# -*- coding: utf-8 -*-
"""tabular_gradio.ipynb

Automatically generated by Colab.

Original file is located at
    https://colab.research.google.com/drive/1-_lDVeqDrMkiSuBA380Q7IC2WYa0NHcF
"""

import os  # For filesystem operations
import shutil  # For directory cleanup
import zipfile  # For extracting model archives
import pathlib  # For path manipulations
import pandas as pd # For tabular data handling
import gradio as gr  # For interactive UI
import huggingface_hub  # For downloading model assets
import autogluon.tabular  # For loading and running AutoGluon predictors

# Settings
MODEL_REPO_ID = "mrob937/2024-24679-tabular-autogluon-predictor"
ZIP_FILENAME  = "autogluon_predictor_dir.zip"
CACHE_DIR = pathlib.Path("hf_assets")
EXTRACT_DIR = CACHE_DIR / "predictor_native"

FEATURE_COLS = [
    "Tgt",
    "Rec",
    "Yds",
    "YBC/R",
    "YAC/R",
    "ADOT",
    "Drop%",
    "Rat"
]
TARGET_COL = "Elite"

def _prepare_predictor_dir() -> str:
    CACHE_DIR.mkdir(parents=True, exist_ok=True)
    local_zip = huggingface_hub.hf_hub_download(
        repo_id=MODEL_REPO_ID,
        filename=ZIP_FILENAME,
        repo_type="model",
        local_dir=str(CACHE_DIR),
        local_dir_use_symlinks=False,
    )
    if EXTRACT_DIR.exists():
        shutil.rmtree(EXTRACT_DIR)
    EXTRACT_DIR.mkdir(parents=True, exist_ok=True)
    with zipfile.ZipFile(local_zip, "r") as zf:
        zf.extractall(str(EXTRACT_DIR))
    contents = list(EXTRACT_DIR.iterdir())
    predictor_root = contents[0] if (len(contents) == 1 and contents[0].is_dir()) else EXTRACT_DIR
    return str(predictor_root)

PREDICTOR_DIR = _prepare_predictor_dir()
PREDICTOR = autogluon.tabular.TabularPredictor.load(PREDICTOR_DIR, require_py_version_match=False)

OUTCOME_MAP = {
    0: "Not Elite",
    1: "Elite",
}

def _human_label(c):
    try:
        ci = int(c)
        if ci in OUTCOME_MAP:
            return OUTCOME_MAP[ci]
    except Exception:
        pass
    return str(c)

def do_predict(player, tgt, rec, yds, ybc_r, yac_r, adot, drop_pct, rat):
    row = {
        "Player": str(player),
        FEATURE_COLS[0]: float(tgt),
        FEATURE_COLS[1]: float(rec),
        FEATURE_COLS[2]: float(yds),
        FEATURE_COLS[3]: float(ybc_r),
        FEATURE_COLS[4]: float(yac_r),
        FEATURE_COLS[5]: float(adot),
        FEATURE_COLS[6]: float(drop_pct),
        FEATURE_COLS[7]: float(rat),
    }
    X = pd.DataFrame([row], columns=["Player"] + FEATURE_COLS)

    pred_series = PREDICTOR.predict(X)
    raw_pred = pred_series.iloc[0]

    try:
        proba = PREDICTOR.predict_proba(X)
        if isinstance(proba, pd.Series):
            proba = proba.to_frame().T
    except Exception:
        proba = None

    pred_label = _human_label(raw_pred)

    proba_dict = None
    if proba is not None:
        row0 = proba.iloc[0]
        tmp = {}
        for cls, val in row0.items():
            key = _human_label(cls)
            tmp[key] = float(val) + float(tmp.get(key, 0.0))
        proba_dict = dict(sorted(tmp.items(), key=lambda kv: kv[1], reverse=True))

    if proba_dict:
        return {f"{player} → {k}": v for k, v in proba_dict.items()}
    else:
        return {f"{player} → {pred_label}": 1.0}

EXAMPLES = [
    ["Justin Jefferson", 8.0, 5.0, 65.0, 6.0, 4.0, 8.5, 1.5, 137.2],
    ["Cooper Kupp", 15.0, 10.0, 140.0, 8.0, 6.0, 9.0, 0.5, 118.5],
    ["Rookie WR", 3.0, 2.0, 12.0, 4.0, 3.0, 5.0, 4.0, 50.5],
    ["Tyreek Hill", 20.0, 14.0, 220.0, 10.0, 8.0, 12.0, 0.2, 130.0],
    ["Bench Player", 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 10.0, 30.0],
]

with gr.Blocks() as demo:
    gr.Markdown("# Football: Will this player be Elite?")
    gr.Markdown("Enter the player's name and stats below. The model will predict whether they're Elite and show probabilities.")

    player_name = gr.Textbox(value="Example Player", label="Player Name")

    with gr.Row():
        tgt = gr.Slider(0, 30, value=8.0, step=0.5, label=FEATURE_COLS[0])
        rec = gr.Slider(0, 20, value=5.0, step=0.5, label=FEATURE_COLS[1])

    with gr.Row():
        yds = gr.Number(value=60.0, precision=1, label=FEATURE_COLS[2])
        ybc_r = gr.Number(value=6.0, precision=2, label=FEATURE_COLS[3])

    with gr.Row():
        yac_r = gr.Number(value=4.0, precision=2, label=FEATURE_COLS[4])
        adot = gr.Number(value=8.0, precision=2, label=FEATURE_COLS[5])

    with gr.Row():
        drop_pct = gr.Slider(0.0, 100.0, value=1.5, step=0.1, label=FEATURE_COLS[6])
        rat = gr.Slider(0.0, 158.0, value=7.2, step=0.1, label=FEATURE_COLS[7])

    proba_pretty = gr.Label(num_top_classes=5, label="Class probabilities")

    inputs = [player_name, tgt, rec, yds, ybc_r, yac_r, adot, drop_pct, rat]
    for comp in inputs:
        comp.change(fn=do_predict, inputs=inputs, outputs=[proba_pretty])

    gr.Examples(
        examples=EXAMPLES,
        inputs=inputs,
        label="Representative examples",
        examples_per_page=5,
        cache_examples=False,
    )

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