# default_exp understat

#hide
from nbdev.showdoc import *

#export
import enum
import re
import json

import requests
import bs4

#export


def fetch_html(url):
    """
    Fetch HTML and decode into a `bs4.BeautifulSoup` object
    """
    r = requests.get(url)
    r.raise_for_status()
    return bs4.BeautifulSoup(str(r.content, 'unicode-escape'), features='html.parser')
    
    
def extract_json(soup, json_var):
    """ Extract a JSON variable from understat HTML. """
    node, *__ = [s for s in soup.select('script') if s.string and json_var in s.string]
    
    # Clean string by removing and newlines (\n) and tabs (\t)
    node_string = ' '.join(node.string.split())
    
    json_value = re.match(f"var {json_var} = JSON\.parse\(\'(?P<json>.*?)\'\)", node_string).group('json')
    return json.loads(json_value)

#export


# 'Competition' might be a better name, but let's stick with understat's terminology
class League(enum.Enum):  
    """
    Understat leagues
    """
    EPL = 'EPL'
    LA_LIGA = 'La_Liga'
    SERIE_A = 'Serie_A'
    BUNDESLIGA = 'Bundesliga'
    LIGUE_1 = 'Ligue_1'
    RPL = 'RPL'

#export


class Understat:
    """
    Fetches understat data webpages
    """
    
    def __init__(self, base_url: str='https://understat.com'):
        self.base_url = base_url
        
    def matches(self, league: League, season: int):
        """ Fetch match data for a given `league` and `season` (start year). """
        league_url = f'{self.base_url}/league/{league.value}/{season}'
        soup = fetch_html(league_url)
        return extract_json(soup, 'datesData')
    
    def shots(self, match_id: int):
        match_url = f'{self.base_url}/match/{match_id}'
        soup = fetch_html(match_url)
        return extract_json(soup, 'shotsData')

understat = Understat()

matches = understat.matches(League.EPL, 2019)[17]
matches

shots = understat.shots(11660)

# Take the home team's 5th shot
shots['h'][5]