import pandas as pd
from IPython.display import display, HTML

# Column names : 'League', 'Country', 'Season', 'Date', 'Home', 'HomeGoals', 'Away', 'AwayGoals'
match_results = pd.read_csv('data/extracted_match_results.csv', parse_dates=['Date'])

# Fix encoding issue : renaming 'Fu\303\237ball-Bundesliga' to 'Bundesliga'
match_results['League'] = match_results['League'].replace('Fu<U+00C3><U+009F>ball-Bundesliga', 'Bundesliga')

# Map country codes to country names
country_map = {'ENG': 'England', 'ITA': 'Italy', 'FRA': 'France', 'GER': 'Germany', 'ESP': 'Spain', 'POR': 'Portugal', 'SCO': 'Scotland', 'POL': 'Poland', 'GRE': 'Greece', 'TUR': 'Turkey', 'SUI': 'Switzerland', 'NED': 'Netherlands', 'BEL': 'Belgium', 'AUT': 'Austria'}
match_results['Country'] = match_results['Country'].map(country_map)

match_results.describe()

match_results.head()

def return_result(goal1, goal2):
    if goal1 > goal2:
        return 'win'
    elif goal1 < goal2:
        return 'loss'
    else:
        return 'draw'

match_results['HomeResult'] = match_results.apply(return_result, axis=1, args=('HomeGoals', 'AwayGoals'))
match_results['AwayResult'] = match_results.apply(return_result, axis=1, args=('AwayGoals', 'HomeGoals'))

home_results = match_results[['League', 'Country', 'Date', 'Home', 'HomeGoals', 'HomeResult']]
home_results = home_results.rename(columns={'Home': 'Team', 'HomeGoals': 'Goals', 'HomeResult': 'Result'})
home_results['isHome'] = True

away_results = match_results[['League', 'Country', 'Away', 'Date', 'AwayGoals', 'AwayResult']]
away_results = away_results.rename(columns={'Away': 'Team', 'AwayGoals': 'Goals', 'AwayResult': 'Result'})
away_results['isHome'] = False

match_results = pd.concat([home_results, away_results], ignore_index=True)

# Column names : 'Team', 'League', 'Country', 'HeadCoach', 'Appointed', 'EndDate', 'Tenure', 'Matches', 'Wins', 'Draws', 'Losses'
head_coach = pd.read_csv('data/extracted_head_coach.csv', parse_dates=['Appointed', 'EndDate'])
head_coach.head()

head_coach.describe()

last_match = match_results['Date'].max()
first_match = match_results['Date'].min()

# Remove head coach that were appointed after 2023 season
head_coach = head_coach[head_coach['Appointed'] <= last_match]
# Keep head coach that were dimissed after 2015 or that are still active
head_coach = head_coach[(head_coach['EndDate'] >= first_match) | (head_coach['EndDate'].isna())]

# Ensuring there is only 1 head coach at a time in any given team.

head_coach_bis = head_coach.copy()
# Sort data by 'Team' and 'Appointed'
head_coach_bis = head_coach_bis.sort_values(['Team', 'Appointed'])
# Fillna with end date of 2022-2023 season
head_coach_bis['EndDate'] = head_coach_bis['EndDate'].fillna('2024-01-14')
# Check if the next appointment is overlapping with the current one
head_coach_bis['OverlapDuration'] = head_coach_bis.groupby('Team')['Appointed'].shift(-1) - head_coach_bis['EndDate']
head_coach_bis['Overlap'] = head_coach_bis['OverlapDuration'].dt.days < 0

# Show team with overlapping appointments
overlapping = head_coach_bis[head_coach_bis['Overlap']]

head_coach_bis.sort_values('OverlapDuration').head()

head_coach_bis[head_coach_bis['Team'].isin(overlapping['Team'])][['Team', 'Appointed', 'EndDate', 'Overlap', 'OverlapDuration']].sort_values(['Team', 'Appointed']).head(10)

head_coach_bis[head_coach_bis['Team'] == 'Stade Reims'][['Team', 'Appointed', 'EndDate', 'Overlap']].sort_values(['Appointed']).iloc[2:6]

