# code to get its "true" rank on the agent course challenge
# - remove all the users that used the code space of someone else (won't work for the forks)
# - set the same rank for all the people with same score

import pandas as pd
from datasets import load_dataset

username = 'guillaumefrd'

dataset = load_dataset('agents-course/unit4-students-scores')
df = pd.DataFrame(dataset['train'])
df.sort_values('score', ascending=False, inplace=True)

# keep only users that submitted from their own code space or than ran locally (with None in code space)
df['to_keep'] = df.apply(lambda row: row['username'] in row['code'] or 'None' in row['code'], axis=1)
df = df[df['to_keep'] == True]

# compute rank (all users with same score have the same rank)
ranks_to_add = 0
for i, score in enumerate(df['score'].unique()):
    df.loc[df['score'] == score, 'rank'] = ranks_to_add + i + 1
    ranks_to_add += len(df[df['score'] == score]) - 1

# find `username` rank
rank = int(df[df['username'] == username]['rank'].values[0])
total = len(df['code'].unique())
print(f"{username} rank: {rank}/{total} (top {rank/total*100:.1f}%)")