6
##########
OPENWEATHER_API_KEY = os.environ.get('OPENWEATHER_API_KEY')
WEATHERAPI_KEY = os.environ.get('WEATHERAPI_KEY')


def openweather_to_result(lat, lon, gmt_time):
    """    
    API docs: https://openweathermap.org/api/one-call-api#current
    
    Parameters
    ------------
    lat [float]: decimal valued latitude
    lon [float]: decimal valued longitude
    gmt_time [datetime object]: time of desired forecast, in gmt and as python datetime object
    
    Returns
    --------
    cloud_pct Tuple(List, List): list of cloud percent and corresponding time for times within 1.5 hours of input GMT time
    """
    exclude_parts = 'current,minutely,daily,alerts'
    request_url = f'https://api.openweathermap.org/data/2.5/onecall?lat={lat}&lon={lon}&exclude={exclude_parts}&appid={OPENWEATHER_API_KEY}'

    response = requests.get(request_url)

    data = response.json()

    cloud_pct = []
    forecast_times = []

    # timeframe around input time to check cloud % for
    timeframe = datetime.timedelta(hours=1, minutes=30)
    for hour in data['hourly']:
        # dt property is unix utc time of forecasted data - convert this to python datetime object
        forecast_time = datetime.datetime.fromtimestamp(
            hour['dt'], tz=datetime.timezone.utc)
        if abs(forecast_time - gmt_time) <= timeframe:
            # cloud pct is stored in each hour at top level
            cloud_pct.append(hour['clouds'])
            forecast_times.append(forecast_time)

    return cloud_pct, forecast_times


def weatherapi_to_result(lat, lon, gmt_time):
    """    
    API docs: https://www.weatherapi.com/docs/
    TODO: implement wrapper instead https://github.com/weatherapicom/weatherapi-Python
    
    Parameters
    ------------
    lat [float]: decimal valued latitude
    lon [float]: decimal values longitude
    gmt_time [datetime object]: time of desired forecast, in gmt and as python datetime object
    
    Returns
    --------
    cloud_pct Tuple(List, List): list of cloud percent and corresponding time for times within 1.5 hours of input GMT time
    """
    request_url = f'http://api.weatherapi.com/v1/forecast.json?key={WEATHERAPI_KEY}&q={lat},{lon}&days=2&alerts=no'
    response = requests.get(request_url)

    data = response.json()

    timezone = data['location']['tz_id']

    cloud_pct = []
    forecast_times = []

    # quick error handling to make sure input time python object has "timezone" property attached
    try:
        gmt_time = gmt_time.astimezone(datetime.timezone.utc)
    except:
        gmt_time = gmt_time.tz_localize('utc')

    # timeframe around input time to check cloud % for
    timeframe = datetime.timedelta(hours=1, minutes=30)

    # this api is first divided into days, then hours
    for day in data['forecast']['forecastday']:
        for hour in day['hour']:
            # time_epoch contains unix epoch time in GMT/UTC
            #forecast_time = datetime.datetime.fromtimestamp(hour['time_epoch'], ZoneInfo(timezone))
            forecast_time = datetime.datetime.fromtimestamp(
                hour['time_epoch'], datetime.timezone.utc)
            if abs(forecast_time - gmt_time) <= timeframe:
                cloud_pct.append(hour['cloud'])
                forecast_times.append(
                    forecast_time.astimezone(datetime.timezone.utc))

    return cloud_pct, forecast_times


def met_to_result(lat, lon, gmt_time):
    """    
    API doc: https://api.met.no/weatherapi/locationforecast/2.0/documentation
    How to: https://api.met.no/doc/locationforecast/HowTO
    
    Parameters
    ------------
    lat [float]: decimal valued latitude
    lon [float]: decimal values longitude
    gmt_time [datetime object]: time of desired forecast, in gmt and as python datetime object
    
    Returns
    --------
    cloud_pct Tuple(List, List): list of cloud percent and corresponding time for times within 1.5 hours of input GMT time
    """

    # set user agent https://stackoverflow.com/questions/10606133/sending-user-agent-using-requests-library-in-python
    # must be unique per API Terms of Service https://api.met.no/doc/TermsOfService
    headers = {
        'User-Agent': 'NASAEarthScienceRemoteSensingUnit [email protected]'}

    request_url = f'https://api.met.no/weatherapi/locationforecast/2.0/compact?lat={lat}&lon={lon}'

    response = requests.get(request_url, headers=headers)

    data = response.json()

    cloud_pct = []
    forecast_times = []