# Check records that are overlapping
head_coach[~head_coach.index.isin(head_coach_bis[head_coach_bis['OverlapDuration'].dt.days <= -20].index)]
# Drop overlapping records with a duration of more than 20 days
head_coach = head_coach[head_coach.index.isin(head_coach_bis[head_coach_bis['OverlapDuration'].dt.days > -20].index)]

# Using cronological information about appointment
# Add a column to head coach records that tells us about how many appointment head coach has done

head_coach = head_coach.sort_values(['HeadCoach', 'Appointed'])
head_coach['AppointmentNumber'] = head_coach.groupby('HeadCoach').cumcount() + 1
head_coach[head_coach['HeadCoach'].isin(head_coach[head_coach['AppointmentNumber'] > 3]['HeadCoach'])].sort_values('HeadCoach').head()

# Check if total_matches = wins + draws + losses
head_coach[head_coach['Matches'] != head_coach['Wins'] + head_coach['Draws'] + head_coach['Losses']].shape[0]

# Display Head Coach with lowest number of match
display(head_coach.sort_values('Matches').head(20))
# Remove head coach records with less than 5 matches
# We consider them to be not relevant as it a short stay of a head coach indicate either a temporary replacement or a very bad performance
head_coach = head_coach[head_coach['Matches'] > 5]

# Compute number of team that are in head_coach but not in match_results
coach_teams = set(head_coach['Team'])
match_teams = set(match_results['Team'])

coach_team_not_in_match = coach_teams - match_teams
match_team_not_in_coach = match_teams - coach_teams

len(coach_team_not_in_match), len(match_team_not_in_coach)

# Group teams by country
coach_teams_by_country = head_coach.groupby('Country')['Team'].unique()
match_teams_by_country = match_results.groupby('Country')['Team'].unique()

coach_teams_by_country, match_teams_by_country

from thefuzz import process

team_name_mapping = {}

# For each country
for country in coach_teams_by_country.index:
    # Get teams for this country
    coach_teams = coach_teams_by_country[country]
    match_teams = match_teams_by_country.get(country, [])

    # For each team in coach_teams
    for coach_team in coach_teams:
        # Find the best match in match_teams
        matching_scores = process.extract(coach_team, match_teams, limit=1)

        if len(matching_scores) != 0 and matching_scores[0][1] >= 60:
            team_name_mapping[coach_team] = matching_scores[0][0]
        else:
            team_name_mapping[coach_team] = None
            print(f"No match found for {coach_team} among {match_teams} in {country}")

name_match = pd.DataFrame(team_name_mapping.items(), columns=['Team in head coach records', 'Team in match results'])
display(HTML(name_match.head().to_html(index=False)))

# Map head_coach['team'] with name_match
head_coach['Team'] = head_coach['Team'].map(team_name_mapping)
head_coach.head()

# Check match with NaN goals
display(match_results[match_results['Goals'].isna()])
# Remove match with NaN goals
match_results = match_results[~match_results['Goals'].isna()]

# Merge head_coach with match_results
match = match_results.merge(head_coach[['Team', 'HeadCoach', 'Appointed', 'EndDate']], on=['Team'], how='left')
# Put aside team that don't have a head coach
no_headcoach = match[match['HeadCoach'].isna()]
match = match[~match['HeadCoach'].isna()]
# Filter match_results_bis to keep only head coach that were appointed before the match and with no end date or end date after the match
match = match[
    (match['Date'] >= match['Appointed']) &
    ((match['Date'] <= match['EndDate']) | match['EndDate'].isna())]
# Join back the team that don't have a head coach
match = pd.concat([match, no_headcoach], ignore_index=True)
# Compute daysInPost
match['DaysInPost'] = (match['Date'] - match['Appointed']).dt.days
match = match.drop(columns=['Appointed', 'EndDate'])
match.head()

# Values count of coach record / no coach record per team
team_hc_count = match.groupby('Team')['HeadCoach'].count()
# Display list of team with no head coach record
team_no_hc_record = team_hc_count[team_hc_count == 0].index.tolist()
print(len(team_no_hc_record))
print(team_no_hc_record)

# Save match_results
match.to_csv('data/match_results.csv', index=False)
head_coach.to_csv('data/head_coach.csv', index=False)