from apscheduler.schedulers.background import BackgroundScheduler
from model_types import MODEL_TYPES, ModelType
from huggingface_hub import HfApi
import matplotlib.pyplot as plt
import bar_chart_race as bcr
import pandas as pd
import gradio as gr
import requests
import os


def restart_space():
    HfApi().restart_space(
        repo_id="IlyasMoutawwakil/llm-bar-race",
        token=os.environ.get("HF_TOKEN", None),
    )


if os.path.exists("open-llm-race-dataset.csv"):
    open_llm_race_dataset = pd.read_csv("open-llm-race-dataset.csv")
else:
    open_llm_race_dataset = pd.read_csv(
        "https://huggingface.co/datasets/IlyasMoutawwakil/open-llm-race-dataset/resolve/main/open-llm-race-dataset.csv"
    )


# resample for ever model to a daily frequency
open_llm_race_dataset["date"] = pd.to_datetime(open_llm_race_dataset["date"])
open_llm_race_dataset = (
    open_llm_race_dataset.set_index("date", drop=True)
    .groupby("model", as_index=False)
    .resample("D", how="last", closed="right", fill_method="ffill")
    .last()
    .reset_index(drop=False)
)
# filter
open_llm_race_dataset["date"] = open_llm_race_dataset["date"].dt.strftime("%Y-%m-%d")
open_llm_race_dataset = open_llm_race_dataset[
    open_llm_race_dataset["date"] >= "2023-07-10"
]
open_llm_race_dataset = open_llm_race_dataset[["date", "score", "model"]]
# drop nan values
open_llm_race_dataset.dropna(inplace=True)
# drop duplicates on model and date
open_llm_race_dataset.drop_duplicates(subset=["model", "date"], inplace=True)
# add the model type
open_llm_race_dataset["type"] = open_llm_race_dataset["model"].apply(
    lambda x: MODEL_TYPES[x].name if x in MODEL_TYPES else ModelType.Unknown.name
)


def get_bar_chart(model_type: str, top_n: int = 10, title: str = ""):
    fig, ax = plt.subplots(figsize=(12, 6))
    ax.set_xlim(0, 100)
    plt.subplots_adjust(left=0.25)

    subset = open_llm_race_dataset[open_llm_race_dataset["type"] == model_type]
    subset = subset.pivot(index="date", columns="model", values="score")
    subset.fillna(0, inplace=True)
    fig = bcr.bar_chart_race(
        subset,
        title=title,
        n_bars=top_n,
        fixed_max=True,
        period_length=1000,
        steps_per_period=20,
        end_period_pause=100,
        bar_texttemplate="{x:.2f}",
        filter_column_colors=True,
        bar_label_font=10,
        tick_label_font=10,
        bar_kwargs={"alpha": 0.2, "ec": "black", "lw": 3},
        fig=fig,
    )
    return gr.HTML(fig)


# Demo interface
demo = gr.Blocks()
with demo:
    # leaderboard title
    gr.HTML("<center><h1>LLM Bar Race 📊🏃‍♂️</h1></center>")

    with gr.Tabs():
        with gr.TabItem(label="Pretrained Models"):
            get_bar_chart(ModelType.PT.name, title="Pretrained Models")
        with gr.TabItem(label="Instructions Finetuned Models"):
            get_bar_chart(ModelType.IFT.name, title="Instructions Finetuned Models")
        with gr.TabItem(label="RLHF Models"):
            get_bar_chart(ModelType.RL.name, top_n=4, title="RLHF Models")


# Restart space every hour
scheduler = BackgroundScheduler()
scheduler.add_job(
    func=restart_space,
    trigger="interval",
    seconds=3600,
)
scheduler.start()
demo.queue(concurrency_count=10).launch()