    # timeframe around input time to check cloud % for
    timeframe = datetime.timedelta(hours=1, minutes=30)

    # walk through json return
    for hour in data['properties']['timeseries']:
        # time is utc formatted time https://api.met.no/doc/locationforecast/FAQ
        forecast_time = datetime.datetime.strptime(
            hour['time'], '%Y-%m-%dT%H:%M:%SZ').replace(tzinfo=datetime.timezone.utc)
        # check if time of forecast is withing "timeframe" of desired time
        if abs(forecast_time - gmt_time) <= timeframe:
            # grab cloud pct from location within the nested json, add to list
            cloud_pct.append(hour['data']['instant']
                             ['details']['cloud_area_fraction'])
            # add time of forecast to list. Should be an "on the hour" time
            forecast_times.append(forecast_time)

    return cloud_pct, forecast_times

################
# generate text
################


def file_to_cloud_listing(input_file, services):
    """

    Args:
        input_file (Union[str, gradio FileType]): input csv file with LAT, LON, SITE, GMT cols
        services (List): list of weather api servies to check 

    Returns:
        str: formatted string with weather predictions for locations
    """
    # this works if the input is from gradio. Then the file has an name property
    try:
        sites = pd.read_csv(input_file.name, parse_dates=['GMT'])
        using_gradio = True
    except:
        # this is for input from a script or command line
        sites = pd.read_csv(input_file, parse_dates=['GMT'])
        using_gradio = False
    start = time.perf_counter()
    date_format = "%H:%M"
    text = ''
    # each row is a site. Get weather data and then print it for each service for each site.
    for row_idx, row in sites.iterrows():
        #time_of_interest = datetime.datetime.strptime(row.GMT, '%m/%d/%y %H:%M')
        text += check_row(row, services, date_format)
        text += f'{"="*60}\n'

    return text


def check_row(row, services, date_format="%H:%M"):
    """Check a row of data (a pd.Series with LAT, LON, GMT, SITE cols)

    Args:
        row (pd.Series): pd.Series with LAT, LON, GMT, SITE cols)
        services (List): List of weather services (['OpenWeather', 'MET (Norwegian)', 'WeatherAPI'] or subset)
        date_format (str, optional): Format for printing time of site pass over. Defaults to "%H:%M".

    Returns:
        str: formatted str of text for weather vals
    """
    text = ""

    text += f'{"Location":13}:\t\t{row.SITE} @ {row["GMT"].strftime(date_format)} GMT\n'

    if not isinstance(row.GMT, datetime.datetime):
        GMT = row["GMT"].to_pydatetime()
    else:
        GMT = row["GMT"]
    GMT = GMT.replace(tzinfo=datetime.timezone.utc)
    if 'OpenWeather' in services:
        try:
            cldp, times = openweather_to_result(row.LAT, row.LON, GMT)
            text += format_cldp_and_time("OpenWeather", cldp=cldp, times=times)
        except Exception as e:
            text += f'OpenWeather:\t\tError {e} in API processing\n'
    if 'MET (Norwegian)' in services:
        try:
            cldp, times = met_to_result(row.LAT, row.LON, GMT)
            text += format_cldp_and_time("Norwegian", cldp=cldp)
        except Exception as e:
            text += f'Norwegian:\t\tError {e} in API processing\n'
    if 'WeatherAPI' in services:
        try:
            cldp, times = weatherapi_to_result(row.LAT, row.LON, GMT)
            text += format_cldp_and_time("WeatherAPI", cldp=cldp)
        except Exception as e:
            text += f'WeatherAPI:\t\tError {e} in API processing\n'

    return text


def format_cldp_and_time(api_name, cldp, times=None):
    """Formats output text for lists of cloud percents and forecast times

    Args:
        api_name ([type]): Name of weather source. 
        cldp (List): List of floating point cloud percentage values.
        times (List, optional): List of forecast times, as datetime objects. Defaults to None.

    Returns:
        str: formatted text for printing
    """
    text = ''
    date_format = "%H:%M"
    if times is not None:
        text += f'{"Forecast Time:":13}\t\t' + '  '.join(time.strftime(date_format)
                                                         for time in times) + "\n"

    text += f'{api_name:13}:\t\t{" ".join(f"{p:<6.0f}" for p in cldp)}\n'
    return text


inputs = [gr.inputs.File(label='Site File with Lat/Lon and GMT Time'), gr.inputs.CheckboxGroup(label='Weather Services',
                                                                                                       choices=['OpenWeather', 'MET (Norwegian)', 'WeatherAPI'], default=['OpenWeather', 'MET (Norwegian)'])]
outputs = gr.outputs.Textbox(label ='Cloud % for hour before, hour of, hour after')
css = """* {font-family: "Lucida Console", "Courier New", monospace !important;/* <-- fonts */
                }"""

gr.Interface(fn=file_to_cloud_listing, inputs=inputs, css=css, outputs=outputs,
                    allow_screenshot=False).launch()