import pandas as pd
import numpy as np
from datasets import load_dataset
from about import results_repo
from about import LB_COLS0

def make_user_clickable(name):
    link =f'https://huggingface.co/{name}'
    return f'<a target="_blank" href="{link}" style="color: var(--link-text-color); text-decoration: underline;text-decoration-style: dotted;">{name}</a>'
def make_tag_clickable(tag):
    return f'<a target="_blank" href="{tag}" style="color: var(--link-text-color); text-decoration: underline;text-decoration-style: dotted;">link</a>'

def fetch_dataset_df():
    dset = load_dataset(results_repo, split='train', download_mode="force_redownload")
    full_df = dset.to_pandas()
    assert all(
        col in full_df.columns for col in LB_COLS0
    ), f"Expected columns {LB_COLS0} not found in {full_df.columns}. Missing columns: {set(LB_COLS0) - set(full_df.columns)}"

    df = full_df.copy()
    df = df[df["user"] != "test"].copy()
    df["submission_time"] = pd.to_datetime(df["submission_time"], errors="coerce")
    df = df.dropna(subset=["submission_time"])

    # Get the most recent submission per user & endpoint
    latest = (
        df.sort_values("submission_time")
          .drop_duplicates(subset=["endpoint", "user"], keep="last")
          .sort_values(["endpoint", "user"])
          .reset_index(drop=True)
    )
    latest.rename(columns={"submission_time": "submission time"}, inplace=True)
    return latest

def metrics_per_ep(pred, true):
    from scipy.stats import spearmanr, kendalltau
    from sklearn.metrics import mean_absolute_error, r2_score
    mae = mean_absolute_error(true, pred)
    rae = mae / np.mean(np.abs(true - np.mean(true)))
    if np.nanstd(true) == 0:
        r2=np.nan
    else:
        r2 = r2_score(true, pred)
    spr, _ = spearmanr(true, pred)
    ktau, _ = kendalltau(true, pred)

    return mae, rae, r2, spr, ktau