import sklearn
import gradio as gr
import joblib
import pandas as pd
import datasets
import requests
import json
import dateutil.parser as dp
import pandas as pd
from huggingface_hub import hf_hub_url, cached_download
import time
from datetime import datetime

def get_row():
    response_tomtom = requests.get(
                    'https://api.tomtom.com/traffic/services/4/flowSegmentData/absolute/10/json?key=azGiX8jKKGxCxdsF1OzvbbWGPDuInWez&point=59.39575,17.98343')
    json_response_tomtom = json.loads(response_tomtom.text)  # get json response
    
    currentSpeed = json_response_tomtom["flowSegmentData"]["currentSpeed"]
    freeFlowSpeed = json_response_tomtom["flowSegmentData"]["freeFlowSpeed"]
    congestionLevel = currentSpeed/freeFlowSpeed
    
    confidence = json_response_tomtom["flowSegmentData"]["confidence"] # Reliability of the traffic data, by percentage
    
    
    # Get weather data from SMHI, updated hourly
    
    response_smhi = requests.get(
                'https://opendata-download-metanalys.smhi.se/api/category/mesan1g/version/2/geotype/point/lon/17.983/lat/59.3957/data.json')
    json_response_smhi = json.loads(response_smhi.text) 
    
    # weather data manual https://opendata.smhi.se/apidocs/metanalys/parameters.html#parameter-wsymb
    # referenceTime = dp.parse(json_response_smhi["referenceTime"]).timestamp()
    
    t             = json_response_smhi["timeSeries"][0]["parameters"][0]["values"][0] # Temperature
    ws            = json_response_smhi["timeSeries"][0]["parameters"][4]["values"][0] # Wind Speed
    prec1h        = json_response_smhi["timeSeries"][0]["parameters"][6]["values"][0] # Precipation last hour
    fesn1h        = json_response_smhi["timeSeries"][0]["parameters"][8]["values"][0] # Snow precipation last hour
    vis           = json_response_smhi["timeSeries"][0]["parameters"][9]["values"][0] # Visibility

    # Use current time
    referenceTime = datetime.fromtimestamp(time.time())
    
    row           ={"referenceTime": referenceTime, 
                        "temperature": t, 
                        "wind speed": ws, 
                        "precipation last hour": prec1h, 
                        "snow precipation last hour": fesn1h, 
                        "visibility": vis, 
                        "confidence of data": confidence}
    
    row = pd.DataFrame([row], columns=row.keys())
    print(row)
    row.dropna(axis=0, inplace=True)
    return row

model = joblib.load(cached_download(
    hf_hub_url("tilos/Traffic_Prediction", "traffic_model.pkl")
))

def infer(input_dataframe):
    serie = input_dataframe["referenceTime"]
    ts = dp.parse(serie.iloc[0]).timestamp()
    input_dataframe["referenceTime"] = ts
    res = pd.DataFrame(model.predict(input_dataframe)).clip(0, 1)
    if res > 0.8:
        status = "Smooth Traffic on E4"
    elif res > 0.5:
        status = "Slight congestion on E4"
    else:
        status = "Total congestion on E4"
    return pd.Dataframe[res, status]

title = "Stoclholm Highway E4 Real Time Traffic Prediction"
description = "Stockholm E4 (59°23'44.7"" N 17°59'00.4""E) highway real time traffic prediction"

# inputs = [gr.Dataframe(row_count = (1, "fixed"), col_count=(7,"fixed"), 
#                        headers=["referenceTime", "t", "ws", "prec1h", "fesn1h", "vis", "confidence"], 
#                        # datatype=["timestamp", "float", "float", "float", "float", "float"],
#                        label="Input Data", interactive=1)]

# outputs = [gr.Dataframe(row_count = (1, "fixed"), col_count=(1, "fixed"), label="Predictions", headers=["Congestion Level"])]

with gr.Blocks() as demo:
    with gr.Row():
        with gr.Column():
            inputs = gr.Dataframe(row_count = (1, "fixed"), col_count=(7,"fixed"), 
                       headers=["referenceTime", "t", "ws", "prec1h", "fesn1h", "vis", "confidence"], 
                       # datatype=["timestamp", "float", "float", "float", "float", "float"],
                       label="Input Data", interactive=1)
        with gr.Column():
            outputs = gr.Dataframe(row_count = (1, "fixed"), col_count=(2, "fixed"), label="Predictions", headers=["Freeflow Level", "Status"])
    
    with gr.Row():
        btn_sub = gr.Button(value="Submit")
    with gr.Row():
        btn_ref = gr.Button(value="Get real-time data")

    
    
    btn_sub.click(infer, inputs = inputs, outputs = outputs)
    btn_ref.click(get_row, inputs = None, outputs = inputs)
    
    #examples = gr.Examples(fn = infer, examples=[get_row()],inputs=inputs,outputs=outputs  ,cache_examples=True)
    examples = gr.Examples(fn = infer, examples=[get_row()] ,inputs=inputs, outputs=outputs, cache_examples=False)

    # demo.load(get_row, inputs = None, outputs = [inputs], every=10)
    demo.load(get_row, inputs = None, outputs = [inputs])

        

    
# interface = gr.Interface(fn = infer, inputs = inputs, outputs = outputs, title=title, description=description, examples=[get_row()], cache_examples=False)
# interface.launch()

if __name__ == "__main__":
    demo.queue().launch()